I have provided a link to the original release location on hugginface in every models version description
edit: in addition to that i just added that clicking on the names of the authors of these models leads to their hugging face account,

@eurotaku Perfect !

@AugmentedRealityCat i also included a thanks to all contributors in the description, the devs and testers especially did a great job the last weeks to get the extension support for all those models working.

@eurotaku I modified my message - basically changed my PLEASE into a THANKS.

As for support for ALL those models, sadly we are not there yet. There are many that will require some extra work, like the latest ones from SargeZT.

Some already work, like the Blur ones from Kohya, but would require some new pre-processor to be convenient to use. Right now you must use a third-party app (or an extension) to pre-blur the image.

But we will get there I'm sure !

@AugmentedRealityCat yeah, i uploaded all models i could find, whether they are already supported or not. i'm also very confident, they all will be eventually :)
btw, feel free to post any important news like e.g. new controlnet models here, as i'm not a regular visitor of huggingface/github.

Yes, waiting for this one!

Kohya blur is a decent substitute while you're waiting. Been using it for a couple days now. Usually end it at 0.85, and combine it with low strength canny, but I've managed full txt2img resamples at double resolution to increase detail. Img2img upscales are fairly reliable up to high denoising strengths. The tone of the image tends to shift more than a tile resample, but it's pretty good at preserving structure and detail.

@xnihil0zer0254 I really love the blur one, that is a great new addition to the CN! Didn't think of using it that way tough, what you are describing is essentially a tile - genius! I will def try it! Thanks!

@xnihil0zer0254 My apologies for not entirely understanding... Would you mind being a little more specific about how you go about using Kohya Blur as a replacement for Tile upscaling? Thank you

@97Buckeye As I understand it, it's not a replacement so much as a stand-in-for...

Lama can be run by itself, its just a kind of preprocessor so instead of img2img with pure noise or existing image it fills with a very neutral inpaint model before inpainting. You can run it alone in https://github.com/hhhzzyang/Comfyui_Lama . You can use diffuser inpaint SDXL model with it.

@Aerilyn235 thanx, man, appreciate it!