Docs

Torch Object Storage

Media processing

Overview

TOS allows you to process video files stored in TOS. You can initiate video processing by including the appropriate parameters in requests made through the GetObject API. Supported video processing operations include extracting video frames and retrieving video metadata.

Constraints and limitations

Category	Description
Format	The video processing feature only supports specific video encoding and container formats. These are listed below: Video encoding formats: MPEG-4 Part 2, HEVC, AVI, VP8, VP9, H264, and H265. Video container formats: avi, mp4, mkv, mov, webm, flv, and ts.
QPS	The QPS limit for media processing operations in each region is 20. tip If you have a compute-intensive workload or usage that exceeds these limits, submit a ticket to consult with our technical team or your designated account manager.
Concurrency	The maximum concurrency for media processing operations in each region is 20. tip If you have a compute-intensive workload or usage that exceeds these limits, submit a ticket to consult with our technical team or your designated account manager.
Server-side encryption	You can use TOS to process videos encrypted with SSE-TOS, SSE-C, or SSE-KMS, but there are some limitations to consider: When processing a video encrypted with SSE-C, you must include the following SSE-C request headers: x-tos-server-side-encryption-customer-algorithm, x-tos-server-side-encryption-customer-key, and x-tos-server-side-encryption-customer-key-MD5. For more information on SSE-C headers, see Using server-side encryption with customer-provided keys (SSE-C). You cannot process videos encrypted with SSE-KMS via anonymous access. To process SSE-KMS-encrypted videos, you must activate Key Management Service (KMS) and have permission to access the master key. Requests must also be signed using valid credentials. You can use pre-signed URLs, APIs, or SDKs for this purpose. For details on using APIs and SDKs for video processing, see the Using REST APIs section below and Video Frame Extraction (Java SDK).
Mirroring	The mirroring feature does not work with the media processing feature.

Video processing methods

You can use REST APIs or file URLs to process videos stored in TOS buckets. This includes capturing a frame from a video, retrieving video metadata, and quickly retrieving processed results.

tip

The processed results are returned immediately and are not saved in your bucket.

Using REST APIs

Example:

Using file URLs

A video processing URL is in the following format: file URL?x-tos-process=video/feature,parameter name_parameter value. Separate multiple parameters in a URL with commas (,). Example:

tip

To process a private video, use an SDK to generate a pre-signed URL that includes the processing parameters. For how to generate pre-signed URLs in various programming languages, see the following documents:

Pre-signed URLs (Java SDK)

Pre-signed URLs (Go SDK)

Pre-signed URLs (C++ SDK)

Pre-signed URLs (Python SDK)

Pre-signed URLs (Node.js SDK)

Pre-signed URLs (Browser.js SDK)

Pre-signed URLs (.NET SDK)

Pre-signed URLs (PHP SDK)

Video processing features

TOS supports the following video processing features. If you specify multiple parameters in a request, TOS will process the video in the order of the parameters.

Feature	Parameter	Description
Snapshot	snapshot	Captures a frame at a specified time and resizes it to a specified size.
Info	info	Returns video metadata.

Last updated: 2026.02.04 17:11:22