Tencent HunyuanVideo · Video

Image / First-Last Frame VideoNEW

📌 What is this：Given a first-frame image (optionally a last frame too), generate a video that starts from the first frame and transitions toward the last.

💡 When to use：Use to animate an image or to precisely control the opening / closing frames. The aspect ratio follows the first frame; aspectRatio is ignored.

Endpoints

Usage	Method	Path
Submit task	POST	/api/cqt/generator/hunyuan
Query Result	GET	/api/cqt/info/hunyuan?id={taskId}

⏱ Recommended polling interval：Generation takes ~1-3 min; poll every 10-15s until status = succeeded / failed.

Sub-models

Sub-models that support this mode. The tiers column is credits/sec (generateAudio → = audio tier, videoUrls → = with-reference-video tier). Resolutions / durations not listed are rejected at submit (no charge).

model	Usage	Materials	Duration (sec)	Tier price (credits/sec)
hunyuan-video-1.5	HunyuanVideo 1.5	1 first frame + ≤1 last frame	5	720P=30 / 1080P=50 / 2K=75 / 4K=112

Request Parameters

Param	Type	Required	Description
model	string	Required	Sub-model: hunyuan-video-1.5 (see the sub-model table below)
prompt	string	—	Video description (optional; materials can drive generation)
firstFrameUrl	string	Required	First-frame image URL (publicly reachable); cannot be combined with imageUrls / videoUrls
lastFrameUrl	string	—	Last-frame image URL (optional; requires firstFrameUrl, produces a first-to-last transition)
negativePrompt	string	—	Negative prompt (optional)
duration	integer	—	Duration (sec), range depends on the sub-model (see table), default 5
resolution	string	—	540P / 720P / 1080P / 2K / 4K (available tiers depend on the sub-model), default 720P (or the lowest tier if 720P is unavailable); 2K / 4K are upscaled
enhancePrompt	boolean	—	Whether to auto-enhance the prompt (optional)
seed	integer	—	Random seed (optional)
returnFirstFrame / returnLastFrame / returnFrameAtSecond	boolean / boolean / number	—	Also return the first / last / given-second frame image of the result video (optional), 1 credit per frame

Request Example

curl -X POST https://api.cqtai.com/api/cqt/generator/hunyuan \
  -H 'Authorization: Bearer <API_KEY>' \
  -H 'Content-Type: application/json' \
  -d '{
    "model": "hunyuan-video-1.5",
    "prompt": "the red car slowly pulls away and drives down the rainy street, camera follows",
    "duration": 5,
    "resolution": "720P",
    "firstFrameUrl": "https://cdn.novapi.ai/seedream/1791950778488_c22c4d053b2a41fbb28193cd319473b1.jpg",
    "lastFrameUrl": "https://cdn.novapi.ai/seedream/1791950778814_11e964cb95be4866838196e0bc6be7f1.jpg"
  }'
# -> { "code":200, "data":"<taskId>" }

Response Example

{
  "code": 200,
  "msg": "success",
  "data": {
    "id": "73e9945331a142db974be37a048594c0",
    "resultUrl": "https://cdn.novapi.ai/hunyuan/1791952497628_9cee9d757d494452818588d3457bc4f2.mp4",
    "status": "succeeded",
    "errorMsg": null,
    "createTime": "2026-10-14 12:32:47",
    "completeTime": "2026-10-14 12:34:58"
  }
}

Response Fields

Field	Type	Description
code	number	Status code; 200 means success
msg	string	Status message; "success" on success
data.id	string	Task ID (taskId); use it to poll GET /api/cqt/info
data.status	string	Task status: succeeded / failed / otherwise still processing
data.resultUrl	string	URL of the generated video; returned when status = succeeded, null otherwise
data.errorMsg	string \| null	Failure reason; null on success
data.createTime	string	Task creation time
data.completeTime	string \| null	Task completion time; null until finished