MiniMax-H3-Realism-People-LoRA

MiniMax H3 Realism People LoRA

Trigger word: r34l1sm

A LoRA adapter for MiniMax H3 specialized in realistic people: faces that hold up in close-up, natural skin texture, believable expressions and gestures, film-style lighting and documentary camera movement.

Same prompt, same seed — base model on the left, this adapter on the right:

<video controls src="https://v3b.fal.media/files/b/0aa5ce26/stD_vdWuX1F90Q2OihK1Q_learn_before_after.mp4"></video>

Before / after

19 pairs, same prompt, same seed, adapter on vs off — the only variable is the LoRA. The trigger word is present on both sides, so it is not doing the work. Each pair plays the base model first, then freezes and dims while the adapted version plays beside it.

<video controls src="https://huggingface.co/fal/MiniMax-H3-Realism-People-LoRA/resolve/main/before-after-comparison.mp4"></video>

Close-up talking faces, arguments, several people speaking at once, weathered skin, children, ritual and travel scenes. Nothing cherry-picked from a larger render batch: these are the pairs that were kept, in order. (download the comparison, 197s, 1920x1080)

What it does

MiniMax H3 is already a strong general video model. This adapter pushes it further on human-centered shots: portraits, faces, hands at work, crowds and everyday characters. Skin keeps its texture instead of smoothing out, eyes and micro-expressions stay coherent, light behaves like it does on a film set, and motion gains a subtle handheld quality. It keeps H3's native synchronized audio.

It is the successor of MiniMax-H3-Realism-LoRA, retrained on a larger dataset focused on people.

How to use

These are plain LoRA weights. Nothing here is tied to a hosted service — run them wherever you run MiniMax H3.

Locally, in ComfyUI

Download the .safetensors and drop it in models/loras/, then insert a Load LoRA node between your model loader and the sampler. No conversion step: the keys are the standard H3 layout (diffusion_model.blocks.N.attn.qkv_proj, fused QKV), the same one other working H3 LoRAs use, so ComfyUI loads it as-is.

wget https://huggingface.co/fal/MiniMax-H3-Realism-People-LoRA/resolve/main/h3-realism-people-t2v-i2v-r2v.safetensors

Start the prompt with the trigger word r34l1sm, then describe the scene. Scale 1.0 is the intended strength; 0.6-0.8 for a lighter touch. It works on text to video, image to video and reference to video, since the adapter only touches the shared attention projections.

In your own code

Any inference stack that can apply a LoRA to H3 will take it. The application is the usual W_eff = W + lora_B @ lora_A.

On a hosted endpoint

If you would rather not run it yourself, fal exposes the H3 LoRA endpoints — but this is one option among others, not a requirement:

• Text to video: fal.ai/models/minimax/h3/text-to-video/lora
• Image to video: fal.ai/models/minimax/h3/image-to-video/lora
• Reference to video: fal.ai/models/minimax/h3/reference-to-video/lora

{
  "prompt": "r34l1sm, a young woman faces the camera in a quiet apartment at dusk, soft window light on her skin, shallow depth of field, subtle handheld sway, cinematic, photorealistic",
  "loras": [
    {
      "path": "https://huggingface.co/fal/MiniMax-H3-Realism-People-LoRA/resolve/main/h3-realism-people-t2v-i2v-r2v.safetensors",
      "scale": 1.0
    }
  ],
  "duration": 5,
  "resolution": "768P"
}

Start the prompt with the trigger word r34l1sm, then describe the scene. A scale of 1.0 is the intended strength; lower it to 0.6-0.8 for a lighter touch.

Files

Direct link:

https://huggingface.co/fal/MiniMax-H3-Realism-People-LoRA/resolve/main/h3-realism-people-t2v-i2v-r2v.safetensors

Training

• Base model: MiniMax H3, trained with the fal H3 trainer.
• Dataset: 176 hand-curated live-action clips centered on people - portraits, faces, workers, athletes and everyday characters - building on the strongest shots from the first Realism dataset. Slow-motion footage was detected and retimed to natural speed, everything normalized to strict 24.000 fps with structured scene captions.
• Sixteen configurations (steps, rank, learning rate, training resolution) were trained and compared through side-by-side human review on same-seed prompt pairs — same prompt, same seed, adapter on vs off.
• The version published here is rank 32, 1500 steps, trained at the high resolution bucket. It was not the highest rank or the longest run that won: training resolution turned out to matter more than either. Skin texture, pores, fine hair and grain live in high spatial frequencies, and at a lower training bucket the latents barely carry them, so there is little for the adapter to learn.

Credits

Created by Lovis Odin at fal.

License

The adapter follows the MiniMax H3 Community License of the base model.