Minimax H3 이미지-비디오: 모델 및 API에 대한 마지막 프레임 제어 기능을 갖춘 768p 및 2K 이미지-비디오 | 달리다편안하다 | Models and API | RunComfy

MiniMax / MiniMax H3 Image To Video#

Minimax H3 Image to Video is MiniMax's H3-series model for turning one picture into moving footage. You hand it the opening frame and describe what should happen; the model resolves motion, camera behavior, and scene development, then renders at 768p or 2K.

Because the first frame is pinned by your upload, subject identity, wardrobe, palette, and framing carry through the clip. Minimax H3 Image to Video therefore fits work where the look is settled and only movement is missing.

Highlights#

• Frame-anchored motion: Your image sets the starting state, so Minimax H3 Image to Video extends what is on screen instead of inventing a subject.

• Optional closing frame: Give Minimax H3 Image to Video a last image and it solves the path between two stills — good for reveals and shot hand-offs.

• 768p or 2K output: Choose 768p for cheaper drafts or 2k when faces, packaging text, and fine texture need to survive a full-screen crop.

• Adjustable length: Minimax H3 Image to Video runs any whole number of seconds from 4 to 15.

• Plain-language direction: One prompt field carries subject action, camera move, lighting, and mood together, so Minimax H3 Image to Video reads intent instead of demanding preset menus.

Parameters#

These are the live fields for Minimax H3 Image to Video on this page.

Parameter	Required	Type	Default	Range / Options	Description
image *	Yes (*)	string (URL)	Sample first frame	256-5760 px per side, aspect ratio 0.4-2.5	Opening frame that anchors subject, composition, and style.
prompt *	Yes (*)	string	Sample motion brief	1-4000 characters	What should move, how the camera behaves, how the scene develops.
last_image	No	string (URL)	None	Same image constraints as image	Closing frame used to steer where the clip ends.
duration *	Yes (*)	integer	5	4-15	Clip length in whole seconds.
resolution *	Yes (*)	string	768p	768p, 2k	Output resolution tier.

Pricing#

Minimax H3 Image to Video bills on resolution and the length of the finished clip:

Resolution	Price per second	5s	10s	15s
768p	$0.11	$0.55	$1.10	$1.65
2K	$0.16	$0.80	$1.60	$2.40

How to Use#

1) Upload the opening frame — Pick an image where the subject reads clearly and the composition matches the shot you want.

2) Write the motion brief — Say what moves, how fast, and what the camera does. Minimax H3 Image to Video answers to verbs and camera language, not adjective lists.

3) Add a closing frame (optional) — Supply a last image when the ending matters, and keep it compatible with the first frame.

4) Set the length — Start at 5 seconds while tuning wording, then extend toward 15 once the motion behaves.

5) Choose the resolution — Use 768p for drafts; switch to 2k for final deliveries that need more detail.

6) Generate — Submit and review the Minimax H3 Image to Video result at full size.

7) Iterate one variable at a time — Swap the prompt, image, duration, or resolution separately so you can tell what changed the take.

Prompt & Reference Tips#

• Lead with the subject and its action, then layer camera, light, and mood behind it, since Minimax H3 Image to Video weights the opening clause heavily.

• Name the camera move outright: slow push-in, orbit, handheld follow.

• Keep the source image close to your target framing; Minimax H3 Image to Video will not recompose a badly cropped subject.

• Match lighting and palette across both stills when using a closing frame.

• Describe progression across longer clips, such as mist thinning before sun breaks through.

• Drop competing style cues; one visual direction per prompt reads cleaner.

How MiniMax H3 Image To Video compares to other models#

• Against text-only video models: A fixed opening frame removes guesswork about who appears, so Minimax H3 Image to Video suits jobs where art direction is locked.

• Against fixed-tier animators: Minimax H3 Image to Video lets you choose 768p or 2K, so you can draft cheaply and deliver at higher detail when needed.

• Against first-frame-only tools: Last-frame guidance controls where a shot ends, so Minimax H3 Image to Video can carry a planned transition.

• Against reference-heavy video models: Minimax H3 Image to Video keeps a short input list, trading broad multimodal conditioning for speed.

More Models to Try#

When Minimax H3 Image to Video is not the right fit, these are worth a pass:

• Hailuo 02 — MiniMax's earlier image-to-video release for quick motion.

• Hailuo 2.3 Pro — a production-tuned Hailuo option.

• Kling 3.0 — structured multi-shot with native audio.

• Seedance 2.0 Pro — reference-heavy video work.