Downloads · 30 days
696
38% of all-time downloads
KennethFal/retro-toon-70s-lora-minimax-h3
retro-toon-70s-lora-minimax-h3 is a image-text-to-video model from KennethFal. Use it for the image-text-to-video task on the model card, and read the license before you ship it in a product. It is set up for minimax-h3. The card lists the license as other.
A style LoRA (rank 32) that renders MiniMax H3 footage as retro 1970s feature-film cel animation — hand-painted characters with bold ink lines over textured, painterly matte backgrounds, limited-animation motion, and…
Downloads · 30 days
696
38% of all-time downloads
All-time downloads
1.8K
Public
Repo size
137 MB
Likes
14
Public
Click a slice to open those files.
.safetensors131 MB · 95%
From the Hugging Face model README
A style LoRA (rank 32) that renders MiniMax H3 footage as retro 1970s feature-film cel animation — hand-painted characters with bold ink lines over textured, painterly matte backgrounds, limited-animation motion, and the grainy, earthy palette of the era. H3 trains audio jointly, so dialogue and ambience come through with the picture.
This repo hosts the weights for the fal endpoint, which is the supported way to use this style — the endpoint applies the tuned trigger phrasing and settings server-side:
minimax/h3-max/lora-gallery/retro-toon-70son fal.ai
<video controls playsinline width="720" src="https://huggingface.co/KennethFal/retro-toon-70s-lora-minimax-h3/resolve/main/samples/wizard-warning.mp4"></video>
Raw single take at 768P / 16:9, picture and voice generated together in one pass — no cherry-picking.
| file | what |
|---|---|
retro-toon-70s.safetensors | the LoRA, rank 32, 6,000 steps |
samples/wizard-warning.mp4 | the example clip above |
Trained with fal's MiniMax H3 t2v trainer (rank 32, 6,000 steps, 16:9, joint video+audio objective) on a small hand-curated set of ~270 three-second clips of vintage 1970s–80s hand-painted feature animation, each clip human-reviewed for style before inclusion. The 6,000-step rank-32 checkpoint was chosen over its rank-32 4k/5k and rank-64 5k siblings by rendering the same complete 30-second trailer with each candidate and comparing the real-world results.
Adapter weights for MiniMaxAI/MiniMax-H3, distributed under the MiniMax H3 Community License.