muapi Logomuapi Logo
muapi Logomuapi Logo

Topaz Upscale Video Precision: Trình chỉnh sửa video AI

Upscale video faithfully with Topaz Proteus, Artemis, Iris, Dione, Gaia, Rhea & Theia models. Try free in the Muapi playground — pay per generation. Topaz Upscale Video Precision enlarges video while staying faithful to the source, using Topaz's full precision model library: Proteus (general-purpose default) and Proteus Natural, Iris and Iris Low Quality, the Dione family (DV/TV/Robust/Dehalo variants) for interlaced and broadcast sources, Artemis for animation and CGI (with High/Medium/Low Quality and halo/aliasing-focused variants), Gaia HQ/CG/2 for computer-generated content, Rhea for general enhancement, and Theia Fine Tune Detail/Fidelity for fine control. Compression, noise, halo reduction, grain, and detail-recovery are all independently adjustable, with upscalefactor up to 4x and frame rate up to 60fps via targetfps. Reach for Topaz Upscale Video Generative instead when the source is too degraded for a faithful pass, or Topaz Upscale Video Creative for a more stylized, generated look.

Điều khiển mô hình tương tác

Upscale video faithfully with Topaz Proteus, Artemis, Iris, Dione, Gaia, Rhea & Theia models. Try free in the Muapi playground — pay per generation.

📝

Tổng quan

Về mô hình này

Topaz Upscale Video Precision enlarges video while staying faithful to the source, using Topaz's full precision model library: Proteus (general-purpose default) and Proteus Natural, Iris and Iris Low Quality, the Dione family (DV/TV/Robust/Dehalo variants) for interlaced and broadcast sources, Artemis for animation and CGI (with High/Medium/Low Quality and halo/aliasing-focused variants), Gaia HQ/CG/2 for computer-generated content, Rhea for general enhancement, and Theia Fine Tune Detail/Fidelity for fine control. Compression, noise, halo reduction, grain, and detail-recovery are all independently adjustable, with upscale_factor up to 4x and frame rate up to 60fps via target_fps. Reach for Topaz Upscale Video Generative instead when the source is too degraded for a faithful pass, or Topaz Upscale Video Creative for a more stylized, generated look.

1Broadcast and interlaced footage: Use the Dione family to clean up interlaced or broadcast-sourced video.
2Animation and CGI: Use Artemis or Gaia variants tuned for animation, game capture, and 3D renders.
3General clean-source upscaling: Use the default Proteus model for straightforward resolution upgrades.
4Fine-grained control: Use Theia Fine Tune Detail/Fidelity when you need to dial in exact compression, noise, and halo settings.
💰

Giá cả & Giá trị

Phân tích chi phí

muapiapp$0.10-$0.60+ per 10 seconds of output depending on resolution (plus margin), billed dynamically

Thanh toán theo mức sử dụng, không cần đăng ký thuê bao. Chi phí được tính theo độ phân giải/thời lượng đầu ra thực tế tại thời điểm gửi yêu cầu.

Fal.aiKhông khả dụng

Đối tác hạ tầng thượng nguồn của muapiapp hiện không cung cấp dưới dạng API công khai độc lập.

ReplicateKhông khả dụng

Hiện chưa được niêm yết trên Replicate.

** Giá của đối thủ cạnh tranh được ước tính dựa trên kiến ​​trúc mô hình và cấp độ sử dụng tương tự.

⚙️

Chi tiết kỹ thuật

Lược đồ cấu hình

Video URLstring

Input video to upscale.

Giá trị mặc địnhhttps://d3adwkbyhxyrtq.cloudfront.net/webassets/videomodels/topaz-upscaler-in.mp4
modelEnum (tùy chọn 21)

Topaz precision upscale model variant (Proteus/Iris/Dione/Artemis/Gaia/Rhea/Theia families).

Giá trị mặc địnhProteus
upscale_factornumber

Factor to upscale the video by.

Giá trị mặc định2
Target FPSint

Target output frame rate.

Giá trị mặc định-
Compressionnumber

Compression-artifact reduction amount (0-1).

Giá trị mặc định-
Noisenumber

Noise reduction amount (0-1).

Giá trị mặc định-
Halonumber

Halo/ringing reduction amount (0-1).

Giá trị mặc định-
Grainnumber

Film-grain amount to add back (0-0.1).

Giá trị mặc định-
Recover Detailnumber

Detail recovery amount (0-1).

Giá trị mặc định-
H264 Outputboolean

Encode output using H.264 instead of the default codec.

Giá trị mặc địnhfalse
📖

Hướng dẫn thực hiện

Tài liệu dành cho nhà phát triển

How to Use Topaz Upscale Video Precision

  1. Provide a video: Set video_url to a publicly accessible video file URL.

  2. Choose a model: Proteus is the default general-purpose choice; use the Dione family for broadcast/interlaced sources, or Artemis/Gaia variants for animation and CGI.

  3. Set the scale: upscale_factor accepts 1-4x.

  4. Tune fidelity controls: compression, noise, halo, grain, and recover_detail (each 0-1, or 0-0.1 for grain) fine-tune the faithful upscale.

  5. Submit and poll: POST to /api/v1/topaz-upscale-video-precision, then poll /api/v1/predictions/{request_id} or use webhook_url.

curl -X POST https://api.muapi.ai/api/v1/topaz-upscale-video-precision \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "video_url": "https://example.com/clip.mp4",
    "model": "Proteus",
    "upscale_factor": 2
  }'
❓

Câu hỏi thường gặp

Thường xuyên hỏi

What does Topaz Upscale Video Precision do?

It enlarges video up to 4x while staying faithful to the source, using Topaz's precision model library (Proteus, Artemis, Iris, Dione, Gaia, Rhea, Theia).

Which model should I use for animation or game footage?

Use one of the Artemis variants (tuned for animation/low-noise sources) or Gaia HQ/CG/2 for computer-generated content.

What is the Dione family for?

The Dione models (DV, TV, Robust, Dehalo, Robust Dehalo) are tuned for interlaced and broadcast-sourced video, with dedicated halo-reduction variants.

How is pricing calculated?

Cost is billed per 10 seconds of output at a 30fps baseline, tiered by output resolution (720p/1080p/4K); Proteus Natural and Gaia 2 have slightly different 4K rates than the rest of the family.

Can I reduce noise and compression artifacts?

Yes — `compression` and `noise` (both 0-1) independently reduce compression artifacts and sensor/video noise as part of the upscale.

Can I add film grain back in?

Yes, `grain` (0-0.1) re-introduces a controlled amount of film grain to the upscaled output.