ComfyUI: Inpainting with Differential Diffusion

The long awaited follow up. Let's begin.

Workflow:

Inpainting with Differential Diffusion is a recent development allowing for more contextually sound inpainting in ComfyUI. This is built into ComfyUI but if you are interested in learning more about the concept the orginal repo is here:

Download Stable diffusion checkpoints from here:

Socials:

[Gabriecielo]

Very cool new function, I tried it with lightning models, also works.

[Artofdrendering]

Thank for the tutorial, really good workflow, keep going with this great stuff

[elowine]

Yeah another tutorial, thanks for sharing! I've run into these issues with inpainting before. Models or controlnets that really understand perspective and size would be a huge step forward in using AI in existing images.

[ShahidShrivastava]

Awesome work bro... Ashutosh this side 👌👌

[nkofr]

nice. definitely interrested in how it works (diff diff)

[zizhdizzabagus]

btw is there any normal way to make workflow pause, wait for mask and then just go on? often its just gona start to generate from the start when using this method of bypassing. i try to use preview chooser, but it doesnt work too, the gaussan blur is not happy with the mask provided

[rkwybzh]

Hi, i thought to write you maybe you can help me.
I want to ask you about some Errors i get with Comfyui. It has nothing to do with this video, but maybe you can help me:
1 . Working with Get Sigma ( from Comfyui Essentials) it shows this error :
Error occurred when executing BNK_GetSigma:
'SDXL' object has no attribute 'get model_object
2working with ReActor i get this:

Error occured when executing ReActorFaceSwap: This ORT build has ['TensorrtExecutionProvider', 'CUDAExecutionProvider', 'CPUExecutionProvider'] enabled. Since ORT 1.9, you are required to explicitly set the providers parametee when instantiating InferenceSession. For example, onnxrunyime.InferenceSession(..., providers=['TensorrtExecutionProvider', CUDAExecutionProvider', ' CPUExutionProvider'], ...)

[houseofcontent]

Do you have a good workflow for matching an object into an existing photo background?

[bloomp]

maybe you can do a tuto about doing this in Pinokio ? good job for the video!

[fillill-]

Thank you for the vid! Could you check the link with workflow, please, i can't find it there (:

[okolenmi]

It work nice even with drawings - it will create prompt in a very similar style (it depends of model sometimes).

[WhatsThisStickyStuff]

What node pack are you using for the gaussian blur mask?

[zizhdizzabagus]

completely doesnt work. have no idea whats wrong, have no errors, using your workflow, resulting image is just same as after upscaling. differential diff is not model dependant, right?

[spearcy]

The workflow link doesn't work. How did you install differential diffusion?

[ceesh]

Thanks a lot, you need better audio for youtube !

[harshitpruthi]

cant find the differntial diffusion node in manager

[WhatsThisStickyStuff]

Where is the comfy node repo? Cant seem to find it