If you like my work then drop a 5 review and hit the heart icon
Character commission is open on Patreon
Join my New Discord Server
This model has no VAE baked in so you don't need to use any vae with the baked version.
Suggested settings:
I had CLIP skip 2 on every image
I had ENSD: 31337 all of them
All of them had highres.fix or img2img at higher resolution.
I don't use restore faces
NOTE: if you find any prompt of the preview images look familiar it's because I've taken them from other model review images, Credits to the original authors. Thanks for the benchmark.
Description
FAQ
Comments (14)
Surprising how well this works at higher resolutions. Other anime/hentai models seem to tap out much earlier.
EDIT: I read the description without paying attention to the release name. Went ahead and applied a VAE eventhough this release has one baked in. Does it technically apply the VAE twice and frying the image? Update the description 馃
Guess I'll wait around for the no VAE version.
you do not have to wait
@Goofy_Ai聽Much appreciated!
really good model, almost all my stuff was made with this model after it came out
this + meina have the best toes/feet i could find as well
Thanks for your feedback 鉂わ笍
I'm new, can anyone tell me what folder should i put it in?
Models/stable-diffusion
pls go to https://spell.novelai.dev/ , upload the model file, it will tell u how to use this model.
the language of this website is Chinese, i dont know if theres an English one, but u can use chrome translation.
Dear Author
This model has a certain amount of peculiarities, but I thought it would be extremely difficult to create a work better than yours that can master these peculiarities.
How many times have I thought, "I don't think I can compete with your sample.
What are those peculiarities. more details would be nice
@Goofy_Ai聽[peculiarities] is a personal expression, which means it has characteristics, both good and bad.
And I guess the point to be said about this model would be "instability of the picture output".
I'm talking about the images posted by [https://civitai.com/bounties/77/entries/992], but as a result of using the TensorRT Extension, the patterns were completely different. I prefer the original different pattern (the first one I used for the thumbnail),
I couldn't use hires.fix, so I removed TensorRT and what came out was a picture that looked like an impressionist. I was surprised, but after a few adjustments, the picture solidified enough that I wasn't worried. (2nd picture)
I guess you could say that fixing the picture is an improvement, but I honestly think it's unnecessary. I think it's just another version at best.
TL;DR: It's not a bad model, but needs more to stand out.
I'd say Goofball Mix V2 is a serviceable checkpoint for anime-styled images. It makes decent anime faces, which are rather proportionate and rather symmetrical. The bodies are aesthetically pleasing. In terms of hands, it does its job on a very average level for SD 1.5 - the hands aren't grotesque, but aren't anatomically correct, either. I'd say that if the model would generate aesthetically pleasing hands more frequently, I would gladly ditch all other anime models. (Does anybody know an embedding or LoRA that can assist the model in the hands department?)
Goofball Mix does fairies with pointy elf-like ears, which I like. But it struggles with the wings - some have a single wing, instead of a pair. "There's a LoRA for that," I guess.
thanks for the feedback