Hi I wanted to ask the following: 1.) Is this LoRA inpainting process better than Fooocus or BrushNet Inpainting? 2.) Is it okay to stack LoRA's using Efficient Loader SDXL to inpaint with this Inpainting process, Fooocus Inpainting, and BrushNet Inpainting. Thank you very much.

i downloaded the workflow and no changes there, but I am getting 60 images that are same, what could be the reason?

Sorry to say that, but given the outcomes you got in the video, I don’t feel an urge to make the change. Tile upscale in SDXL is still very primitive, and I’d rather to use Ultimate SD instead, which normally gives me much stable and better results, under both A1111 and Comfyui. Actually for the image generated from SDXL, I always use high res fix with an upscale model like 4xultrasharp, it works great in terms of keeping the original details. However, for upscaling a random images using SDXL, I would prefer SUPIR as it gives me more clean and creative results. One alternative way of refine the images generated from either SD 1.5, SDXL or MJ, is using SD3 with a low denoising without a control net. I find the results are much better than the original version. But be aware it may introduce too much details, and some of them we may not want.

Nice video, I used to use Supir quite a lot, but then I moved onto using Mcboaty upscaler/refiner it uses tiling (can even edit the prompt/denoise per tile) and can use controlnet models with it as well, it's from marascott nodes.

after the generation my storage decreased from 50 gb to 30 gb , do you know why?

Error occurred when executing RIFE VFI: RIFE_VFI.vfi() missing 1 required positional argument: 'frames' File "H:\MachineLearning\ComfyUI_windows_portable\ComfyUI\execution.py", line 152, in recursive_execute output_data, output_ui = get_output_data(obj, input_data_all) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "H:\MachineLearning\ComfyUI_windows_portable\ComfyUI\execution.py", line 82, in get_output_data return_values = map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "H:\MachineLearning\ComfyUI_windows_portable\ComfyUI\execution.py", line 75, in map_node_over_list results.append(getattr(obj, func)(**slice_dict(input_data_all, i))) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Error occurred when executing ImageScaleBy: ImageScaleBy.upscale() missing 1 required positional argument: fimage' File "C:\Users\akash\Documents\ComfyUI_windows_portable\ComfyUI\execution.py", line 151, in

class

where can i download depth sdxl.safetensors??

OK

Can you provide the workflow in the intro?

I got this error what i do Error occurred when executing LoraLoader: 'NoneType' object has no attribute 'lower' File "C:\Users\akash\Documents\ComfyUI_windows_portable\ComfyUI\execution.py", line 151, in recursive_execute output_data, output_ui = get_output_data(obj, input_data_all) File "C:\Users\akash\Documents\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 "C:\Users\akash\Documents\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 "C:\Users\akash\Documents\ComfyUI_windows_portable\ComfyUI odes.py", 11, in load lora

where can i find a lastest version , i saw in ur subcribe but it not lastest version <3

Hi sir i need help about comfyui😊

I've also had bad results with the old SDXL canny. I've always wondered why it's worse than the SD 1.5 canny. Good to know that the Canny in the Controlnet Union model doesn't have such problems. Thanks for the demonstration!

Thank you for this, another great video. I like how you go over the workflow, highlighting different steps. And no background music to disract me! You sound like a wise professor.

thx!

Great content and great workflow. Thanks for sharing.

It is amazing

I have a question to ask about the commercial use of comfyui workflow, is it possible to naturally put its products on the models? At present, I seem to know only clothes and shoes are relatively large products, but jewelry, such as earrings, bracelets, necklaces, etc. Although midjourney can be used to process the modified photos, the appearance of the product cannot be controlled, but the appearance control of very small objects in sd seems not to be strong, at least the thin chain will be difficult, I wonder if you have a solution? Thank you very much

how do i get depth_sdxl.safetensors for depth anything?

bit big diference actually

I feel like V2 actually has LESS details 🤔

Thank you, I use this workflow to generate a picture, everything seems good, but only the upload image didn't affect the color which I masked.

sdxl depth controlnet keeps giving error -> Error occurred when executing ACN_AdvancedControlNetApply: 'ControlNet' object has no attribute 'latent_format' Do you know anything about this, or can you please give the link to the exact dept and scribble files for ControlNet that you are using

comfy is so annoying. The developer really needs to make it more stable. Could not install this. And I have installed LLM's even dependencies for faceswap, dlib and anyone knows that isn't straightforward but this? No go... sigh. I give up and not reinstalling again.

Thank you very much for your information. For me it's quite surprising to have a more detailed depth map with V2, but more or less the same results. I guess canny or scribble is of help to overcome that lack of precision of depth map V1.

Amazing work🙌🙌🥳🔥

how to increase the number of frames?

Thank you very much for posting the workflow! for anybody curious, I ran CG Pixel's default workflow and prompt on an NVidia P40. Image generation was 25 seconds and video generation was 9 minutes and 11 seconds. I have a 3090 on the way lol.

Interesting video. Did you base this on the ipiv workflow? As only the upscaling seems to differ.

thanks a lot brother! this is work flow that I'm looking for, you are my superhero ! XD

Interesting comparison but it's a bit of an apple to oranges one as the fine tuned models have the benefit of a much greater data set and development. Not seen anyone compare it to SDXL Base yet which would be more of an accurate check. SD3's main issue that I can see is it appears to have quite a limited training data set as poses all look very similar etc. Really looking forward to seeing what the community do with it.

thank you so much

There is a annoying problem When i zoom out the fog on the moon disappears from my vision How can i increase the view, so the fog doesn't disappear? Please help me

Can we do all the upscale stuff in ComfyUI directly?

so great, thank you so much

Thank you for presenting this tool. Seems to be really interesting and could be quite helpful regarding compositing!

Your comparison between IC-light and IP-Adapter is really a good idea. I have the feeling that you have more control of the final result with IP-Adapter in selecting a base image. With IC-light you always have a quite heavy color shift. Is the mask still playing a role if you are using IP-Adapter?

Possible to get your light type images?

Great job

Thx sir❤

Merci, c'est nickel 👍

this is good.

thanks a lot. I respect you, sir!

hey, I really like this workflow and concept, but I can't seem to run it. I keep getting this error Error occurred when executing KSampler: 'ModuleList' object has no attribute '1' And in the console I see WARNING SHAPE MISMATCH diffusion_model.input_blocks.0.0.weight WEIGHT NOT MERGED torch.Size([320, 8, 3, 3]) != torch.Size([320, 4, 3, 3]) IC-Light: Merged with diffusion_model.input_blocks.0.0.weight channel changed from torch.Size([320, 4, 3, 3]) to [320, 8, 3, 3] !!! Exception during processing!!! 'ModuleList' object has no attribute '1' I didn't touch anything, and watched the IC light installation video before. I completely re-installed comfyui and installed only modules used in this workflow, and still I get this error... any ideas?

Could you help me on this case? Please An error happened while trying to locate the file on the Hub and we cannot find the requested files in the local cache. Please check your connection and try again or make sure your Internet connection is on. File "D:\AI\ComfyUI-aki-v1.3\execution.py", line 151, in recursive_execute output_data, output_ui = get_output_data(obj, input_data_all) File "D:\AI\ComfyUI-aki-v1.3\execution.py", line 81, in get_output_data return_values = map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True) File "D:\AI\ComfyUI-aki-v1.3\execution.py", line 74, in map_node_over_list results.append(getattr(obj, func)(**slice_dict(input_data_all, i))) File "D:\AI\ComfyUI-aki-v1.3\custom_nodes\comfyui_controlnet_aux ode_wrappers\depth_anything.py", line 19, in execute model = DepthAnythingDetector.from_pretrained(filename=ckpt_name).to(model_management.get_torch_device()) File "D:\AI\ComfyUI-aki-v1.3\custom_nodes\comfyui_controlnet_aux\src\controlnet_aux\depth_anything\__init__.py", line 40, in from_pretrained model_path = custom_hf_download(pretrained_model_or_path, filename, subfolder="checkpoints", repo_type="space") File "D:\AI\ComfyUI-aki-v1.3\custom_nodes\comfyui_controlnet_aux\src\controlnet_aux\util.py", line 324, in custom_hf_download model_path = hf_hub_download(repo_id=pretrained_model_or_path, File "", line 52, in hf_hub_download_wrapper_inner File "D:\AI\ComfyUI-aki-v1.3\python\lib\site-packages\huggingface_hub\utils\_validators.py", line 118, in _inner_fn return fn(*args, **kwargs) File "D:\AI\ComfyUI-aki-v1.3\python\lib\site-packages\huggingface_hub\file_download.py", line 1371, in hf_hub_download raise LocalEntryNotFoundError( I'm put the checkpoint file in D:\AI\sd-webui-aki-v4.5\models\Depth-Anything and Add the file link address in comfyui as: Add the file link address in comfyui as,then i'm put 3 pth files in D:\AI\sd-webui-aki-v4.5\extensions\sd-webui-controlnet\models and mark the same adress on comfyui yaml file

wonderful tutorials, could you please make a video work with people subjects? ( IClight and IPADAPTER ERROR with face and body) thanks

I can't find this vedio for 16:9 Version in your channel