Models/Topaz Upscale Video

Topaz Upscale Video API — Creative, Generative & Precision

Live3 models

Muapi exposes Topaz's full video-upscaling library through one REST API: Creative (Astra 2) for a stylized diffusion upscale that generates new detail up to 4K, Generative (Starlight Precise/HQ/Mini/Sharp/Fast 2) for reconstructing clean, temporally consistent detail from degraded footage, and Precision (Proteus, Artemis, Iris, Dione, Gaia, Rhea, Theia) for a faithful enlargement tuned to broadcast, animation, and CGI sources. All three use the same submit-and-poll pattern as every other model on the platform, billed dynamically per 10 seconds of output from $0.10.

3 of 3 models
TopazCreative

Topaz Upscale Video Creative

Enlarge video up to 4x and 4K with Topaz's Astra 2 diffusion model, generating new, temporally consistent detail rather than simple pixel interpolation.

Video in, video out
$3.00–$5.00+ / 10s
Try Model
TopazGenerative

Topaz Upscale Video Generative

Reconstruct clean, temporally consistent detail from low-resolution or degraded footage using Topaz's Starlight model family.

Video in, video out
$0.60–$2.60+ / 10s
Try Model
TopazPrecision

Topaz Upscale Video Precision

Enlarge video faithfully with Topaz's full precision library — Proteus, Artemis, Iris, Dione, Gaia, Rhea, and Theia — tuned for clean broadcast, CGI, and animation sources.

Video in, video out
$0.10–$0.60+ / 10s
Try Model

What is the Topaz Upscale Video API?

Topaz's video-upscaling models increase resolution and reconstruct detail in existing footage using models trained specifically for temporal consistency across frames, rather than upscaling each frame independently. Three families cover a spectrum from generated stylized detail to a faithful, source-preserving enlargement, scaling up to 4x and up to 4K.

Muapi exposes all three Topaz video families through the same unified REST pattern used across the platform: one API key, one request-and-poll flow, and dynamic pay-as-you-go pricing per 10 seconds of output. Choose Creative for a more stylized, generated look, Generative when the source is heavily degraded, or Precision when the source is already clean and needs a faithful enlargement.

Key Capabilities

Creative Diffusion Upscaling (Astra 2)

Topaz Upscale Video Creative generates new, temporally consistent detail while enlarging up to 4x and up to 4K, with tunable creativity, realism, and sharpness controls.

Degraded-Footage Reconstruction (Starlight)

Topaz Upscale Video Generative's five Starlight models reconstruct clean detail from old, compressed, or low-resolution source video, including a faster/cheaper Starlight Fast 2 tier.

Faithful Precision Upscaling

Topaz Upscale Video Precision's model library — Proteus, Artemis, Iris, Dione, Gaia, Rhea, and Theia — covers broadcast, interlaced, animation, and CGI sources without inventing new detail.

Frame-Rate Uplift

All three families support target_fps (16-60) to raise the output frame rate alongside resolution, priced proportionally to the increase.

Up to 4x and 4K Enlargement

Every family supports an adjustable upscale_factor up to 4x, with Creative and Generative able to reach a 4K output tier.

Pay Per Generation

No subscription or minimum commitment — pricing is calculated dynamically per 10 seconds of real output duration and resolution at request time.

Topaz Video Upscale Model Comparison

ModelProviderPriceBest For
Topaz Upscale Video CreativeTopaz$3.00–$5.00+ / 10sA stylized diffusion upscale with new generated detail, up to 4K
Topaz Upscale Video GenerativeTopaz$0.60–$2.60+ / 10sReconstructing clean detail from degraded or low-resolution footage
Topaz Upscale Video PrecisionTopaz$0.10–$0.60+ / 10sA faithful enlargement of already-clean broadcast, CGI, or animation sources

Pricing is billed dynamically per 10 seconds of output at a 30fps baseline and tiered by output resolution — see each model's playground page for the exact per-model rate.

How to Upscale a Video with Topaz via API

  1. Pick a family. Use Creative for stylized generated detail, Generative for degraded or low-resolution source footage, or Precision for a faithful enlargement of clean sources.
  2. Host your source video. Upload via POST /api/v1/upload_file or use any publicly reachable video URL.
  3. Submit the request. Call POST /api/v1/topaz-upscale-video-creative, /topaz-upscale-video-generative, or /topaz-upscale-video-precision with video_url and upscale_factor.
  4. Poll for completion. Check GET /api/v1/predictions/{request_id}/result until status is completed — video jobs can take several minutes — then download the upscaled video.
curl -X POST https://api.muapi.ai/api/v1/topaz-upscale-video-creative \
  -H "Content-Type: application/json" \
  -H "x-api-key: YOUR_API_KEY" \
  -d '{
    "video_url": "https://example.com/clip.mp4",
    "upscale_factor": 2,
    "creativity": 0.5
  }'
import requests

response = requests.post(
    "https://api.muapi.ai/api/v1/topaz-upscale-video-creative",
    headers={"x-api-key": "YOUR_API_KEY"},
    json={
        "video_url": "https://example.com/clip.mp4",
        "upscale_factor": 2,
        "creativity": 0.5,
    },
)
request_id = response.json()["request_id"]

result = requests.get(
    f"https://api.muapi.ai/api/v1/predictions/{request_id}/result",
    headers={"x-api-key": "YOUR_API_KEY"},
)
print(result.json())

Frequently Asked Questions

What is the Topaz Upscale Video API?

Muapi's Topaz video upscale API increases a video's resolution using Topaz's Creative (Astra 2), Generative (Starlight), and Precision (Proteus/Artemis/Iris/Dione/Gaia/Rhea/Theia) model families, all through a unified REST workflow.

How is Creative different from Generative or Precision?

Creative (Astra 2) leans furthest toward generating new stylized detail. Generative (Starlight) reconstructs realistic detail for degraded footage. Precision stays closest to the source and does not invent new detail.

Which model should I use for old or heavily compressed footage?

Topaz Upscale Video Generative's Starlight models are built for reconstructing clean, temporally consistent detail from degraded or low-resolution sources.

How is pricing calculated?

Cost is billed per 10 seconds of output at a 30fps baseline, tiered by output resolution — Creative costs the most due to its diffusion model, Precision the least.

Can I raise the frame rate at the same time?

Yes, set target_fps (16-60) on any of the three families to output at a higher frame rate than the source; higher fps increases cost proportionally.

Is there a webhook for completion?

Yes, pass webhook_url in the request body or as a ?webhook= query parameter and Muapi will POST the final result when processing completes.