nodes/kijai-ComfyUI-CCSR
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Autodetect fp16 use, separate model selector, keep model loaded optionally

Browse Source
This commit is contained in:
kijai
2024-01-14 14:00:15 +02:00
parent e3dee409e4
commit 77092a1e0a
1 changed files with 43 additions and 19 deletions
Download Patch File Download Diff File Expand all files Collapse all files
nodes.py
+43 -19
View File
@@ -10,13 +10,16 @@ from .model.ccsr_stage1 import ControlLDM
from .utils.common import instantiate_from_config, load_state_dict
import comfy.model_management
import folder_paths
script_directory = os.path.dirname(os.path.abspath(__file__))
class CCSR_Upscale:
@classmethod
def INPUT_TYPES(s):
return {"required": {
return {"required": {
"ccsr_model": ("CCSRMODEL", ),
"image": ("IMAGE", ),
"steps": ("INT", {"default": 45, "min": 1, "max": 4096, "step": 1}),
"t_max": ("FLOAT", {"default": 0.6667,"min": 0, "max": 1, "step": 0.01}),
@@ -32,7 +35,7 @@ class CCSR_Upscale:
], {
"default": 'adain'
}),
"use_fp16": ("BOOLEAN", {"default": True}),
"keep_model_loaded": ("BOOLEAN", {"default": False}),
},
}
@@ -44,22 +47,21 @@ class CCSR_Upscale:
CATEGORY = "CCSR"
@torch.no_grad()
def process(self, image, steps, t_max, t_min, tiled,tile_size, tile_stride, color_fix_type, use_fp16):
checkpoint_path = os.path.join(script_directory, "../../models/checkpoints/real-world_ccsr.ckpt")
config_path = os.path.join(script_directory, "configs/model/ccsr_stage2.yaml")
config = OmegaConf.load(config_path)
model = instantiate_from_config(config)
def process(self, ccsr_model, image, steps, t_max, t_min, tiled,tile_size, tile_stride, color_fix_type, keep_model_loaded):
device = comfy.model_management.get_torch_device()
config_path = os.path.join(script_directory, "configs/model/ccsr_stage2.yaml")
dtype = torch.float16 if comfy.model_management.should_use_fp16() else torch.float32
if not hasattr(self, "model") or self.model is None:
config = OmegaConf.load(config_path)
self.model = instantiate_from_config(config)
load_state_dict(self.model, torch.load(ccsr_model, map_location="cpu"), strict=True)
# reload preprocess model if specified
load_state_dict(model, torch.load(checkpoint_path, map_location="cpu"), strict=True)
# reload preprocess model if specified
model.freeze()
model.to(device)
if (use_fp16):
model.half()
sampler = SpacedSampler(model, var_type="fixed_small")
self.model.freeze()
self.model.to(device, dtype=dtype)
sampler = SpacedSampler(self.model, var_type="fixed_small")
# Assuming 'image' is a PyTorch tensor with shape [B, H, W, C] and you want to resize it.
B, H, W, C = image.shape
@@ -77,12 +79,12 @@ class CCSR_Upscale:
# Move the tensor to the GPU.
resized_image = resized_image.to(device)
strength = 1.0
model.control_scales = [strength] * 13
self.model.control_scales = [strength] * 13
cond_fn = None
height, width = resized_image.size(-2), resized_image.size(-1)
shape = (1, 4, height // 8, width // 8)
x_T = torch.randn(shape, device=model.device, dtype=torch.float32)
with torch.autocast(comfy.model_management.get_autocast_device(device), dtype=model.dtype) if use_fp16 else nullcontext():
x_T = torch.randn(shape, device=self.model.device, dtype=dtype)
with torch.autocast(comfy.model_management.get_autocast_device(device), dtype=dtype) if dtype == torch.float16 else nullcontext():
if not tiled:
samples = sampler.sample_ccsr(
steps=steps, t_max=t_max, t_min=t_min, shape=shape, cond_img=resized_image,
@@ -116,11 +118,33 @@ class CCSR_Upscale:
# If necessary, rearrange from [B, C, H, W] back to [B, H, W, C] and move to CPU
resized_back_image = resized_back_image.permute(0, 2, 3, 1).cpu()
if not keep_model_loaded:
self.model = None
comfy.model_management.soft_empty_cache()
return(resized_back_image,)
class CCSR_Model_Select:
@classmethod
def INPUT_TYPES(s):
return {"required": {
"ckpt_name": (folder_paths.get_filename_list("checkpoints"),),
}}
RETURN_TYPES = ("CCSRMODEL",)
RETURN_NAMES = ("ccsr_model",)
FUNCTION = "load_ccsr_checkpoint"
CATEGORY = "CCSR"
def load_ccsr_checkpoint(self, ckpt_name):
ckpt_path = folder_paths.get_full_path("checkpoints", ckpt_name)
return (ckpt_path,)
NODE_CLASS_MAPPINGS = {
"CCSR_Upscale": CCSR_Upscale,
"CCSR_Model_Select": CCSR_Model_Select
}
NODE_DISPLAY_NAME_MAPPINGS = {
"CCSR_Upscale": "CCSR_Upscale",
"CCSR_Model_Select": "CCSR_Model_Select"
}