Update process_watcher.py

.gitignore

@@ -15,6 +15,6 @@ dev
 scepter.egg-info
 .readthedocs.yml
 1.9
-MANIFEST.in
+#MANIFEST.in
 *resources
 *.ipynb_checkpoints*

docs/en/conf.py

@@ -1,4 +1,5 @@
 # -*- coding: utf-8 -*-
+# Copyright (c) Alibaba, Inc. and its affiliates.
 # Configuration file for the Sphinx documentation builder.
 #
 # This file only contains a selection of the most common options. For a full

docs/en/tasks/stylebooth.md

@@ -29,8 +29,21 @@ Given an original image, image editing aims to generate an image that align with
     <td><img src="../../../asset/images/stylebooth/retrogame.jpeg" width="240"></td>
     <td><img src="../../../asset/images/stylebooth/vangogh.jpeg" width="240"></td>
   </tr>
+  <tr>
+    <td><strong>Origin Image</strong></td>
+    <td><strong>Lowpoly</strong></td>
+    <td><strong>Colored Pencil Art</strong></td>
+    <td><strong>Watercolor</strong></td>
+    <td><strong>misc-disco</strong></td>
+  </tr>
+  <tr>
+    <td><img src="../../../asset/images/stylebooth/mountain.jpg" width="240"></td>
+    <td><img src="../../../asset/images/stylebooth/lowpoly.jpg" width="240"></td>
+    <td><img src="../../../asset/images/stylebooth/colorpencil.jpeg" width="240"></td>
+    <td><img src="../../../asset/images/stylebooth/watercolor.jpeg" width="240"></td>
+    <td><img src="../../../asset/images/stylebooth/disco.jpeg" width="240"></td>
+  </tr>
 </table>
-
 ## Features
 
 | **Text-Based** | **Exemplar-Based** |
@@ -92,7 +105,7 @@ class DiffusionInferenceTest(unittest.TestCase):
         output = diff_infer({'prompt': 'Let this image be in the style of sai-lowpoly'},
                             style_edit_image=Image.open('asset/images/inpainting_text_ref/ex4_scene_im.jpg'),
                             style_guide_scale_text=7.5,
-                            style_guide_scale_image=0.5)
+                            style_guide_scale_image=1.5)
         save_path = os.path.join(self.tmp_dir,
                                  'stylebooth_test_lowpoly_cute_dog.png')
         save_image(output['images'], save_path)

docs/en/tutorials/dataset.md

@@ -5,7 +5,7 @@ Below are examples for each format, illustrating their details and basic usage.
 
 ## Modelscope Format
 
-We use a [custom-stylized dataset](https://modelscope.cn/datasets/damo/style_custom_dataset/summary), which included classes 3D, anime, flat illustration, oil painting, sketch, and watercolor, each with 30 image-text pairs.
+We use a [custom-stylized dataset](https://modelscope.cn/datasets/iic/style_custom_dataset/summary), which included classes 3D, anime, flat illustration, oil painting, sketch, and watercolor, each with 30 image-text pairs.
 
 ```python
 # pip install modelscope
@@ -16,7 +16,11 @@ print(next(iter(ms_train_dataset)))
 
 ## CSV Format
 
-For the data format used by SCEPTER Studio, please refer to [3D_example_csv.zip](https://www.modelscope.cn/api/v1/models/iic/scepter/repo?Revision=master&FilePath=datasets/3D_example_csv.zip).
+For the data format used by SCEPTER Studio, please refer to [3D_example_csv.zip](https://www.modelscope.cn/api/v1/models/iic/scepter/repo?Revision=master&FilePath=datasets/3D_example_csv.zip) and [hed_pair.zip](https://www.modelscope.cn/api/v1/models/iic/scepter/repo?Revision=master&FilePath=datasets%2Fhed_pair.zip).
+```shell
+mkdir -p cache/datasets/ && wget 'https://www.modelscope.cn/api/v1/models/iic/scepter/repo?Revision=master&FilePath=datasets/3D_example_csv.zip' -O cache/datasets/3D_example_csv.zip && unzip cache/datasets/3D_example_csv.zip -d cache/datasets/ && rm cache/datasets/3D_example_csv.zip
+mkdir -p cache/datasets/ && wget 'https://www.modelscope.cn/api/v1/models/iic/scepter/repo?Revision=master&FilePath=datasets/hed_pair.zip' -O cache/datasets/hed_pair.zip && unzip cache/datasets/hed_pair.zip -d cache/datasets/ && rm cache/datasets/hed_pair.zip
+```
 
 ## TXT Format
 
@@ -24,3 +28,4 @@ To facilitate starting training in command-line mode, you can use a dataset in t
 
 ```shell
 mkdir -p cache/datasets/ && wget 'https://www.modelscope.cn/api/v1/models/iic/scepter/repo?Revision=master&FilePath=datasets/3D_example_txt.zip' -O cache/datasets/3D_example_txt.zip && unzip cache/datasets/3D_example_txt.zip -d cache/datasets/ && rm cache/datasets/3D_example_txt.zip
+```

docs/en/tutorials/inference.md

@@ -40,6 +40,6 @@ python scepter/tools/run_inference.py --cfg scepter/methods/scedit/t2i/sd15_512_
 
 - SCEdit
 ```shell
-python scepter/tools/run_inference.py --cfg scepter/methods/scedit/ctr/sd21_768_sce_ctr_canny.yaml --num_samples 1 --prompt 'a single flower is shown in front of a tree' --save_folder 'test_flower_canny' --image_size 768 --task control --image 'asset/images/flower.jpg' --control_mode canny --pretrained_model ms://damo/scepter_scedit@controllable_model/SD2.1/canny_control/0_SwiftSCETuning/pytorch_model.bin   # canny
+python scepter/tools/run_inference.py --cfg scepter/methods/scedit/ctr/sd21_768_sce_ctr_canny.yaml --num_samples 1 --prompt 'a single flower is shown in front of a tree' --save_folder 'test_flower_canny' --image_size 768 --task control --image 'asset/images/flower.jpg' --control_mode canny --pretrained_model ms://iic/scepter_scedit@controllable_model/SD2.1/canny_control/0_SwiftSCETuning/pytorch_model.bin   # canny
-python scepter/tools/run_inference.py --cfg scepter/methods/scedit/ctr/sd21_768_sce_ctr_pose.yaml --num_samples 1 --prompt 'super mario' --save_folder 'test_mario_pose' --image_size 768 --task control --image 'asset/images/pose_source.png' --control_mode source --pretrained_model ms://damo/scepter_scedit@controllable_model/SD2.1/pose_control/0_SwiftSCETuning/pytorch_model.bin   # pose
+python scepter/tools/run_inference.py --cfg scepter/methods/scedit/ctr/sd21_768_sce_ctr_pose.yaml --num_samples 1 --prompt 'super mario' --save_folder 'test_mario_pose' --image_size 768 --task control --image 'asset/images/pose_source.png' --control_mode source --pretrained_model ms://iic/scepter_scedit@controllable_model/SD2.1/pose_control/0_SwiftSCETuning/pytorch_model.bin   # pose
 ```

docs/zh_cn/conf.py

@@ -1,4 +1,5 @@
 # -*- coding: utf-8 -*-
+# Copyright (c) Alibaba, Inc. and its affiliates.
 # Configuration file for the Sphinx documentation builder.
 #
 # This file only contains a selection of the most common options. For a full

environment.yaml

@@ -6,4 +6,5 @@ dependencies:
   - pip>=20.3
   - numpy>=1.23.1
   - pip:
+      - -r requirements/recommended.txt
       - -r requirements.txt

example/classifier/classifier_dataset.py

@@ -2,7 +2,6 @@
 # Copyright (c) Alibaba, Inc. and its affiliates.
 import numpy as np
 import torchvision
-
 from scepter.modules.data.dataset.base_dataset import BaseDataset
 from scepter.modules.data.dataset.registry import DATASETS
 from scepter.modules.utils.config import dict_to_yaml

readme.md

@@ -18,6 +18,7 @@ SCEPTER offers 3 core components:
 
 
 ## 🎉 News
+- [2024.05]: Introducing SCEPTER v1, supporting customized image edit tasks! Simply provide 10 image pairs, SCEPTER will tune an edit tuner for your own Image-to-Image tasks, like `Clay Style`, `De-Text`, `Segmentation`, etc.
 - [2024.04]: New [StyleBooth](https://ali-vilab.github.io/stylebooth-page/) demo on SCEPTER Studio for`Text-Based Style Editing`.
 - [2024.03]: We optimize the training UI and checkpoint management. New [LAR-Gen](https://arxiv.org/abs/2403.19534) model has been added on SCEPTER Studio, supporting `zoom-out`, `virtual try on`, `inpainting`.
 - [2024.02]: We release new SCEdit controllable image synthesis models for SD v2.1 and SD XL. Multiple strategies applied to accelerate inference time for SCEPTER Studio.
@@ -29,59 +30,60 @@ SCEPTER offers 3 core components:
 
 ## 🖼 Gallery for Recent Works
 
-### StyleBooth
+### Edit Tuners
-<table>
-  <tr>
-    <td><strong>Origin Image</strong><br>Gold Dragon Tuner</td>
-    <td><strong>Graffiti Art</strong></td>
-    <td><strong>Adorable Kawaii</strong></td>
-    <td><strong>game-retro game</strong></td>
-    <td><strong>Vincent van Gogh</strong></td>
-  </tr>
-  <tr>
-    <td><img src="asset/images/scedit/tuner_gold_dragon.jpeg" width="240"></td>
-    <td><img src="asset/images/stylebooth/graffiti.jpeg" width="240"></td>
-    <td><img src="asset/images/stylebooth/kawaii.jpeg" width="240"></td>
-    <td><img src="asset/images/stylebooth/retrogame.jpeg" width="240"></td>
-    <td><img src="asset/images/stylebooth/vangogh.jpeg" width="240"></td>
-  </tr>
-</table>
 
-<table>
+Simply provide 10 image pairs, SCEPTER will tune an edit tuner for your own Image-to-Image tasks, like `Clay Style`, `De-Text`, `Segmentation`, etc.
-  <tr>
+Try our official few-shot datasets: [De-Text](https://www.modelscope.cn/api/v1/models/iic/scepter/repo?Revision=master&FilePath=datasets%2Fdetext.zip), [Image2Hed](https://www.modelscope.cn/api/v1/models/iic/scepter/repo?Revision=master&FilePath=datasets%2Fhed_pair.zip), [Image2Depth](https://www.modelscope.cn/api/v1/models/iic/scepter/repo?Revision=master&FilePath=datasets%2Fimage2depth.zip), [Depth2Image](https://www.modelscope.cn/api/v1/models/iic/scepter/repo?Revision=master&FilePath=datasets%2Fdepth2image.zip).
-    <td><strong>Origin Image</strong></td>
-    <td><strong>Lowpoly</strong></td>
-    <td><strong>Colored Pencil Art</strong></td>
-    <td><strong>Watercolor</strong></td>
-    <td><strong>misc-disco</strong></td>
-  </tr>
-  <tr>
-    <td><img src="asset/images/stylebooth/mountain.jpg" width="240"></td>
-    <td><img src="asset/images/stylebooth/lowpoly.jpg" width="240"></td>
-    <td><img src="asset/images/stylebooth/colorpencil.jpeg" width="240"></td>
-    <td><img src="asset/images/stylebooth/watercolor.jpeg" width="240"></td>
-    <td><img src="asset/images/stylebooth/disco.jpeg" width="240"></td>
-  </tr>
-</table>
 
 
+<table><tbody>
+  <tr>
+    <th align="center" colspan="4">Clay Style<br>Prompt: "Convert this image into clay style"</th>
+  </tr>
+  <tr>
+    <td><img src="asset/images/edit_tuner/vermeer.jpeg" width="300"></td>
+    <td><img src="asset/images/edit_tuner/clay_vermeer.jpeg" width="300"></td>
+    <td><img src="asset/images/edit_tuner/cat_512.jpg" width="300"></td>
+    <td><img src="asset/images/edit_tuner/clay_cat.jpeg" width="300"></td>
+  </tr>
+  <tr>
+    <th align="center" colspan="2">De-Text<br>Prompt: "Remove the texts"</th>
+    <th align="center" colspan="2">Image2Hed<br>Prompt: "Convert to an edge map"</th>
+  </tr>
+  <tr>
+    <td><img src="asset/images/edit_tuner/text.jpg" width="300"></td>
+    <td><img src="asset/images/edit_tuner/detext.jpeg" width="300"></td>
+    <td><img src="asset/images/edit_tuner/cat_512.jpg" width="300"></td>
+    <td><img src="asset/images/edit_tuner/hed.jpeg" width="300"></td>
+  </tr>
+    <tr>
+    <th align="center" colspan="2">Image2Depth<br>Prompt: "Calculate the depth map"</th>
+    <th align="center" colspan="2">Depth2Image<br>Prompt: "Convert depth map into color image"</th>
+  </tr>
+  <tr>
+    <td><img src="asset/images/edit_tuner/house.jpg" width="300"></td>
+    <td><img src="asset/images/edit_tuner/image2depth.jpeg" width="300"></td>
+    <td><img src="asset/images/edit_tuner/depth.jpg" width="300"></td>
+    <td><img src="asset/images/edit_tuner/depth2image.jpeg" width="300"></td>
+  </tr>
+</tbody>
+</table>
+Note: Left image is input and right image is output.
+
 ## 🛠️ Installation
 
-- Create new environment
+- Create new environment with `conda` command:
 
 ```shell
 conda env create -f environment.yaml
 conda activate scepter
 ```
-- We recommend installing the specific version of PyTorch and accelerate toolbox [xFormers](https://pypi.org/project/xformers/). You can install these recommended version by pip:
 
+- Install with `pip` command:
+
+We recommend installing the specific version of PyTorch and accelerate toolbox [xFormers](https://pypi.org/project/xformers/). You can install these recommended version by pip:
 ```shell
 pip install -r requirements/recommended.txt
-```
-
-- Install SCEPTER by the `pip` command:
-
-```shell
 pip install scepter
 ```
 
@@ -132,8 +134,6 @@ The startup of **SCEPTER Studio** eliminates the need for manual downloading and
 Depending on the network and hardware situation, the initial startup usually requires 15-60 minutes, primarily involving the download and processing of SDv1.5, SDv2.1, and SDXL models.
 Therefore, subsequent startups will become much faster (about one minute) as downloading is no longer required.
 
-To support the sharing and downloading of models,
-please make sure that you have installed **zip** and Git Large File Storage (**git lfs**).
 ### Usage Demo
 
 |              [Image Editing](https://www.modelscope.cn/api/v1/models/iic/scepter/repo?Revision=master&FilePath=assets%2Fscepter_studio%2Fimage_editing_20240419.webm)              |                [Training](https://www.modelscope.cn/api/v1/models/iic/scepter/repo?Revision=master&FilePath=assets%2Fscepter_studio%2Ftraining_20240419.webm)                 |              [Model Sharing](https://www.modelscope.cn/api/v1/models/iic/scepter/repo?Revision=master&FilePath=assets%2Fscepter_studio%2Fmodel_sharing_20240419.webm)               |             [Model Inference](https://www.modelscope.cn/api/v1/models/iic/scepter/repo?Revision=master&FilePath=assets%2Fscepter_studio%2Fmodel_inference_20240419.webm)              |             [Data Management](https://www.modelscope.cn/api/v1/models/iic/scepter/repo?Revision=master&FilePath=assets%2Fscepter_studio%2Fdata_management_20240419.webm)              |
@@ -142,7 +142,7 @@ please make sure that you have installed **zip** and Git Large File Storage (**g
 
 ### Modelscope Studio & Huggingface Space
 
-We deploy a work studio on Modelscope that includes only the inference tab, please refer to [ms_scepter_studio](https://www.modelscope.cn/studios/damo/scepter_studio/summary) and [hf_scepter_studio](https://huggingface.co/spaces/modelscope/scepter_studio)
+We deploy a work studio on Modelscope that includes only the inference tab, please refer to [ms_scepter_studio](https://www.modelscope.cn/studios/iic/scepter_studio/summary) and [hf_scepter_studio](https://huggingface.co/spaces/modelscope/scepter_studio)
 
 
 ## 🔍 Learn More

scepter/methods/edit/edit_512_lora.yaml

@@ -0,0 +1,250 @@
+ENV:
+  BACKEND: nccl
+SOLVER:
+  NAME: LatentDiffusionSolver
+  RESUME_FROM:
+  LOAD_MODEL_ONLY: True
+  USE_FSDP: False
+  SHARDING_STRATEGY:
+  USE_AMP: True
+  DTYPE: float16
+  CHANNELS_LAST: True
+  MAX_STEPS: 2000
+  MAX_EPOCHS: -1
+  NUM_FOLDS: 1
+  ACCU_STEP: 1
+  EVAL_INTERVAL: 100
+  #
+  WORK_DIR: ./cache/save_data/edit_512_lora
+  LOG_FILE: std_log.txt
+  #
+  FILE_SYSTEM:
+    NAME: "ModelscopeFs"
+    TEMP_DIR: "./cache/data"
+  #
+  TUNER:
+    -
+      NAME: SwiftLoRA
+      R: 64
+      LORA_ALPHA: 64
+      LORA_DROPOUT: 0.0
+      BIAS: "none"
+      TARGET_MODULES: model.*(to_q|to_k|to_v|to_out.0|net.0.proj|net.2)$
+  #
+  MODEL:
+    NAME: LatentDiffusionEdit
+    PARAMETERIZATION: eps
+    TIMESTEPS: 1000
+    MIN_SNR_GAMMA:
+    ZERO_TERMINAL_SNR: False
+    PRETRAINED_MODEL: ms://iic/stylebooth@models/stylebooth-tb-5000-0.bin
+    IGNORE_KEYS: [ ]
+    CONCAT_NO_SCALE_FACTOR: True
+    SCALE_FACTOR: 0.18215
+    SIZE_FACTOR: 8
+    # DEFAULT_N_PROMPT: 'lowres, error, worst quality, low quality, jpeg artifacts, ugly, duplicate, morbid, mutilated, out of frame, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, mutation, deformed, blurry, dehydrated, bad anatomy, bad proportions, extra limbs, cloned face, disfigured, gross proportions, malformed limbs, missing arms, missing legs, extra arms, extra legs, fused fingers, too many fingers, long neck, username, watermark, signature'
+    DEFAULT_N_PROMPT:
+    SCHEDULE_ARGS:
+      "NAME": "scaled_linear"
+      "BETA_MIN": 0.00085
+      "BETA_MAX": 0.012
+    USE_EMA: False
+    #
+    DIFFUSION_MODEL:
+      NAME: DiffusionUNet
+      IN_CHANNELS: 8
+      OUT_CHANNELS: 4
+      MODEL_CHANNELS: 320
+      NUM_HEADS: 8
+      NUM_RES_BLOCKS: 2
+      ATTENTION_RESOLUTIONS: [ 4, 2, 1 ]
+      CHANNEL_MULT: [ 1, 2, 4, 4 ]
+      CONV_RESAMPLE: True
+      DIMS: 2
+      USE_CHECKPOINT: False
+      USE_SCALE_SHIFT_NORM: False
+      RESBLOCK_UPDOWN: False
+      USE_SPATIAL_TRANSFORMER: True
+      TRANSFORMER_DEPTH: 1
+      CONTEXT_DIM: 768
+      DISABLE_MIDDLE_SELF_ATTN: False
+      USE_LINEAR_IN_TRANSFORMER: False
+      PRETRAINED_MODEL:
+      IGNORE_KEYS: []
+    #
+    FIRST_STAGE_MODEL:
+      NAME: AutoencoderKL
+      EMBED_DIM: 4
+      PRETRAINED_MODEL:
+      IGNORE_KEYS: []
+      BATCH_SIZE: 4
+      #
+      ENCODER:
+        NAME: Encoder
+        CH: 128
+        OUT_CH: 3
+        NUM_RES_BLOCKS: 2
+        IN_CHANNELS: 3
+        ATTN_RESOLUTIONS: [ ]
+        CH_MULT: [ 1, 2, 4, 4 ]
+        Z_CHANNELS: 4
+        DOUBLE_Z: True
+        DROPOUT: 0.0
+        RESAMP_WITH_CONV: True
+      #
+      DECODER:
+        NAME: Decoder
+        CH: 128
+        OUT_CH: 3
+        NUM_RES_BLOCKS: 2
+        IN_CHANNELS: 3
+        ATTN_RESOLUTIONS: [ ]
+        CH_MULT: [ 1, 2, 4, 4 ]
+        Z_CHANNELS: 4
+        DROPOUT: 0.0
+        RESAMP_WITH_CONV: True
+        GIVE_PRE_END: False
+        TANH_OUT: False
+    #
+    TOKENIZER:
+      NAME: ClipTokenizer
+      PRETRAINED_PATH: ms://AI-ModelScope/clip-vit-large-patch14
+      LENGTH: 77
+      CLEAN: True
+    #
+    COND_STAGE_MODEL:
+      NAME: FrozenCLIPEmbedder
+      FREEZE: True
+      LAYER: last
+      PRETRAINED_MODEL: ms://AI-ModelScope/clip-vit-large-patch14
+    #
+    LOSS:
+      NAME: ReconstructLoss
+      LOSS_TYPE: l2
+  #
+  SAMPLE_ARGS:
+    SAMPLER: ddim
+    SAMPLE_STEPS: 50
+    SEED: 2023
+    GUIDE_SCALE: #7.5
+      image: 1.5
+      text: 7.5
+    GUIDE_RESCALE: 0.5
+    DISCRETIZATION: trailing
+    IMAGE_SIZE: [512, 512]
+    RUN_TRAIN_N: False
+  #
+  OPTIMIZER:
+    NAME: AdamW
+    LEARNING_RATE: 0.064
+    BETAS: [ 0.9, 0.999 ]
+    EPS: 1e-8
+    WEIGHT_DECAY: 1e-2
+    AMSGRAD: False
+  #
+  TRAIN_DATA:
+    NAME: ImageTextPairMSDataset
+    MODE: train
+    MS_DATASET_NAME: cache/datasets/hed_pair
+    MS_DATASET_NAMESPACE: ""
+    MS_DATASET_SPLIT: "train"
+    MS_DATASET_SUBNAME: ""
+    PROMPT_PREFIX: ""
+    REPLACE_STYLE: False
+    PIN_MEMORY: True
+    BATCH_SIZE: 1
+    NUM_WORKERS: 4
+    SAMPLER:
+      NAME: LoopSampler
+    TRANSFORMS:
+      - NAME: LoadImageFromFileList
+        FILE_KEYS: ['img_path', 'src_path']
+        RGB_ORDER: RGB
+        BACKEND: pillow
+      - NAME: FlexibleResize
+        INTERPOLATION: bilinear
+        SIZE: [ 512, 512 ]
+        INPUT_KEY: [ 'img', 'src' ]
+        OUTPUT_KEY: [ 'img', 'src' ]
+        BACKEND: pillow
+      - NAME: FlexibleCenterCrop
+        SIZE: [ 512, 512 ]
+        INPUT_KEY: [ 'img', 'src' ]
+        OUTPUT_KEY: [ 'img', 'src' ]
+        BACKEND: pillow
+      - NAME: ImageToTensor
+        INPUT_KEY: [ 'img', 'src' ]
+        OUTPUT_KEY: [ 'img', 'src' ]
+        BACKEND: pillow
+      - NAME: Normalize
+        MEAN: [ 0.5,  0.5,  0.5 ]
+        STD: [ 0.5,  0.5,  0.5 ]
+        INPUT_KEY: [ 'img', 'src' ]
+        OUTPUT_KEY: [ 'image', 'condition_cat' ]
+        BACKEND: torchvision
+      - NAME: Select
+        KEYS: [ 'image', 'condition_cat', 'prompt' ]
+        META_KEYS: [ 'data_key' ]
+  #
+  TRAIN_HOOKS:
+    -
+      NAME: BackwardHook
+      PRIORITY: 0
+    -
+      NAME: LogHook
+      LOG_INTERVAL: 50
+    -
+      NAME: CheckpointHook
+      INTERVAL: 1000
+    -
+      NAME: ProbeDataHook
+      PROB_INTERVAL: 100
+
+  EVAL_DATA:
+    NAME: Text2ImageDataset
+    MODE: eval
+    PROMPT_FILE:
+    PROMPT_DATA: [ "Convert to an edge map#;#cache/datasets/hed_pair/images/src_001.jpeg" ]
+    IMAGE_SIZE: [ 512, 512 ]
+    FIELDS: [ "prompt", "src_path" ]
+    DELIMITER: '#;#'
+    PROMPT_PREFIX: ''
+    PIN_MEMORY: True
+    BATCH_SIZE: 1
+    NUM_WORKERS: 4
+    TRANSFORMS:
+      - NAME: LoadImageFromFileList
+        FILE_KEYS: [ 'src_path' ]
+        RGB_ORDER: RGB
+        BACKEND: pillow
+      - NAME: FlexibleResize
+        INTERPOLATION: bilinear
+        SIZE: [ 512, 512 ]
+        INPUT_KEY: [ 'src' ]
+        OUTPUT_KEY: [ 'src' ]
+        BACKEND: pillow
+      - NAME: FlexibleCenterCrop
+        SIZE: [ 512, 512 ]
+        INPUT_KEY: [ 'src' ]
+        OUTPUT_KEY: [ 'src' ]
+        BACKEND: pillow
+      - NAME: ImageToTensor
+        INPUT_KEY: [ 'src' ]
+        OUTPUT_KEY: [ 'src' ]
+        BACKEND: pillow
+      - NAME: Normalize
+        MEAN: [ 0.5,  0.5,  0.5 ]
+        STD: [ 0.5,  0.5,  0.5 ]
+        INPUT_KEY: [ 'src' ]
+        OUTPUT_KEY: [ 'condition_cat' ]
+        BACKEND: torchvision
+      - NAME: Select
+        KEYS: [ 'condition_cat', 'prompt' ]
+        META_KEYS: [ 'image_size' ]
+  EVAL_HOOKS:
+    -
+      NAME: ProbeDataHook
+      PROB_INTERVAL: 100
+      SAVE_LAST: True
+      SAVE_NAME_PREFIX: 'step'
+      SAVE_PROBE_PREFIX: 'image'

scepter/methods/scedit/ctr/sd15_512_sce_ctr_hed.yaml

@@ -127,7 +127,7 @@ SOLVER:
         DOWN_RATIO: 1.0
     CONTROL_ANNO:
       NAME: HedAnnotator
-      PRETRAINED_MODEL: ms://damo/scepter_scedit@annotator/ckpts/ControlNetHED.pth
+      PRETRAINED_MODEL: ms://iic/scepter_scedit@annotator/ckpts/ControlNetHED.pth
   #
   SAMPLE_ARGS:
     SAMPLER: ddim

scepter/methods/scedit/ctr/sd21_768_sce_ctr_pose.yaml

@@ -125,8 +125,8 @@ SOLVER:
         DOWN_RATIO: 1.0
     CONTROL_ANNO:
       NAME: OpenposeAnnotator
-      BODY_MODEL_PATH: ms://damo/scepter_scedit@annotator/ckpts/body_pose_model.pth
+      BODY_MODEL_PATH: ms://iic/scepter_scedit@annotator/ckpts/body_pose_model.pth
-      HAND_MODEL_PATH: ms://damo/scepter_scedit@annotator/ckpts/hand_pose_model.pth
+      HAND_MODEL_PATH: ms://iic/scepter_scedit@annotator/ckpts/hand_pose_model.pth
   #
   SAMPLE_ARGS:
     SAMPLER: ddim

scepter/methods/scedit/ctr/sdxl_1024_sce_ctr_depth.yaml

@@ -241,7 +241,7 @@ SOLVER:
         DOWN_RATIO: 1.0
     CONTROL_ANNO:
       NAME: MidasDetector
-      PRETRAINED_MODEL: ms://damo/scepter_scedit@annotator/ckpts/dpt_hybrid-midas-501f0c75.pt
+      PRETRAINED_MODEL: ms://iic/scepter_scedit@annotator/ckpts/dpt_hybrid-midas-501f0c75.pt
   #
   SAMPLE_ARGS:
     SAMPLER: ddim

scepter/methods/studio/extensions/controllers/official_controllers.yaml

@@ -5,59 +5,59 @@ CONTROLLERS:
     DESCRIPTION:
     BASE_MODEL: SD2.1
     TYPE: Canny
-    MODEL_PATH: ms://damo/scepter_scedit@controllable_model/SD2.1/canny_control/
+    MODEL_PATH: ms://iic/scepter_scedit@controllable_model/SD2.1/canny_control/
   - NAME: openpose
     NAME_ZH:
     DESCRIPTION:
     BASE_MODEL: SD2.1
     TYPE: Openpose
-    MODEL_PATH: ms://damo/scepter_scedit@controllable_model/SD2.1/pose_control/
+    MODEL_PATH: ms://iic/scepter_scedit@controllable_model/SD2.1/pose_control/
   - NAME: color
     NAME_ZH:
     DESCRIPTION:
     BASE_MODEL: SD2.1
     TYPE: Color
-    MODEL_PATH: ms://damo/scepter_scedit@controllable_model/SD2.1/color_control/
+    MODEL_PATH: ms://iic/scepter_scedit@controllable_model/SD2.1/color_control/
   - NAME: hed
     NAME_ZH:
     DESCRIPTION:
     BASE_MODEL: SD2.1
     TYPE: Hed
-    MODEL_PATH: ms://damo/scepter_scedit@controllable_model/SD2.1/hed_control
+    MODEL_PATH: ms://iic/scepter_scedit@controllable_model/SD2.1/hed_control
   - NAME: depth
     NAME_ZH:
     DESCRIPTION:
     BASE_MODEL: SD2.1
     TYPE: Midas
-    MODEL_PATH: ms://damo/scepter_scedit@controllable_model/SD2.1/depth_control
+    MODEL_PATH: ms://iic/scepter_scedit@controllable_model/SD2.1/depth_control
   # SD_XL1.0
   - NAME: canny
     NAME_ZH:
     DESCRIPTION:
     BASE_MODEL: SD_XL1.0
     TYPE: Canny
-    MODEL_PATH: ms://damo/scepter_scedit@controllable_model/SD_XL1.0/canny_control
+    MODEL_PATH: ms://iic/scepter_scedit@controllable_model/SD_XL1.0/canny_control
   - NAME: color
     NAME_ZH:
     DESCRIPTION:
     BASE_MODEL: SD_XL1.0
     TYPE: Color
-    MODEL_PATH: ms://damo/scepter_scedit@controllable_model/SD_XL1.0/color_control
+    MODEL_PATH: ms://iic/scepter_scedit@controllable_model/SD_XL1.0/color_control
   - NAME: depth
     NAME_ZH:
     DESCRIPTION:
     BASE_MODEL: SD_XL1.0
     TYPE: Midas
-    MODEL_PATH: ms://damo/scepter_scedit@controllable_model/SD_XL1.0/depth_control
+    MODEL_PATH: ms://iic/scepter_scedit@controllable_model/SD_XL1.0/depth_control
   - NAME: hed
     NAME_ZH:
     DESCRIPTION:
     BASE_MODEL: SD_XL1.0
     TYPE: Hed
-    MODEL_PATH: ms://damo/scepter_scedit@controllable_model/SD_XL1.0/hed_control
+    MODEL_PATH: ms://iic/scepter_scedit@controllable_model/SD_XL1.0/hed_control
   - NAME: openpose
     NAME_ZH:
     DESCRIPTION:
     BASE_MODEL: SD_XL1.0
     TYPE: Openpose
-    MODEL_PATH: ms://damo/scepter_scedit@controllable_model/SD_XL1.0/pose_control
+    MODEL_PATH: ms://iic/scepter_scedit@controllable_model/SD_XL1.0/pose_control

scepter/methods/studio/extensions/tuners/official_tuners.yaml

@@ -1,11 +1,20 @@
 TUNERS:
+  - NAME: Clay-Style-Editing
+    NAME_ZH: 黏土风
+    SOURCE: scepter
+    DESCRIPTION: None
+    BASE_MODEL: EDIT
+    MODEL_PATH: ms://iic/stylebooth@tuners/clay_style_edit/
+    IMAGE_PATH: ms://iic/stylebooth@tuners/clay_style_edit/image.jpg
+    TUNER_TYPE: LORA
+    PROMPT_EXAMPLE: Convert this image into clay style
   - NAME: Azure-Dragon
     NAME_ZH: 青龙
     SOURCE: scepter
     DESCRIPTION: None
     BASE_MODEL: SD_XL1.0
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/azure_dragon/
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/azure_dragon/
-    IMAGE_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/azure_dragon/xl_azure_dragon.png
+    IMAGE_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/azure_dragon/xl_azure_dragon.png
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: Azure Dragon, 8K, high quality,Ultra High Detail.One of the Four Divine Creatures in Charge of Water.
   - NAME: Gold-Dragon
@@ -13,8 +22,8 @@ TUNERS:
     SOURCE: scepter
     DESCRIPTION: None
     BASE_MODEL: SD_XL1.0
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/gold_dragon/
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/gold_dragon/
-    IMAGE_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/gold_dragon/xl_gold_dragon.png
+    IMAGE_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/gold_dragon/xl_gold_dragon.png
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: Chinese Gold Dragon in the clouds. Translucent Texture. Zbrush.  Fuzzy Art. Exquisite Craftsmanship. 3D. 8K. Ultra High Detail
   - NAME: SpringFestival-Dragon
@@ -22,8 +31,8 @@ TUNERS:
     SOURCE: scepter
     DESCRIPTION: None
     BASE_MODEL: SD_XL1.0
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/spring_festival_dragon/
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/spring_festival_dragon/
-    IMAGE_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/spring_festival_dragon/xl_spring_festival_dragon.png
+    IMAGE_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/spring_festival_dragon/xl_spring_festival_dragon.png
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: Chinese dragon. Spring Festival.Festive.Street.Lanterns.32K.High quality.expressive, dramatic, dreamlike and mysterious, Surrealism
   - NAME: Red-Dragon
@@ -31,8 +40,8 @@ TUNERS:
     SOURCE: scepter
     DESCRIPTION: None
     BASE_MODEL: SD_XL1.0
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/red_dragon/
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/red_dragon/
-    IMAGE_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/red_dragon/xl_red_dragon.png
+    IMAGE_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/red_dragon/xl_red_dragon.png
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: Traditional Red Dragon of China. Low Water Level. Studio Ghibli Style. Mural Illustration. White Background. High Detail
   - NAME: ChinesePunk-Dragon
@@ -40,8 +49,8 @@ TUNERS:
     SOURCE: scepter
     DESCRIPTION: None
     BASE_MODEL: SD_XL1.0
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/chinese_punk_dragon/
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/chinese_punk_dragon/
-    IMAGE_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/chinese_punk_dragon/xl_chinese_punk_dragon.png
+    IMAGE_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/chinese_punk_dragon/xl_chinese_punk_dragon.png
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: uhd Image,Dragon,Chinese Dragon, Dunhuang Mural Style, Traditional Maritime Art Style
   - NAME: Cute-Dragon
@@ -49,8 +58,8 @@ TUNERS:
     SOURCE: scepter
     DESCRIPTION: None
     BASE_MODEL: SD_XL1.0
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/cute_dragon/
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/cute_dragon/
-    IMAGE_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/cute_dragon/xl_kawaii_dragon.png
+    IMAGE_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/cute_dragon/xl_kawaii_dragon.png
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: China Kawaii Dragon. Contest Winner. Minimalist Illustration. White Background. Flat Style. Digital Painting Style. Red. 32k uhd. Fun Comics. Fuzzy Art. Bold. Comic-Inspired Characters
   - NAME: Dragon-Baby
@@ -58,8 +67,8 @@ TUNERS:
     SOURCE: scepter
     DESCRIPTION: None
     BASE_MODEL: SD_XL1.0
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/baby_dragon/
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/baby_dragon/
-    IMAGE_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/baby_dragon/xl_baby_dragon.png
+    IMAGE_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/baby_dragon/xl_baby_dragon.png
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: Warm Colors, Soft,Chinese Dragon Baby, Felt Style,Dragon Baby, Best Quality, 3D Doll, Macaron Tones, Glittering Big Eyes, Winter,Dragon
   - NAME: Sloppy-Dragon
@@ -67,8 +76,8 @@ TUNERS:
     SOURCE: scepter
     DESCRIPTION: None
     BASE_MODEL: SD_XL1.0
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/sloppy_dragon/
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/sloppy_dragon/
-    IMAGE_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/sloppy_dragon/xl_sloppy_dragon.png
+    IMAGE_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/sloppy_dragon/xl_sloppy_dragon.png
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: Messy Chinese Dragon,Cute, Wu Guanzhong, Rough
   -
@@ -77,8 +86,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD_XL1.0
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD_XL1.0/894f40ed44b37c3372e6a22b8ae577a4.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD_XL1.0/894f40ed44b37c3372e6a22b8ae577a4.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/Caricature
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/Caricature
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -87,8 +96,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD2.1
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD2.1/894f40ed44b37c3372e6a22b8ae577a4.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD2.1/894f40ed44b37c3372e6a22b8ae577a4.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD2.1/Caricature
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD2.1/Caricature
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -97,8 +106,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD1.5
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD1.5/894f40ed44b37c3372e6a22b8ae577a4.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD1.5/894f40ed44b37c3372e6a22b8ae577a4.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD1.5/Caricature
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD1.5/Caricature
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -107,8 +116,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD_XL1.0
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD_XL1.0/80e5b4075c572c04cbb4e48c37b8366b.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD_XL1.0/80e5b4075c572c04cbb4e48c37b8366b.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/ColorFieldPainting
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/ColorFieldPainting
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -117,8 +126,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD2.1
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD2.1/80e5b4075c572c04cbb4e48c37b8366b.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD2.1/80e5b4075c572c04cbb4e48c37b8366b.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD2.1/ColorFieldPainting
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD2.1/ColorFieldPainting
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -127,8 +136,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD1.5
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD1.5/80e5b4075c572c04cbb4e48c37b8366b.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD1.5/80e5b4075c572c04cbb4e48c37b8366b.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD1.5/ColorFieldPainting
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD1.5/ColorFieldPainting
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -137,8 +146,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD_XL1.0
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD_XL1.0/9ae235d7f1a7c2a4edab52a5e9f9cbae.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD_XL1.0/9ae235d7f1a7c2a4edab52a5e9f9cbae.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/ColoredPencilArt
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/ColoredPencilArt
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -147,8 +156,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD2.1
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD2.1/9ae235d7f1a7c2a4edab52a5e9f9cbae.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD2.1/9ae235d7f1a7c2a4edab52a5e9f9cbae.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD2.1/ColoredPencilArt
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD2.1/ColoredPencilArt
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -157,8 +166,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD1.5
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD1.5/9ae235d7f1a7c2a4edab52a5e9f9cbae.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD1.5/9ae235d7f1a7c2a4edab52a5e9f9cbae.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD1.5/ColoredPencilArt
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD1.5/ColoredPencilArt
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -167,8 +176,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD_XL1.0
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD_XL1.0/3da915da2f5cedaf243e57e08163f35b.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD_XL1.0/3da915da2f5cedaf243e57e08163f35b.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/DarkMoodyAtmosphere
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/DarkMoodyAtmosphere
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -177,8 +186,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD2.1
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD2.1/3da915da2f5cedaf243e57e08163f35b.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD2.1/3da915da2f5cedaf243e57e08163f35b.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD2.1/DarkMoodyAtmosphere
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD2.1/DarkMoodyAtmosphere
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -187,8 +196,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD1.5
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD1.5/3da915da2f5cedaf243e57e08163f35b.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD1.5/3da915da2f5cedaf243e57e08163f35b.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD1.5/DarkMoodyAtmosphere
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD1.5/DarkMoodyAtmosphere
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -197,8 +206,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD_XL1.0
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD_XL1.0/69fd81f5983107acc3d334af62915851.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD_XL1.0/69fd81f5983107acc3d334af62915851.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/DrippingPaintSplatterArt
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/DrippingPaintSplatterArt
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -207,8 +216,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD2.1
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD2.1/69fd81f5983107acc3d334af62915851.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD2.1/69fd81f5983107acc3d334af62915851.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD2.1/DrippingPaintSplatterArt
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD2.1/DrippingPaintSplatterArt
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -217,8 +226,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD1.5
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD1.5/69fd81f5983107acc3d334af62915851.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD1.5/69fd81f5983107acc3d334af62915851.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD1.5/DrippingPaintSplatterArt
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD1.5/DrippingPaintSplatterArt
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -227,8 +236,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD_XL1.0
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD_XL1.0/f152edb4b3ca6248758b48115258ddfa.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD_XL1.0/f152edb4b3ca6248758b48115258ddfa.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/FadedPolaroidPhoto
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/FadedPolaroidPhoto
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -237,8 +246,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD2.1
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD2.1/f152edb4b3ca6248758b48115258ddfa.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD2.1/f152edb4b3ca6248758b48115258ddfa.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD2.1/FadedPolaroidPhoto
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD2.1/FadedPolaroidPhoto
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -247,8 +256,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD1.5
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD1.5/f152edb4b3ca6248758b48115258ddfa.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD1.5/f152edb4b3ca6248758b48115258ddfa.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD1.5/FadedPolaroidPhoto
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD1.5/FadedPolaroidPhoto
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -257,8 +266,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD_XL1.0
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD_XL1.0/940cfd34155634cf051e1b2942cca426.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD_XL1.0/940cfd34155634cf051e1b2942cca426.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/Flat2DArt
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/Flat2DArt
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -267,8 +276,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD2.1
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD2.1/940cfd34155634cf051e1b2942cca426.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD2.1/940cfd34155634cf051e1b2942cca426.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD2.1/Flat2DArt
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD2.1/Flat2DArt
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -277,8 +286,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD1.5
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD1.5/940cfd34155634cf051e1b2942cca426.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD1.5/940cfd34155634cf051e1b2942cca426.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD1.5/Flat2DArt
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD1.5/Flat2DArt
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -287,8 +296,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD_XL1.0
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD_XL1.0/57b751b11564cb22cd49ef21f2004a5f.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD_XL1.0/57b751b11564cb22cd49ef21f2004a5f.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/GraffitiArt
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/GraffitiArt
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -297,8 +306,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD2.1
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD2.1/57b751b11564cb22cd49ef21f2004a5f.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD2.1/57b751b11564cb22cd49ef21f2004a5f.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD2.1/GraffitiArt
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD2.1/GraffitiArt
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -307,8 +316,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD1.5
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD1.5/57b751b11564cb22cd49ef21f2004a5f.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD1.5/57b751b11564cb22cd49ef21f2004a5f.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD1.5/GraffitiArt
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD1.5/GraffitiArt
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -317,8 +326,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD_XL1.0
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD_XL1.0/0312b673dc6858a9864d7f45f0c5c1fc.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD_XL1.0/0312b673dc6858a9864d7f45f0c5c1fc.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/Impressionism
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/Impressionism
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -327,8 +336,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD2.1
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD2.1/0312b673dc6858a9864d7f45f0c5c1fc.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD2.1/0312b673dc6858a9864d7f45f0c5c1fc.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD2.1/Impressionism
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD2.1/Impressionism
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -337,8 +346,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD1.5
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD1.5/0312b673dc6858a9864d7f45f0c5c1fc.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD1.5/0312b673dc6858a9864d7f45f0c5c1fc.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD1.5/Impressionism
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD1.5/Impressionism
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -347,8 +356,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD_XL1.0
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD_XL1.0/9aa040b0c60d289da9610c91ad9b7c7e.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD_XL1.0/9aa040b0c60d289da9610c91ad9b7c7e.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/LogoDesign
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/LogoDesign
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -357,8 +366,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD2.1
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD2.1/9aa040b0c60d289da9610c91ad9b7c7e.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD2.1/9aa040b0c60d289da9610c91ad9b7c7e.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD2.1/LogoDesign
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD2.1/LogoDesign
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -367,8 +376,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD1.5
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD1.5/9aa040b0c60d289da9610c91ad9b7c7e.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD1.5/9aa040b0c60d289da9610c91ad9b7c7e.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD1.5/LogoDesign
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD1.5/LogoDesign
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -377,8 +386,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD_XL1.0
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD_XL1.0/a9056e1eac85e5e4fe96a93917d4cce4.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD_XL1.0/a9056e1eac85e5e4fe96a93917d4cce4.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/PencilSketchDrawing
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/PencilSketchDrawing
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -387,8 +396,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD2.1
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD2.1/a9056e1eac85e5e4fe96a93917d4cce4.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD2.1/a9056e1eac85e5e4fe96a93917d4cce4.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD2.1/PencilSketchDrawing
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD2.1/PencilSketchDrawing
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -397,8 +406,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD1.5
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD1.5/a9056e1eac85e5e4fe96a93917d4cce4.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD1.5/a9056e1eac85e5e4fe96a93917d4cce4.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD1.5/PencilSketchDrawing
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD1.5/PencilSketchDrawing
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -407,8 +416,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD_XL1.0
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD_XL1.0/568777f447fc02510b618152726d5002.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD_XL1.0/568777f447fc02510b618152726d5002.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/SilhouetteArt
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/SilhouetteArt
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -417,8 +426,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD2.1
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD2.1/568777f447fc02510b618152726d5002.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD2.1/568777f447fc02510b618152726d5002.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD2.1/SilhouetteArt
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD2.1/SilhouetteArt
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -427,8 +436,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD1.5
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD1.5/568777f447fc02510b618152726d5002.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD1.5/568777f447fc02510b618152726d5002.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD1.5/SilhouetteArt
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD1.5/SilhouetteArt
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -437,8 +446,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD_XL1.0
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD_XL1.0/07d7b27cd73f2d43684003563511c15b.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD_XL1.0/07d7b27cd73f2d43684003563511c15b.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/Steampunk2
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/Steampunk2
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -447,8 +456,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD2.1
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD2.1/07d7b27cd73f2d43684003563511c15b.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD2.1/07d7b27cd73f2d43684003563511c15b.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD2.1/Steampunk2
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD2.1/Steampunk2
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -457,8 +466,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD1.5
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD1.5/07d7b27cd73f2d43684003563511c15b.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD1.5/07d7b27cd73f2d43684003563511c15b.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD1.5/Steampunk2
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD1.5/Steampunk2
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -467,8 +476,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD_XL1.0
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD_XL1.0/2d1e9867058db2c57f2fe47530de3243.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD_XL1.0/2d1e9867058db2c57f2fe47530de3243.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/StickerDesigns
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/StickerDesigns
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -477,8 +486,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD2.1
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD2.1/2d1e9867058db2c57f2fe47530de3243.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD2.1/2d1e9867058db2c57f2fe47530de3243.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD2.1/StickerDesigns
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD2.1/StickerDesigns
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -487,8 +496,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD1.5
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD1.5/2d1e9867058db2c57f2fe47530de3243.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD1.5/2d1e9867058db2c57f2fe47530de3243.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD1.5/StickerDesigns
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD1.5/StickerDesigns
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -497,8 +506,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD_XL1.0
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD_XL1.0/8859d532ae5901cc8457d6118fb9b7da.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD_XL1.0/8859d532ae5901cc8457d6118fb9b7da.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/Watercolor2
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/Watercolor2
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -507,8 +516,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD2.1
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD2.1/8859d532ae5901cc8457d6118fb9b7da.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD2.1/8859d532ae5901cc8457d6118fb9b7da.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD2.1/Watercolor2
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD2.1/Watercolor2
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -517,8 +526,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: diva
     BASE_MODEL: SD1.5
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD1.5/8859d532ae5901cc8457d6118fb9b7da.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD1.5/8859d532ae5901cc8457d6118fb9b7da.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD1.5/Watercolor2
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD1.5/Watercolor2
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -527,8 +536,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: mre
     BASE_MODEL: SD_XL1.0
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD_XL1.0/5895d78cf58c1ca05178991f37cc48ff.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD_XL1.0/5895d78cf58c1ca05178991f37cc48ff.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/mre-elemental-art
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/mre-elemental-art
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -537,8 +546,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: mre
     BASE_MODEL: SD2.1
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD2.1/5895d78cf58c1ca05178991f37cc48ff.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD2.1/5895d78cf58c1ca05178991f37cc48ff.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD2.1/mre-elemental-art
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD2.1/mre-elemental-art
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -547,8 +556,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: mre
     BASE_MODEL: SD1.5
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD1.5/5895d78cf58c1ca05178991f37cc48ff.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD1.5/5895d78cf58c1ca05178991f37cc48ff.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD1.5/mre-elemental-art
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD1.5/mre-elemental-art
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -557,8 +566,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: mre
     BASE_MODEL: SD_XL1.0
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD_XL1.0/a08149bc8e50f6bc65c0010d4cd416f8.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD_XL1.0/a08149bc8e50f6bc65c0010d4cd416f8.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/mre-anime
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/mre-anime
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -567,8 +576,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: mre
     BASE_MODEL: SD2.1
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD2.1/a08149bc8e50f6bc65c0010d4cd416f8.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD2.1/a08149bc8e50f6bc65c0010d4cd416f8.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD2.1/mre-anime
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD2.1/mre-anime
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -577,8 +586,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: mre
     BASE_MODEL: SD1.5
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD1.5/a08149bc8e50f6bc65c0010d4cd416f8.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD1.5/a08149bc8e50f6bc65c0010d4cd416f8.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD1.5/mre-anime
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD1.5/mre-anime
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -587,8 +596,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: mre
     BASE_MODEL: SD_XL1.0
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD_XL1.0/48c65cebf1fa4284d7b8feb619412e65.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD_XL1.0/48c65cebf1fa4284d7b8feb619412e65.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/mre-comic
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/mre-comic
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -597,8 +606,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: mre
     BASE_MODEL: SD2.1
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD2.1/48c65cebf1fa4284d7b8feb619412e65.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD2.1/48c65cebf1fa4284d7b8feb619412e65.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD2.1/mre-comic
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD2.1/mre-comic
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -607,8 +616,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: mre
     BASE_MODEL: SD1.5
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD1.5/48c65cebf1fa4284d7b8feb619412e65.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD1.5/48c65cebf1fa4284d7b8feb619412e65.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD1.5/mre-comic
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD1.5/mre-comic
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -617,8 +626,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: sai
     BASE_MODEL: SD_XL1.0
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD_XL1.0/8fc51113f725f27326c4398a7457cd6d.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD_XL1.0/8fc51113f725f27326c4398a7457cd6d.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/sai-craftclay
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/sai-craftclay
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -627,8 +636,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: sai
     BASE_MODEL: SD2.1
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD2.1/8fc51113f725f27326c4398a7457cd6d.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD2.1/8fc51113f725f27326c4398a7457cd6d.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD2.1/sai-craftclay
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD2.1/sai-craftclay
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -637,8 +646,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: sai
     BASE_MODEL: SD1.5
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD1.5/8fc51113f725f27326c4398a7457cd6d.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD1.5/8fc51113f725f27326c4398a7457cd6d.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD1.5/sai-craftclay
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD1.5/sai-craftclay
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -647,8 +656,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: sai
     BASE_MODEL: SD_XL1.0
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD_XL1.0/a6f8d92afcd5803dfb2ebecbc92091b6.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD_XL1.0/a6f8d92afcd5803dfb2ebecbc92091b6.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/sai-fantasyart
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/sai-fantasyart
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -657,8 +666,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: sai
     BASE_MODEL: SD2.1
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD2.1/a6f8d92afcd5803dfb2ebecbc92091b6.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD2.1/a6f8d92afcd5803dfb2ebecbc92091b6.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD2.1/sai-fantasyart
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD2.1/sai-fantasyart
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -667,8 +676,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: sai
     BASE_MODEL: SD1.5
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD1.5/a6f8d92afcd5803dfb2ebecbc92091b6.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD1.5/a6f8d92afcd5803dfb2ebecbc92091b6.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD1.5/sai-fantasyart
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD1.5/sai-fantasyart
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -677,8 +686,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: sai
     BASE_MODEL: SD_XL1.0
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD_XL1.0/034a51b0dd34b018be8859bf45b4f7ed.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD_XL1.0/034a51b0dd34b018be8859bf45b4f7ed.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/sai-lineart
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/sai-lineart
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -687,8 +696,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: sai
     BASE_MODEL: SD2.1
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD2.1/034a51b0dd34b018be8859bf45b4f7ed.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD2.1/034a51b0dd34b018be8859bf45b4f7ed.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD2.1/sai-lineart
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD2.1/sai-lineart
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -697,8 +706,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: sai
     BASE_MODEL: SD1.5
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD1.5/034a51b0dd34b018be8859bf45b4f7ed.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD1.5/034a51b0dd34b018be8859bf45b4f7ed.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD1.5/sai-lineart
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD1.5/sai-lineart
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -707,8 +716,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: sai
     BASE_MODEL: SD_XL1.0
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD_XL1.0/7e9ed25bb34008beb5f417df63c4b2fe.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD_XL1.0/7e9ed25bb34008beb5f417df63c4b2fe.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/sai-neonpunk
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/sai-neonpunk
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -717,8 +726,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: sai
     BASE_MODEL: SD2.1
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD2.1/7e9ed25bb34008beb5f417df63c4b2fe.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD2.1/7e9ed25bb34008beb5f417df63c4b2fe.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD2.1/sai-neonpunk
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD2.1/sai-neonpunk
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -727,8 +736,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: sai
     BASE_MODEL: SD1.5
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD1.5/7e9ed25bb34008beb5f417df63c4b2fe.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD1.5/7e9ed25bb34008beb5f417df63c4b2fe.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD1.5/sai-neonpunk
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD1.5/sai-neonpunk
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -737,8 +746,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: sai
     BASE_MODEL: SD_XL1.0
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD_XL1.0/924f46a8f276011a0953d7988e90ee25.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD_XL1.0/924f46a8f276011a0953d7988e90ee25.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/sai-origami
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/sai-origami
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -747,8 +756,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: sai
     BASE_MODEL: SD2.1
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD2.1/924f46a8f276011a0953d7988e90ee25.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD2.1/924f46a8f276011a0953d7988e90ee25.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD2.1/sai-origami
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD2.1/sai-origami
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -757,8 +766,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: sai
     BASE_MODEL: SD1.5
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD1.5/924f46a8f276011a0953d7988e90ee25.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD1.5/924f46a8f276011a0953d7988e90ee25.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD1.5/sai-origami
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD1.5/sai-origami
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -767,8 +776,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: sai
     BASE_MODEL: SD_XL1.0
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD_XL1.0/a5ab89c0960be8c1216e65c98d92ae4a.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD_XL1.0/a5ab89c0960be8c1216e65c98d92ae4a.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD_XL1.0/sai-pixelart
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD_XL1.0/sai-pixelart
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -777,8 +786,8 @@ TUNERS:
     DESCRIPTION:
     SOURCE: sai
     BASE_MODEL: SD2.1
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD2.1/a5ab89c0960be8c1216e65c98d92ae4a.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD2.1/a5ab89c0960be8c1216e65c98d92ae4a.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD2.1/sai-pixelart
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD2.1/sai-pixelart
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket
   -
@@ -787,7 +796,7 @@ TUNERS:
     DESCRIPTION:
     SOURCE: sai
     BASE_MODEL: SD1.5
-    IMAGE_PATH: ms://damo/scepter@mantra_images/SD1.5/a5ab89c0960be8c1216e65c98d92ae4a.png
+    IMAGE_PATH: ms://iic/scepter@mantra_images_jpg/SD1.5/a5ab89c0960be8c1216e65c98d92ae4a.jpg
-    MODEL_PATH: ms://damo/scepter_scedit@tuners_model/SD1.5/sai-pixelart
+    MODEL_PATH: ms://iic/scepter_scedit@tuners_model/SD1.5/sai-pixelart
     TUNER_TYPE: SwiftSCE
     PROMPT_EXAMPLE: a boy wearing green jacket

scepter/methods/studio/home/home.yaml

@@ -10,7 +10,7 @@ DESC_INFO:
     <table align="center">
       <tr>
       <td>
-        <img src="https://modelscope.cn/api/v1/models/damo/scepter/repo?Revision=master&FilePath=assets/scepter_studio/scepter_studio_banner.jpg">
+        <img src="https://modelscope.cn/api/v1/models/iic/scepter/repo?Revision=master&FilePath=assets/scepter_studio/scepter_studio_banner.jpg">
         <h3><center>SCEPTER Studio是基于开源基模型和自研微调编辑算法构建的生成定制和编辑工具箱，提供围绕生成、微调、编辑、数据处理等一系列的工具和插件。</center><h3>
       </td>
       </tr>
@@ -22,7 +22,7 @@ DESC_INFO:
     <table align="center">
       <tr>
       <td>
-        <img src="https://modelscope.cn/api/v1/models/damo/scepter/repo?Revision=master&FilePath=assets/scepter_studio/scepter_studio_banner.jpg">
+        <img src="https://modelscope.cn/api/v1/models/iic/scepter/repo?Revision=master&FilePath=assets/scepter_studio/scepter_studio_banner.jpg">
         <h3><center>SCEPTER Studio is a customized generation and editing toolkit built on the open-source base models and proprietary fine-tuning editing algorithms, offering a range of tools and plugins centered around generation, fine-tuning, editing, and data processing.</center><h3>
       </td>
       </tr>

scepter/methods/studio/inference/edit/stylebooth_tb_pro.yaml

@@ -2,7 +2,7 @@ NAME: EDIT
 IS_DEFAULT: False
 DEFAULT_PARAS:
   PARAS:
-    RESOLUTIONS: [[1024, 1024]]
+    RESOLUTIONS: [[512, 512], [1024, 1024]]
   INPUT:
     IMAGE:
     PROMPT: ""
@@ -49,7 +49,7 @@ DEFAULT_PARAS:
           INPUT: ["PROMPT", "NEGATIVE_PROMPT"]
 
 MODEL:
-  PRETRAINED_MODEL: ms://damo/stylebooth@models/stylebooth-tb-5000-0.bin
+  PRETRAINED_MODEL: ms://iic/stylebooth@models/stylebooth-tb-5000-0.bin
   SCHEDULE:
     PARAMETERIZATION: "eps"
     TIMESTEPS: 1000

scepter/methods/studio/inference/inference.yaml

@@ -63,7 +63,8 @@ DIFFUSION_PARAS:
     MAX: 1.0
     DEFAULT: 0.15
   RESOLUTIONS:
-    VALUES: [[704, 1408], [704, 1344], [768, 1344],
+    VALUES: [ [512, 512], [768, 768],
+              [704, 1408], [704, 1344], [768, 1344],
              [720, 1280],
              [768, 1280], [832, 1216], [832, 1152],
              [896, 1152], [896, 1088], [960, 1088],
@@ -87,18 +88,18 @@ CONTROLABLE_ANNOTATORS:
     IS_DEFAULT: True
   -
     NAME: "HedAnnotator"
-    PRETRAINED_MODEL: "ms://damo/scepter_scedit@annotator/ckpts/ControlNetHED.pth"
+    PRETRAINED_MODEL: "ms://iic/scepter_scedit@annotator/ckpts/ControlNetHED.pth"
     TYPE: Hed
     IS_DEFAULT: False
   -
     NAME: "OpenposeAnnotator"
-    BODY_MODEL_PATH: "ms://damo/scepter_scedit@annotator/ckpts/body_pose_model.pth"
+    BODY_MODEL_PATH: "ms://iic/scepter_scedit@annotator/ckpts/body_pose_model.pth"
-    HAND_MODEL_PATH: "ms://damo/scepter_scedit@annotator/ckpts/hand_pose_model.pth"
+    HAND_MODEL_PATH: "ms://iic/scepter_scedit@annotator/ckpts/hand_pose_model.pth"
     TYPE: Openpose
     IS_DEFAULT: False
   -
     NAME: "MidasDetector"
-    PRETRAINED_MODEL: "ms://damo/scepter_scedit@annotator/ckpts/dpt_hybrid-midas-501f0c75.pt"
+    PRETRAINED_MODEL: "ms://iic/scepter_scedit@annotator/ckpts/dpt_hybrid-midas-501f0c75.pt"
     TYPE: Midas
     IS_DEFAULT: False
   -

scepter/methods/studio/inference/largen/largen_pro.yaml

@@ -68,7 +68,7 @@ DEFAULT_PARAS:
           INPUT: ["ORIGINAL_SIZE_AS_TUPLE", "AESTHETIC_SCORE", "NEGATIVE_AESTHETIC_SCORE", "CROP_COORDS_TOP_LEFT", "PROMPT", "NEGATIVE_PROMPT"]
 
 MODEL:
-  PRETRAINED_MODEL: ms://damo/LARGEN@models/largen_ckpt_s22k.pth
+  PRETRAINED_MODEL: ms://iic/LARGEN@models/largen_ckpt_s22k.pth
   # SCHEDULE_ARGS DESCRIPTION:  TYPE:  default: ''
   SCHEDULE:
     PARAMETERIZATION: "eps"
@@ -245,8 +245,8 @@ MODEL:
         LEGACY_UCG_VALUE:
       -
         NAME: IPAdapterPlusEmbedder
-        CLIP_DIR: ms://damo/LARGEN@models/clip_encoder/
+        CLIP_DIR: ms://iic/LARGEN@models/clip_encoder/
-        PRETRAINED_MODEL: ms://damo/LARGEN@models/ip-adapter-plus_sdxl_vit-h.bin
+        PRETRAINED_MODEL: ms://iic/LARGEN@models/ip-adapter-plus_sdxl_vit-h.bin
         INPUT_KEYS: [ "ref_ip", "ref_detail" ]
         IN_DIM: 1280
         HEADS: 20

scepter/methods/studio/scepter_ui.yaml

@@ -48,11 +48,11 @@ BANNER: |
       </div>
       <div class="qr-codes">
           <div class="qr-code-container">
-              <img src="https://modelscope.cn/api/v1/models/damo/scepter/repo?Revision=master&FilePath=assets/scepter_studio/ms_scepter_studio_qr.png" alt="ms_scepter_studio_qr">
+              <img src="https://modelscope.cn/api/v1/models/iic/scepter/repo?Revision=master&FilePath=assets/scepter_studio/ms_scepter_studio_qr.png" alt="ms_scepter_studio_qr">
               <div class="caption"><a href="https://www.modelscope.cn/studios/iic/scepter_studio">Modelscope Studio</a></div>
           </div>
           <div class="qr-code-container">
-              <img src="https://modelscope.cn/api/v1/models/damo/scepter/repo?Revision=master&FilePath=assets/scepter_studio/scepter_github_qr.png" alt="scepter_github_qr">
+              <img src="https://modelscope.cn/api/v1/models/iic/scepter/repo?Revision=master&FilePath=assets/scepter_studio/scepter_github_qr.png" alt="scepter_github_qr">
               <div class="caption"><a href="https://github.com/modelscope/scepter">Github</a></div>
           </div>
       </div>

scepter/methods/studio/self_train/edit/stylebooth_tb_pro.yaml

@@ -0,0 +1,376 @@
+ENV:
+  BACKEND: nccl
+META:
+  VERSION: 'EDIT'
+  DESCRIPTION: "EDIT"
+  IS_DEFAULT: False
+  IS_SHARE: True
+  INFERENCE_PARAS:
+    INFERENCE_BATCH_SIZE: 1
+    INFERENCE_PREFIX: ""
+    DEFAULT_SAMPLER: "ddim"
+    DEFAULT_SAMPLE_STEPS: 40
+    INFERENCE_N_PROMPT: ""
+    RESOLUTION: [512, 512]
+  PARAS:
+    -
+      TASK: 'Image Editing'
+      TRAIN_BATCH_SIZE: 1
+      TRAIN_PREFIX: ""
+      TRAIN_N_PROMPT: ""
+      RESOLUTION: [512, 512]
+      MEMORY: 29000
+      EPOCHS: 50
+      SAVE_INTERVAL: 25
+      EPSEC: 0.818
+      LEARNING_RATE: 0.0001
+      IS_DEFAULT: False
+      TUNER: FULL
+    -
+      TASK: 'Image Editing'
+      TRAIN_BATCH_SIZE: 1
+      TRAIN_PREFIX: ""
+      TRAIN_N_PROMPT: ""
+      RESOLUTION: [512, 512]
+      MEMORY: 29000
+      EPOCHS: 50
+      SAVE_INTERVAL: 25
+      EPSEC: 0.818
+      LEARNING_RATE: 0.0001
+      IS_DEFAULT: True
+      TUNER: LORA
+    -
+      TASK: 'Image Editing'
+      TRAIN_BATCH_SIZE: 1
+      TRAIN_PREFIX: ""
+      TRAIN_N_PROMPT: ""
+      RESOLUTION: [512, 512]
+      MEMORY: 29000
+      EPOCHS: 50
+      SAVE_INTERVAL: 25
+      EPSEC: 0.818
+      LEARNING_RATE: 0.0001
+      IS_DEFAULT: False
+      TUNER: SCE
+
+    -
+      TASK: 'Image Editing'
+      TRAIN_BATCH_SIZE: 1
+      TRAIN_PREFIX: ""
+      TRAIN_N_PROMPT: ""
+      RESOLUTION: [512, 512]
+      MEMORY: 29000
+      EPOCHS: 50
+      SAVE_INTERVAL: 25
+      EPSEC: 0.818
+      LEARNING_RATE: 0.0001
+      IS_DEFAULT: False
+      TUNER: TEXT_SCE
+
+    -
+      TASK: 'Image Editing'
+      TRAIN_BATCH_SIZE: 4
+      TRAIN_PREFIX: ""
+      TRAIN_N_PROMPT: ""
+      RESOLUTION: [512, 512]
+      MEMORY: 29000
+      EPOCHS: 50
+      SAVE_INTERVAL: 25
+      EPSEC: 0.818
+      LEARNING_RATE: 0.0001
+      IS_DEFAULT: False
+      TUNER: TEXT_LORA
+
+  TUNERS:
+    LORA:
+      -
+        NAME: SwiftLoRA
+        R: 256
+        LORA_ALPHA: 256
+        LORA_DROPOUT: 0.0
+        BIAS: "none"
+        TARGET_MODULES: "model.*(to_q|to_k|to_v|to_out.0|net.0.proj|net.2)$"
+    TEXT_LORA:
+      -
+        NAME: SwiftLoRA
+        R: 256
+        LORA_ALPHA: 256
+        LORA_DROPOUT: 0.0
+        BIAS: "none"
+        TARGET_MODULES: "(cond_stage_model.*(q_proj|k_proj|v_proj|out_proj|mlp.fc1|mlp.fc2))|(model.*(to_q|to_k|to_v|to_out.0|net.0.proj|net.2))$"
+    SCE:
+      -
+        NAME: SwiftSCETuning
+        DIMS: [1280, 1280, 1280, 1280, 1280, 640, 640, 640, 320, 320, 320, 320]
+        DOWN_RATIO: 1.0
+        TARGET_MODULES: model.lsc_identity\.\d+$
+        TUNER_MODE: identity
+    TEXT_SCE:
+      -
+        NAME: SwiftSCETuning
+        DIMS: [ 1280, 1280, 1280, 1280, 1280, 640, 640, 640, 320, 320, 320, 320 ]
+        DOWN_RATIO: 1.0
+        TARGET_MODULES: model.lsc_identity\.\d+$
+        TUNER_MODE: identity
+      -
+        NAME: SwiftLoRA
+        R: 256
+        LORA_ALPHA: 256
+        LORA_DROPOUT: 0.0
+        BIAS: "none"
+        TARGET_MODULES: "cond_stage_model.*(q_proj|k_proj|v_proj|out_proj|mlp.fc1|mlp.fc2)$"
+
+  MODIFY_PARAS:
+    TEXT_LORA:
+      TRAIN:
+        SOLVER.MODEL.COND_STAGE_MODEL.USE_GRAD: True
+    TEXT_SCE:
+      TRAIN:
+        SOLVER.MODEL.COND_STAGE_MODEL.USE_GRAD: True
+
+SOLVER:
+  NAME: LatentDiffusionSolver
+  RESUME_FROM:
+  LOAD_MODEL_ONLY: True
+  USE_FSDP: False
+  SHARDING_STRATEGY:
+  USE_AMP: True
+  DTYPE: float16
+  CHANNELS_LAST: True
+  MAX_STEPS: 1000
+  MAX_EPOCHS: -1
+  NUM_FOLDS: 1
+  ACCU_STEP: 1
+  EVAL_INTERVAL: -1
+  #
+  WORK_DIR:
+  LOG_FILE: std_log.txt
+  #
+  FILE_SYSTEM:
+    NAME: "ModelscopeFs"
+    TEMP_DIR: "./cache/data"
+  #
+  FREEZE:
+  #
+  TUNER:
+  #
+  MODEL:
+    NAME: LatentDiffusionEdit
+    PARAMETERIZATION: eps
+    TIMESTEPS: 1000
+    MIN_SNR_GAMMA:
+    ZERO_TERMINAL_SNR: False
+    PRETRAINED_MODEL: ms://iic/stylebooth@models/stylebooth-tb-5000-0.bin
+    IGNORE_KEYS: [ ]
+    CONCAT_NO_SCALE_FACTOR: True
+    SCALE_FACTOR: 0.18215
+    SIZE_FACTOR: 8
+    # DEFAULT_N_PROMPT: 'lowres, error, worst quality, low quality, jpeg artifacts, ugly, duplicate, morbid, mutilated, out of frame, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, mutation, deformed, blurry, dehydrated, bad anatomy, bad proportions, extra limbs, cloned face, disfigured, gross proportions, malformed limbs, missing arms, missing legs, extra arms, extra legs, fused fingers, too many fingers, long neck, username, watermark, signature'
+    DEFAULT_N_PROMPT:
+    SCHEDULE_ARGS:
+      "NAME": "scaled_linear"
+      "BETA_MIN": 0.00085
+      "BETA_MAX": 0.012
+    USE_EMA: False
+    #
+    DIFFUSION_MODEL:
+      NAME: DiffusionUNet
+      IN_CHANNELS: 8
+      OUT_CHANNELS: 4
+      MODEL_CHANNELS: 320
+      NUM_HEADS: 8
+      NUM_RES_BLOCKS: 2
+      ATTENTION_RESOLUTIONS: [ 4, 2, 1 ]
+      CHANNEL_MULT: [ 1, 2, 4, 4 ]
+      CONV_RESAMPLE: True
+      DIMS: 2
+      USE_CHECKPOINT: False
+      USE_SCALE_SHIFT_NORM: False
+      RESBLOCK_UPDOWN: False
+      USE_SPATIAL_TRANSFORMER: True
+      TRANSFORMER_DEPTH: 1
+      CONTEXT_DIM: 768
+      DISABLE_MIDDLE_SELF_ATTN: False
+      USE_LINEAR_IN_TRANSFORMER: False
+      IGNORE_KEYS: []
+    #
+    FIRST_STAGE_MODEL:
+      NAME: AutoencoderKL
+      EMBED_DIM: 4
+      IGNORE_KEYS: []
+      BATCH_SIZE: 4
+      #
+      ENCODER:
+        NAME: Encoder
+        CH: 128
+        OUT_CH: 3
+        NUM_RES_BLOCKS: 2
+        IN_CHANNELS: 3
+        ATTN_RESOLUTIONS: [ ]
+        CH_MULT: [ 1, 2, 4, 4 ]
+        Z_CHANNELS: 4
+        DOUBLE_Z: True
+        DROPOUT: 0.0
+        RESAMP_WITH_CONV: True
+      #
+      DECODER:
+        NAME: Decoder
+        CH: 128
+        OUT_CH: 3
+        NUM_RES_BLOCKS: 2
+        IN_CHANNELS: 3
+        ATTN_RESOLUTIONS: [ ]
+        CH_MULT: [ 1, 2, 4, 4 ]
+        Z_CHANNELS: 4
+        DROPOUT: 0.0
+        RESAMP_WITH_CONV: True
+        GIVE_PRE_END: False
+        TANH_OUT: False
+    #
+    TOKENIZER:
+      NAME: ClipTokenizer
+      PRETRAINED_PATH: ms://AI-ModelScope/clip-vit-large-patch14
+      LENGTH: 77
+      CLEAN: True
+    #
+    COND_STAGE_MODEL:
+      NAME: FrozenCLIPEmbedder
+      FREEZE: True
+      USE_GRAD: False
+      LAYER: last
+      PRETRAINED_MODEL: ms://AI-ModelScope/clip-vit-large-patch14
+    #
+    LOSS:
+      NAME: ReconstructLoss
+      LOSS_TYPE: l2
+  #
+  SAMPLE_ARGS:
+    SAMPLER: ddim
+    SAMPLE_STEPS: 50
+    SEED: 2023
+    GUIDE_SCALE: #7.5
+      image: 1.5
+      text: 7.5
+    GUIDE_RESCALE: 0.5
+    DISCRETIZATION: trailing
+    IMAGE_SIZE: [512, 512]
+    RUN_TRAIN_N: False
+  #
+  OPTIMIZER:
+    NAME: AdamW
+    LEARNING_RATE: 0.064
+    BETAS: [ 0.9, 0.999 ]
+    EPS: 1e-8
+    WEIGHT_DECAY: 1e-2
+    AMSGRAD: False
+  #
+  TRAIN_DATA:
+    NAME: ImageTextPairMSDataset
+    MODE: train
+    MS_DATASET_NAME: cache/save_data/delogo/
+    MS_DATASET_NAMESPACE: ""
+    MS_DATASET_SPLIT: "train"
+    MS_DATASET_SUBNAME: ""
+    PROMPT_PREFIX: ""
+    REPLACE_STYLE: False
+    PIN_MEMORY: True
+    BATCH_SIZE: 1
+    NUM_WORKERS: 4
+    SAMPLER:
+      NAME: LoopSampler
+    TRANSFORMS:
+      - NAME: LoadImageFromFileList
+        FILE_KEYS: ['img_path', 'src_path']
+        RGB_ORDER: RGB
+        BACKEND: pillow
+      - NAME: FlexibleResize
+        INTERPOLATION: bilinear
+        SIZE: [ 512, 512 ]
+        INPUT_KEY: [ 'img', 'src' ]
+        OUTPUT_KEY: [ 'img', 'src' ]
+        BACKEND: pillow
+      - NAME: FlexibleCenterCrop
+        SIZE: [ 512, 512 ]
+        INPUT_KEY: [ 'img', 'src' ]
+        OUTPUT_KEY: [ 'img', 'src' ]
+        BACKEND: pillow
+      - NAME: ImageToTensor
+        INPUT_KEY: [ 'img', 'src' ]
+        OUTPUT_KEY: [ 'img', 'src' ]
+        BACKEND: pillow
+      - NAME: Normalize
+        MEAN: [ 0.5,  0.5,  0.5 ]
+        STD: [ 0.5,  0.5,  0.5 ]
+        INPUT_KEY: [ 'img', 'src' ]
+        OUTPUT_KEY: [ 'image', 'condition_cat' ]
+        BACKEND: torchvision
+      - NAME: Select
+        KEYS: [ 'image', 'condition_cat', 'prompt' ]
+        META_KEYS: [ 'data_key' ]
+  #
+  TRAIN_HOOKS:
+    -
+      NAME: BackwardHook
+      PRIORITY: 0
+    -
+      NAME: LogHook
+      LOG_INTERVAL: 10
+      SHOW_GPU_MEM: True
+    -
+      NAME: TensorboardLogHook
+    -
+      NAME: CheckpointHook
+      INTERVAL: 10000
+      PRIORITY: 200
+      SAVE_LAST: True
+      SAVE_NAME_PREFIX: 'step'
+      DISABLE_SNAPSHOT: True
+  #
+  EVAL_DATA:
+    NAME: Text2ImageDataset
+    MODE: eval
+    PROMPT_FILE:
+    PROMPT_DATA: [  ]
+    IMAGE_SIZE: [ 512, 512 ]
+    FIELDS: [ "prompt", "src_path" ]
+    DELIMITER: '#;#'
+    PROMPT_PREFIX: ''
+    PIN_MEMORY: True
+    BATCH_SIZE: 1
+    NUM_WORKERS: 4
+    TRANSFORMS:
+      - NAME: LoadImageFromFileList
+        FILE_KEYS: [ 'src_path' ]
+        RGB_ORDER: RGB
+        BACKEND: pillow
+      - NAME: FlexibleResize
+        INTERPOLATION: bilinear
+        SIZE: [ 512, 512 ]
+        INPUT_KEY: [ 'src' ]
+        OUTPUT_KEY: [ 'src' ]
+        BACKEND: pillow
+      - NAME: FlexibleCenterCrop
+        SIZE: [ 512, 512 ]
+        INPUT_KEY: [ 'src' ]
+        OUTPUT_KEY: [ 'src' ]
+        BACKEND: pillow
+      - NAME: ImageToTensor
+        INPUT_KEY: [ 'src' ]
+        OUTPUT_KEY: [ 'src' ]
+        BACKEND: pillow
+      - NAME: Normalize
+        MEAN: [ 0.5,  0.5,  0.5 ]
+        STD: [ 0.5,  0.5,  0.5 ]
+        INPUT_KEY: [ 'src' ]
+        OUTPUT_KEY: [ 'condition_cat' ]
+        BACKEND: torchvision
+      - NAME: Select
+        KEYS: [ 'condition_cat', 'prompt' ]
+        META_KEYS: [ 'image_size' ]
+  EVAL_HOOKS:
+    -
+      NAME: ProbeDataHook
+      PROB_INTERVAL: 100
+      SAVE_LAST: True
+      SAVE_NAME_PREFIX: 'step'
+      SAVE_PROBE_PREFIX: 'image'

scepter/methods/studio/tuner_manager/readme_en.md

@@ -1,7 +1,7 @@
 ---
 frameworks:
 - Pytorch
-license: Apache License 2.0
+license: apache-2.0
 tasks:
 - efficient-diffusion-tuning
 ---

scepter/methods/studio/tuner_manager/readme_zh.md

@@ -1,7 +1,7 @@
 ---
 frameworks:
 - Pytorch
-license: Apache License 2.0
+license: apache-2.0
 tasks:
 - efficient-diffusion-tuning
 ---

scepter/methods/studio/tuner_manager/tuner_manager.yaml

@@ -1,14 +1,15 @@
 WORK_DIR: "tuner_manager"
-SELF_TRAIN_DIR: "self_train"
 EXPORT_DIR: "export_model"
 TUNER_LIST_YAML: "tuner_list.yaml"
 README_EN: "scepter/methods/studio/tuner_manager/readme_en.md"
 README_ZH: "scepter/methods/studio/tuner_manager/readme_zh.md"
 
 BASE_MODEL_VERSION:
-  - BASE_MODEL: 'SD_XL1.0'
-    TUNER_TYPE: [ 'TEXT_SCE', 'SCE', 'LORA', 'TEXT_LORA', 'FULL' ]
   - BASE_MODEL: 'SD1.5'
-    TUNER_TYPE: [ 'TEXT_SCE', 'SCE', 'LORA', 'TEXT_LORA', 'FULL' ]
+    TUNER_TYPE: [ 'TEXT_LORA', 'TEXT_SCE', 'LORA', 'SCE', 'FULL' ]
+  - BASE_MODEL: 'SD_XL1.0'
+    TUNER_TYPE: [ 'TEXT_LORA', 'TEXT_SCE', 'LORA', 'SCE', 'FULL' ]
   - BASE_MODEL: 'SD2.1'
-    TUNER_TYPE: [ 'SCE', 'LORA', 'FULL' ]
+    TUNER_TYPE: [ 'LORA', 'SCE', 'FULL' ]
+  - BASE_MODEL: 'EDIT'
+    TUNER_TYPE: [ 'LORA', 'SCE', 'FULL' ]

scepter/modules/annotator/base_annotator.py

@@ -4,7 +4,6 @@ from abc import ABCMeta
 
 import torch
 import torch.nn as nn
-
 from scepter.modules.annotator.registry import ANNOTATORS
 from scepter.modules.model.base_model import BaseModel
 from scepter.modules.utils.config import dict_to_yaml

scepter/modules/annotator/canny.py

@@ -6,7 +6,6 @@ import cv2
 import numpy as np
 import torch
 from PIL import Image
-
 from scepter.modules.annotator.base_annotator import BaseAnnotator
 from scepter.modules.annotator.registry import ANNOTATORS
 from scepter.modules.utils.config import dict_to_yaml

scepter/modules/annotator/color.py

@@ -6,7 +6,6 @@ import cv2
 import numpy as np
 import torch
 from PIL import Image
-
 from scepter.modules.annotator.base_annotator import BaseAnnotator
 from scepter.modules.annotator.registry import ANNOTATORS
 from scepter.modules.utils.config import dict_to_yaml

scepter/modules/annotator/hed.py

@@ -11,13 +11,14 @@ from abc import ABCMeta
 import cv2
 import numpy as np
 import torch
+import torchvision
 from einops import rearrange
-
 from scepter.modules.annotator.base_annotator import BaseAnnotator
 from scepter.modules.annotator.registry import ANNOTATORS
 from scepter.modules.utils.config import dict_to_yaml
 from scepter.modules.utils.distribute import we
 from scepter.modules.utils.file_system import FS
+from torchvision.transforms import InterpolationMode
 
 
 def nms(x, t, s):
@@ -123,6 +124,8 @@ class HedAnnotator(BaseAnnotator, metaclass=ABCMeta):
             if len(image.shape) == 3:
                 image = rearrange(image, 'h w c -> 1 c h w')
                 B, C, H, W = image.shape
+            elif len(image.shape) == 4:
+                B, C, H, W = image.shape
             else:
                 raise "Unsurpport input image's shape"
         elif isinstance(image, np.ndarray):
@@ -130,22 +133,22 @@ class HedAnnotator(BaseAnnotator, metaclass=ABCMeta):
             if len(image.shape) == 3:
                 image = rearrange(image, 'h w c -> 1 c h w')
                 B, C, H, W = image.shape
+            elif len(image.shape) == 4:
+                B, C, H, W = image.shape
             else:
                 raise "Unsurpport input image's shape"
         else:
             raise "Unsurpport input image's type"
+        transform = torchvision.transforms.Resize(
+            (H, W), interpolation=InterpolationMode.BILINEAR, antialias=True)
         edges = self.netNetwork(image.to(we.device_id))
-        edges = [
+        edges = [transform(e) for e in edges]
-            e.detach().cpu().numpy().astype(np.float32)[0, 0] for e in edges
+        edges = torch.cat(edges, dim=1)
-        ]
+        edges = 1 / (1 +
-        edges = [
+                     torch.exp(-torch.mean(edges, dim=1).type(torch.float)))
-            cv2.resize(e, (W, H), interpolation=cv2.INTER_LINEAR)
+        edges = edges.cpu().numpy()
-            for e in edges
+        edges = 255 - (edges * 255.0).clip(0, 255).astype(np.uint8)
-        ]
+        return edges[..., None].repeat(3, -1)
-        edges = np.stack(edges, axis=2)
-        edge = 1 / (1 + np.exp(-np.mean(edges, axis=2).astype(np.float64)))
-        edge = 255 - (edge * 255.0).clip(0, 255).astype(np.uint8)
-        return edge[..., None].repeat(3, 2)
 
     @staticmethod
     def get_config_template():

scepter/modules/annotator/midas/__init__.py

@@ -0,0 +1,2 @@
+# -*- coding: utf-8 -*-
+# Copyright (c) Alibaba, Inc. and its affiliates.

scepter/modules/annotator/midas/api.py

@@ -1,4 +1,5 @@
 # -*- coding: utf-8 -*-
+# Copyright (c) Alibaba, Inc. and its affiliates.
 # based on https://github.com/isl-org/MiDaS
 
 import cv2

scepter/modules/annotator/midas/base_model.py

@@ -1,4 +1,5 @@
 # -*- coding: utf-8 -*-
+# Copyright (c) Alibaba, Inc. and its affiliates.
 import torch
 
 

scepter/modules/annotator/midas/blocks.py

@@ -1,4 +1,5 @@
 # -*- coding: utf-8 -*-
+# Copyright (c) Alibaba, Inc. and its affiliates.
 import torch
 import torch.nn as nn
 

scepter/modules/annotator/midas/dpt_depth.py

@@ -1,4 +1,5 @@
 # -*- coding: utf-8 -*-
+# Copyright (c) Alibaba, Inc. and its affiliates.
 import torch
 import torch.nn as nn
 

scepter/modules/annotator/midas/midas_net.py

@@ -1,4 +1,5 @@
 # -*- coding: utf-8 -*-
+# Copyright (c) Alibaba, Inc. and its affiliates.
 """MidashNet: Network for monocular depth estimation trained by mixing several datasets.
 This file contains code that is adapted from
 https://github.com/thomasjpfan/pytorch_refinenet/blob/master/pytorch_refinenet/refinenet/refinenet_4cascade.py

scepter/modules/annotator/midas/midas_net_custom.py

@@ -1,4 +1,5 @@
 # -*- coding: utf-8 -*-
+# Copyright (c) Alibaba, Inc. and its affiliates.
 """MidashNet: Network for monocular depth estimation trained by mixing several datasets.
 This file contains code that is adapted from
 https://github.com/thomasjpfan/pytorch_refinenet/blob/master/pytorch_refinenet/refinenet/refinenet_4cascade.py

scepter/modules/annotator/midas/transforms.py

@@ -1,4 +1,5 @@
 # -*- coding: utf-8 -*-
+# Copyright (c) Alibaba, Inc. and its affiliates.
 import math
 
 import cv2

scepter/modules/annotator/midas/utils.py

@@ -1,4 +1,5 @@
 # -*- coding: utf-8 -*-
+# Copyright (c) Alibaba, Inc. and its affiliates.
 """Utils for monoDepth."""
 import re
 import sys

scepter/modules/annotator/midas/vit.py

@@ -1,4 +1,5 @@
 # -*- coding: utf-8 -*-
+# Copyright (c) Alibaba, Inc. and its affiliates.
 import math
 import types
 

scepter/modules/annotator/midas_op.py

@@ -9,7 +9,6 @@ import numpy as np
 import torch
 from einops import rearrange
 from PIL import Image
-
 from scepter.modules.annotator.base_annotator import BaseAnnotator
 from scepter.modules.annotator.midas.api import MiDaSInference
 from scepter.modules.annotator.registry import ANNOTATORS

scepter/modules/annotator/mlsd/__init__.py

@@ -0,0 +1,2 @@
+# -*- coding: utf-8 -*-
+# Copyright (c) Alibaba, Inc. and its affiliates.

scepter/modules/annotator/mlsd/mbv2_mlsd_large.py

@@ -1,4 +1,5 @@
 # -*- coding: utf-8 -*-
+# Copyright (c) Alibaba, Inc. and its affiliates.
 
 import torch
 import torch.nn as nn

scepter/modules/annotator/mlsd/mbv2_mlsd_tiny.py

@@ -1,4 +1,5 @@
 # -*- coding: utf-8 -*-
+# Copyright (c) Alibaba, Inc. and its affiliates.
 import torch
 import torch.nn as nn
 import torch.utils.model_zoo as model_zoo

scepter/modules/annotator/mlsd/utils.py

@@ -1,4 +1,5 @@
 # -*- coding: utf-8 -*-
+# Copyright (c) Alibaba, Inc. and its affiliates.
 
 # modified by  lihaoweicv
 # pytorch version

scepter/modules/annotator/mlsd_op.py

@@ -11,7 +11,6 @@ import cv2
 import numpy as np
 import torch
 from PIL import Image
-
 from scepter.modules.annotator.base_annotator import BaseAnnotator
 from scepter.modules.annotator.mlsd.mbv2_mlsd_large import MobileV2_MLSD_Large
 from scepter.modules.annotator.mlsd.utils import pred_lines

scepter/modules/annotator/openpose.py

@@ -15,13 +15,12 @@ import numpy as np
 import torch
 import torch.nn as nn
 from PIL import Image
-from scipy.ndimage.filters import gaussian_filter
-from skimage.measure import label
-
 from scepter.modules.annotator.base_annotator import BaseAnnotator
 from scepter.modules.annotator.registry import ANNOTATORS
 from scepter.modules.utils.config import dict_to_yaml
 from scepter.modules.utils.file_system import FS
+from scipy.ndimage.filters import gaussian_filter
+from skimage.measure import label
 
 os.environ['KMP_DUPLICATE_LIB_OK'] = 'TRUE'
 

scepter/modules/annotator/utils.py

@@ -37,30 +37,30 @@ class AnnotatorProcessor():
     hed_cfg = {
         'NAME': 'HedAnnotator',
         'PRETRAINED_MODEL':
-        'ms://damo/scepter_scedit@annotator/ckpts/ControlNetHED.pth',
+        'ms://iic/scepter_scedit@annotator/ckpts/ControlNetHED.pth',
         'INPUT_KEYS': ['img'],
         'OUTPUT_KEYS': ['hed']
     }
     openpose_cfg = {
         'NAME': 'OpenposeAnnotator',
         'BODY_MODEL_PATH':
-        'ms://damo/scepter_scedit@annotator/ckpts/body_pose_model.pth',
+        'ms://iic/scepter_scedit@annotator/ckpts/body_pose_model.pth',
         'HAND_MODEL_PATH':
-        'ms://damo/scepter_scedit@annotator/ckpts/hand_pose_model.pth',
+        'ms://iic/scepter_scedit@annotator/ckpts/hand_pose_model.pth',
         'INPUT_KEYS': ['img'],
         'OUTPUT_KEYS': ['openpose']
     }
     midas_cfg = {
         'NAME': 'MidasDetector',
         'PRETRAINED_MODEL':
-        'ms://damo/scepter_scedit@annotator/ckpts/dpt_hybrid-midas-501f0c75.pt',
+        'ms://iic/scepter_scedit@annotator/ckpts/dpt_hybrid-midas-501f0c75.pt',
         'INPUT_KEYS': ['img'],
         'OUTPUT_KEYS': ['depth']
     }
     mlsd_cfg = {
         'NAME': 'MLSDdetector',
         'PRETRAINED_MODEL':
-        'ms://damo/scepter_scedit@annotator/ckpts/mlsd_large_512_fp32.pth',
+        'ms://iic/scepter_scedit@annotator/ckpts/mlsd_large_512_fp32.pth',
         'INPUT_KEYS': ['img'],
         'OUTPUT_KEYS': ['mlsd']
     }

scepter/modules/data/dataset/base_dataset.py

@@ -3,14 +3,13 @@
 
 from abc import ABCMeta, abstractmethod
 
-from torch.utils.data import Dataset
-
 from scepter.modules.transform.registry import TRANSFORMS, build_pipeline
 from scepter.modules.utils.config import dict_to_yaml
 from scepter.modules.utils.distribute import we
 from scepter.modules.utils.file_system import FS
 from scepter.modules.utils.logger import get_logger
 from scepter.modules.utils.registry import old_python_version
+from torch.utils.data import Dataset
 
 
 class BaseDataset(Dataset, metaclass=ABCMeta):

scepter/modules/data/dataset/dataset.py

@@ -8,7 +8,6 @@ from collections.abc import Iterable
 
 import numpy as np
 import torchvision
-
 from scepter.modules.data.dataset.base_dataset import BaseDataset
 from scepter.modules.data.dataset.registry import DATASETS
 from scepter.modules.utils.config import dict_to_yaml
@@ -272,6 +271,11 @@ class Text2ImageDataset(BaseDataset):
                     item['prompt'] = prompt_prefix + value
                 elif key in ['oss_key', 'path', 'img_path', 'target_img_path']:
                     item['meta']['img_path'] = os.path.join(path_prefix, value)
+                elif key in [
+                        'src_oss_key', 'src_path', 'src_img_path',
+                        'src_target_img_path'
+                ]:
+                    item['meta']['src_path'] = os.path.join(path_prefix, value)
                 elif key in ['width', 'height']:
                     item['meta'][key] = int(value)
                 else:

scepter/modules/data/dataset/ms_dataset.py

@@ -101,14 +101,13 @@ class ImageTextPairMSDataset(BaseDataset):
             if isinstance(self.output_size, numbers.Number):
                 self.output_size = [self.output_size, self.output_size]
         # Use modelscope dataset
-
         if not ms_dataset_name:
             raise (
                 'Your must set MS_DATASET_NAME as modelscope dataset or your local dataset orignized '
                 'as modelscope dataset.')
         if FS.exists(ms_dataset_name):
             ms_dataset_name = FS.get_dir_to_local_dir(ms_dataset_name)
-            ms_remap_path = ms_dataset_name
+            # ms_remap_path = ms_dataset_name
         try:
             self.data = MsDataset.load(str(ms_dataset_name),
                                        namespace=ms_dataset_namespace,
@@ -133,9 +132,11 @@ class ImageTextPairMSDataset(BaseDataset):
         if ms_remap_path:
 
             def map_func(example):
-                example['Target:FILE'] = os.path.join(ms_remap_path,
+                return {
-                                                      example['Target:FILE'])
+                    k: os.path.join(ms_remap_path, v)
-                return example
+                    if k.endswith(':FILE') else v
+                    for k, v in example.items()
+                }
 
             self.data = self.data.ds_instance.map(map_func)
         self.real_number = len(self.data)
@@ -149,9 +150,12 @@ class ImageTextPairMSDataset(BaseDataset):
     def _get(self, index: int):
         current_data = self.data[index % len(self.data)]
         # print(current_data.keys())
-        image_path = current_data['Target:FILE']
+        image_path = current_data[
-        prompt = current_data['Prompt']
+            'Target:FILE'] if 'Target:FILE' in current_data else ''
+        prompt = current_data.get('Prompt', current_data.get('prompt', ''))
         style = current_data['Style'] if 'Style' in current_data else ''
+        src_image_path = current_data[
+            'Source:FILE'] if 'Source:FILE' in current_data else ''
         # print(prompt, style)
         if self.replace_style and not style == '':
             prompt = prompt.replace(style, f'<{self.keywords_sign}>')
@@ -166,6 +170,7 @@ class ImageTextPairMSDataset(BaseDataset):
         ret_item = {
             'meta': {
                 'img_path': image_path,
+                'src_path': src_image_path,
                 'data_key': style,
                 'data_num': self.real_number
             },
@@ -173,6 +178,9 @@ class ImageTextPairMSDataset(BaseDataset):
         }
         if self.output_size is not None:
             ret_item['meta']['image_size'] = self.output_size
+        for key in current_data:
+            if key not in ret_item['meta']:
+                ret_item['meta'][key] = current_data[key]
         return ret_item
 
     @staticmethod
@@ -241,19 +249,18 @@ class ImageTextPairFolderDataset(BaseDataset):
         data_folder = FS.get_dir_to_local_dir(data_folder)
         all_lines = open(os.path.join(data_folder, 'train.csv'),
                          'r').read().split('\n')
-        assert all_lines[0] == 'Target:FILE,Prompt'
+        header = all_lines[0].split(',')
         self.data = []
         for line in all_lines[1:]:
             line = line.strip()
             if line == '':
                 continue
-            self.data.append({
+            record = dict(zip(header, line.split(',', len(header) - 1)))
-                'Target:FILE':
+            record = {
-                os.path.join(data_folder,
+                k: os.path.join(data_folder, v) if k.endswith(':FILE') else v
-                             line.split(',', 1)[0]),
+                for k, v in record.items()
-                'Prompt':
+            }
-                line.split(',', 1)[1]
+            self.data.append(record)
-            })
         self.real_number = len(self.data)
 
     def __len__(self):
@@ -265,9 +272,12 @@ class ImageTextPairFolderDataset(BaseDataset):
     def _get(self, index: int):
         current_data = self.data[index % len(self.data)]
         # print(current_data.keys())
-        image_path = current_data['Target:FILE']
+        image_path = current_data[
-        prompt = current_data['Prompt']
+            'Target:FILE'] if 'Target:FILE' in current_data else ''
+        prompt = current_data.get('Prompt', current_data.get('prompt', ''))
         style = current_data['Style'] if 'Style' in current_data else ''
+        src_image_path = current_data[
+            'Source:FILE'] if 'Source:FILE' in current_data else ''
         # print(prompt, style)
         if self.replace_style and not style == '':
             prompt = prompt.replace(style, f'<{self.keywords_sign}>')
@@ -282,6 +292,7 @@ class ImageTextPairFolderDataset(BaseDataset):
         ret_item = {
             'meta': {
                 'img_path': image_path,
+                'src_path': src_image_path,
                 'data_key': style,
                 'data_num': self.real_number
             },
@@ -289,6 +300,9 @@ class ImageTextPairFolderDataset(BaseDataset):
         }
         if self.output_size is not None:
             ret_item['meta']['image_size'] = self.output_size
+        for key in current_data:
+            if key not in ret_item['meta']:
+                ret_item['meta'][key] = current_data[key]
         return ret_item
 
     @staticmethod

scepter/modules/data/dataset/registry.py

@@ -7,13 +7,12 @@ import re
 from functools import partial
 
 import torch
-from torch.utils.data import DataLoader, DistributedSampler
-
 from scepter.modules.data.sampler import (SAMPLERS, MixtureOfSamplers,
                                           MultiFoldDistributedSampler,
                                           MultiLevelBatchSampler)
 from scepter.modules.utils.registry import (Registry, deep_copy,
                                             old_python_version)
+from torch.utils.data import DataLoader, DistributedSampler
 
 string_classes = (str, bytes)
 int_classes = (int, )

scepter/modules/data/sampler/base_sampler.py

@@ -1,10 +1,9 @@
 # -*- coding: utf-8 -*-
 # Copyright (c) Alibaba, Inc. and its affiliates.
 
-from torch.utils.data.sampler import Sampler
-
 from scepter.modules.utils.config import dict_to_yaml
 from scepter.modules.utils.distribute import we
+from torch.utils.data.sampler import Sampler
 
 
 class BaseSampler(Sampler):

scepter/modules/data/sampler/sampler.py

@@ -12,7 +12,6 @@ from typing import List, Optional
 import numpy as np
 import torch
 import torch.distributed as dist
-
 from scepter.modules.data.sampler.base_sampler import BaseSampler
 from scepter.modules.data.sampler.registry import SAMPLERS
 from scepter.modules.data.utils.data_bucket import (BucketBatchIndex,

scepter/modules/inference/control_inference.py

@@ -8,7 +8,6 @@ import torch
 import torch.nn as nn
 import torchvision.transforms as TT
 from PIL.Image import Image
-
 from scepter.modules.model.registry import TUNERS
 from scepter.modules.utils.config import Config
 from scepter.modules.utils.distribute import we

scepter/modules/inference/diffusion_inference.py

@@ -8,7 +8,6 @@ from collections import OrderedDict
 import torch
 import torch.nn.functional as F
 from PIL.Image import Image
-
 from scepter.modules.model.network.diffusion.diffusion import GaussianDiffusion
 from scepter.modules.model.network.diffusion.schedules import noise_schedule
 from scepter.modules.model.registry import (BACKBONES, EMBEDDERS, MODELS,
@@ -277,7 +276,7 @@ class DiffusionInference():
                     module = self.load(module)
                     self.loaded_model[name] = module
                     return module
-                elif module['device'] == 'cpu' or module['device'] == "offline":
+                elif module['device'] == 'cpu' or module['device'] == 'offline':
                     module = self.load(module)
                     return module
                 else:
@@ -629,7 +628,7 @@ class DiffusionInference():
                         }, {
                             'cond': refine_null_context
                         }],
-                        steps=value_input.get('sample_steps', 50),
+                        steps=value_input.get('refine_sample_steps', 50),
                         guide_scale=value_input.get('refine_guide_scale', 7.5),
                         guide_rescale=value_input.get('refine_guide_rescale',
                                                       0.5),

scepter/modules/inference/stylebooth_inference.py

@@ -120,7 +120,7 @@ class StyleboothInference(DiffusionInference):
                  cat_uc=True,
                  tuner_model=None,
                  control_model=None,
-                 stylebooth_state=False,
+                 stylebooth_state=True,
                  style_edit_image=None,
                  style_exemplar_image=None,
                  style_guide_scale_text=None,

scepter/modules/inference/tuner_inference.py

@@ -7,7 +7,6 @@ import os
 import warnings
 
 import torch
-
 from scepter.modules.utils.file_system import FS
 
 try:

scepter/modules/model/backbone/autoencoder/ae_module.py

@@ -4,14 +4,13 @@
 import torch
 import torch.nn as nn
 from einops import repeat
-from torch.utils.checkpoint import checkpoint
-
 from scepter.modules.model.backbone.autoencoder.ae_utils import (
     XFORMERS_IS_AVAILBLE, AttnBlock, Downsample, MemoryEfficientAttention,
     Normalize, ResnetBlock, Upsample, nonlinearity)
 from scepter.modules.model.base_model import BaseModel
 from scepter.modules.model.registry import BACKBONES
 from scepter.modules.utils.config import dict_to_yaml
+from torch.utils.checkpoint import checkpoint
 
 
 @BACKBONES.register_class()

scepter/modules/model/backbone/image/mlp.py

@@ -1,7 +1,6 @@
 # -*- coding: utf-8 -*-
 # Copyright (c) Alibaba, Inc. and its affiliates.
 import torch.nn as nn
-
 from scepter.modules.model.base_model import BaseModel
 from scepter.modules.model.registry import BACKBONES
 from scepter.modules.utils.config import dict_to_yaml

scepter/modules/model/backbone/image/utils/clip.py

@@ -3,14 +3,14 @@
 from collections import OrderedDict
 from typing import List, Tuple, Union
 
+from pkg_resources import packaging
+
 import numpy as np
 import torch
 import torch.nn.functional as F
-from pkg_resources import packaging
-from torch import nn
-
 from scepter.modules.model.backbone.image.utils.simple_tokenizer import \
     SimpleTokenizer as _Tokenizer
+from torch import nn
 
 _tokenizer = _Tokenizer()
 

scepter/modules/model/backbone/image/vit_modify.py

@@ -2,7 +2,6 @@
 # Copyright (c) Alibaba, Inc. and its affiliates.
 import torch
 import torch.nn as nn
-
 from scepter.modules.model.backbone.image.utils.vit import (
     MULTI_HEAD_VIT_MODEL, VIT, VIT_MODEL, MULTI_HEAD_VIT_MODEL_Split)
 from scepter.modules.model.base_model import BaseModel

scepter/modules/model/backbone/unet/unet_module.py

@@ -6,7 +6,6 @@ from collections import OrderedDict
 
 import torch
 import torch.nn as nn
-
 from scepter.modules.model.backbone.unet.unet_utils import (
     BasicTransformerBlock, Downsample, ResBlock, SpatialTransformer,
     SpatialTransformerV2, Timestep, TimestepEmbedSequential,

scepter/modules/model/backbone/unet/unet_utils.py

@@ -6,16 +6,16 @@ import warnings
 from abc import abstractmethod
 from importlib import find_loader
 
+from packaging import version
+
 import numpy as np
 import torch
 import torch.nn as nn
 import torch.nn.functional as F
 import torchvision.transforms.functional as TF
 from einops import rearrange, repeat
-from packaging import version
-from torch.utils.checkpoint import checkpoint
-
 from scepter.modules.model.utils.basic_utils import default, exists
+from torch.utils.checkpoint import checkpoint
 
 try:
     import xformers

scepter/modules/model/backbone/video/bricks/csn_branch.py

@@ -2,7 +2,6 @@
 # Copyright (c) Alibaba, Inc. and its affiliates.
 
 import torch.nn as nn
-
 from scepter.modules.model.backbone.video.bricks.base_branch import BaseBranch
 from scepter.modules.model.registry import BRICKS
 from scepter.modules.utils.config import Config, dict_to_yaml

scepter/modules/model/backbone/video/bricks/non_local.py

@@ -5,7 +5,6 @@
 import torch
 import torch.nn as nn
 import torch.nn.functional as F
-
 from scepter.modules.model.backbone.video.bricks.visualize_3d_module import \
     Visualize3DModule
 from scepter.modules.model.registry import BRICKS

scepter/modules/model/backbone/video/bricks/r2d3d_branch.py

@@ -2,7 +2,6 @@
 # Copyright (c) Alibaba, Inc. and its affiliates.
 
 import torch.nn as nn
-
 from scepter.modules.model.backbone.video.bricks.base_branch import BaseBranch
 from scepter.modules.model.registry import BRICKS
 from scepter.modules.utils.config import Config, dict_to_yaml

scepter/modules/model/backbone/video/bricks/r2plus1d_branch.py

@@ -4,7 +4,6 @@
 import math
 
 import torch.nn as nn
-
 from scepter.modules.model.backbone.video.bricks.base_branch import BaseBranch
 from scepter.modules.model.registry import BRICKS
 from scepter.modules.utils.config import Config, dict_to_yaml

scepter/modules/model/backbone/video/bricks/stems/base_2d_stem.py

@@ -2,7 +2,6 @@
 # Copyright (c) Alibaba, Inc. and its affiliates.
 
 import torch.nn as nn
-
 from scepter.modules.model.backbone.video.bricks.visualize_3d_module import \
     Visualize3DModule
 from scepter.modules.model.registry import STEMS

scepter/modules/model/backbone/video/bricks/stems/base_3d_stem.py

@@ -2,7 +2,6 @@
 # Copyright (c) Alibaba, Inc. and its affiliates.
 
 import torch.nn as nn
-
 from scepter.modules.model.backbone.video.bricks.visualize_3d_module import \
     Visualize3DModule
 from scepter.modules.model.registry import STEMS

scepter/modules/model/backbone/video/bricks/stems/down_sample_stem.py

@@ -2,7 +2,6 @@
 # Copyright (c) Alibaba, Inc. and its affiliates.
 
 import torch.nn as nn
-
 from scepter.modules.model.backbone.video.bricks.stems.base_3d_stem import \
     Base3DStem
 from scepter.modules.model.registry import STEMS

scepter/modules/model/backbone/video/bricks/stems/embedding_stem.py

@@ -2,7 +2,6 @@
 # Copyright (c) Alibaba, Inc. and its affiliates.
 
 import torch.nn as nn
-
 from scepter.modules.model.backbone.video.bricks.visualize_3d_module import \
     Visualize3DModule
 from scepter.modules.model.registry import STEMS

scepter/modules/model/backbone/video/bricks/stems/r2plus1d_stem.py

@@ -4,7 +4,6 @@
 import math
 
 import torch.nn as nn
-
 from scepter.modules.model.backbone.video.bricks.stems.base_3d_stem import \
     Base3DStem
 from scepter.modules.model.registry import STEMS

scepter/modules/model/backbone/video/bricks/tada_conv.py

@@ -8,7 +8,6 @@ from itertools import repeat
 import torch
 import torch.nn as nn
 import torch.nn.functional as F
-
 from scepter.modules.model.backbone.video.bricks.base_branch import BaseBranch
 from scepter.modules.model.registry import BRICKS
 from scepter.modules.utils.config import Config, dict_to_yaml

scepter/modules/model/backbone/video/bricks/transformer_branch.py

@@ -22,7 +22,6 @@ import torch
 import torch.nn as nn
 import torch.nn.functional as F
 from einops import rearrange, repeat
-
 from scepter.modules.model.backbone.video.bricks.visualize_3d_module import \
     Visualize3DModule
 from scepter.modules.model.registry import BRICKS

scepter/modules/model/backbone/video/bricks/visualize_3d_module.py

@@ -4,7 +4,6 @@
 import os
 
 import torch.nn as nn
-
 from scepter.modules.utils.config import dict_to_yaml
 
 

scepter/modules/model/backbone/video/resnet_3d.py

@@ -2,7 +2,6 @@
 # Copyright (c) Alibaba, Inc. and its affiliates.
 
 import torch.nn as nn
-
 from scepter.modules.model.backbone.video.bricks.non_local import NonLocal
 from scepter.modules.model.backbone.video.init_helper import \
     _init_convnet_weights

scepter/modules/model/backbone/video/video_transformer.py

@@ -33,7 +33,6 @@ import torch
 import torch.nn as nn
 import torch.nn.functional
 from einops import rearrange
-
 from scepter.modules.model.backbone.video.init_helper import (
     _init_transformer_weights, trunc_normal_)
 from scepter.modules.model.registry import BACKBONES, BRICKS, STEMS

scepter/modules/model/base_model.py

@@ -3,7 +3,6 @@
 import copy
 
 import torch.nn as nn
-
 from scepter.modules.utils.config import dict_to_yaml
 from scepter.modules.utils.distribute import gather_data, we
 from scepter.modules.utils.probe import (ProbeData, merge_gathered_probe,

scepter/modules/model/embedder/embedder.py

@@ -11,8 +11,6 @@ import torch
 import torch.nn as nn
 import torch.utils.dlpack
 from einops import rearrange
-from torch.utils.checkpoint import checkpoint
-
 # to check
 from scepter.modules.model.backbone.unet.unet_utils import Timestep
 from scepter.modules.model.registry import EMBEDDERS
@@ -20,6 +18,7 @@ from scepter.modules.model.utils.basic_utils import expand_dims_like
 from scepter.modules.utils.config import dict_to_yaml
 from scepter.modules.utils.distribute import we
 from scepter.modules.utils.file_system import FS
+from torch.utils.checkpoint import checkpoint
 
 from .base_embedder import BaseEmbedder
 from .resampler import Resampler

scepter/modules/model/head/classifier_head.py

@@ -7,11 +7,10 @@ from collections import OrderedDict
 import torch
 import torch.nn as nn
 import torch.nn.functional as F
-from torch.nn.parameter import Parameter
-
 from scepter.modules.model.base_model import BaseModel
 from scepter.modules.model.registry import HEADS
 from scepter.modules.utils.config import dict_to_yaml
+from torch.nn.parameter import Parameter
 
 
 @HEADS.register_class()

scepter/modules/model/loss/base_losses.py

@@ -1,12 +1,12 @@
 # -*- coding: utf-8 -*-
 # Copyright (c) Alibaba, Inc. and its affiliates.
 
-import torch.nn as nn
 from packaging import version
-from torch.version import __version__ as torch_version
 
+import torch.nn as nn
 from scepter.modules.model.registry import LOSSES
 from scepter.modules.utils.config import dict_to_yaml
+from torch.version import __version__ as torch_version
 
 
 @LOSSES.register_class()

scepter/modules/model/loss/rec_loss.py

@@ -1,10 +1,9 @@
 # -*- coding: utf-8 -*-
 # Copyright (c) Alibaba, Inc. and its affiliates.
 import torch
-from torch import nn
-
 from scepter.modules.model.registry import LOSSES
 from scepter.modules.utils.config import dict_to_yaml
+from torch import nn
 
 
 @LOSSES.register_class()

scepter/modules/model/metric/classification.py

@@ -5,7 +5,6 @@ from collections import OrderedDict
 
 import numpy as np
 import torch
-
 from scepter.modules.model.metric.base_metric import BaseMetric
 from scepter.modules.model.metric.registry import METRICS
 from scepter.modules.utils.config import dict_to_yaml

scepter/modules/model/neck/global_average_pooling.py

@@ -2,7 +2,6 @@
 # Copyright (c) Alibaba, Inc. and its affiliates.
 
 import torch.nn as nn
-
 from scepter.modules.model.base_model import BaseModel
 from scepter.modules.model.registry import NECKS
 from scepter.modules.utils.config import dict_to_yaml

scepter/modules/model/network/autoencoder/ae_kl.py

@@ -5,7 +5,6 @@ from collections import OrderedDict
 
 import numpy as np
 import torch
-
 from scepter.modules.model.network.train_module import TrainModule
 from scepter.modules.model.registry import BACKBONES, LOSSES, MODELS
 from scepter.modules.utils.config import dict_to_yaml

scepter/modules/model/network/classifier.py

@@ -1,17 +1,16 @@
 # -*- coding: utf-8 -*-
-# Copyright 2021 Alibaba Group Holding Limited. All Rights Reserved.
+# Copyright (c) Alibaba, Inc. and its affiliates.
 
 from collections import OrderedDict
 from functools import partial
 
 import torch.nn as nn
-from torch.nn.functional import sigmoid, softmax
-
 from scepter.modules.model.metric.registry import METRICS
 from scepter.modules.model.network.train_module import TrainModule
 from scepter.modules.model.registry import (BACKBONES, HEADS, LOSSES, MODELS,
                                             NECKS)
 from scepter.modules.utils.config import Config, dict_to_yaml
+from torch.nn.functional import sigmoid, softmax
 
 _ACTIVATE_MAPPER = {'softmax': partial(softmax, dim=1), 'sigmoid': sigmoid}
 

scepter/modules/model/network/diffusion/diffusion.py

@@ -10,10 +10,12 @@ import random
 import torch
 
 from .schedules import karras_schedule
-from .solvers import (sample_ddim, sample_dpm_2, sample_dpm_2_ancestral,
+from .solvers import (
-                      sample_dpmpp_2m, sample_dpmpp_2m_sde,
+    sample_ddim, sample_dpm_2, sample_dpm_2_ancestral, sample_dpmpp_2m,
-                      sample_dpmpp_2s_ancestral, sample_dpmpp_sde,
+    sample_dpmpp_2m_sde, sample_dpmpp_2m_sde_lcm, sample_dpmpp_2s_ancestral,
-                      sample_euler, sample_euler_ancestral, sample_heun)
+    sample_dpmpp_sde, sample_euler, sample_euler_ancestral, sample_heun,
+    sample_onestep, stochastic_iterative_sampler,
+    stochastic_iterative_sampler2, stochastic_iterative_sampler3)
 
 __all__ = ['GaussianDiffusion']
 
@@ -427,7 +429,12 @@ class GaussianDiffusion(object):
             'dpmpp_2s_ancestral_karras': sample_dpmpp_2s_ancestral,
             'dpmpp_2m_karras': sample_dpmpp_2m,
             'dpmpp_sde_karras': sample_dpmpp_sde,
-            'dpmpp_2m_sde_karras': sample_dpmpp_2m_sde
+            'dpmpp_2m_sde_karras': sample_dpmpp_2m_sde,
+            'onestep': sample_onestep,
+            'multistep': stochastic_iterative_sampler,
+            'multistep2': stochastic_iterative_sampler2,
+            'multistep3': stochastic_iterative_sampler3,
+            'dpmpp_2m_sde_lcm': sample_dpmpp_2m_sde_lcm,
         }[solver]
 
         # options
@@ -620,6 +627,44 @@ class GaussianDiffusion(object):
                   | torch.isinf(log_sigma)] = float('inf')
         return log_sigma.exp()
 
+    @torch.no_grad()
+    def stochastic_encode(self, x0, t, steps):
+        # fast, but does not allow for exact reconstruction
+        # t serves as an index to gather the correct alphas
+
+        t_max = None
+        t_min = None
+
+        # discretization method
+        discretization = 'trailing' if self.prediction_type == 'v' else 'leading'
+
+        # timesteps
+        if isinstance(steps, int):
+            t_max = self.num_timesteps - 1 if t_max is None else t_max
+            t_min = 0 if t_min is None else t_min
+            steps = discretize_timesteps(t_max, t_min, steps, discretization)
+        steps = torch.as_tensor(steps).round().long().flip(0).to(x0.device)
+        # steps = torch.as_tensor(steps).round().long().to(x0.device)
+
+        # self.alphas_bar = torch.cumprod(1 - self.sigmas ** 2, dim=0)
+        # print('sigma: ', self.sigmas, len(self.sigmas))
+        # print('alpha_bar: ', self.alphas_bar, len(self.alphas_bar))
+        # print('steps: ', steps, len(steps))
+        # sqrt_alphas_cumprod = torch.sqrt(self.alphas_bar).to(x0.device)[steps]
+        # sqrt_one_minus_alphas_cumprod = torch.sqrt(1 - self.alphas_bar).to(x0.device)[steps]
+
+        sqrt_alphas_cumprod = self.alphas.to(x0.device)[steps]
+        sqrt_one_minus_alphas_cumprod = self.sigmas.to(x0.device)[steps]
+        # print('sigma: ', self.sigmas, len(self.sigmas))
+        # print('alpha: ', self.alphas, len(self.alphas))
+        # print('steps: ', steps, len(steps))
+
+        noise = torch.randn_like(x0)
+        return (
+            extract_into_tensor(sqrt_alphas_cumprod, t, x0.shape) * x0 +
+            extract_into_tensor(sqrt_one_minus_alphas_cumprod, t, x0.shape) *
+            noise)
+
 
 def extract_into_tensor(a, t, x_shape):
     b, *_ = t.shape
@@ -642,3 +687,19 @@ def discretize_timesteps(t_max, t_min, steps, discretization):
         raise NotImplementedError(
             f'{discretization} discretization not implemented')
     return steps.clamp_(t_min, t_max)
+
+
+def get_scalings_for_boundary_condition(sigma):
+    sigma_data = 0.5
+    c_skip = (1 -
+              sigma**2)**0.5 * sigma_data**2 / (sigma**2 +
+                                                (1 - sigma**2) * sigma_data**2)
+    c_out = (sigma * sigma_data / (sigma**2 +
+                                   (1 - sigma**2) * sigma_data**2)**0.5)
+    return c_skip, c_out
+
+
+def v_to_x0(v, t, x_t, diffusion):
+    sigmas = _i(diffusion.sigmas, t, v)
+    alphas = _i(diffusion.alphas, t, v)
+    return alphas * x_t - sigmas * v

scepter/modules/model/network/diffusion/solvers.py

@@ -12,13 +12,17 @@ q(x_t | x_0) = N(x_t | alpha_t x_0, sigma_t^2 I),
 
 where 0 <= sigma_t <= 1 and alpha_t^2 = 1 - sigma_t^2.
 """
-import torch
 from tqdm.auto import trange
 
+import torch
+
 __all__ = [
     'sample_euler', 'sample_euler_ancestral', 'sample_heun', 'sample_dpm_2',
     'sample_dpm_2_ancestral', 'sample_dpmpp_2s_ancestral', 'sample_dpmpp_sde',
-    'sample_dpmpp_2m', 'sample_dpmpp_2m_sde', 'sample_ddim'
+    'sample_dpmpp_2m', 'sample_dpmpp_2m_sde', 'sample_ddim',
+    'sample_dpmpp_2m_sde_lcm', 'sample_onestep',
+    'stochastic_iterative_sampler', 'stochastic_iterative_sampler2',
+    'stochastic_iterative_sampler3'
 ]
 
 # -------------------- variation exploding (VE) solver --------------------#
@@ -630,3 +634,183 @@ def sample_img2img_euler_ancestral(noise,
             if sigmas[i + 1] > 0:
                 x = x + torch.randn_like(x) * s_noise * sigma_up
     return x
+
+
+# --------- LCM ------------
+
+
+@torch.no_grad()
+def sample_dpmpp_2m_sde_lcm(noise,
+                            model,
+                            sigmas,
+                            eta=1.,
+                            s_noise=1.,
+                            solver_type='midpoint',
+                            show_progress=True,
+                            total_sample_steps=50,
+                            sample_steps=5,
+                            **kwargs):
+    """
+    DPM-Solver++ (2M) SDE.
+    """
+    assert solver_type in {'heun', 'midpoint'}
+
+    dm_steps = sample_steps
+    x = noise * (sigmas[-dm_steps - 1]**2 + 1)**0.5
+    sigma_min, sigma_max = sigmas[sigmas > 0].min(), sigmas[
+        sigmas < float('inf')].max()
+    noise_sampler = BrownianTreeNoiseSampler(x, sigma_min, sigma_max)
+    old_denoised = None
+    h_last = None
+
+    for i in trange(len(sigmas) - dm_steps - 1,
+                    len(sigmas) - 1,
+                    disable=not show_progress):
+        if sigmas[i] == float('inf'):
+            # Euler method
+            denoised = model(noise, sigmas[i])
+            x = denoised + sigmas[i + 1] * noise
+        else:
+            _, c_in = get_scalings(sigmas[i])
+            denoised = model(x * c_in, sigmas[i])
+            if sigmas[i + 1] == 0:
+                # Denoising step
+                x = denoised
+            else:
+                # DPM-Solver++(2M) SDE
+                t, s = -sigmas[i].log(), -sigmas[i + 1].log()
+                h = s - t
+                eta_h = eta * h
+
+                x = sigmas[i + 1] / sigmas[i] * (-eta_h).exp() * x + \
+                    (-h - eta_h).expm1().neg() * denoised
+
+                if old_denoised is not None:
+                    r = h_last / h
+                    if solver_type == 'heun':
+                        x = x + ((-h - eta_h).expm1().neg() / (-h - eta_h) + 1) * \
+                            (1 / r) * (denoised - old_denoised)
+                    elif solver_type == 'midpoint':
+                        x = x + 0.5 * (-h - eta_h).expm1().neg() * \
+                            (1 / r) * (denoised - old_denoised)
+
+                x = x + noise_sampler(sigmas[i], sigmas[i + 1]) * sigmas[
+                    i + 1] * (-2 * eta_h).expm1().neg().sqrt() * s_noise
+
+                old_denoised = denoised
+                h_last = h
+    return x
+
+
+@torch.no_grad()
+def sample_onestep(noise, model, sigmas, show_progress=True, **kwargs):
+    """
+    CM one step solver.
+    """
+    x = noise
+    denoised = model(x, sigmas[0])
+    return denoised
+
+
+@torch.no_grad()
+def stochastic_iterative_sampler(noise,
+                                 model,
+                                 sigmas,
+                                 show_progress=True,
+                                 **kwargs):
+    """
+    CM multiple steps solver.
+    """
+    x = noise
+    sigmas_vp = (sigmas**2 / (1 + sigmas**2))**0.5
+    sigmas_vp[sigmas == float('inf')] = 1.
+    for i in trange(len(sigmas) - 1, disable=not show_progress):
+        denoised = model(x, sigmas[i])
+
+        sigma = sigmas_vp[i + 1]
+        alpha = torch.sqrt(1 - sigma**2)
+        noises = torch.randn_like(x)
+        x = alpha * denoised + sigma * noises
+    return x
+
+
+@torch.no_grad()
+def stochastic_iterative_sampler2(noise,
+                                  model,
+                                  sigmas,
+                                  show_progress=True,
+                                  repeat=1,
+                                  ema=0.05,
+                                  **kwargs):
+    """
+    CM multiple steps solver.
+    """
+    # denoised_s = []
+    # for _ in range(repeat):
+    #     noise = torch.randn_like(noise)
+    #     denoised_s.append(model(noise, sigmas[0]))
+
+    # denoised_s = torch.cat(denoised_s, dim = 0)
+    # denoised = torch.mean(denoised_s, dim=0, keepdim=True)
+    denoised = model(noise, sigmas[0])
+    denoised_ema = denoised
+
+    sigmas_vp = (sigmas**2 / (1 + sigmas**2))**0.5
+    sigmas_vp[sigmas == float('inf')] = 1.
+    for i in trange(1, len(sigmas) - 1, disable=not show_progress):
+        sigma = sigmas_vp[i]
+        alpha = torch.sqrt(1 - sigma**2)
+        noise = torch.randn_like(noise)
+        x = alpha * denoised + sigma * noise
+        denoised = model(x, sigmas[i])
+        denoised_ema = ema * denoised_ema + (1.0 - ema) * denoised
+    return denoised_ema
+
+
+@torch.no_grad()
+def stochastic_iterative_sampler3(noise,
+                                  model,
+                                  sigmas,
+                                  show_progress=True,
+                                  repeat=1,
+                                  solver_eta=0.2,
+                                  solver_ema=0.05,
+                                  **kwargs):
+    """
+    CM multiple steps solver.
+    """
+    eta = solver_eta
+    ema = solver_ema
+    # denoised_s = []
+    # for _ in range(repeat):
+    #     noise = torch.randn_like(noise)
+    #     denoised_s.append(model(noise, sigmas[0]))
+
+    # denoised_s = torch.cat(denoised_s, dim = 0)
+    # denoised = torch.mean(denoised_s, dim=0, keepdim=True)
+    denoised = model(noise, sigmas[0])
+    denoised_ema = denoised
+    eta = float(eta)
+
+    sigmas_vp = (sigmas**2 / (1 + sigmas**2))**0.5
+    sigmas_vp[sigmas == float('inf')] = 1.
+    noises = [
+        noise.clone(),
+    ]
+    for i in trange(1, len(sigmas) - 1, disable=not show_progress):
+        sigma = sigmas_vp[i]
+        alpha = torch.sqrt(1 - sigma**2)
+        # x = alpha * denoised + sigma * torch.randn_like(noise)
+        # x = alpha * denoised + sigma * (eta * noise + (1 - eta ** 2) ** 0.5 * torch.randn_like(noise))
+        # eta = 0.2*sigma / (sigma + 1.0)
+        # noise = eta * noise + (1 - eta ** 2) ** 0.5 * torch.randn_like(noise)
+        noise2 = 0
+        for n in noises:
+            noise2 += eta * n
+        noise2 += (1 - (eta * len(noises))**2)**0.5 * torch.randn_like(noise)
+        noises.append(noise2.clone())
+        # noise = torch.randn_like(noise)
+        x = alpha * denoised + sigma * noise2
+        denoised = model(x, sigmas[i])
+        denoised_ema = ema * denoised_ema + (1.0 - ema) * denoised
+    return denoised_ema

scepter/modules/model/network/ldm/__init__.py

@@ -1,6 +1,7 @@
 # -*- coding: utf-8 -*-
 # Copyright (c) Alibaba, Inc. and its affiliates.
 from scepter.modules.model.network.ldm.ldm import LatentDiffusion
+from scepter.modules.model.network.ldm.ldm_edit import LatentDiffusionEdit
 from scepter.modules.model.network.ldm.ldm_sce import (
     LatentDiffusionSCEControl, LatentDiffusionSCETuning,
     LatentDiffusionXLSCEControl, LatentDiffusionXLSCETuning)