README.md

metadata

title: MiniMax H3 Prompt Enhancer
emoji: 🎬
colorFrom: purple
colorTo: pink
sdk: gradio
sdk_version: 5.49.1
app_file: app.py
python_version: '3.10'
startup_duration_timeout: 1h
preload_from_hub:
  - geocine/minimax-video-prompt-enhancer-350m-gguf
pinned: false
license: other
short_description: MiniMax H3 video prompts via CPU GGUF
tags:
  - text-generation
  - video
  - prompt
  - minimax
  - liquid
  - lfm2
  - llama.cpp
  - cpu

MiniMax H3 Prompt Enhancer

Not another “make my prompt longer” toy.

This Space runs a purpose-trained small language model that rewrites rough ideas into production-ready MiniMax H3 video prompts — the same kind of structured audiovisual brief MiniMax expects: shots, camera language, diegetic sound, and score, not a bag of adjectives.

Type something half-formed like “baker opens a shop at dawn”. The model doesn’t slap on “cinematic, 8k, masterpiece.” It reasons about scene, motion, and audio, then emits the exact field layout used for MiniMax video generation.

Want more creativity? This Space runs the fast 350M model. For a richer, more creative enhancer (style / camera / dialogue / music), open the 2.6B creative Space (ZeroGPU).

---

Why this is different

Typical prompt “enhancers”	This one
Templates, regex, or a generic chat model	Fine-tuned LLM dedicated to this job
Mood words and hashtag soup	Shot lists, camera moves, soundscape, music
Guesswork about “what sounds good”	Aligned to MiniMax H3 video prompt writing guidelines
One free-form paragraph	Structured output MiniMax can actually consume

Under the hood: LoRA SFT of LiquidAI/LFM2.5-350M — transformers weights at geocine/minimax-video-prompt-enhancer-350m, Q4_K_M GGUF at geocine/minimax-video-prompt-enhancer-350m-gguf, served via llama.cpp on CPU.

---

Runtime

Backend	llama.cpp (llama-cpp-python)
Quant	Q4_K_M (~219 MB)
Hardware	CPU — no ZeroGPU quota, no login required for inference
Typical latency	A few seconds for a short T2VA brief on free CPU

Weights are downloaded from the model repo at boot (or read from models/ if present).

---

What you get

Paste a rough idea (and optional reference notes). Choose duration. Hit Generate.

You get a full brief, for example:

• integrated_multimodal_description — style, framing, action, camera, dialogue tags
• overall_soundscape — ambience and physical sound
• non_diegetic_music — audience-only score (or N/A)
• For image/video refs: alignment lines, subjects, retention-style structure in full-reference mode

Built for people who actually ship MiniMax clips — not for prompt-cosplay.

---

How to use

1. Write a rough user prompt (dialogue, shot notes, or a one-liner).
2. Optionally fill References & frames (first/last frame, subjects, clips, audio). The Space picks the task for you.
3. Set duration. Generate. Copy the structured prompt into MiniMax.

Tip: Short, real-user prompts work best. The model expands them into the full MiniMax format.

---

Model & license

Model (HF)	geocine/minimax-video-prompt-enhancer-350m
GGUF (CPU)	geocine/minimax-video-prompt-enhancer-350m-gguf
Base	LiquidAI/LFM2.5-350M
License	Liquid lfm1.0 — read the base card before commercial use

This is a text prompt specialist, not a video generator. It writes the brief; MiniMax (or your pipeline) makes the frames.