MiniMax H3 Reference to Video
minimax/h3/reference-to-video
MiniMax H3 Reference to Video generates coherent 2K videos from natural-language prompts and multimodal references, including images, videos, and audio, guiding subject consistency, motion, timing, visual style, and scene continuity. Ready-to-use REST inference API, best performance, no coldstarts, affordable pricing.
Examples
Parameters
| Name | Type | Default | Constraints | Description |
|---|---|---|---|---|
| prompt *Prompt | textarea | — | ≤ 7000 chars | Text description of the desired video scene and interactions. |
| imagesReference Images | image_upload_group | — | image/* · 1–9 items | Reference image URLs. At least one reference image or video is required. |
| audiosReference Audios | audio_upload_group | — | audio/* · 0–3 items | Optional reference audio URLs. Audio cannot be provided alone. |
| aspect_ratioAspect Ratio | select | adaptive | 21:9 | 16:9 | 4:3 | 1:1 | 3:4 | 9:16 | adaptive | Output aspect ratio. |
| resolutionResolution | select | 2k | 2k | Output video resolution. |
| durationDuration | slider | 5 | 4 ~ 15 · step 1 | Output video duration in seconds. |
Output fields
| Field | Type | Description |
|---|---|---|
| videos | array<string> | Generated video URLs |
API
Call this model through one unified REST API. Get a key on the API Keys page.
cURL
# 1) Submit — returns { "task_uuid": "..." }
curl -X POST "https://www.namifusion.com/api/v1/marketplace/run/minimax/h3/reference-to-video" \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"input": {
"prompt": "A cinematic portrait, dramatic rim light, shallow depth of field",
"aspect_ratio": "adaptive",
"resolution": "2k",
"duration": 5
}
}'
# 2) Poll until status is "completed", then read the output URLs
curl "https://www.namifusion.com/api/v1/marketplace/run/tasks/TASK_UUID" \
-H "Authorization: Bearer YOUR_API_KEY"Python
import time, requests
API_KEY = "YOUR_API_KEY"
HEADERS = {"Authorization": f"Bearer {API_KEY}"}
# 1) Submit
resp = requests.post(
"https://www.namifusion.com/api/v1/marketplace/run/minimax/h3/reference-to-video",
headers=HEADERS,
json={
"input": {
"prompt": "A cinematic portrait, dramatic rim light, shallow depth of field",
"aspect_ratio": "adaptive",
"resolution": "2k",
"duration": 5
}
},
)
resp.raise_for_status() # 401/402/429/5xx stop here instead of polling a bad task
task = resp.json()
# 2) Poll until a terminal state (completed / failed / cancelled).
# This model is allowed up to 300s server-side.
deadline = time.time() + 360
while task.get("status") not in ("completed", "failed", "cancelled"):
if time.time() > deadline:
raise TimeoutError(f"still {task.get('status')} — keep the task_uuid and poll later")
time.sleep(3)
poll = requests.get(f"https://www.namifusion.com/api/v1/marketplace/run/tasks/{task['task_uuid']}", headers=HEADERS)
poll.raise_for_status()
task = poll.json()
print(task["status"], task.get("output"))JavaScript
const API_KEY = "YOUR_API_KEY";
const HEADERS = { Authorization: `Bearer ${API_KEY}` };
// 1) Submit
const resp = await fetch("https://www.namifusion.com/api/v1/marketplace/run/minimax/h3/reference-to-video", {
method: "POST",
headers: { ...HEADERS, "Content-Type": "application/json" },
body: JSON.stringify({
"input": {
"prompt": "A cinematic portrait, dramatic rim light, shallow depth of field",
"aspect_ratio": "adaptive",
"resolution": "2k",
"duration": 5
}
}),
});
if (!resp.ok) throw new Error(`submit failed: ${resp.status} ${await resp.text()}`);
let task = await resp.json();
// 2) Poll until a terminal state (completed / failed / cancelled).
// This model is allowed up to 300s server-side.
const deadline = Date.now() + 360 * 1000;
while (!["completed", "failed", "cancelled"].includes(task.status)) {
if (Date.now() > deadline) throw new Error(`still ${task.status} — keep the task_uuid and poll later`);
await new Promise((r) => setTimeout(r, 3000));
const poll = await fetch(`https://www.namifusion.com/api/v1/marketplace/run/tasks/${task.task_uuid}`, { headers: HEADERS });
if (!poll.ok) throw new Error(`poll failed: ${poll.status}`);
task = await poll.json();
}
console.log(task.status, task.output);Documentation
MiniMax H3 Reference to Video
Reference-guided 2K video generation for stronger consistency, motion control, and scene continuity.
MiniMax H3 Reference to Video is a multimodal image-to-video model that generates coherent 2K videos from natural-language prompts plus reference media. By combining text with images, videos, and optional audio, it offers stronger control over subject identity, motion, timing, visual style, and overall scene progression—while also delivering no cold start behavior and cost-efficient production.
🚀 Key Features
- Multimodal reference guidance: Generate videos from a Prompt combined with reference images, reference videos, and optional reference audio for more controllable results.
- Stronger subject consistency: Use reference images to preserve characters, products, objects, or brand visuals across the generated video.
- Motion and camera guidance: Reference videos help steer movement, pacing, interactions, and camera behavior for more directed outputs.
- Native 2K output: Produces high-resolution 2K videos suitable for polished creative, commercial, and concept work.
- Flexible Aspect Ratio support: Supports 21:9, 16:9, 4:3, 1:1, 3:4, 9:16, and Adaptive for cinematic, social, square, and vertical formats.
- No cold start, production-friendly pricing: Well suited for reliable generation workflows, rapid iteration, and scalable content creation.
🛠️ Technical Specifications
| Item | Details |
|---|---|
| Model Name | MiniMax H3 Reference to Video |
| Model ID | minimax/h3/reference-to-video |
| Model Architecture | Reference-guided image-to-video generation model with multimodal conditioning |
| Task Type | reference-to-video |
| Input Format | Prompt + reference images and/or reference videos, with optional reference audio |
| Output Format | Video |
| Resolution | 2K（fixed） |
| Duration | 4–15 seconds |
| Aspect Ratio | 21:9, 16:9, 4:3, 1:1, 3:4, 9:16, Adaptive |
| Reference Images | Up to 9 |
| Reference Videos | Up to 3 |
| Reference Audio | Up to 3 files（cannot be used alone） |
| Input Requirement | At least one reference image or reference video is required |
| Latency | No cold start highlighted by the provider |
| Primary Strengths | Subject consistency, motion guidance, style control, timing, scene continuity |
Sample Prompts
A cinematic woman in a futuristic jacket walks through a neon-lit street after rain, slow camera push-in, reflective pavement, dramatic lighting, highly detailed.Create a premium product commercial using the reference product images, with smooth orbit camera movement, clean studio background, elegant lighting, and refined motion.Using the reference character image and motion video, generate a sunset beach scene where the character runs forward, turns back, and smiles, with gentle tracking camera movement.
💰 Pricing
| Pricing Item | Price |
|---|---|
| Base Price | $0.13 per output second |
| 5s output | $0.65 |
| 10s output | $1.30 |
| 15s output | $1.95 |
Billing Rules
| Item | Rule |
|---|---|
| Reference images | First 5 images are free; each additional image costs $0.04 |
| Reference videos | Billed at $0.13 per second based on combined duration |
| Reference audio | Supported audio inputs are free |
| Reference video limits | Up to 3 videos, 15 seconds total per request |
💡 Best Use Cases
- Commercial and e-commerce video creation: Turn product shots, brand assets, and reference footage into polished short-form marketing videos.
- Character and IP consistency workflows: Maintain a stable visual identity for people, avatars, mascots, or branded subjects across generated clips.
- Storyboarding and concept development: Rapidly prototype scenes, camera moves, and visual direction from written ideas plus reference media.
- Social and mobile-first content: Produce vertical, square, or widescreen videos tailored to different publishing channels.
🔗 Related Models
- MiniMax H3 Text-to-Video: Best for generating high-resolution video directly from text-only prompts.
- MiniMax H3 Image-to-Video: Best for generating video from a first-frame image plus a Prompt.
Best Practices
- Use clear, relevant reference images when subject identity, product detail, or style consistency matters most.
- Use reference videos when motion, pacing, interaction, or camera movement is the priority.
- Keep the Prompt focused on visible action, scene progression, camera language, lighting, and mood.
- Start with shorter Duration values for fast iteration, then extend length for more developed scenes.
- Match the Aspect Ratio to the target channel: 16:9 for standard widescreen, 9:16 for mobile vertical, and 1:1 for square social content.
Related models
Gemini Omni Flash Reference to Video API
Gemini Omni Flash Reference to Video creates short AI videos with synchronized audio from one or more reference images and a text prompt, preserving visual identity and following the provided references for guided multimodal video generation. Ready-to-use REST inference API, best performance, no coldstarts, affordable pricing.
Seedance 2.0 Mini Reference to Video
Seedance 2.0 Mini Reference-to-Video is a high-speed, cost-effective video generation model developed by ByteDance, engineered specifically for creating multi-shot videos with a cinematic feel. By seamlessly combining reference images with text prompts, it generates highly narrative video segments. The model features advanced AI camera control and ensures robust character consistency across different scenes. It supports multiple output resolutions from 480P up to 4K, durations ranging from 4 to 15 seconds, and flexible aspect ratios. Complete with an out-of-the-box REST inference API, it delivers exceptional performance with zero cold starts, offering an affordable yet premium solution for professional video creation.
Seedance 2.0 Fast reference-to-video
Seedance 2.0 Fast (Video-Edit) edits an input video from a natural-language prompt at a faster, cheaper tier. Built on ByteDance Seed's unified multimodal architecture, it preserves subject identity, composition, and motion while rewriting lighting, style, weather, environment, or specific elements as instructed. Ready-to-use REST API, best performance, no coldstarts, affordable pricing.