1. Write the camera clause first
State the camera path before the subject — a push-in, a pull-back, an orbit, a pan. This gives MiniMax H3 a stable frame to build the rest of the shot around.
Open the prompt box →Keeps reference identity and detail close to the source.
Use images for identity, video for movement, and audio for timing. The completed generation appears here.
Camera-led prompting for MiniMax H3: state the camera move, the subject, and add references only where you need tighter control.
MiniMax H3 already reads camera-movement language when it’s written clearly into a prompt. Director framing organizes that language into repeatable clauses — camera path, subject action, composition, lighting — and layers in image, video, and audio references when a shot needs stronger identity or motion control.
It runs on the same reference-capable generator as Reference to Video above, just framed around directing a shot rather than around the reference types themselves.
State the camera path before the subject — a push-in, a pull-back, an orbit, a pan. This gives MiniMax H3 a stable frame to build the rest of the shot around.
Open the prompt box →Drop in an image for identity, a video for a camera move you want to reuse, or audio for timing — each reference does one job.
Add a reference →Name the subject, the environment, and one clear action. Keep it readable inside the duration you pick instead of stacking several events.
Test the camera move at a shorter duration and 768P before committing to a longer or 2K pass — camera behavior is usually the first thing worth confirming.
Submit a test job →Open the generator above and write a camera clause first, subject second.
Copy a prompt below or open it directly in the generator above.
A slow dolly push-in toward a matte black perfume bottle on a dark reflective surface, single soft studio light from the upper left, camera settles on a tight close-up, no camera shake, steady pacing.
Locks a product close-up with one continuous camera move and a fixed light source.
A low tracking shot follows a woman in a red coat walking through a rain-soaked city street at night, neon reflections on wet pavement, camera stays at waist height and keeps her centered in frame, steady walking pace.
Separates subject action from camera height and framing for a controlled reveal.
A slow 180-degree orbit around a lone lighthouse on a rocky coastline at dawn, wide establishing frame, soft golden light, locked horizon, no sudden zoom.
Uses an orbit clause plus explicit horizon and zoom constraints for a stable wide shot.
An orbit or push-in around a hero product protects geometry and branding while adding cinematic motion a static image can’t deliver.
A tracking shot or slow reveal introduces a character with intentional pacing — useful for trailers, teasers, and openers.
Reference an existing clip’s camera path as a video reference, so a new scene inherits the same move without copying its subject.
A slow, steady camera move paired with an audio reference for pacing produces atmosphere-led results where timing matters as much as the shot.
Combining an orbit, a push-in, and a whip pan in one clip usually produces a less controlled result than testing each move separately.
Keep the camera clause and the subject’s action as two distinct parts of the prompt so the model doesn’t merge them into one ambiguous instruction.
An image reference should carry identity, a video reference should carry motion — mixing intents into a single reference weakens both.
A locked horizon, no sudden zoom, or steady pacing all reduce drift on longer or more complex camera moves.
It’s a camera-led way of using MiniMax H3: write the camera move first, then the subject, and add image, video, or audio references when you need tighter control over identity, motion, or timing.
No. This page runs the same reference-capable MiniMax H3 generator as Reference to Video, framed around camera direction. There’s no extra toggle to find.
Camera work and motion, independent of what’s on screen. Point it at an existing clip whose camera move you want to reuse on a new scene.
Yes — an image reference for identity or appearance and a video reference for camera motion can be used together in the same generation.
Long enough to state the camera path, the subject and action, the environment, and lighting, without stacking multiple unrelated events. One focused move is usually more reliable than several.
The same live credit estimate as every other MiniMax H3 workflow, based on duration and resolution, shown before you generate.
Write the camera clause first, the subject second, and test at a short duration before committing to a longer pass.
Move from research to a relevant generator, prompt resource, or pricing page without restarting your workflow.
Create with image, video, and audio references.
Write and test prompts for specific video goals.
Follow the complete generation workflow.
Understand the speed-focused search intent and availability.
Transform an existing clip with a video-first reference generator.
Reapply movement from a video reference onto a new subject.
See every online input path in one workflow map.
Understand LoRA availability and the closest online option.
See what a ComfyUI-based path involves and the online alternative.
What API access means here and the online generator alternative.
VRAM, GPU, Mac, AMD, and model size questions in one place.
What running MiniMax H3 locally would involve, mapped out.
Where to look for repositories, workflows, and implementations.
What to look for on Hugging Face: weights, variants, and LoRA.
Open the generator with MiniMax H3 selected.
Compare plans and one-time credit packs.