V3.1 Uploaded a better epoch of the newly trained model, performs alot better and works on realistic furrys as well! Setting the MPS lora to around 0.7 seems to work best. Always experiment with strengths to achieve the best results!

This model requires the Wan fun 2.1 1.3b InP model to work correctly place the "diffusion_pytorch_model.safetensors" in your comfyUI diffusion_models folder. You can download it here: https://huggingface.co/alibaba-pai/Wan2.1-Fun-V1.1-1.3B-InP/tree/main just rename it to WanFunInpv1.1.safetensors for convenience

This is for image 2 video I used the wan fun 1.3b inp v1.1 model since it preforms the best at image 2 video I have not tested inpainting

Set the furry titfuck motion enhancement Lora at a 0.01 strength and the MPS reward Lora at a 1.0 strength. These combined create the best quality generations with the most consistent motion and facial animations!

The MPS reward Lora significantly enhances prompt following and can be used on other wan models and loras as well!

You need the furry titfuck motion enhancement Lora you can get it here: https://civarchive.com/models/1698933?modelVersionId=1922737 and the Wan2.1-Fun-1.3B-InP-MPS.safetensors Lora you can get it here: https://huggingface.co/alibaba-pai/Wan2.1-Fun-Reward-LoRAs/tree/main

I strongly recommend downloading this node pack (all though not required) https://github.com/ClownsharkBatwing/RES4LYF to get the best possible generations with the Wan 2.1 models it really enhances the quality of generations! Make sure your ComfyUI is updated. Workflow is in the training data folder only the workflow is in there civit just renames the workflow to that for whatever reason.

This was trained to get as close as possible to wan 14b's quality and does very well at this! This allows those who do not have 24gb of VRAM to reliably generate longer 81+ frame videos. A 480x480 81 frame video takes 140 seconds on a 3090 at 8 steps using qin_zhang_2s sampler with the beta57 scheduler. These are 2 custom sampler/schedulers in the recommended comfyUI nodes I linked they MASSIVLY improve the consistency and quality of all wan model generations including the 14b as well while being quicker than using other traditional samplers. res_2s with 20-30 steps and the beta57 scheduler also produces great results!

Works on 8gb of VRAM without any issues!

Made for i2v T2V is untested

The generation of Eipril's OC Elizabeth is about as close to "fan art" as I can make lol

Prompting example remember to experiment for the best results!

A lifelike 2.5D animation depicts a scene from a first-person perspective. A cute nude furry cat woman with white and yellow fur and ginger hair is positioned between a man’s legs giving him a titfuck, her large, elastic breasts fully enclosing his erect penis. (Her hands guide her breasts in rapid continuous, bouncing, fluid up and down motions going the entire length of his penis, their soft curves stretching and contracting with each movement causing the soft tissue of her breasts to ripple and oscillate from the impact like gelatin without stopping:1.2). The warm sunlight of the forest casts gentle shadows through the room that trace the contours of her fur and skin, emphasizing the texture of her white fur and the smoothness of her skin. Her breasts move with rhythmic, (jiggle physics:1.1), their bouncing creating a dynamic interplay with the man’s body. She has a loving expression on her face showing arousal. The background is softly blurred, its details fading into a hazy glow, drawing focus to the intimate interaction between them. It has detailed 2.5D depth. The scene is rendered with a high frame rate, capturing every nuance of her movement as her fur and skin shimmer under the soft light, their textures blending with the motions to create a lifelike, seamless flow.

Negative prompt example:

Unclear and blurry images, overexposure, static, blurred details, subtitles, paintings, pictures, still, overall gray, worst quality, low quality, jpeg compression residue, ugly, mutilated, redundant fingers, poorly painted hands, poorly painted faces, deformed, disfigured, deformed limbs, fused fingers, cluttered background, three legs, upside down, white flashlights, glow light effect, post edit, ugly distorted hands, scrambled fingers, wild arm movement, text, icons, logo, post production, a lot of people in the background, motion blur, camera movement, stiff, stuttering, jerky, choppy, impossible anatomy, freezing, frozen frame, poor scene composition, talking, amateur animation, weightless, floaty, no gravity, weak impact, no collision, mechanical, passing through objects, clipping, intersecting

All characters are 18+. This is intended for mature audiences only you and you alone are responsible for what you generate. 18+ ONLY!