Please consider supporting me so I can continue to make more Models / LoRA's for the community! --> Buy Me A Coffee
This model was trained on Furry Vixens v2.0
This model is mainly trained for Anthro's, but it does small Chibi's too. It's trained for both male / female Braixens, so it should do both fairly easy.
Be sure to use the tags "gynomorph, breasts" for generating females, and "andromorph, flat chest" when you want to generate male Braixens. The tag "penis" is interchangeable for both, and there's also some training for "bukkake, many dicks" if you want to have some group fun.
It works best at a weight of 0.7, though you can use 0.4 if you want to create more generic looking furries instead.
You can use these tags when generating images:
Braixen, Braixen ears, Braixen tail, red eyes
Be sure to use these settings!
Steps: 32
Sampler: DPM++ 2M Karras
CFG scale: 7
Size: 384x512
Model: Furry Vixens v2
Denoising strength: 0.6
Clip skip: 2
Hires upscale: 2
Hires upscaler: R-ESRGAN 4x+ Anime6B
Here's an example prompt, though you can find others in the gallery.
Positive Prompt
(Braixen, Braixen ears, Braixen tail), (gynomorph:1.1), breasts, head turn, blushing, open mouth, tongue out, smile, red eyes, autumn, forest, trees <lora:Braixen v3-000006:0.7>Negative prompt
(worst quality:1.4), (low quality:1.4), (easynegative:1.2)Description
Initial release.
Trained for 7 epochs on Furry Vixens v2.0
FAQ
Comments (6)
Wow 0_0 like actually wow this lora is absolutely 100% top! you did an amazing job with this one! :D
Thank you very much! It took a lot of effort to generate and paint over the dataset by hand, so I'm glad you're enjoying it! 😊
I have no idea what im doing wrong here.
DPM++ 2m karras, guidance scale at 7, steps are at 32, using the provided positive and negative prompts and yet half my attempts end up with pictures that look like deep fried photos of single colors. Clip skip is on.
Even if i somehow generate a picture it doesnt seem right.
Are these setting work the same way in easy diffusion?
I'd like to second this ^^
Most images are upscaled and inpainted afterwards with lots of batches to get closer and closer to your idea. I use controlnet to get more finer control over important things like pose, depth, etc. If you check my profile the Loona art I did was mostly done using this technique
Also I had the same issue when I had xformers or low vram options enabled