Digital watermarking is a technology that embeds digital information, known as a digital watermark, into media files such as audio and video. This process does not affect the visual quality or integrity of the content and is used for copyright protection, leak tracing, and file verification. This topic explains how to add digital watermarks to your video files to protect your copyright.
How it works
How copyright watermarks work
This method embeds custom text information that is invisible to the naked eye into a video. After the video is distributed or processed, you can extract and identify the embedded text to determine the video's copyright ownership.
How tracing watermarks work
At playback, the system uses pre-generated A/B stream segments of the same time interval to arrange and combine them in real time. This process generates a unique manifest (M3U8) file. The sequence of the A/B streams in the file represents a unique identifier.
Scenarios
Digital watermarks help you combat piracy and protect your video copyright. They are effective against various tampering methods. This helps you secure revenue from your content and protects the interests of everyone from content creators to distributors.
Copyright protection: Use the video copyright watermark technology. You can add a copyright watermark, such as a unique ID or a text label, to your copyrighted videos. If a copyright dispute occurs, you can use extraction and identification tools to retrieve the ID and confirm ownership.
Tracing leaks from organizations: Use the video copyright watermark technology. You can provide different organizations with video copies that have unique watermarks embedded. If a video is leaked, you can detect the watermark to identify the source of the leak.
NoteTracing leaks from organizations is a scenario for tracing copyright leaks within a small group, such as tracing leaks during the distribution of a movie before its official release. A film distributor sends copies to several organizations, typically a few to a dozen. If the movie is leaked and illegally distributed, the copyright holder needs to determine which organization was the source. This allows them to stop the leak and hold the responsible parties accountable.
Tracing leaks from users: Use the video tracing watermark (A/B stream watermark) technology. This method generates a unique video stream for each user by combining A/B stream segments. If a video is pirated, you can detect the watermark to trace the user who leaked it and take appropriate action.
NoteA common type of video leakage occurs when end users use technical methods to download or record videos. Digital watermarking technology can trace leaks back to the specific user.
Benefits
Visually appealing: The watermark is invisible and does not affect the normal playback of the media.
Content security: The watermark is imperceptible, which makes it difficult to attack directly. The embedded information is specially encoded, which makes it hard to tamper with or forge.
Accurate extraction: The specially encoded watermark ensures accuracy during extraction. The extracted information is both unique and correct.
Robust: The watermark is strong and can withstand common piracy attacks. These attacks include video compression, rotation, resolution changes, frame rate changes, screen recording, cropping, and scaling.
Encrypted: The watermark information is specially encoded and encrypted. This makes it difficult to modify and ensures its accuracy upon extraction.
Features
Feature | Description | Related API operations | Console operations |
Transparent image watermark | Generates a transparent image containing an invisible watermark. Place this image as the top layer on a page to defend against attacks such as screenshots. | Not supported | |
Video copyright watermark | Embeds specified invisible text information into a video. The embedded digital watermark is hidden in the media file and does not affect the video's visual quality or integrity. The embedded text can still be extracted and identified even after the video is distributed or processed. | Not supported | |
Video tracing watermark | Transcodes a video (only videos of 3 minutes or longer are supported) into two HLS files: stream A and stream B. Different tracing watermark content is embedded into each stream. When a user requests playback, the system randomly selects and combines segments from the A and B streams to generate a unique M3U8 file in real time. This file corresponds to the user's identity. If video piracy occurs, you can detect the watermark to trace the user who leaked the video and take appropriate action. | Not supported |
Limits
The digital watermark feature is available only in the China (Shanghai), China (Beijing), Singapore, and US (Silicon Valley) regions.
The following requirements apply to the input video when you extract a digital watermark:
The robustness of the watermark extraction algorithm, including its resistance to compression bitrates and support for screen recording, depends on the duration of the video being extracted. The requirements in the following table apply to videos with relatively static scenes, such as online courses. This is because video characteristics such as motion speed and texture complexity affect watermark robustness. For videos with highly dynamic scenes, such as action movies, you must use a video bitrate higher than the values in the table.
Video duration (minutes)
6 ≤ duration < 10
10 ≤ duration < 20
duration ≥ 20
Video bitrate (Kbps, recommended)
≥ 2000
≥ 1000
≥ 500
Supports extraction from screen-recorded videos (recorded by phones, cameras, etc.)
The video duration is too short. Extraction from screen-recorded videos is not recommended.
Screen recording is a strong attack that can severely damage the watermark. For extraction from screen-recorded videos, we recommend a video bitrate greater than 3000 Kbps.
Billing
For more information about billing, see Digital watermark billing.
How to use
You can add or extract watermarks as needed. The following sections describe the end-to-end process, from adding a watermark to extracting it.
Use video copyright watermarks
Add a video copyright watermark
Before you distribute a video, you can follow these steps to add a copyright watermark:
Submit a job to add a copyright watermark.
You can call the SubmitCopyrightJob operation to submit the video URL and the watermark information to embed. After the watermark is added, the new video is saved to the location specified by the Output parameter. For more information, see Add a copyright watermark to a video.
You can retrieve the result of the video copyright watermarking task.
You can call the QueryCopyrightJob operation to query the status and result of the job. The response includes the save location of the new video. For more information, see Query a video copyright watermark job.
If you specified a webhook address in Step a, a request that contains the execution result is sent to the specified address after the job is complete. The callback result includes the save location of the new video.
Extract the copyright watermark from a video file
When you need to confirm copyright ownership, you can follow these steps to extract the watermark from a video file:
Submit a copyright watermark extraction job.
You can call the SubmitCopyrightExtractJob operation to submit information about the video file from which to extract the watermark. For more information, see Extract a copyright watermark from a video.
You can retrieve the result of the video copyright information extraction task.
You can call the QueryCopyrightExtractJob operation to retrieve the result of the extraction job. The result contains the watermark information. For more information, see Query a copyright watermark extraction job.
If you specified a webhook address in Step a, a request that contains the execution result is sent to the specified address after the job is complete. The execution result contains the watermark information.
Use video tracing watermarks
Generate A/B streams
Submit a job to generate A/B streams for a tracing watermark.
You can call the SubmitTraceAbJob operation to submit the video URL and the A/B stream watermark information to embed. After the watermark is added, the new video is saved to the location specified by the Output parameter. For more information, see Add an A/B stream tracing watermark to a video.
Retrieve the result of the A/B stream job.
You can call the QueryTraceAbJob operation to query the status and result of the A/B stream job. You can query by task ID or media ID. For more information, see Query a video tracing watermark A/B stream job.
Request a unique M3U8 sequence at playback
Submit an M3U8 job for a tracing watermark.
When a user requests playback, you can call the SubmitTraceM3u8Job operation. You can use the media ID of the generated A/B streams to assemble a unique M3U8 watermarked video for the user to play. For more information, see Add an M3U8 tracing watermark to a video.
Retrieve the M3U8 result for the video tracing watermark.
Query the job result. You can call the QueryTraceM3u8Job operation to query the status and result of the M3U8 job. You can query by task ID. For more information, see Query a video tracing watermark M3U8 job.
If you specified a webhook address in Step a, a request that contains the execution result is sent to the specified address after the M3U8 job is complete. The execution result contains the watermark information.
Extract the tracing watermark from a video file
Submit a tracing watermark extraction job.
For a video that is uploaded to OSS or a video URL from a third-party site, you can call the SubmitTraceExtractJob operation to submit a tracing watermark extraction job. For more information, see Extract a tracing watermark from a video.
Retrieve the job result.
You can call the QueryTraceExtractJob operation to retrieve the result of the tracing watermark extraction job. The result contains the watermark information. For more information, see Query a tracing watermark extraction job.
If you specified a webhook address in Step a, a request that contains the execution result is sent to the specified address after the extraction job is complete. The execution result contains the watermark information.