seargedp / seargesdxl Goto Github PK

Would you please explain, what is "prompt G" and "prompt L" and “prompt R”?

I am using google collab pro.

After generating any image, when trying to make a second one, the following message appears. Failed to validate prompt for output 364:
Output will be ignored

After this the notebook disconnects and I have to run the last cell again

Hello,
I followed the installation guide and when I loaded the json file into ComfyUI after restart. I got following errors:

When loading the graph, the following node types were not found:
SeargeInput2
SeargeInput5
SeargeInput3
SeargeInput4
SeargeIntegerScaler
SeargeOutput5
SeargeOutput7
SeargeSDXLPromptEncoder
SeargeSDXLBasePromptEncoder
SeargeSDXLRefinerPromptEncoder
SeargeInput6
SeargeOutput2
SeargeOutput1
SeargePromptText
SeargeGenerated1
SeargeLatentMuxer3
SeargeInput7
SeargeLoraLoader
SeargeOutput4
SeargeCheckpointLoader
SeargeVAELoader
SeargeUpscaleModelLoader
SeargeInput1
SeargeParameterProcessor
SeargeStylePreprocessor
SeargeSDXLSampler2
SeargeImageSave
SeargeOutput3
SeargeSDXLImage2ImageSampler2
SeargeOutput6
SeargeConditioningMuxer5
SeargeConditioningMuxer2
SeargeIntegerConstant
Nodes that have failed to load will show as red on the graph.

P.S. ComfyUI is installed on Windows

I am getting this error. How to fix that. Thanks

They release a few new schedulers and samplers and it's causing an issue:

Prompt outputs failed validation
SeargeSDXLSampler:

• Return type mismatch between linked nodes: sampler_name, SAMPLER_NAME != ['euler', 'euler_ancestral', 'heun', 'dpm_2', 'dpm_2_ancestral', 'lms', 'dpm_fast', 'dpm_adaptive', 'dpmpp_2s_ancestral', 'dpmpp_sde', 'dpmpp_sde_gpu', 'dpmpp_2m', 'dpmpp_2m_sde', 'dpmpp_2m_sde_gpu', 'dpmpp_3m_sde', 'dpmpp_3m_sde_gpu', 'ddim', 'uni_pc', 'uni_pc_bh2']
• Return type mismatch between linked nodes: scheduler, SCHEDULER_NAME != ['normal', 'karras', 'exponential', 'sgm_uniform', 'simple', 'ddim_uniform']

Hello,

I get this error message when i'm trying to run a workflow with SeargeSDXL.
Am I missing something ?

`Error occurred when executing SeargeSDXLPromptEncoder:

list indices must be integers or slices, not str

File "E:\Utilitaires\ComfyUI_windows_portable\ComfyUI\execution.py", line 144, in recursive_execute
output_data, output_ui = get_output_data(obj, input_data_all)
File "E:\Utilitaires\ComfyUI_windows_portable\ComfyUI\execution.py", line 74, in get_output_data
return_values = map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True)
File "E:\Utilitaires\ComfyUI_windows_portable\ComfyUI\execution.py", line 67, in map_node_over_list
results.append(getattr(obj, func)(**slice_dict(input_data_all, i)))
File "E:\Utilitaires\ComfyUI_windows_portable\ComfyUI\custom_nodes\SeargeSDXL\searge_sdxl_sampler_node.py", line 116, in encode
tokens1["l"] = base_clip.tokenize(pos_l)["l"]`

thanks

Can you add workflow for sd 1.5 too?

I plan to make a tutorial hopefully and will show your repo too

I get this error when the "HiRes Fix Sampler" node is rendered.

Error occurred when executing SeargeSDXLSampler2:

mat1 and mat2 shapes cannot be multiplied (308x2048 and 1280x768)

File "E:\ComfyUI_windows_portable\ComfyUI\execution.py", line 151, in recursive_execute
output_data, output_ui = get_output_data(obj, input_data_all)
File "E:\ComfyUI_windows_portable\ComfyUI\execution.py", line 81, in get_output_data
return_values = map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True)
File "E:\ComfyUI_windows_portable\ComfyUI\execution.py", line 74, in map_node_over_list
results.append(getattr(obj, func)(**slice_dict(input_data_all, i)))
File "E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\SeargeSDXL\modules\sampling.py", line 192, in sample
precondition_result = nodes.common_ksampler(refiner_model, noise_seed + 2, steps, cfg, sampler_name, scheduler, refiner_positive, refiner_negative, latent_image, denoise=denoise, disable_noise=False, start_step=steps - refiner_prep_steps, last_step=steps, force_full_denoise=False)
File "E:\ComfyUI_windows_portable\ComfyUI\nodes.py", line 1176, in common_ksampler
samples = comfy.sample.sample(model, noise, steps, cfg, sampler_name, scheduler, positive, negative, latent_image,
File "E:\ComfyUI_windows_portable\ComfyUI\comfy\sample.py", line 88, in sample
samples = sampler.sample(noise, positive_copy, negative_copy, cfg=cfg, latent_image=latent_image, start_step=start_step, last_step=last_step, force_full_denoise=force_full_denoise, denoise_mask=noise_mask, sigmas=sigmas, callback=callback, disable_pbar=disable_pbar, seed=seed)
File "E:\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 716, in sample
samples = getattr(k_diffusion_sampling, "sample_{}".format(self.sampler))(self.model_k, noise, sigmas, extra_args=extra_args, callback=k_callback, disable=disable_pbar)
File "E:\ComfyUI_windows_portable\python_embeded\lib\site-packages\torch\utils_contextlib.py", line 115, in decorate_context
return func(*args, **kwargs)
File "E:\ComfyUI_windows_portable\ComfyUI\comfy\k_diffusion\sampling.py", line 613, in sample_dpmpp_2m_sde
denoised = model(x, sigmas[i] * s_in, **extra_args)
File "E:\ComfyUI_windows_portable\python_embeded\lib\site-packages\torch\nn\modules\module.py", line 1501, in _call_impl
return forward_call(*args, **kwargs)
File "E:\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 319, in forward
out = self.inner_model(x, sigma, cond=cond, uncond=uncond, cond_scale=cond_scale, cond_concat=cond_concat, model_options=model_options, seed=seed)
File "E:\ComfyUI_windows_portable\python_embeded\lib\site-packages\torch\nn\modules\module.py", line 1501, in call_impl
return forward_call(*args, **kwargs)
File "E:\ComfyUI_windows_portable\ComfyUI\comfy\k_diffusion\external.py", line 125, in forward
eps = self.get_eps(input * c_in, self.sigma_to_t(sigma), **kwargs)
File "E:\ComfyUI_windows_portable\ComfyUI\comfy\k_diffusion\external.py", line 151, in get_eps
return self.inner_model.apply_model(*args, **kwargs)
File "E:\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 307, in apply_model
out = sampling_function(self.inner_model.apply_model, x, timestep, uncond, cond, cond_scale, cond_concat, model_options=model_options, seed=seed)
File "E:\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 285, in sampling_function
cond, uncond = calc_cond_uncond_batch(model_function, cond, uncond, x, timestep, max_total_area, cond_concat, model_options)
File "E:\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 262, in calc_cond_uncond_batch
output = model_function(input_x, timestep, **c).chunk(batch_chunks)
File "E:\ComfyUI_windows_portable\ComfyUI\comfy\model_base.py", line 61, in apply_model
return self.diffusion_model(xc, t, context=context, y=c_adm, control=control, transformer_options=transformer_options).float()
File "E:\ComfyUI_windows_portable\python_embeded\lib\site-packages\torch\nn\modules\module.py", line 1501, in _call_impl
return forward_call(*args, **kwargs)
File "E:\ComfyUI_windows_portable\ComfyUI\comfy\ldm\modules\diffusionmodules\openaimodel.py", line 620, in forward
h = forward_timestep_embed(module, h, emb, context, transformer_options)
File "E:\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-animatediff\animatediff\nodes.py", line 37, in forward_timestep_embed
x = layer(x, context, transformer_options)
File "E:\ComfyUI_windows_portable\python_embeded\lib\site-packages\torch\nn\modules\module.py", line 1501, in _call_impl
return forward_call(*args, **kwargs)
File "E:\ComfyUI_windows_portable\ComfyUI\comfy\ldm\modules\attention.py", line 695, in forward
x = block(x, context=context[i], transformer_options=transformer_options)
File "E:\ComfyUI_windows_portable\python_embeded\lib\site-packages\torch\nn\modules\module.py", line 1501, in _call_impl
return forward_call(*args, **kwargs)
File "E:\ComfyUI_windows_portable\ComfyUI\comfy\ldm\modules\attention.py", line 527, in forward
return checkpoint(self._forward, (x, context, transformer_options), self.parameters(), self.checkpoint)
File "E:\ComfyUI_windows_portable\ComfyUI\comfy\ldm\modules\diffusionmodules\util.py", line 123, in checkpoint
return func(*inputs)
File "E:\ComfyUI_windows_portable\ComfyUI\comfy\ldm\modules\attention.py", line 627, in _forward
n = self.attn2(n, context=context_attn2, value=value_attn2)
File "E:\ComfyUI_windows_portable\python_embeded\lib\site-packages\torch\nn\modules\module.py", line 1501, in _call_impl
return forward_call(*args, **kwargs)
File "E:\ComfyUI_windows_portable\ComfyUI\comfy\ldm\modules\attention.py", line 422, in forward
k = self.to_k(context)
File "E:\ComfyUI_windows_portable\python_embeded\lib\site-packages\torch\nn\modules\module.py", line 1501, in _call_impl
return forward_call(*args, **kwargs)
File "E:\ComfyUI_windows_portable\ComfyUI\comfy\ops.py", line 18, in forward
return torch.nn.functional.linear(input, self.weight, self.bias)

I've just tried the inpainting, for the first time, to remove a "spare hand". It worked, but it also changed the overall image; the person's skin was smoothed out and lost detail, as well as other minor changes throughout the image.
These are the before (with the extra hand) and after (correct number of hands, but lacking the detail of the original) images:

I also tried an upscaled version, with slightly less denoise:

Perhaps I just misunderstand how inpainting works, but I expected it to only change the masked area and leave the rest intact.

#Prompt outputs failed validation
CheckpointLoaderSimple:

• Return type mismatch between linked nodes: ckpt_name, CHECKPOINT_NAME != ['Deliberate-inpainting.safetensors', 'analogMadness_v50.safetensors', 'clarity_-inpainting.safetensors', 'clarity_2.safetensors', 'cyberrealistic_v31-inpainting.safetensors', 'cyberrealistic_v31.safetensors', 'deliberate_v2.safetensors', 'dreamlikePhotoreal20_dreamlikePhotoreal20.safetensors', 'dreamshaperXL10_alpha2Xl10.safetensors', 'dreamshaper_4-inpainting.safetensors', 'dreamshaper_4BakedVae.safetensors', 'dreamshaper_7.safetensors', 'f222.ckpt', 'moreRealistic_v10.ckpt', 'protogenX34Photorealism_1.safetensors', 'protogenX53Photorealism_10.safetensors', 'realisticVisionV13_v13.safetensors', 'rundiffusionXL_beta.safetensors', 'sd_xl_base_1.0.safetensors', 'sd_xl_refiner_1.0.safetensors', 'talmendoxlSDXL_v11Beta.safetensors', 'v1-5-pruned-emaonly.ckpt', 'v2-1_768-ema-pruned.safetensors']
  VAELoader:
• Return type mismatch between linked nodes: vae_name, VAE_NAME != ['sdxl_vae.safetensors', 'vae-ft-mse-840000-ema-pruned.ckpt', 'vae-ft-mse-840000-ema-pruned.safetensors']
  CheckpointLoaderSimple:
• Return type mismatch between linked nodes: ckpt_name, CHECKPOINT_NAME != ['Deliberate-inpainting.safetensors', 'analogMadness_v50.safetensors', 'clarity_-inpainting.safetensors', 'clarity_2.safetensors', 'cyberrealistic_v31-inpainting.safetensors', 'cyberrealistic_v31.safetensors', 'deliberate_v2.safetensors', 'dreamlikePhotoreal20_dreamlikePhotoreal20.safetensors', 'dreamshaperXL10_alpha2Xl10.safetensors', 'dreamshaper_4-inpainting.safetensors', 'dreamshaper_4BakedVae.safetensors', 'dreamshaper_7.safetensors', 'f222.ckpt', 'moreRealistic_v10.ckpt', 'protogenX34Photorealism_1.safetensors', 'protogenX53Photorealism_10.safetensors', 'realisticVisionV13_v13.safetensors', 'rundiffusionXL_beta.safetensors', 'sd_xl_base_1.0.safetensors', 'sd_xl_refiner_1.0.safetensors', 'talmendoxlSDXL_v11Beta.safetensors', 'v1-5-pruned-emaonly.ckpt', 'v2-1_768-ema-pruned.safetensors']
  UpscaleModelLoader:
• Return type mismatch between linked nodes: model_name, UPSCALER_NAME != ['4x-UltraSharp.pth', '4x_NMKD-Siax_200k.pth', 'ESRGAN_4x.pth', 'RealESRGAN_x4plus.pth', 'SwinIR_4x.pth']
  UpscaleModelLoader:
• Return type mismatch between linked nodes: model_name, UPSCALER_NAME != ['4x-UltraSharp.pth', '4x_NMKD-Siax_200k.pth', 'ESRGAN_4x.pth', 'RealESRGAN_x4plus.pth', 'SwinIR_4x.pth']

I tried updating from 3.0 to 3.2 with git pull or overwrite existing files, both successfuly, but after restarting comfy its still 3.0 version.

The extension fails to import since 8f59149. Getting this

Traceback (most recent call last):
  File "/sd/ComfyUI/nodes.py", line 1647, in load_custom_node
    module_spec.loader.exec_module(module)
  File "<frozen importlib._bootstrap_external>", line 855, in exec_module
  File "<frozen importlib._bootstrap>", line 228, in _call_with_frames_removed
  File "/sd/ComfyUI/custom_nodes/SeargeSDXL/__init__.py", line 1, in <module>
    from .searge_sdxl_sampler_node import NODE_CLASS_MAPPINGS, NODE_DISPLAY_NAME_MAPPINGS
  File "/sd/ComfyUI/custom_nodes/SeargeSDXL/searge_sdxl_sampler_node.py", line 615
    match save_to:
          ^
SyntaxError: invalid syntax

Cannot import /sd/ComfyUI/custom_nodes/SeargeSDXL module for custom nodes: invalid syntax (searge_sdxl_sampler_node.py, line 615)

I'm running ComfyUI in a docker container with Python 3.9.5 on an amd card if that helps.

Running text2image using the provided workflow throws this error.

Stacktrace:

DDIM Sampler: 100%|██████████████████████████████████████████████████████████████████████| 6/6 [00:01<00:00,  3.42it/s]
!!! Exception during processing !!!
Traceback (most recent call last):
  File "C:\Users\sutir\comfyui\ComfyUI\execution.py", line 144, in recursive_execute
    output_data, output_ui = get_output_data(obj, input_data_all)
  File "C:\Users\sutir\comfyui\ComfyUI\execution.py", line 74, in get_output_data
    return_values = map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True)
  File "C:\Users\sutir\comfyui\ComfyUI\execution.py", line 67, in map_node_over_list
    results.append(getattr(obj, func)(**slice_dict(input_data_all, i)))
  File "C:\Users\sutir\comfyui\ComfyUI\custom_nodes\SeargeSDXL\searge_sdxl_sampler_node.py", line 631, in save_images
    if not args.disable_metadata:
AttributeError: 'Namespace' object has no attribute 'disable_metadata'

I followed the steps to install and restart the UI, but the interface was empty. The two JSON files in the manual load example would get stuck. Why is it that my comfyui was not installed properly?

I'm getting this message on SDXL Sampler v2

making attention of type 'vanilla-xformers' with 512 in_channels
building MemoryEfficientAttnBlock with 512 in_channels...
Working with z of shape (1, 4, 32, 32) = 4096 dimensions.
making attention of type 'vanilla-xformers' with 512 in_channels
building MemoryEfficientAttnBlock with 512 in_channels...
missing {'cond_stage_model.clip_g.transformer.text_model.embeddings.position_ids'}
torch.Size([1, 1280]) 4096 4096 0 0 4096 4096
torch.Size([1, 1280]) 4096 4096 0 0 4096 4096
100%|██████████████████████████████████████████████████████████████████████████████████| 16/16 [00:13<00:00, 1.22it/s]
torch.Size([1, 1280]) 2048 2048 0 0 6.0
torch.Size([1, 1280]) 2048 2048 0 0 2.5
100%|████████████████████████████████████████████████████████████████████████████████████| 4/4 [00:03<00:00, 1.03it/s]
Warning torch.load doesn't support weights_only on this pytorch version, loading unsafely.
Prompt executed in 70.55 seconds

I would like to make a grid of x/y samplers and schedulers. in node there is a possibility to convert a widget to input, but how to give there the list of samplers and schedulers. can someone tell me this point?

Not sure if I've done something wrong, but my main prompt isn't being used at all with 'Failed to validate prompt for output 115:'. I'm using the reborn workflow. Do you have any recommendations on what the problem might be?

Hi great job! Is it possible it integrate the loopback wave script I've been using in auto1111. I've found comfyui better performance for sdxl but lacking some off the ease of use that came with auto1111. Which you've obviously done a lot of work to make this easy to use with your work flow. Thanks for sharing.

Hi there,

Great work on the workflow! I enjoy using it immensely, thank you for creating and maintaining it.

I noticed that hires-fix doesn't work when using inpainting - it just outputs the same image. This looks like a bug because it works with text2image and image2image modes.

To work around this issue, I create my image with inpainting first and then I go through an image2image workflow with denoise set to 0 to get my hires-fixed version. However, this workaround will not occur to many users and shouldn't be needed.

Tangentially related to this, currently there is no hires-fix only operation_mode. I sometimes create multiple images with low steps to find something I like and then I copy its seed to work on it further and only then I'd like to go for hires-fix. Adding hires-fix only mode would be helpful for such a workflow.

Thanks.

I tried to load an image with mask, using VAE encoder(for inpainting), and then latent is output to SeargeSDXLSampler

It seems only base model gets the latent. when I increase the step percentage for refiner model, the image get blurry.

maybe I am doing something wrong. Can someone provide an example of inpainting using SeargeSDXLSampler

I tried the example workflows, and it kept failing because it couldn't find the lora. I converted the node's "lora_name" to a widget and tried browsing for my SDXL loras (they're all in a subdirectory) and none were shown. All loras not in the subdirectory are shown.

I've tried updating both with git pull (I originally installed with git) and then also by extracting the .zip

I keep getting this error when loading your workflows (specifically img2img and inpainting that I've tried)

When loading the graph, the following node types were not found:
SeargeSDXLImage2ImageSampler
Nodes that have failed to load will show as red on the graph.

Hi Searge.
Having issue at the end of the prompt rendering and the final images are always black.

So this is the entire log message :

E:...\ComfyUI_windows_portable>.\python_embeded\python.exe -s ComfyUI\main.py --windows-standalone-build --normalvram --fp16-vae
** ComfyUI start up time: 2023-08-12 19:40:13.776386
Prestartup times for custom nodes:
0.0 seconds: E:...\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Manager

Total VRAM 8192 MB, total RAM 98211 MB
xformers version: 0.0.20
Set vram state to: NORMAL_VRAM
Device: cuda:0 Quadro M4000 : native
Using xformers cross attention

Loading: ComfyUI-Manager (V0.21)

ComfyUI Revision: 1274 [2bc12d3d]

Import times for custom nodes:
0.0 seconds: E:...\ComfyUI_windows_portable\ComfyUI\custom_nodes\SeargeSDXL
0.5 seconds: E:...\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-Manager

Starting server

To see the GUI go to: http://127.0.0.1:8188
got prompt
Failed to validate prompt for output 364:

• ImageUpscaleWithModel 66:
  • Required input is missing: image
• ImageUpscaleWithModel 67:
  • Required input is missing: image
    Output will be ignored
    Failed to validate prompt for output 243:
    Output will be ignored
    Setting up MemoryEfficientCrossAttention. Query dim is 640, context_dim is None and using 10 heads.
    ....
    Setting up MemoryEfficientCrossAttention. Query dim is 640, context_dim is 2048 and using 10 heads.
    model_type EPS
    adm 2816
    making attention of type 'vanilla-xformers' with 512 in_channels
    building MemoryEfficientAttnBlock with 512 in_channels...
    Working with z of shape (1, 4, 32, 32) = 4096 dimensions.
    making attention of type 'vanilla-xformers' with 512 in_channels
    building MemoryEfficientAttnBlock with 512 in_channels...
    missing {'cond_stage_model.clip_g.transformer.text_model.embeddings.position_ids'}
    Setting up MemoryEfficientCrossAttention. Query dim is 768, context_dim is None and using 12 heads.
    .....
    Setting up MemoryEfficientCrossAttention. Query dim is 768, context_dim is 1280 and using 12 heads.
    model_type EPS
    adm 2560
    making attention of type 'vanilla-xformers' with 512 in_channels
    building MemoryEfficientAttnBlock with 512 in_channels...
    Working with z of shape (1, 4, 32, 32) = 4096 dimensions.
    making attention of type 'vanilla-xformers' with 512 in_channels
    building MemoryEfficientAttnBlock with 512 in_channels...
    missing {'cond_stage_model.clip_g.transformer.text_model.embeddings.position_ids'}
    making attention of type 'vanilla-xformers' with 512 in_channels
    building MemoryEfficientAttnBlock with 512 in_channels...
    Working with z of shape (1, 4, 32, 32) = 4096 dimensions.
    making attention of type 'vanilla-xformers' with 512 in_channels
    building MemoryEfficientAttnBlock with 512 in_channels...
    100%|#####################################################| 24/24 [04:36<00:00, 11.52s/it]
    100%|#####################################################| 6/6 [01:08<00:00, 11.40s/it]
    E:...\ComfyUI_windows_portable\ComfyUI\nodes.py:1269: RuntimeWarning: invalid value encountered in cast
    img = Image.fromarray(np.clip(i, 0, 255).astype(np.uint8))
    E:...\ComfyUI_windows_portable\ComfyUI\custom_nodes\SeargeSDXL\searge_sdxl_sampler_node.py:359: RuntimeWarning: invalid value encountered in cast
    img = Image.fromarray(np.clip(i, 0, 255).astype(np.uint8))
    E:...\ComfyUI_windows_portable\python_embeded\lib\site-packages\torch_utils.py:776: UserWarning: TypedStorage is deprecated. It will be removed in the future and UntypedStorage will be the only storage class. This should only matter to you if you are using storages directly. To access UntypedStorage directly, use tensor.untyped_storage() instead of tensor.storage()
    return self.fget.get(instance, owner)()
    Prompt executed in 423.31 seconds

End of log.

I'm using in ops mod : txt2img, prompt style : simple with default settings and all models are loaded correctly.
Thanks in advance for your response.

Any possible reason for this not showing up on Comfy? Installed it correctly and updated https://i.imgur.com/PpXRMeE.png Comfy working ok. but no Searge. tia https://i.imgur.com/wQe4R5E.png never mind figured it out load workflow

Just a small heads-up to anyone struggling with this, I can't remember if I loaded 3.3 or 3.4 but it was one of them. For me having followed the instructions when trying to generate the default image I got an error talking about 'image' and 'sdxl_vae.safetensors'.

It turns out that the main page instructions has you downloading:

https://huggingface.co/madebyollin/sdxl-vae-fp16-fix/resolve/main/sdxl_vae.safetensors

After you have put this in your models/vae folder you need to rename or copy it to:

sdxl-vae.safetensors

The default image of the cat now generates correctly. Issue was a dash needed rather than an underscore. I guess this will be fixed in the next version but just posting in case anyone today is having the issue.

Right now the node is hard-coded to always refine. It would be helpful if a toggle could be added to switch the refiner on/off by the user. This way you could do your testing quickly on the base model and then switch on the refiner for the final gen.

Hi first go with the workflow trying to run jusst txt2img and simple, getting this error at the min, ive got extra model paths .yamal for comfyui. ive tried copying the upscaler models into the defualt comfyui directory but still getting this error am i doing something wrong maybe.
If i add the missing link for the upscaling section it runs sucessfully. but im vetting black output images

Error occurred when executing ImageUpscaleWithModel:

ImageUpscaleWithModel.upscale() missing 1 required positional argument: 'image'

File "E:\ComfyUI_windows_portable_nvidia_cu118_or_cpu\ComfyUI_windows_portable\ComfyUI\execution.py", line 151, in recursive_execute
output_data, output_ui = get_output_data(obj, input_data_all)
File "E:\ComfyUI_windows_portable_nvidia_cu118_or_cpu\ComfyUI_windows_portable\ComfyUI\execution.py", line 81, in get_output_data
return_values = map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True)
File "E:\ComfyUI_windows_portable_nvidia_cu118_or_cpu\ComfyUI_windows_portable\ComfyUI\execution.py", line 74, in map_node_over_list
results.append(getattr(obj, func)(**slice_dict(input_data_all, i)))

The repo has an example workflow json file that uses a missing node. Could you include the node, or remove it from the workflow if it's not required, please?

Can someone help me to find the node containing imageupscalewith model?
i am using v3.2

This image was created with commit 5e18a21 from Jul 28, 2023
https://storage.googleapis.com/comfy-images/iron-man-lava-suit-5e18a21aa6ac80d3eb1fcc28b19cf34273ed71dd.png

This image was created from commit 5a61f08
https://storage.googleapis.com/comfy-images/Iron-man-lava-suit-5a61f08e464c54aead5bf920dc4b6d96af25b1c4.png

These images should have the metadata attached.

Something has changed. I wanted you to be aware of the issue as I have to stay on commit 5e18a21 until this is fixed because I've created a whole bunch of tutorials that need the images to recreate.

Error occurred when executing SeargeImageSave:

'Namespace' object has no attribute 'disable_metadata'

File "X:\Games\ComfyUI_windows_portable_nvidia_or_cpu\ComfyUI_windows_portable\ComfyUI\execution.py", line 144, in recursive_execute
output_data, output_ui = get_output_data(obj, input_data_all)
File "X:\Games\ComfyUI_windows_portable_nvidia_or_cpu\ComfyUI_windows_portable\ComfyUI\execution.py", line 74, in get_output_data
return_values = map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True)
File "X:\Games\ComfyUI_windows_portable_nvidia_or_cpu\ComfyUI_windows_portable\ComfyUI\execution.py", line 67, in map_node_over_list
results.append(getattr(obj, func)(**slice_dict(input_data_all, i)))
File "X:\Games\ComfyUI_windows_portable_nvidia_or_cpu\ComfyUI_windows_portable\ComfyUI\custom_nodes\SeargeSDXL\searge_sdxl_sampler_node.py", line 361, in save_images
if args.disable_metadata is None or not args.disable_metadata:

when opening either the image or json file in ComfyUI the nodes to the left of the prompts dont appear, I tried refreshing and it doesnt work. Inpainting works and I tried to do it that way for img2img but Im a complete noob at nodes so it wouldnt work for me to do an img2img through the inpainting workflow.

Latest git pull against this project indicated that it had failed to automatically identify my eMail address. Is this normal behaviour ?
Other Git pulls on different projects are not reporting the same alert.

Error occurred when executing SeargeSDXLSampler:

sigma_min and sigma_max must not be 0

File "F:\ComfyUI\ComfyUI_windows_portable\ComfyUI\execution.py", line 145, in recursive_execute
output_data, output_ui = get_output_data(obj, input_data_all)
File "F:\ComfyUI\ComfyUI_windows_portable\ComfyUI\execution.py", line 75, in get_output_data
return_values = map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True)
File "F:\ComfyUI\ComfyUI_windows_portable\ComfyUI\execution.py", line 68, in map_node_over_list
results.append(getattr(obj, func)(**slice_dict(input_data_all, i)))
File "F:\ComfyUI\ComfyUI_windows_portable\ComfyUI\custom_nodes\SeargeSDXL\searge_sdxl_sampler_node.py", line 107, in sample
base_result = nodes.common_ksampler(base_model, noise_seed, steps, cfg, sampler_name, scheduler, base_positive, base_negative, input_latent, denoise=denoise, disable_noise=False, start_step=start_at_step, last_step=base_steps, force_full_denoise=True)
File "F:\ComfyUI\ComfyUI_windows_portable\ComfyUI\nodes.py", line 1052, in common_ksampler
samples = comfy.sample.sample(model, noise, steps, cfg, sampler_name, scheduler, positive, negative, latent_image,
File "F:\ComfyUI\ComfyUI_windows_portable\ComfyUI\comfy\sample.py", line 88, in sample
samples = sampler.sample(noise, positive_copy, negative_copy, cfg=cfg, latent_image=latent_image, start_step=start_step, last_step=last_step, force_full_denoise=force_full_denoise, denoise_mask=noise_mask, sigmas=sigmas, callback=callback, disable_pbar=disable_pbar, seed=seed)
File "F:\ComfyUI\ComfyUI_windows_portable\ComfyUI\comfy\samplers.py", line 669, in sample
samples = k_diffusion_sampling.sample_dpm_adaptive(self.model_k, noise, sigma_min, sigmas[0], extra_args=extra_args, callback=k_callback, disable=disable_pbar)
File "F:\ComfyUI\ComfyUI_windows_portable\python_embeded\lib\site-packages\torch\utils_contextlib.py", line 115, in decorate_context
return func(*args, **kwargs)
File "F:\ComfyUI\ComfyUI_windows_portable\ComfyUI\comfy\k_diffusion\sampling.py", line 507, in sample_dpm_adaptive
raise ValueError('sigma_min and sigma_max must not be 0')

In your latest reborn workflow, the switch for upscaling is visually confusing, because it looks like it is enabled when it isn't.

May I suggest changing the orientation, so it is visually more obvious?

When trying the 3.4 version it always let to this issue, doesn't matter what image, or when convert to mask/image

Failed to validate prompt for output 243:

• ImageUpscaleWithModel 66:
  • Required input is missing: image
• ImageUpscaleWithModel 67:
  • Required input is missing: image
    Output will be ignored
    Failed to validate prompt for output 364:
    Output will be ignored

As presented on Huggingface ControlNet SDXL, it's available now in ComfyUI. Would you like to add this feature?

I noted this with 3.0 and spent some time fixing it last night. Should have reported it here, but did so on reddit.
The fix that was working for me is attached and was working, prior to downloading 3.1 and running into another issue.

The change is an import of torch, and then to modify SeargeLatentMuxer3 to pass in the batch_size and then, for inpaint and img2img modes only, to use torch.repeat to fill out the latents to the batch size.

searge_sdxl_sampler_node.zip

sorry i realized its wrong reported issue , please delete my issue.

Can we also have black and white here to upload mask images from another source. (Both Black and white since it is easier to swap channels and experiment in PS and GIMP and bring it here quickly as masks)

Example -

Hello, this is an awesome workflow setup but I am hoping that it would be possible along with existing options to add the new control-loras in this user friendly workflow:
https://huggingface.co/stabilityai/control-lora

(I know this is probably not an easy thing to do and you already have a massive to-do list but I just thought I'd ask)

What the hell i feel like total noob, i did not changed anything at all just input my own prompt into main prompt window, is there a video on how to use it for various txt2img img2img inpaint modes cause theyre really a must , this node setup is damn scary and big
No clue how to go to img2img
No clue how to go to inpaint
No clue why theres not a single info about it next to install instructions , i mean explain it like were 5 years old, this might be comfortable for you but not for newcomers, i do know how nodes work but i have no clue what to do anways, and on top of that it ignores my prompts

Ok looks like most recent json file is borked
I got older one and it works , still i would like to know how to use img2img or inpaint , that info should be on mainpage

Would greatly appreciate help in resolving this!

MetadataIncompleteBuffer error when using SDXL with Google Colab (A100 and T4)

Prompt mode: Simple, inpainting

Models used:
https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/resolve/main/sd_xl_base_1.0_0.9vae.safetensors
https://huggingface.co/stabilityai/stable-diffusion-xl-refiner-1.0/resolve/main/sd_xl_refiner_1.0_0.9vae.safetensors
https://huggingface.co/madebyollin/sdxl-vae-fp16-fix/resolve/main/sdxl_vae.safetensors
https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/resolve/main/sd_xl_offset_example-lora_1.0.safetensors (strength was put to 0 so should be off)
https://huggingface.co/sberbank-ai/Real-ESRGAN/resolve/main/RealESRGAN_x4.pth (though not connected)

Workflow (based on SeargeSDXL 3.4):
workflow.zip

Colab Link:
https://colab.research.google.com/drive/1alJNsalf87i5W8QV43TofT1Bnu505yde?usp=sharing

Error Message:
Traceback (most recent call last):
File "/content/drive/MyDrive/ComfyUI/execution.py", line 151, in recursive_execute
output_data, output_ui = get_output_data(obj, input_data_all)
File "/content/drive/MyDrive/ComfyUI/execution.py", line 81, in get_output_data
return_values = map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True)
File "/content/drive/MyDrive/ComfyUI/execution.py", line 74, in map_node_over_list
results.append(getattr(obj, func)(**slice_dict(input_data_all, i)))
File "/content/drive/MyDrive/ComfyUI/custom_nodes/SeargeSDXL/searge_sdxl_sampler_node.py", line 416, in load_vae
return self.vae_loader.load_vae(vae_name)
File "/content/drive/MyDrive/ComfyUI/nodes.py", line 557, in load_vae
vae = comfy.sd.VAE(ckpt_path=vae_path)
File "/content/drive/MyDrive/ComfyUI/comfy/sd.py", line 576, in init
sd = utils.load_torch_file(ckpt_path)
File "/content/drive/MyDrive/ComfyUI/comfy/utils.py", line 11, in load_torch_file
sd = safetensors.torch.load_file(ckpt, device=device.type)
File "/usr/local/lib/python3.10/dist-packages/safetensors/torch.py", line 309, in load_file
with safe_open(filename, framework="pt", device=device) as f:
safetensors_rust.SafetensorError: Error while deserializing header: MetadataIncompleteBuffer

Thank you very much!

After install SeargeSDXL ,my search menu like this. Unistall SeargeSDXL the search menu fixed.

I tried to link output condition of controlnet with input base_positive of SeargeSDXLSampler, but got error.

`Prompt outputs failed validation
SeargeSDXLSampler:

• Required input is missing: sampler_name
• Required input is missing: scheduler
  SeargeSDXLImage2ImageSampler:
• Required input is missing: sampler_name
• Required input is missing: scheduler`
  I look at the sampler box and there is nothing its just a green box with no input field nothing to change and all I did was launch your json

normal node image_save can use prefix about say as here https://blenderneko.github.io/ComfyUI-docs/Interface/SaveFileFormatting/

can use value from variable %node.widget% , your version of this node has more customization, but lost old functionality.