449 lines
23 KiB
Python
449 lines
23 KiB
Python
# Copyright 2023 The HuggingFace Team. All rights reserved.
|
|
#
|
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
# you may not use this file except in compliance with the License.
|
|
# You may obtain a copy of the License at
|
|
#
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
#
|
|
# Unless required by applicable law or agreed to in writing, software
|
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
# See the License for the specific language governing permissions and
|
|
# limitations under the License.
|
|
|
|
import inspect
|
|
from typing import Callable, List, Optional, Union
|
|
|
|
import PIL.Image
|
|
import torch
|
|
from packaging import version
|
|
from transformers import CLIPImageProcessor, CLIPVisionModelWithProjection
|
|
|
|
from ...configuration_utils import FrozenDict
|
|
from ...image_processor import VaeImageProcessor
|
|
from ...models import AutoencoderKL, UNet2DConditionModel
|
|
from ...schedulers import KarrasDiffusionSchedulers
|
|
from ...utils import deprecate, logging
|
|
from ...utils.torch_utils import randn_tensor
|
|
from ..pipeline_utils import DiffusionPipeline
|
|
from . import StableDiffusionPipelineOutput
|
|
from .safety_checker import StableDiffusionSafetyChecker
|
|
|
|
|
|
logger = logging.get_logger(__name__) # pylint: disable=invalid-name
|
|
|
|
|
|
class StableDiffusionImageVariationPipeline(DiffusionPipeline):
|
|
r"""
|
|
Pipeline to generate image variations from an input image using Stable Diffusion.
|
|
|
|
This model inherits from [`DiffusionPipeline`]. Check the superclass documentation for the generic methods
|
|
implemented for all pipelines (downloading, saving, running on a particular device, etc.).
|
|
|
|
Args:
|
|
vae ([`AutoencoderKL`]):
|
|
Variational Auto-Encoder (VAE) model to encode and decode images to and from latent representations.
|
|
image_encoder ([`~transformers.CLIPVisionModelWithProjection`]):
|
|
Frozen CLIP image-encoder ([clip-vit-large-patch14](https://huggingface.co/openai/clip-vit-large-patch14)).
|
|
text_encoder ([`~transformers.CLIPTextModel`]):
|
|
Frozen text-encoder ([clip-vit-large-patch14](https://huggingface.co/openai/clip-vit-large-patch14)).
|
|
tokenizer ([`~transformers.CLIPTokenizer`]):
|
|
A `CLIPTokenizer` to tokenize text.
|
|
unet ([`UNet2DConditionModel`]):
|
|
A `UNet2DConditionModel` to denoise the encoded image latents.
|
|
scheduler ([`SchedulerMixin`]):
|
|
A scheduler to be used in combination with `unet` to denoise the encoded image latents. Can be one of
|
|
[`DDIMScheduler`], [`LMSDiscreteScheduler`], or [`PNDMScheduler`].
|
|
safety_checker ([`StableDiffusionSafetyChecker`]):
|
|
Classification module that estimates whether generated images could be considered offensive or harmful.
|
|
Please refer to the [model card](https://huggingface.co/runwayml/stable-diffusion-v1-5) for more details
|
|
about a model's potential harms.
|
|
feature_extractor ([`~transformers.CLIPImageProcessor`]):
|
|
A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
|
|
"""
|
|
|
|
# TODO: feature_extractor is required to encode images (if they are in PIL format),
|
|
# we should give a descriptive message if the pipeline doesn't have one.
|
|
_optional_components = ["safety_checker"]
|
|
model_cpu_offload_seq = "image_encoder->unet->vae"
|
|
_exclude_from_cpu_offload = ["safety_checker"]
|
|
|
|
def __init__(
|
|
self,
|
|
vae: AutoencoderKL,
|
|
image_encoder: CLIPVisionModelWithProjection,
|
|
unet: UNet2DConditionModel,
|
|
scheduler: KarrasDiffusionSchedulers,
|
|
safety_checker: StableDiffusionSafetyChecker,
|
|
feature_extractor: CLIPImageProcessor,
|
|
requires_safety_checker: bool = True,
|
|
):
|
|
super().__init__()
|
|
|
|
if safety_checker is None and requires_safety_checker:
|
|
logger.warn(
|
|
f"You have disabled the safety checker for {self.__class__} by passing `safety_checker=None`. Ensure"
|
|
" that you abide to the conditions of the Stable Diffusion license and do not expose unfiltered"
|
|
" results in services or applications open to the public. Both the diffusers team and Hugging Face"
|
|
" strongly recommend to keep the safety filter enabled in all public facing circumstances, disabling"
|
|
" it only for use-cases that involve analyzing network behavior or auditing its results. For more"
|
|
" information, please have a look at https://github.com/huggingface/diffusers/pull/254 ."
|
|
)
|
|
|
|
if safety_checker is not None and feature_extractor is None:
|
|
raise ValueError(
|
|
"Make sure to define a feature extractor when loading {self.__class__} if you want to use the safety"
|
|
" checker. If you do not want to use the safety checker, you can pass `'safety_checker=None'` instead."
|
|
)
|
|
|
|
is_unet_version_less_0_9_0 = hasattr(unet.config, "_diffusers_version") and version.parse(
|
|
version.parse(unet.config._diffusers_version).base_version
|
|
) < version.parse("0.9.0.dev0")
|
|
is_unet_sample_size_less_64 = hasattr(unet.config, "sample_size") and unet.config.sample_size < 64
|
|
if is_unet_version_less_0_9_0 and is_unet_sample_size_less_64:
|
|
deprecation_message = (
|
|
"The configuration file of the unet has set the default `sample_size` to smaller than"
|
|
" 64 which seems highly unlikely .If you're checkpoint is a fine-tuned version of any of the"
|
|
" following: \n- CompVis/stable-diffusion-v1-4 \n- CompVis/stable-diffusion-v1-3 \n-"
|
|
" CompVis/stable-diffusion-v1-2 \n- CompVis/stable-diffusion-v1-1 \n- runwayml/stable-diffusion-v1-5"
|
|
" \n- runwayml/stable-diffusion-inpainting \n you should change 'sample_size' to 64 in the"
|
|
" configuration file. Please make sure to update the config accordingly as leaving `sample_size=32`"
|
|
" in the config might lead to incorrect results in future versions. If you have downloaded this"
|
|
" checkpoint from the Hugging Face Hub, it would be very nice if you could open a Pull request for"
|
|
" the `unet/config.json` file"
|
|
)
|
|
deprecate("sample_size<64", "1.0.0", deprecation_message, standard_warn=False)
|
|
new_config = dict(unet.config)
|
|
new_config["sample_size"] = 64
|
|
unet._internal_dict = FrozenDict(new_config)
|
|
|
|
self.register_modules(
|
|
vae=vae,
|
|
image_encoder=image_encoder,
|
|
unet=unet,
|
|
scheduler=scheduler,
|
|
safety_checker=safety_checker,
|
|
feature_extractor=feature_extractor,
|
|
)
|
|
self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
|
|
self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
|
|
self.register_to_config(requires_safety_checker=requires_safety_checker)
|
|
|
|
def _encode_image(self, image, device, num_images_per_prompt, do_classifier_free_guidance):
|
|
dtype = next(self.image_encoder.parameters()).dtype
|
|
|
|
if not isinstance(image, torch.Tensor):
|
|
image = self.feature_extractor(images=image, return_tensors="pt").pixel_values
|
|
|
|
image = image.to(device=device, dtype=dtype)
|
|
image_embeddings = self.image_encoder(image).image_embeds
|
|
image_embeddings = image_embeddings.unsqueeze(1)
|
|
|
|
# duplicate image embeddings for each generation per prompt, using mps friendly method
|
|
bs_embed, seq_len, _ = image_embeddings.shape
|
|
image_embeddings = image_embeddings.repeat(1, num_images_per_prompt, 1)
|
|
image_embeddings = image_embeddings.view(bs_embed * num_images_per_prompt, seq_len, -1)
|
|
|
|
if do_classifier_free_guidance:
|
|
negative_prompt_embeds = torch.zeros_like(image_embeddings)
|
|
|
|
# For classifier free guidance, we need to do two forward passes.
|
|
# Here we concatenate the unconditional and text embeddings into a single batch
|
|
# to avoid doing two forward passes
|
|
image_embeddings = torch.cat([negative_prompt_embeds, image_embeddings])
|
|
|
|
return image_embeddings
|
|
|
|
# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.run_safety_checker
|
|
def run_safety_checker(self, image, device, dtype):
|
|
if self.safety_checker is None:
|
|
has_nsfw_concept = None
|
|
else:
|
|
if torch.is_tensor(image):
|
|
feature_extractor_input = self.image_processor.postprocess(image, output_type="pil")
|
|
else:
|
|
feature_extractor_input = self.image_processor.numpy_to_pil(image)
|
|
safety_checker_input = self.feature_extractor(feature_extractor_input, return_tensors="pt").to(device)
|
|
image, has_nsfw_concept = self.safety_checker(
|
|
images=image, clip_input=safety_checker_input.pixel_values.to(dtype)
|
|
)
|
|
return image, has_nsfw_concept
|
|
|
|
# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.decode_latents
|
|
def decode_latents(self, latents):
|
|
deprecation_message = "The decode_latents method is deprecated and will be removed in 1.0.0. Please use VaeImageProcessor.postprocess(...) instead"
|
|
deprecate("decode_latents", "1.0.0", deprecation_message, standard_warn=False)
|
|
|
|
latents = 1 / self.vae.config.scaling_factor * latents
|
|
image = self.vae.decode(latents, return_dict=False)[0]
|
|
image = (image / 2 + 0.5).clamp(0, 1)
|
|
# we always cast to float32 as this does not cause significant overhead and is compatible with bfloat16
|
|
image = image.cpu().permute(0, 2, 3, 1).float().numpy()
|
|
return image
|
|
|
|
# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.prepare_extra_step_kwargs
|
|
def prepare_extra_step_kwargs(self, generator, eta):
|
|
# prepare extra kwargs for the scheduler step, since not all schedulers have the same signature
|
|
# eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers.
|
|
# eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502
|
|
# and should be between [0, 1]
|
|
|
|
accepts_eta = "eta" in set(inspect.signature(self.scheduler.step).parameters.keys())
|
|
extra_step_kwargs = {}
|
|
if accepts_eta:
|
|
extra_step_kwargs["eta"] = eta
|
|
|
|
# check if the scheduler accepts generator
|
|
accepts_generator = "generator" in set(inspect.signature(self.scheduler.step).parameters.keys())
|
|
if accepts_generator:
|
|
extra_step_kwargs["generator"] = generator
|
|
return extra_step_kwargs
|
|
|
|
def check_inputs(self, image, height, width, callback_steps):
|
|
if (
|
|
not isinstance(image, torch.Tensor)
|
|
and not isinstance(image, PIL.Image.Image)
|
|
and not isinstance(image, list)
|
|
):
|
|
raise ValueError(
|
|
"`image` has to be of type `torch.FloatTensor` or `PIL.Image.Image` or `List[PIL.Image.Image]` but is"
|
|
f" {type(image)}"
|
|
)
|
|
|
|
if height % 8 != 0 or width % 8 != 0:
|
|
raise ValueError(f"`height` and `width` have to be divisible by 8 but are {height} and {width}.")
|
|
|
|
if (callback_steps is None) or (
|
|
callback_steps is not None and (not isinstance(callback_steps, int) or callback_steps <= 0)
|
|
):
|
|
raise ValueError(
|
|
f"`callback_steps` has to be a positive integer but is {callback_steps} of type"
|
|
f" {type(callback_steps)}."
|
|
)
|
|
|
|
# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.prepare_latents
|
|
def prepare_latents(self, batch_size, num_channels_latents, height, width, dtype, device, generator, latents=None):
|
|
shape = (batch_size, num_channels_latents, height // self.vae_scale_factor, width // self.vae_scale_factor)
|
|
if isinstance(generator, list) and len(generator) != batch_size:
|
|
raise ValueError(
|
|
f"You have passed a list of generators of length {len(generator)}, but requested an effective batch"
|
|
f" size of {batch_size}. Make sure the batch size matches the length of the generators."
|
|
)
|
|
|
|
if latents is None:
|
|
latents = randn_tensor(shape, generator=generator, device=device, dtype=dtype)
|
|
else:
|
|
latents = latents.to(device)
|
|
|
|
# scale the initial noise by the standard deviation required by the scheduler
|
|
latents = latents * self.scheduler.init_noise_sigma
|
|
return latents
|
|
|
|
# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.enable_freeu
|
|
def enable_freeu(self, s1: float, s2: float, b1: float, b2: float):
|
|
r"""Enables the FreeU mechanism as in https://arxiv.org/abs/2309.11497.
|
|
|
|
The suffixes after the scaling factors represent the stages where they are being applied.
|
|
|
|
Please refer to the [official repository](https://github.com/ChenyangSi/FreeU) for combinations of the values
|
|
that are known to work well for different pipelines such as Stable Diffusion v1, v2, and Stable Diffusion XL.
|
|
|
|
Args:
|
|
s1 (`float`):
|
|
Scaling factor for stage 1 to attenuate the contributions of the skip features. This is done to
|
|
mitigate "oversmoothing effect" in the enhanced denoising process.
|
|
s2 (`float`):
|
|
Scaling factor for stage 2 to attenuate the contributions of the skip features. This is done to
|
|
mitigate "oversmoothing effect" in the enhanced denoising process.
|
|
b1 (`float`): Scaling factor for stage 1 to amplify the contributions of backbone features.
|
|
b2 (`float`): Scaling factor for stage 2 to amplify the contributions of backbone features.
|
|
"""
|
|
if not hasattr(self, "unet"):
|
|
raise ValueError("The pipeline must have `unet` for using FreeU.")
|
|
self.unet.enable_freeu(s1=s1, s2=s2, b1=b1, b2=b2)
|
|
|
|
# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.disable_freeu
|
|
def disable_freeu(self):
|
|
"""Disables the FreeU mechanism if enabled."""
|
|
self.unet.disable_freeu()
|
|
|
|
@torch.no_grad()
|
|
def __call__(
|
|
self,
|
|
image: Union[PIL.Image.Image, List[PIL.Image.Image], torch.FloatTensor],
|
|
height: Optional[int] = None,
|
|
width: Optional[int] = None,
|
|
num_inference_steps: int = 50,
|
|
guidance_scale: float = 7.5,
|
|
num_images_per_prompt: Optional[int] = 1,
|
|
eta: float = 0.0,
|
|
generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
|
|
latents: Optional[torch.FloatTensor] = None,
|
|
output_type: Optional[str] = "pil",
|
|
return_dict: bool = True,
|
|
callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None,
|
|
callback_steps: int = 1,
|
|
):
|
|
r"""
|
|
The call function to the pipeline for generation.
|
|
|
|
Args:
|
|
image (`PIL.Image.Image` or `List[PIL.Image.Image]` or `torch.FloatTensor`):
|
|
Image or images to guide image generation. If you provide a tensor, it needs to be compatible with
|
|
[`CLIPImageProcessor`](https://huggingface.co/lambdalabs/sd-image-variations-diffusers/blob/main/feature_extractor/preprocessor_config.json).
|
|
height (`int`, *optional*, defaults to `self.unet.config.sample_size * self.vae_scale_factor`):
|
|
The height in pixels of the generated image.
|
|
width (`int`, *optional*, defaults to `self.unet.config.sample_size * self.vae_scale_factor`):
|
|
The width in pixels of the generated image.
|
|
num_inference_steps (`int`, *optional*, defaults to 50):
|
|
The number of denoising steps. More denoising steps usually lead to a higher quality image at the
|
|
expense of slower inference. This parameter is modulated by `strength`.
|
|
guidance_scale (`float`, *optional*, defaults to 7.5):
|
|
A higher guidance scale value encourages the model to generate images closely linked to the text
|
|
`prompt` at the expense of lower image quality. Guidance scale is enabled when `guidance_scale > 1`.
|
|
num_images_per_prompt (`int`, *optional*, defaults to 1):
|
|
The number of images to generate per prompt.
|
|
eta (`float`, *optional*, defaults to 0.0):
|
|
Corresponds to parameter eta (η) from the [DDIM](https://arxiv.org/abs/2010.02502) paper. Only applies
|
|
to the [`~schedulers.DDIMScheduler`], and is ignored in other schedulers.
|
|
generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
|
|
A [`torch.Generator`](https://pytorch.org/docs/stable/generated/torch.Generator.html) to make
|
|
generation deterministic.
|
|
latents (`torch.FloatTensor`, *optional*):
|
|
Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for image
|
|
generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
|
|
tensor is generated by sampling using the supplied random `generator`.
|
|
output_type (`str`, *optional*, defaults to `"pil"`):
|
|
The output format of the generated image. Choose between `PIL.Image` or `np.array`.
|
|
return_dict (`bool`, *optional*, defaults to `True`):
|
|
Whether or not to return a [`~pipelines.stable_diffusion.StableDiffusionPipelineOutput`] instead of a
|
|
plain tuple.
|
|
callback (`Callable`, *optional*):
|
|
A function that calls every `callback_steps` steps during inference. The function is called with the
|
|
following arguments: `callback(step: int, timestep: int, latents: torch.FloatTensor)`.
|
|
callback_steps (`int`, *optional*, defaults to 1):
|
|
The frequency at which the `callback` function is called. If not specified, the callback is called at
|
|
every step.
|
|
|
|
Returns:
|
|
[`~pipelines.stable_diffusion.StableDiffusionPipelineOutput`] or `tuple`:
|
|
If `return_dict` is `True`, [`~pipelines.stable_diffusion.StableDiffusionPipelineOutput`] is returned,
|
|
otherwise a `tuple` is returned where the first element is a list with the generated images and the
|
|
second element is a list of `bool`s indicating whether the corresponding generated image contains
|
|
"not-safe-for-work" (nsfw) content.
|
|
|
|
Examples:
|
|
|
|
```py
|
|
from diffusers import StableDiffusionImageVariationPipeline
|
|
from PIL import Image
|
|
from io import BytesIO
|
|
import requests
|
|
|
|
pipe = StableDiffusionImageVariationPipeline.from_pretrained(
|
|
"lambdalabs/sd-image-variations-diffusers", revision="v2.0"
|
|
)
|
|
pipe = pipe.to("cuda")
|
|
|
|
url = "https://lh3.googleusercontent.com/y-iFOHfLTwkuQSUegpwDdgKmOjRSTvPxat63dQLB25xkTs4lhIbRUFeNBWZzYf370g=s1200"
|
|
|
|
response = requests.get(url)
|
|
image = Image.open(BytesIO(response.content)).convert("RGB")
|
|
|
|
out = pipe(image, num_images_per_prompt=3, guidance_scale=15)
|
|
out["images"][0].save("result.jpg")
|
|
```
|
|
"""
|
|
# 0. Default height and width to unet
|
|
height = height or self.unet.config.sample_size * self.vae_scale_factor
|
|
width = width or self.unet.config.sample_size * self.vae_scale_factor
|
|
|
|
# 1. Check inputs. Raise error if not correct
|
|
self.check_inputs(image, height, width, callback_steps)
|
|
|
|
# 2. Define call parameters
|
|
if isinstance(image, PIL.Image.Image):
|
|
batch_size = 1
|
|
elif isinstance(image, list):
|
|
batch_size = len(image)
|
|
else:
|
|
batch_size = image.shape[0]
|
|
device = self._execution_device
|
|
# here `guidance_scale` is defined analog to the guidance weight `w` of equation (2)
|
|
# of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1`
|
|
# corresponds to doing no classifier free guidance.
|
|
do_classifier_free_guidance = guidance_scale > 1.0
|
|
|
|
# 3. Encode input image
|
|
image_embeddings = self._encode_image(image, device, num_images_per_prompt, do_classifier_free_guidance)
|
|
|
|
# 4. Prepare timesteps
|
|
self.scheduler.set_timesteps(num_inference_steps, device=device)
|
|
timesteps = self.scheduler.timesteps
|
|
|
|
# 5. Prepare latent variables
|
|
num_channels_latents = self.unet.config.in_channels
|
|
latents = self.prepare_latents(
|
|
batch_size * num_images_per_prompt,
|
|
num_channels_latents,
|
|
height,
|
|
width,
|
|
image_embeddings.dtype,
|
|
device,
|
|
generator,
|
|
latents,
|
|
)
|
|
|
|
# 6. Prepare extra step kwargs. TODO: Logic should ideally just be moved out of the pipeline
|
|
extra_step_kwargs = self.prepare_extra_step_kwargs(generator, eta)
|
|
|
|
# 7. Denoising loop
|
|
num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
|
|
with self.progress_bar(total=num_inference_steps) as progress_bar:
|
|
for i, t in enumerate(timesteps):
|
|
# expand the latents if we are doing classifier free guidance
|
|
latent_model_input = torch.cat([latents] * 2) if do_classifier_free_guidance else latents
|
|
latent_model_input = self.scheduler.scale_model_input(latent_model_input, t)
|
|
|
|
# predict the noise residual
|
|
noise_pred = self.unet(latent_model_input, t, encoder_hidden_states=image_embeddings).sample
|
|
|
|
# perform guidance
|
|
if do_classifier_free_guidance:
|
|
noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
|
|
noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)
|
|
|
|
# compute the previous noisy sample x_t -> x_t-1
|
|
latents = self.scheduler.step(noise_pred, t, latents, **extra_step_kwargs).prev_sample
|
|
|
|
# call the callback, if provided
|
|
if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
|
|
progress_bar.update()
|
|
if callback is not None and i % callback_steps == 0:
|
|
step_idx = i // getattr(self.scheduler, "order", 1)
|
|
callback(step_idx, t, latents)
|
|
|
|
self.maybe_free_model_hooks()
|
|
|
|
if not output_type == "latent":
|
|
image = self.vae.decode(latents / self.vae.config.scaling_factor, return_dict=False)[0]
|
|
image, has_nsfw_concept = self.run_safety_checker(image, device, image_embeddings.dtype)
|
|
else:
|
|
image = latents
|
|
has_nsfw_concept = None
|
|
|
|
if has_nsfw_concept is None:
|
|
do_denormalize = [True] * image.shape[0]
|
|
else:
|
|
do_denormalize = [not has_nsfw for has_nsfw in has_nsfw_concept]
|
|
|
|
image = self.image_processor.postprocess(image, output_type=output_type, do_denormalize=do_denormalize)
|
|
|
|
self.maybe_free_model_hooks()
|
|
|
|
if not return_dict:
|
|
return (image, has_nsfw_concept)
|
|
|
|
return StableDiffusionPipelineOutput(images=image, nsfw_content_detected=has_nsfw_concept)
|