xinyun/ComfyUI-KJNodes
1
0
Fork 0
mirror of https://git.datalinker.icu/kijai/ComfyUI-KJNodes.git synced 2026-06-01 16:57:06 +08:00
Code Issues Packages Projects Releases Wiki Activity

Allow using latents with intristic loras

Browse Source
This commit is contained in:
kijai 2024-02-27 01:16:37 +02:00
parent 1040f0950d
commit e9fcd376a6
1 changed files with 34 additions and 28 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

62
nodes.py
View File

@ -3670,44 +3670,50 @@ class Intrinsic_lora_sampling:
@classmethod @classmethod
def INPUT_TYPES(s): def INPUT_TYPES(s):
return {"required": { "model": ("MODEL",), return {"required": { "model": ("MODEL",),
"lora_name": (folder_paths.get_filename_list("intristic_loras"), ), "lora_name": (folder_paths.get_filename_list("intristic_loras"), ),
"task": ( "task": (
[ [
'depth map', 'depth map',
'surface normals', 'surface normals',
'albedo', 'albedo',
'shading', 'shading',
], ],
{ {
"default": 'depth map' "default": 'depth map'
}), }),
"text": ("STRING", {"multiline": True, "default": ""}), "text": ("STRING", {"multiline": True, "default": ""}),
"clip": ("CLIP", ), "clip": ("CLIP", ),
"vae": ("VAE", ), "vae": ("VAE", ),
"image": ("IMAGE",), "per_batch": ("INT", {"default": 16, "min": 1, "max": 4096, "step": 1}),
"per_batch": ("INT", {"default": 16, "min": 1, "max": 4096, "step": 1}), },
"optional": {
}} "image": ("IMAGE",),
"optional_latent": ("LATENT",),
},
}
RETURN_TYPES = ("IMAGE",) RETURN_TYPES = ("IMAGE", "LATENT",)
FUNCTION = "onestepsample" FUNCTION = "onestepsample"
CATEGORY = "KJNodes" CATEGORY = "KJNodes"
def onestepsample(self, model, lora_name, clip, vae, image, text, task, per_batch): def onestepsample(self, model, lora_name, clip, vae, text, task, per_batch, image=None, optional_latent=None):
pbar = comfy.utils.ProgressBar(3) pbar = comfy.utils.ProgressBar(3)
image_list = [] if optional_latent is None:
for start_idx in range(0, image.shape[0], per_batch): image_list = []
sub_pixels = vae.vae_encode_crop_pixels(image[start_idx:start_idx+per_batch]) for start_idx in range(0, image.shape[0], per_batch):
image_list.append(vae.encode(sub_pixels[:,:,:,:3])) sub_pixels = vae.vae_encode_crop_pixels(image[start_idx:start_idx+per_batch])
image_list.append(vae.encode(sub_pixels[:,:,:,:3]))
sample = torch.cat(image_list, dim=0) sample = torch.cat(image_list, dim=0)
else:
sample = optional_latent["samples"]
noise = torch.zeros(sample.size(), dtype=sample.dtype, layout=sample.layout, device="cpu") noise = torch.zeros(sample.size(), dtype=sample.dtype, layout=sample.layout, device="cpu")
prompt = task + "," + text prompt = task + "," + text
positive, = CLIPTextEncode.encode(self, clip, prompt) positive, = CLIPTextEncode.encode(self, clip, prompt)
pbar.update(1)
negative = positive #negative shouldn't do anything in this scenario negative = positive #negative shouldn't do anything in this scenario
pbar.update(1)
#custom model sampling to pass latent through as it is #custom model sampling to pass latent through as it is
class X0_PassThrough(comfy.model_sampling.EPS): class X0_PassThrough(comfy.model_sampling.EPS):
def calculate_denoised(self, sigma, model_output, model_input): def calculate_denoised(self, sigma, model_output, model_input):
@ -3755,7 +3761,7 @@ class Intrinsic_lora_sampling:
else: else:
image_out = image_out.clamp(-1.,1.) image_out = image_out.clamp(-1.,1.)
return (image_out, ) return (image_out, samples,)
class RemapMaskRange: class RemapMaskRange:
@classmethod @classmethod