Yiff Toolkit
On this page

Yiff Toolkit

Introduction

The yiff_toolkit is a comprehensive set of tools designed to enhance your creative process in the realm of furry art. From refining artist styles to generating unique characters, the Yiff Toolkit provides a range of tools to help you cum.

Toolkit Sections

Data Preparation and Management

SDXL Furry Bible

ResAdapter

ResAdapter [Paper] enhances a model’s ability to generate images outside their trained resolution domains. What does this mean? It means you can generate images that are higher resolution than 1024x1024. Which sounds great on paper, but your milage might vary, it will also help a bit with the consistency of your generations a bit, which is an even bigger boon!

Just don’t forget to turn it off every now and then so you can compare the results! Sometimes, you might get a better picture without it, especially with a fierce model like Pony, with that particular model I have so far only used v1 and it has been working out pretty good. It also works with ControlNet + IPAdapter which is just great!

CompassMix

The new kid on the block, based on SeaArt Furry with some GAN magic by Lodestone, this mix tried to improve upon the original model and succeeded! My recommendation for training a LoRA for usage with it is to not treat it as a regular mix. My experiments ([#1] [Dataset Reference], [#2] [Example Output]) show that LoRAs trained directly on Compass produces much better result, add the new compass_optimizer to your training script instead of AdamW while you are there, which will help with generalization and overall training, if you set it up correctly, (I can highly recommend the one with normalization using the default parameters I left in it’s __init__) and don’t forget to shrink it down a notch.

Pony Diffusion V6

Pony Diffusion V6 XL is a specialized Stable Diffusion XL model optimized for generating anthropomorphic and feral creature artwork, with particular expertise in the ‘My Little Pony’ aesthetic. Built on the SDXL architecture, it combines high-resolution output capabilities with precise character generation, supporting both SFW and NSFW content creation.

For detailed prompting tips and techniques, see the Pony Diffusion V6 Prompting Guide.

Related Posts

LoRA Training Guide

The LoRA Training Guide explains Low-Rank Adaptation (LoRA), a technique for fine-tuning large language and diffusion models efficiently by introducing small, trainable low-rank matrices instead of modifying all model parameters. This approach keeps the original model weights frozen and injects two additional matrices into each layer to learn necessary adjustments. LoRA is lightweight, making it feasible to train multiple adaptations without hefty storage requirements. The guide also compares LoRA with LyCORIS, an advanced extension that offers more control and flexibility, and introduces LoKr, which uses Kronecker products for matrix decomposition, enhancing memory efficiency and control over the adaptation process.

10 Minute LoRA Training for the Ultimate Degenerates

An offensive, unethical and biased guide on how to poop out awesome LoRAs with the bare minimal of effort and compute time.

NoobAI Notes

Comprehensive guide for training with NoobAI SDXL model, including optimal VRAM settings, training parameters, and prompting techniques with period-specific tags.