NoobAI-XL ControlNet
This is the ControlNet collection of the NoobAI-XL models.
How to use?
Version name is formatted as "<prediction_type>-<preprocessor_type>", where "<prediction_type>" is either "v" for "v prediction" or "eps" for "epsilon prediction", and "<preprocessor_type>" is the full name of the preprocessor.
You HAVE TO match the preprocessor type and this ControlNet model.
You MAY NEED TO match the prediction type of your base model and this ControlNet model.
Where is Openpose?
Find it here!
Description
FAQ
Comments (28)
What does eps-blur do? just blur the image?
was just wondering the same. and why use cnet for that?
Destroys details through blurring and reconstructs them.
@Euge_ I'm still confused about the use case. Is there a benefit of eps-blur that photoshop's blur don't have?
@TinySandwich The model is used to remove blur from input blurry image.
@Euge_ Ahh! I understand now. Thanks you! ❤️
和在tile模型里使用blur方法有什么区别吗?
can anyone help me how to rename the CN models please help me how to use and where to put, please explain in simple way and if there is any Youtube tutorial available please share
I'm not quite understanding something about the tile model. I set it in ControlNet, use the 'tile_resample' preprocessor, select Prefer ControlNet, and apply Ultimate SD Upscale with a denoise strength of 0.6... and I get something worse than if I had set it without ControlNet but with a denoise strength of 0.25. Could this be the influence of using an Illustrious-based model instead of a NoobAI-based one? (WEB UI reForge)
Yes, it may happen. And it also depends on parameters and a lot of things (that is AI).
I'm having the exact same issue.
I'm looking for a way to just add details, the same as "inpaint masked" is doing.
Did you find any solutions to your problem?
MetallicAchu Unfortunately, I just gave up. And in the end, I just don't use ControlNet.
i downloaded it but i cant enable it in a1111 under extensions it just doesnt appear. i put the .safetensors file into the right folder but still nothing.
The thing you call "a1111" is a very outdated UI
I tried this scribble pidinet cnet in krita AI but i don't know which works: black background with white strokes, or white background with black strokes ? It looks like both gives different results but seem to work so idk.
black background with white strokes
Thanks for these useful models. It would be great to have a controlnet union as with base SDXL, to save disk space and simplify things. Are you considering that option?
Why is openpose not here? If I use it, I cannot attribute your model in my images, that reduces the visibility of the model.
Thanks for your like :)
You may refer to here: NoobAI-XL ControlNet-OPENPOSE - openpose model | NoobAI Controlnet | Civitai
It's trained by another great trainer!
Why does the image I generated have a deep red background and messy white lines after adding this ControlNet model?
This happened to me as well :(
Do these models work for v-pred as well as eps, or are they just for eps models?
What is the difference between manga and anime ?
Basically, anime is animated manga.
@Pilanoku Pika! THAT ISN'T THE KIND OF DIFFERENCE THEY WANTED BAKA! =p It depends, but manga is usually more line centric with varying design differences. Anime really depends upon the quality of the source. Sometimes Manga will get you cleaner, more sophisticated designs. The real difference comes down to different artists/directors and their design aesthetics as well as production quality/cost. Like, if you want something on the level of Akira or its spiritual sequel, then be prepared to have a VERY limited selection [it comes down to the quality of the between images]. A LOT of the style differences between TV anime and film is the quality of the art/budget. Artists have more time to do more comprehensive work vs busting shit out ASAP. Hopefully, that helps? The best way to tell is to look at some of the examples and print. Unfortunately, you won't get much help otherwise. =P This is a trial by fire subject, I'm afraid...
Could you help me? I need to upload a picture and have a new image created based on the original reference. Is there some kind of ControlNet model for this? I think it's called Reference, but when I upload the original image and use the Reference tab, there's no window to load the model. The image doesn't change at all, and selecting a reference doesn't work.