IL v2.3 : New version, more flexibility. Better than IL v1
This version don't have a baked in VAE. I recommend to use SDXL VAE
Please consider supporting me so I can continue to make more models --> https://www.patreon.com/BB95FurryMix
Don't forget to join the Furry Diffusion discord server --> https://discord.gg/furrydiffusion
Since v3, this model uses e621 tags.
This model is a mix of various furry models.
It's doing well on generating photorealistic male and female anthro, SFW and NSFW.
I HIGHLY recommend to use Hires Fix to have better results
Below is an example prompt for the v7/v6/v5/v4/v3/v2.
Positive:
anthro (white wolf), male, adult, muscular, veiny muscles, shorts, tail, (realistic fur, detailed fur texture:1.2), detailed background, outside background, photorealistic, hyperrealistic, ultradetailed,
Negative:
I recommend to use deformityv6 and boring_e621_fluffyrock_v4, you can add bad-hands-v5 if you want
Settings 1.5 :
Steps: 30-150
Sampler: DPM++ 3M Karras, DDIM, UniPC or Euler A
CFG scale: 7-14
Size: from 512x512 to 750x750 (only v4/v5/v6/v7/v8/v9), up to 900x900 for v10/v11/v12
Denoising strength: 0.6
Clip skip: 1
Hires upscale: 2
Hires steps: 30-150
Hires upscaler: Latent (nearest)
Settings XL :
Steps: Below 30
Sampler: Euler A
CFG scale: 3-6
Size: 1024x1024
Denoising strength: 0.2
Clip skip: 1
Hires upscale: 2
Hires steps: Below 30
Hires upscaler: Remacri
Description
This version fix some mistakes v8 made, can generates at higher resolution than v8. I think this is the best version I made. More realistic, better fur, better clothes, better NSFW !
This version doesn't use VAE and doesn't need one.
FAQ
Comments (22)
this is very good. I was also rather impressed with v8
Thank you !
amazing model , thank you ❤️
You're welcome !
A new model? Oh, well... I didn't need that free time anyway xD
Can you use character tags by a certain artist? If so, how? How are tags used which include ( ) parentheses?
You have to write it like that : chris \(meesh\),
@Bumblebee95 Thank you.
@Bumblebee95 So I tried this and it does not seem to work. Tried it with multiple characters across multiple popular artists. With an empty prompt and only the character tag in the prompt, it gives me nothing related at all. Just a random image in the style of the artist.
@Minase460 It depends on how much the character have images on e621
Found out that you can achieve more realistic generations using the MJ eta noise seed delta (31337) and that when the model gets overwhelm by complicated and heavy prompts changing CLIP skip to 2 or in rare cases 3 helps with coherency.
Thanks for your experiment !
I will try that
what's an mj eta noise seed delta? where do you set that?
@angangang Here's where you can change the ENSD value
How to fix it so that the model does not draw real animals, but anthropomorphic ones? For a couple of days I've been trying to pick up tags and CFG Scale, but it doesn't work :(
You need to add ‘anthro’ tag in your prompt and add feral in negatives.
If you try to make some anthro horses it’s normal to have sometimes feral horses
1. Increase the sampling steps (samplers like UniPC like it when you increase it to around 60)
2. Increase the height
3. Remove unnecessary tags
4. Preferably use e621 tags
Adjusting Clip Skip to 2 can also help. But just make sure you put anthro very early on in the prompt.