Data Requirements
Our software is optimized for the following data specifications.
Faces
- Size: Faces need to have an area of at least 200 pixels (e.g. 10x20 pixels) for both DNAT and Speed-optimized Blur. Precision Blur is exempt from these limitations.
License Plates
- Size: License plates need to have an area of at least 100 pixels (e.g. 20x5 pixels) for both DNAT and Speed-optimized Blur. Precision Blur is exempt from these limitations.
- Supported countries: Optimized for license plates from Europe, China, Japan and US. Other regions are available upon request.
Supported Input Formats
Type | Codec/format | Filetype |
---|---|---|
Archive | tarball (default) | .tar (default) |
Image | jpeg | .jpeg, .jpg |
Image | png (default) | .png (default) |
Image | bitmap | .bmp |
Video | h264 (default) | .mp4 (default), .avi, .mov, .mkv, .ts, .mts |
Video | mpeg4 | .mp4, .avi, .ts |
Video | mjpeg | .mp4, .avi |
Video | hevc | .mp4, .mov, .mkv, .ts |
Video | vp9 | .webm, .mkv |
Archive
Please use tar only for video files without subfolders, with all frames belonging to the video in the top-level directory, in order of video. Frames inside the archives are processed alphabetically.
E.g.
cd video_folder
tar cf ../video.tar *.png
On macOS devices, it may be necessary to add COPYFILE_DISABLE=1
E.g.
COPYFILE_DISABLE=1 tar cf ../video.tar *.png
(Official support for videos up to 1920 × 1920 pixel. Shorter or split videos decrease the processing time.)
Supported Audio Codecs
- AAC
- MP3
- OPUS
Supported Input / Output Combinations
Input / Output | image | video | archive | overlay |
---|---|---|---|---|
image | Yes | No | No | Yes |
video | No | Yes | Yes | Yes |
archive | No | Yes | Yes | Yes |
Updated 14 days ago