Access Tencent Cloud MPS video enhancement through EvoLink's unified API. Choose from 7 modes for footage, animation or faces, upscale existing clips to 4K, and use the same API key for video generation and enhancement.
Add a video, pick an enhancement mode and a target resolution, check the estimate, and compare the result before you wire up the async API.
Enhance a Video
Add one video, then choose the mode, resolution and frame-rate limit.One MP4 or MOV video, up to 60 seconds and up to 4K. Upload a file or paste a public URL.
General real-world footage. The default choice for filmed and photorealistic AI video.
A limit, not a target: a source at or below it keeps its own frame rate and is never interpolated upward. Source follows the video up to 120 fps.
Enhanced videos are kept for 24 hours — download anything you want to keep.
One per-second rate, set by two things: the target resolution and the frame rate of the delivered video. Seconds are rounded up to a whole second, and all seven enhancement modes cost the same.
First Test Cost
5s clip - 720p - up to 30 fpsOne 5s clip enhanced to 720p
About 546 tests with $10 in credits.
Check the look on a short clip at 720p first, then move to 1080p, 2K or 4K once the enhancement mode suits your footage.
Testing Budget Guide
Pick an amount based on how many tests you expect.Good for a first validation.
Good for comparing modes and resolutions.
Good before production integration.
Totals for one run with the clip length shown. Your final cost follows the length and frame rate of the delivered video.
tencent-video-upscale| Output frame rate | 720p | 1080p | 2K | 4K |
|---|---|---|---|---|
| Up to 30 fpsMost AI-generated and filmed video | $0.0035/s-15% 0.238 cr/s$0.0042 official price | $0.0080/s-12% 0.544 cr/s$0.0092 official price | $0.014/s-15% 0.952 cr/s$0.017 official price | $0.032/s-13% 2.176 cr/s$0.037 official price |
| Up to 60 fpsOutput above 30 fps | $0.0070/s-8% 0.476 cr/s$0.0077 official price | $0.016/s-9% 1.088 cr/s$0.018 official price | $0.028/s-11% 1.904 cr/s$0.032 official price | $0.064/s-9% 4.352 cr/s$0.071 official price |
| Up to 120 fpsOutput above 60 fps | $0.014/s-10% 0.952 cr/s$0.016 official price | $0.032/s-8% 2.176 cr/s$0.035 official price | $0.056/s-10% 3.808 cr/s$0.063 official price | $0.128/s-8% 8.704 cr/s$0.139 official price |
Per output second. Every cell is the same base rate multiplied by the resolution and frame-rate factors, so a 60 fps output costs twice the 30 fps rate and a 120 fps output four times.
Official comparisons use Tencent Cloud's international Diffusion Transformer Enhancement rates plus Mainland China H.264 TSC transcoding, converted from per minute to per second. Regional rates, codecs and rounding up per billing cycle can change the actual Tencent Cloud invoice. Tencent Cloud pricing reference
Seconds round up
The delivered length is rounded up to a whole second, with a 1-second minimum. A 5.04-second clip is billed as 6 seconds.
Frame rate follows the output
The tier is set by the delivered video, not by the limit you send. A 24 fps source stays on the base tier even with a 60 fps limit.
Cap it to save
Set model_params.max_fps to 30 and a 60 or 120 fps source is delivered at 30 fps on the base rate.
Same price for every mode
All seven enhancement modes share one rate, so choose the mode by how the result looks.
Tencent Video Upscale is the AI video enhancement service of Tencent Cloud Media Processing Service. On EvoLink, one model ID takes an existing video and returns an enhanced copy at 720p, 1080p, 2K or 4K. You choose one of seven enhancement modes for real footage, animation, faces or detail, and the output keeps the source frame rate unless you cap it. There is no prompt: the request is a video URL plus a few parameters, handled as an asynchronous task with live cost estimation and one API key shared with other models.
tencent-video-upscaleOne model ID covers every enhancement mode and target resolution. Send the video in video_urls, then set mode and quality; add model_params.max_fps only when you want to cap the frame rate.
Use the EvoLink API for product integration and batch jobs, or call it from Codex, Claude, or Gemini for fast creative and development workflows. Both paths share the same EvoLink API key, balance, model routes, and task history.
Best for: product backends, batch jobs, automated workflows
Call EvoLink’s unified video API from your server and control the model ID, parameters, task queue, callbacks, and result storage.
Best for: creative and development tasks in Codex, Claude, and Gemini
Give the Agent your output goal, assets, and acceptance criteria. It can choose the route, assemble the request, track the task, and return the result without requiring you to hand-code every step.
Raise a 480p or 720p clip to 1080p, 2K or 4K without regenerating or re-shooting it. To stay close to the original size, pick the lowest option that is not below your source.
The output keeps the source frame rate and never adds interpolated frames. A frame-rate limit lets you deliver high-frame-rate sources at 30 or 60 fps when cost matters more.
Submit once, track the task by polling or callback, then persist the enhanced video before its temporary result URL expires.
realisticGeneral real-world footage and photorealistic AI video. The default mode.
realistic_distantReal-world footage shot from a distance, tuned to keep small faces clean.
animationAnimation, anime and cartoon content with flat colors and line art.
animation_distantAnimation with distant characters, tuned to keep small faces clean.
face_fidelityKeeps facial identity and features as close to the source as possible.
detail_boostPushes texture and edge detail harder for a crisper result.
softA gentler result with less sharpening, for footage that looks over-processed.
Generate drafts at 480p or 720p to keep iteration cheap, then enhance the takes you keep to 1080p or higher for delivery.
Use this setupUse the animation modes to clean up line art and flat color when moving older or low-resolution animation to 2K and 4K screens.
Use this setupTry face fidelity for interviews, presenters and avatar video. Compare facial features and consistency across frames with the source before keeping the result; preservation depends on the input clip.
Use this setupThis example shows the shortest runnable flow: create a task, save the returned task ID, then poll or wait for an HTTPS callback. Open the API tab for the complete parameter and response reference.
curl -X POST https://api.evolink.ai/v1/videos/generations \
-H "Authorization: Bearer $EVOLINK_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "tencent-video-upscale",
"video_urls": ["https://example.com/my-video.mp4"],
"mode": "realistic",
"quality": "1080p",
"model_params": { "max_fps": "30" },
"callback_url": "https://your-domain.com/webhooks/video-task-completed"
}'
# Save the returned task ID, then query:
curl https://api.evolink.ai/v1/tasks/{task_id} \
-H "Authorization: Bearer $EVOLINK_API_KEY"Check every value against the parameter cards above: ranges, allowed options, asset counts, and file types must match the selected route.
Check the Authorization bearer token and confirm the available balance and task billing in the console.
Review likeness rights, brand assets, readable text, sensitive content, and input-asset compliance.
A Tencent Video Upscale task that reaches a final failed state is not charged. Keep the task ID, inspect the final state, and identify whether the cause is validation, review, or execution before retrying.
callback_url must be a public HTTPS URL, not localhost or a private network address. Always keep the task ID as a polling fallback.
| Feature | Tencent Video Upscale | Production impact | Choose another route when |
|---|---|---|---|
| Input contract | 1 public video URL, MP4 / MOV | The server reads the file before accepting the task, so the URL must be directly downloadable. | Your source is WebM, a stream, or a private link. |
| Length & source size | Up to 60 seconds, source up to 4K | Split longer videos into segments before submitting. | You need to process long-form video in one request. |
| Target resolution | 720p, 1080p, 2K or 4K by short side | The target cannot be lower than the source; the lowest allowed option keeps the picture close to its original size. | You need an arbitrary scale factor or a custom pixel size. |
| Frame rate | Follows the source, optional 30 / 60 / 120 fps limit | The limit only lowers the frame rate; billing follows the delivered video. | You need frame interpolation to a higher frame rate. |
| Delivery | Async task with polling or callback | Queue tasks and make terminal handling idempotent. | Your workflow requires an immediate synchronous result. |
| Billing | Per output second by resolution and frame-rate tier | Estimate each task from the clip length; seconds round up. | You need a fixed cost independent of length. |
The request carries one video URL and no prompt. The server checks the length, resolution and frame rate first, and rejects files it cannot read, videos over 60 seconds, and sources above 4K.
model_params.max_fps is a limit. A source at or below it keeps its own frame rate; a source above it is reduced. Sources above 120 fps are always delivered at 120 fps.
Submit a task, poll or receive a callback, and persist the enhanced video. EvoLink result URLs are available for 24 hours.
See the per-second rate for every resolution and frame-rate tier before you submit a video.
Call video upscaling through the same endpoint, authentication and task flow as the video generation models.
Manage credits, spending, and model usage in the same EvoLink account.
Track pending, processing, completed, and failed tasks through a consistent async workflow.
Create a clip with any video model on EvoLink and pass the result URL straight into the upscaler with the same API key.
Compare the video upscalers available under one EvoLink account and API key, then choose by output control, source limits, and cost.
Tencent Video Upscale
Enhance existing video with 7 modes and output up to 4K. Billed per output second.
View API
Topaz Video Upscale
AI video super-resolution with 1x, 2x and 4x scale factors, billed by input video length.
View API
Yes. Tencent Video Upscale on EvoLink accesses the AI video enhancement capability of Tencent Cloud Media Processing Service (MPS). It enhances an existing video rather than generating a new scene from a prompt. Use tencent-video-upscale through EvoLink; the input limits, modes and billing on this page describe the EvoLink route, not every Tencent Cloud MPS or SDK feature.
Use tencent-video-upscale in EvoLink requests. The same model ID covers all seven enhancement modes and all four target resolutions.
No. The request needs a video in video_urls. mode, quality and model_params.max_fps are optional and fall back to realistic, 1080p and the source frame rate.
Send one MP4 or MOV video through a publicly downloadable URL. The video can be up to 60 seconds long and up to 4K. The server reads the file before accepting the task and rejects one it cannot read.
Use realistic for filmed and photorealistic footage and animation for anime and cartoon content. The two distant modes are tuned for small faces in wide shots, face fidelity keeps a person's features close to the source, detail boost sharpens harder, and soft gives a gentler result. All modes cost the same.
quality sets the target resolution by the short side of the video: 720p, 1080p, 2k (1440 px) or 4k (2160 px). It cannot be lower than the source video. Choosing the lowest option that is not below the source keeps the picture close to its original size.
By default the output keeps the source frame rate. model_params.max_fps sets a limit of 30, 60 or 120: a source above the limit is reduced to it, and a source at or below the limit is left unchanged. Frames are never interpolated upward.
EvoLink bills per output second. The rate depends on the target resolution and on the frame rate of the delivered video, and the length is rounded up to a whole second. Use the live pricing table and the playground estimate for the current price instead of relying on a copied static figure.
Higher frame rates mean more frames to enhance. Output of up to 30 fps is billed at the base rate, 31-60 fps at twice that rate, and 61-120 fps at four times. Set model_params.max_fps to 30 if you want a high-frame-rate source delivered on the base rate.
The delivered length is rounded up to a whole second, with a minimum of 1 second. A 5.04-second clip is billed as 6 seconds and a 10-second clip as 10 seconds.
Not in a single request. Split the video into segments of 60 seconds or less, submit one task per segment, and join the results.
Processing is asynchronous. Submit a task, poll its status or provide an HTTPS callback, and persist the returned video when the task completes.
EvoLink result URLs are available for 24 hours. Copy completed videos to your own object storage instead of using the temporary URL in production.