SmoothMix Animations Minimax H3
"Epic stuff or lewd stuff... make your choice!"
This LoRA is packed with content. Here’s a general idea of whats inside:
Poses & movement with bounce Dances, running, climbing, stripping, crawling, and more. I tried to include as much variety as possible. Everything has that extra little bounce. haha
Wide clothing variety Multiple sexy takes on gothic, Egyptian, Greek, armor, bikinis, uniforms, and more. All with rich extra details.
Fantasy races Easily generate fox, cat, oni, demon, and elf characters (both male and female). Note: Fox and cat versions are humanoid, not furry. Oni have pointed horns on the forehead and no tails, unlike the demons.
Appearance variety Lots of hairstyles, body types, and skin tones for both men and women.
About NSFW content
No XXX animations were used during training. However, images of various poses were included. The goal is to provide stronger support for animation LoRAs that use those poses — especially with I2V — and for the upcoming SmoothMix XXX Animations LoRA for MinimaxH3 I plan to make.
On the showcase there is little bit of whats possible in terms of visual style and generation: I has T2V, I2V using only the First Frame, I2V using First to Last Frame and I2V using only the Last Frame.
I hope it gives some cool ideas for your generations.
Trigger Words
The Lora uses the following trigger words for visual style:
'SmoothMixAnime' and 'Anime Style' - Will make the video have the style of SmoothMix Illustrious/NoobAI/Illustrious2+NoobAI, I recommend using both;
'SmoothMixRealism' and 'Realistic Style' - Will make the video have the style of SmoothMix Realism, I recommend using both.
ComfyUI Workflow Settings
All videos on all Showcases used this setup:
->Workflow: [
->Model: [minimax_h3_fl2va_pruned_int8_convrot.safetensors]
->Turbo Lora: [minimax_h3_fl2v_turbo_8step_v1.0_768p_comfyui_bf16]
With SmoothMix MinimaxH3 set to 1.0 weight.
Img2Vid and Txt2Vid
Its primary usage is for T2V but works very well on I2V. Minimax H3 has an enormous library of data in it so SmoothMix works great in it regardless of what you use it for. ^^
Limitations
Unfortunately SmoothMix Minimax H3 cannot FULLY uncensor, or rather, provide a high quality uncensorship yet:
The penis heads: I'll be dammed. How many fucking videos and images of penis Minimax H3 need to fully generate one?? Well, despite the huge amount of penis on the dataset the Lora CAN'T generate the head of the penis correctly most of the time. It can still look good in most angles but its can look terrible most of the times. >.<
Anus: This one hurts. A lot. The Lora seems unable to properly render the anus. I can guarantee that there are MORE THAN ENOUGH pics and videos of assholes in the dataset and yet it was not enough. This sucks. T_T
I guess I need to make either a v2.0 or an equivalent of my Futas and Males Animations model for WAN 2.2. Dammit! x_x
As always this is only possible with your support! =)
So thank you for all your support!
Thats all for now!
Have fun! =)