Topic: Which checkpoints are y'all using?

Posted under AI Art Tutorials

DRLa

Member

I've been using NovaFurryXL V4 since I started, because that's what I saw recommended somewhere once. It's been working fine for me, and I learned it's strengths and weaknesses. At this point I'm so used to it I gave up when trying V6 because I didn't feel like relearning the subtleties of the prompts.

Now I heard that Flux is the next gen stuff, but it seems it's mostly being used for realistic humans. So I thought I'd ask here: any suggestions for stylized furry/feral capable models?

drla said:
Any suggestions for stylized furry/feral capable models?

Depends on what you're doing with the model. If you're using art mediums (watercolor, oil painting, marker etc), using artists or artist mixes, or want more of just the stock look of what a particular model does. Honestly it's hard to go wrong with any noob/illustrious model that's out right now. I've personally been bouncing between 4 models (3wolfmond, bananasplitz, manticoretradtional, and personalmerge). 3wolfmond does good 2.5d gens baseline, highly detailed stuff going into but not quite realistic (unless you specifically prompt for it). Bananasplitz does toony looking gens by default, manticoretraditional has a high focus on traditional art mediums, and personalmerge has been my go to for a long time now especially when I'm using artist mixes. Although all 4 models can do artists/artist mixes no problem.

drla said:
I've been using NovaFurryXL V4 since I started, because that's what I saw recommended somewhere once. It's been working fine for me, and I learned it's strengths and weaknesses. At this point I'm so used to it I gave up when trying V6 because I didn't feel like relearning the subtleties of the prompts.

Now I heard that Flux is the next gen stuff, but it seems it's mostly being used for realistic humans. So I thought I'd ask here: any suggestions for stylized furry/feral capable models?

I mostly focus on using NovaFurryXL Illustrious. Playing around with v8b right now and really loving it.

I definitely find though that for the styles I want, LORAs are doing 90% of the work for aesthetic consistency.

Some of my AI art team friends are working in Chroma now and have good things to say about it. Others are using Perchance's free services. I really love Midjourney for SFW painterly stuff.

I'd say with modern hard drives being so generous of space, download a bunch of models and try them out, and compare evolutions of models!

Slop

Janitor

I personally use StableMondAI. This is a Noobai based merge and I've found it better at using styles/style Loras than most. Next on the list would be NoobaiCyberFix. Again, a noob merge I found more consistent then most.

I just pretend to use AI. In reality I have an artist chained up in my basement.

A lot of people say my pictures are full of soul. My secret? I bought a jar containing a trapped artist's soul on Etsy and inject a little into each one of my images.

Updated

I've been using ZoinksNoob for quite a while, but recently I switched to YiffyMix, it's an interesting model.

Have used Yiffymix for ages, although all the recent v6 models push humans like crazy and you have to really prompt them out of there. Certain species look off too, and at times I have to go back to the v5.4 model.

I'm using a custom checkpoint, which is just a fine-tune of Nova Furry XL Illustrious v6a. Planning to do a new fine-tune since v8a/b have been released on my next vacation. Might also take a look at Noob and Yiff in Hell then.