MiniMax H3 - UGC Product Video

Turn a model photo and a product photo into a UGC-style product review video. Write a script, and MiniMax H3 generates a vertical clip where your person picks up, demonstrates, and talks about your product, with voice and ambient audio generated in the same pass.

Upload two reference images, write your scene beat by beat, and hit run. You get back an MP4 with stereo sound. The default setup produces a 15-second 9:16 vertical clip at about 720p, ready for TikTok, Reels, or Shorts.

How do you generate UGC product videos with MiniMax H3?

Upload a photo of your person and a photo of your product. Write a timestamped script that describes what happens second by second. MiniMax H3 reads both reference images and generates a vertical video where the person interacts with your product, with native stereo audio built into the output.

Model (reference image 1)
This is your person. Upload a clear, well-lit photo of whoever should appear in the video. MiniMax H3 carries their face, body, clothing, and features into the generated clip.

Product (reference image 2)
This is your product. Upload a clean product shot with good lighting and visible detail. The model will place it into the scene and maintain its design, color, and shape throughout the video.

Prompt
This is your script. The default prompt is a 15-second UGC product review broken into timestamped beats: 0-2s hook, 2-5s product detail, 5-8s feature demo, and so on. MiniMax H3 responds well to this timeline format.

Want a talking-head review? Describe the person picking up the product, pointing to features, and reacting on camera. Need a lifestyle demo? Describe the product in use with a specific setting and camera movement. Always include style notes at the end: lighting, camera feel (iPhone handheld vs. studio), and what to keep consistent between references.

Specific prompts give specific results. "Person uses product" gives you something generic. "She picks up the bottle, tilts it toward the camera, reads the label, then sets it down with an approving nod" gives you a directed scene.

Duration (seconds)
Controls how long your video runs. Default is 15 seconds. Range is roughly 4 to 15 seconds. Shorter clips finish faster and tend to hold visual consistency better between the reference images and the output.

Resolution (aspect ratio and megapixels)
Default is 9:16 portrait at 0.9 megapixels (about 736 x 1280). This is the right starting point for vertical social content.

Want landscape for YouTube or website embeds? Switch to 16:9. For faster test runs, drop megapixels to 0.3-0.5. For higher detail, push to 1.2-1.5, but expect longer generation times.

Seed
Default is random. Lock the seed when you want to compare script changes without randomness shifting everything else. Find a take you like, note the seed, and iterate your prompt from there.

What are MiniMax H3 UGC product videos good for?

MiniMax H3 UGC videos let you generate short-form product content from two photos and a script. You get a vertical video where a specific person demonstrates a specific product, with audio, without booking a shoot or editing footage.

E-commerce teams can generate video variations across product lines without rebooking talent. Upload the same model photo with five different product shots, write five scripts, and produce a batch of UGC-style clips in an afternoon.

Social media teams can use this to test video concepts before a real shoot. Write three hooks for the same product, run all three, and compare which angle lands before spending production budget on the winning version.

Content creators prototyping ad formats can iterate on script structure, camera direction, and pacing without reshooting. Change the beat-by-beat timing, swap the hook line, or try a different product angle, then regenerate.

This workflow exposes two reference image slots. MiniMax H3 supports up to nine reference images per generation, so for multi-product scenes or setups with multiple people, you would need a different workflow configuration.

The open-weights version generates at up to about 768p short-side. For 2K output, MiniMax offers a separate API-based pipeline.

FAQ

What resolution do MiniMax H3 UGC videos generate at?
The open-weights model generates at up to 768p short-side. In this workflow, the default is 9:16 portrait at 0.9 megapixels, about 736 x 1280 pixels. You can adjust megapixels from 0.2 (352p quick preview) up to 2.0 (roughly 1080p), trading speed for detail.

Does MiniMax H3 generate voice and audio with the video?
Yes. MiniMax H3 produces native stereo audio in the same generation pass as the video. There is no separate audio model or post-processing step. The audio is decoded from a dedicated audio VAE and muxed into the final MP4 automatically.

How long can the generated UGC videos be?
This workflow defaults to 15 seconds at 24fps. You can shorten it to about 4 seconds. Shorter clips generate faster and tend to maintain better consistency between your reference images and the output. For longer content, you would chain multiple clips together.

How should I write prompts for UGC product videos?
Use a timestamped beat-by-beat structure. Break your video into 2-3 second segments and describe what happens in each one: camera position, person action, product interaction, spoken lines. End with style notes covering lighting, camera feel, and consistency instructions. The more specific your timeline, the more directed the output.

How do I run MiniMax H3 UGC video generation online?
You can run MiniMax H3 UGC video generation online through Floyo. No installation, no setup. Open the workflow in your browser, upload your reference images, write your script, and hit run. Free to try.