Skip to main content
NewB · Video
中文Start creating

Video with sound · 4 min read

Minimax H3: create a short film with sound at 768p

Choose a 6-, 10- or 15-second shot, combine image, video and audio references, and write a clear prompt for visuals and sound.

01Choose a duration that fits the action

Minimax H3 768p in NewB Video offers 6, 10 and 15 seconds. Six seconds suits a simple action or product shot. Ten seconds can fit an action and a short line of dialogue. Fifteen seconds gives you room for a setup, action and ending. Keeping the scene within the time available makes movement and speech easier to plan.

768p is the current H3 setting; it is not 1920×1080. If your delivery requires a specific resolution, plan the necessary post-production and verify the dimensions after your final export.

02Combine image, video and audio references

  1. Add up to 9 reference images for characters, settings, clothing or important props. Explain the role of each image.
  2. Add up to 1 MP4 or MOV reference video, no larger than 50 MB on this site. Describe the movement, camera work or sound you want to reference. A video reference can also be used without images.
  3. Add up to 3 WAV or MP3 audio references. Each must be 2–15 seconds and no larger than 15 MB; their combined duration must not exceed 15 seconds. Audio references must be accompanied by an image or video reference.
  4. Use the reference tags shown in the studio, such as @图片1, @视频1 and @音频1. Distinguish reference audio from the dialogue and environmental sounds you want the model to generate.

An audio reference is not a guaranteed finished voiceover. Review the output, especially the dialogue, pronunciation, lip synchronization and background volume.

03Give each prompt one clear scene

Describe the subject and framing, lighting, main action, camera movement and sound in that order. Keep dialogue concise and identify who speaks each line. Several simultaneous speakers, complex transitions and dense action make a short clip harder to review.

For a sense of presence, specify subtle footsteps, paper sounds or rain. If you do not want music or narration, say so directly. Use the images to establish the character and setting, leaving the prompt to describe how the scene moves.

Example prompt

Use the character in @图片1 and the classroom in @图片2 as references. A static medium shot. Soft morning light from the window illuminates the desks. The character gently sets down a book, looks at the camera, smiles and says: “Think it through, then speak with confidence.” Keep movement natural, speech clear and mouth movements synchronized with the dialogue. Include a quiet sound as the book is placed down. No background music or additional narration.

04Review before and after submitting

  1. Before submitting, check 768p, the duration, aspect ratio and every reference tag. Check the video count and file sizes, and confirm the combined audio duration and accompanying image or video.
  2. Review the displayed cost. Without membership: 6 seconds costs 6 credits; 10 seconds costs 8 credits; 15 seconds costs 10 credits. H3 is included in the unlimited benefits of active Lite and Premium memberships.
  3. Follow the task status in your generation history. When the clip is ready, review character appearance, hands, dialogue and sound before choosing a version to keep.
  4. For a sequence of shots, save the character and scene references, add the next shot to your storyboard, and use the previous shot’s last frame to help maintain continuity.

Unlimited benefits apply while membership is active. Task concurrency, queues, model availability and content rules still apply.