MiniMax H3 turns a written scene, a starting image with an optional ending frame, or combined visual and sound references into a 5–15-second 2K video. Describe the subject, action, camera movement and sound, then choose the workflow that matches your material. Frame images establish the opening and ending, while reference images, video and audio provide separate cues for appearance, movement and ambience. Text and reference tasks offer framing controls; each request uses the selected output duration, with reference-media charges included when applicable.
MiniMax H3 is developed by MiniMax. Vidgo provides access to the model. Model names and trademarks belong to their respective owners.Use Text to Video to develop a scene from a written brief. Choose Image to Video when an approved starting frame defines the subject and composition, adding an ending frame to guide the final pose. Choose Reference to Video when images or clips should guide a new scene, with optional audio for its soundscape. All three tasks use fixed 2K output; compare their complete costs with your intended duration and reference material.
MiniMax H3 on Vidgo API with separate text-to-video, start/end-frame image-to-video, and multimodal reference-to-video workflows at fixed 2K output.
MiniMax H3 on Vidgo API with separate text-to-video, start/end-frame image-to-video, and multimodal reference-to-video workflows at fixed 2K output.
MiniMax H3 on Vidgo API with separate text-to-video, start/end-frame image-to-video, and multimodal reference-to-video workflows at fixed 2K output.
A frame-based request uses one or two ordered image_urls. A reference request uses up to nine images, three videos and three audio clips, with at least one image or video. State which source defines the subject, which supplies motion and which establishes sound; the prompt then describes how those cues should form one new scene. Use the workflow table to compare the original task descriptions before opening the corresponding task.
| Model / task | Input and workflow |
|---|---|
| MiniMax H3 Text-to-Video | MiniMax H3 on Vidgo API with separate text-to-video, start/end-frame image-to-video, and multimodal reference-to-video workflows at fixed 2K output. |
| MiniMax H3 Image-to-Video | MiniMax H3 on Vidgo API with separate text-to-video, start/end-frame image-to-video, and multimodal reference-to-video workflows at fixed 2K output. |
| MiniMax H3 Reference-to-Video | MiniMax H3 on Vidgo API with separate text-to-video, start/end-frame image-to-video, and multimodal reference-to-video workflows at fixed 2K output. |
Existing records show a text-defined osprey dive, a two-frame Arctic fox jump and a kingfisher scene guided by appearance, motion video and rain audio. Inspect these distinct requests to understand how the prompt and media contribute to the result. Duration and framing help fit the sequence to the intended presentation.
Describe the subject, environment, action order, camera path and audible events in a prompt of up to 2,000 characters. Select 5–15 whole seconds and a supported ratio for a 2K clip; the recorded osprey request connects flight, water entry and ascent in one documentary-style shot.
Place the starting image first and an optional ending image second. Describe how the subject moves between them so an approved composition and final pose can guide a transition, reveal or physical action. The output framing follows the first image.
Name the appearance, setting and lighting to retain, then specify the changing pose, physical contact and camera movement. The fox request uses both endpoint images and asks for plausible weight and landing snow, giving you concrete details to review in the actual result.
Use reference images to establish a subject and reference video to describe movement or a camera path. Explain each role and how the motion should adapt to the new subject. The kingfisher case uses its own appearance image with an osprey dive video to guide a new stream scene.
Add up to three audio references alongside a visual reference and describe their role in the generated sound. In the kingfisher request, rain audio supplies ambience and timing. Listen to the source and result together to assess how the sound supports the visible action.
The fox case supplies a crouched opening and a landing endpoint, then requests a continuous jump with stable identity, glacier and lighting. The kingfisher case supplies its appearance image, an osprey motion clip and rain audio, then requests a dive-and-return scene adapted to the smaller bird. Read each complete prompt and compare every source with its own generated video; the cases demonstrate different workflows.
Continue the same Arctic fox from the start frame as it crouches, pushes off powerfully, clears the narrow blue-ice crevasse, and lands in the exact pose and composition of the end frame. One continuous eye-level wildlife documentary shot, anatomically plausible motion, stable white coat and gray shoulder marking, natural body weight, and a small spray of powder snow under the landing paws. Preserve the glacier, lighting, camera position, and fox identity. Natural stereo sound: cold wind, paws scraping ice, soft snow impact. No text, no logos, no products, no watermark.
Use Image 1 only for the common kingfisher species, turquoise-blue and orange plumage, bill shape, feather detail, and body proportions. Use Video 1 only for the takeoff-to-dive motion, water interaction, and continuous camera path, adapting the movement naturally to the smaller kingfisher. Use Audio 1 as the rain soundscape and timing cue; do not reuse Video 1 audio. Generate one continuous wildlife documentary shot beside a temperate forest stream: the kingfisher leaves its wet alder branch, follows the referenced aerial dive, enters the stream, then returns to the branch while preserving its appearance. Natural documentary realism. No text, no logos, no products, no watermark.
Prepare a scene description, approved frame pair or references for the subject, motion and sound. Describe the action sequence and details to preserve, then generate a 2K clip with the matching task. Reviewed results can support wildlife concepts, character animation, planned transitions, moving illustrations, camera studies and audiovisual scene proposals.
Describe the animal, habitat and ordered activity, including camera distance and natural sounds. Generate a clip and review anatomy, water or ground contact and action progression for a documentary-style storyboard.
Prepare a starting illustration or photograph and describe a gesture, expression or physical action. Inspect how the generated video retains appearance and setting while adding movement for a character-development study.
Supply starting and ending frames with compatible subject and scene details. Describe the intervening action and final pose, then inspect continuity and endpoint framing for an animation or storyboard transition.
Prepare a recognizable subject image and a movement clip, assign their roles and describe a new environment. Review identity and adapted action for recurring-character scenes or a visual narrative proposal.
Describe a continuous track, push or change in camera distance using text or a motion reference. Review framing and subject visibility through the result to explore how the sequence communicates an action.
Combine a visual reference with rain, wind or another appropriate audio source and describe its contribution. Listen while watching the generated scene to review ambience and action timing for a narrative concept.
Text and frame-based generation use the current per-second rate multiplied by the selected output duration. For reference generation, each reference video is rounded down to whole seconds separately and those seconds are added to output duration at the same rate. Reference images after the first five add a per-image usage charge. The dynamic table shows each task’s starting rate; prepare the complete references and inspect the task estimate to plan the full request cost.
| Model | Provider | Workflow | Input material | Output specification | Starting rate |
|---|---|---|---|---|---|
| MiniMax H3 Text-to-Video | MiniMax | Text to Video | Text prompt | 2K | $0.105 / second |
| MiniMax H3 Image-to-Video | MiniMax | Image to Video | 1–2 starting/ending frames | 2K | $0.105 / second |
| MiniMax H3 Reference-to-Video | MiniMax | Reference to Video | Image / video; optional audio | 2K | $0.105 / second |
Test your creative brief online, then bring the verified request into your application.
Open the chosen task from the model list above. Describe the action and camera, supply its required images or reference media, and set duration and supported framing. Review the full estimate, generate, then inspect the result alongside every source for appearance, motion, endpoint composition and sound.
Choose a task, inspect its settings and cost, and follow the result from an online test to an authenticated API request.
Choose Text to Video for a written scene, Image to Video for one starting frame and an optional ending frame, or Reference to Video for subject and movement sources. Reference generation requires an image or video and can include audio. Select 5–15 whole seconds, fixed 2K and the framing control offered by the chosen task.
Open the chosen task from the model list above. Describe the action and camera, supply its required images or reference media, and set duration and supported framing. Review the full estimate, generate, then inspect the result alongside every source for appearance, motion, endpoint composition and sound.
Plan the output duration first. For reference requests, count images beyond the first five and let the task read each video duration before reviewing the full estimate. Compare the complete amount with the number of tests planned for the project.
Read the actual prompt and inspect every source image, video or audio alongside its generated result. Use the recorded examples to understand how to describe input roles, action progression and the details to preserve in your own request.
Retain the task ID returned by the API to query status and retrieve completed file URLs. Use your online generation history to review earlier results and the creative direction you tested.
Use the documentation linked from each task for its exact model identifier, accepted inputs, authentication and response format. Integrate the same task you tested online so its media and settings match the request.
MiniMax H3 is a video family on Vidgo with text, start/end-frame animation and multimodal-reference tasks. Each generates 5–15-second video at fixed 2K. Reference generation combines image or video guidance with optional audio; task choice determines the accepted media and applicable reference charges.
Explore model introductions, API guides and platform comparisons to understand input preparation, request integration and cost planning for your project.
