Minimax H3 · Ref2VA

1 Reference Image · Full Custom

Turbo LoRA + 3 Custom LoRAs

Freedom / Unlocked — Strict control for power users

Drive cinematic video from a single reference image with MiniMax H3 Ref2VA. Lock identity, outfit, and style to <Picture 1>, swap Turbo and custom LoRA IDs freely (public, private, or hidden), and tune duration, steps, and detail without losing the subject.

Cite the upload in your prompt as <Picture 1>. Output is 48fps for smoother motion.

━━━━━━━━━━━━━━━━━━━━

DURATION (enter frame count, not seconds)

~5s → 124

~8s → 192

~10s → 243

~12s → 294

~15s → 362

Tip: H3 grid only — length ≥ 5 and length % 17 = 5.

RESOLUTION · MEGAPIXELS

0.4 — lighter / lower cost

0.6 — balanced (recommended start)

0.8 — sharper detail

1.0 — maximum clarity · higher credit use

Aspect ratio follows your upload. Higher MP = more detail and higher cost.

SAMPLING STEPS

4–6 → faster / lighter (turbo-friendly)

8 → balanced / sweet spot

10–12 → more refine (slower, higher cost)

With Turbo LoRA, about 6–8 steps is usually enough.

PROMPT TIP

Always reference <Picture 1> for face, body, outfit, and art style.

Example: “<Picture 1> walks through a rainy street at night, same face and clothing, natural motion.”

━━━━━━━━━━━━━━━━━━━━

Research & responsible use

This tool is a general-purpose AI generation and research utility for experimentation, testing, and creative work.

Use it in line with applicable laws. Do not generate or share illegal, harmful, or rights-infringing content. You are responsible for your prompts, uploads, and how outputs are used.

By using this tool, you accept its experimental nature and agree to use it responsibly.