Lossless depth buffer compression designed and tested for Microsoft Azure Kinect DK. Based on the Facebook Zstd library.
There is a lossy version of this library here that produces files half the size, in trade for slower compression and lower quality: https://github.com/catid/ZdepthLossy
The compressor defines a file format and performs full input checking. Supports temporal back-references similar to other video formats.
We do not use H.264 or other video codecs because the main goal is to measure the limits of lossless real-time compression of depth data. Another goal is to be 2x smaller than the best published software (RVL).
Algorithm:
- Quantize depth based on sensor accuracy at range.
- Compress run-lengths of zeroes with Zstd.
- For each 8x8 block of the depth image, determine the best predictor.
- Zig-zag encode, 12-bit pack, and compress the residuals with Zstd.
- Store residuals near edges in a different context than on surface.
The predictors are similar to the way the PNG format works.
For streaming depth compression, I believe this is currently the best open-source option available online.
It runs in 1-2 milliseconds and produces compressed lossless depth video at 4-5 Mbps. This is similar to the bandwidth needed for HD video (1080p@30FPS).
Detailed benchmarks and test vectors available at the end of the document.
This is useful as part of a larger real-time volumetric video streaming system enabled by the Microsoft Azure Kinect DK. The only data that needs to be sent for each frame are the compressed depth frames generated by this library, and the JPEGs from the camera.
The video recorder reads a 16-bit depth image that is compressed with this library. These depth images are transmitted to the video player computer over a network. The video parameter set must include the calibration data for the depth sensor. The video player uses this library to decompress the depth image and then uses the calibration data to generate xyz, uv, and indices for rendering the mesh.
There is software here that can convert from 16-bit depth to xyzuv/indices in only 3 milliseconds here: https://github.com/catid/ZdepthLossy
There's example usage in the tests folder.
This project provides a CMakeLists.txt that generates a zdepth target. Zdepth supports CMake find_package().
# Example CMake ExternalProject include
include(ExternalProject)
ExternalProject_Add(
zdepth
GIT_REPOSITORY https://github.com/catid/Zdepth.git
GIT_TAG master
GIT_SHALLOW TRUE
SOURCE_DIR ${CMAKE_SOURCE_DIR}/extern/src/zdepth
CMAKE_ARGS -DCMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE} -DCMAKE_INSTALL_PREFIX=${CMAKE_SOURCE_DIR}/extern/bin/zdepth
)
Link to the target and include the header:
#include "zdepth.hpp"
Compress 16-bit depth image to vector of bytes:
zdepth::DepthCompressor compressor;
...
std::vector<uint8_t> compressed;
const bool is_keyframe = true;
compressor.Compress(Width, Height, frame, compressed, is_keyframe);
Re-use the same DepthCompressor object for multiple frames for best performance.
Setting is_keyframe = false will use ~10% less bandwidth (e.g. 4 Mbps instead of 5 Mbps)
but has limitations just like normal video: You must guarantee delivery of all the frames
instead of dropping frames opportunistically.
Decompress vector of bytes back to 16-bit depth image:
zdepth::DepthCompressor decompressor;
int width, height;
std::vector<uint16_t> depth;
zdepth::DepthResult result = decompressor.Decompress(compressed, width, height, depth);
if (result != zdepth::DepthResult::Success) {
// Handle input error
}
You must use different objects for the compressor and decompressor, or it will fail to decode.
Format Magic is used to quickly check that the file is of this format.
Words are stored in little-endian byte order.
0: <Format Magic = 202 (1 byte)>
1: <Flags (1 byte)>
2: <Frame Number (2 bytes)>
4: <Width (2 bytes)>
6: <Height (2 bytes)>
8: <Zeroes Uncompressed Bytes (4 bytes)>
12: <Zeroes Compressed Bytes (4 bytes)>
16: <Blocks Uncompressed Bytes (4 bytes)>
20: <Blocks Compressed Bytes (4 bytes)>
24: <Edges Uncompressed Bytes (4 bytes)>
28: <Edges Compressed Bytes (4 bytes)>
32: <Surfaces Uncompressed Bytes (4 bytes)>
36: <Surfaces Compressed Bytes (4 bytes)>
Followed by compressed Zeroes, then Blocks, then Edges, then Surfaces.
The compressed and uncompressed sizes are of packed data for Zstd.
Flags = 1 for I-frames and 0 for P-frames.
The P-frames are able to use predictors that reference the previous frame.
The decoder keeps track of the previously decoded Frame Number and rejects
frames that cannot be decoded due to a missing previous frame.
For more details on algorithms and format please check out the source code. Feel free to modify the format for your data to improve the performance.
-------------------------------------------------------------------
Test vector: Room
-------------------------------------------------------------------
===================================================================
+ Test: Frame 0 Keyframe=true compression
===================================================================
Zdepth Compression: 184320 bytes -> 22626 bytes (ratio = 8.14638:1) (5.43024 Mbps @ 30 FPS)
Zdepth Speed: Compressed in 1.674 msec. Decompressed in 0.804 msec
Quantization+RVL Compression: 184320 bytes -> 40312 bytes (ratio = 4.57234:1) (9.67488 Mbps @ 30 FPS)
Quantization+RVL Speed: Compressed in 0.379 msec. Decompressed in 0.493 msec
Quantization+RVL+Zstd Compression: 184320 bytes -> 29460 bytes (ratio = 6.25662:1) (7.0704 Mbps @ 30 FPS)
Quantization+RVL+Zstd Speed: Compressed in 0.609 msec. Decompressed in 0.592 msec
===================================================================
+ Test: Frame 1 Keyframe=false compression
===================================================================
Zdepth Compression: 184320 bytes -> 18703 bytes (ratio = 9.8551:1) (4.48872 Mbps @ 30 FPS)
Zdepth Speed: Compressed in 1.471 msec. Decompressed in 0.679 msec
Quantization+RVL Compression: 184320 bytes -> 40188 bytes (ratio = 4.58644:1) (9.64512 Mbps @ 30 FPS)
Quantization+RVL Speed: Compressed in 0.446 msec. Decompressed in 0.491 msec
Quantization+RVL+Zstd Compression: 184320 bytes -> 29291 bytes (ratio = 6.29272:1) (7.02984 Mbps @ 30 FPS)
Quantization+RVL+Zstd Speed: Compressed in 0.683 msec. Decompressed in 0.593 msec
-------------------------------------------------------------------
Test vector: Ceiling
-------------------------------------------------------------------
===================================================================
+ Test: Frame 0 Keyframe=true compression
===================================================================
Zdepth Compression: 184320 bytes -> 16499 bytes (ratio = 11.1716:1) (3.95976 Mbps @ 30 FPS)
Zdepth Speed: Compressed in 1.463 msec. Decompressed in 0.845 msec
Quantization+RVL Compression: 184320 bytes -> 42084 bytes (ratio = 4.37981:1) (10.1002 Mbps @ 30 FPS)
Quantization+RVL Speed: Compressed in 0.376 msec. Decompressed in 0.461 msec
Quantization+RVL+Zstd Compression: 184320 bytes -> 23251 bytes (ratio = 7.9274:1) (5.58024 Mbps @ 30 FPS)
Quantization+RVL+Zstd Speed: Compressed in 0.69 msec. Decompressed in 0.583 msec
===================================================================
+ Test: Frame 1 Keyframe=false compression
===================================================================
Zdepth Compression: 184320 bytes -> 15495 bytes (ratio = 11.8955:1) (3.7188 Mbps @ 30 FPS)
Zdepth Speed: Compressed in 1.447 msec. Decompressed in 0.667 msec
Quantization+RVL Compression: 184320 bytes -> 41948 bytes (ratio = 4.39401:1) (10.0675 Mbps @ 30 FPS)
Quantization+RVL Speed: Compressed in 0.374 msec. Decompressed in 0.465 msec
Quantization+RVL+Zstd Compression: 184320 bytes -> 23143 bytes (ratio = 7.9644:1) (5.55432 Mbps @ 30 FPS)
Quantization+RVL+Zstd Speed: Compressed in 0.658 msec. Decompressed in 0.586 msec
-------------------------------------------------------------------
Test vector: Person
-------------------------------------------------------------------
===================================================================
+ Test: Frame 0 Keyframe=true compression
===================================================================
Zdepth Compression: 184320 bytes -> 23665 bytes (ratio = 7.78872:1) (5.6796 Mbps @ 30 FPS)
Zdepth Speed: Compressed in 1.442 msec. Decompressed in 0.69 msec
Quantization+RVL Compression: 184320 bytes -> 45016 bytes (ratio = 4.09454:1) (10.8038 Mbps @ 30 FPS)
Quantization+RVL Speed: Compressed in 0.422 msec. Decompressed in 0.506 msec
Quantization+RVL+Zstd Compression: 184320 bytes -> 33524 bytes (ratio = 5.49815:1) (8.04576 Mbps @ 30 FPS)
Quantization+RVL+Zstd Speed: Compressed in 0.67 msec. Decompressed in 0.61 msec
===================================================================
+ Test: Frame 1 Keyframe=false compression
===================================================================
Zdepth Compression: 184320 bytes -> 19848 bytes (ratio = 9.28658:1) (4.76352 Mbps @ 30 FPS)
Zdepth Speed: Compressed in 1.514 msec. Decompressed in 0.718 msec
Quantization+RVL Compression: 184320 bytes -> 45120 bytes (ratio = 4.08511:1) (10.8288 Mbps @ 30 FPS)
Quantization+RVL Speed: Compressed in 0.4 msec. Decompressed in 0.514 msec
Quantization+RVL+Zstd Compression: 184320 bytes -> 33667 bytes (ratio = 5.4748:1) (8.08008 Mbps @ 30 FPS)
Quantization+RVL+Zstd Speed: Compressed in 0.642 msec. Decompressed in 0.619 msec
Zdepth uses the Zstd library (slightly modified).
Inspired by prior work:
- Microsoft Fast Lossless Depth Image Compression (RVL library) available here
- Edges/Surface split context was inspired by BCIF image format.
- Trying different predictors was inspired by the PNG image format.
- Encoding zero run lengths mask was inspired by my previous work on the GCIF image format.
Software by Christopher A. Taylor [email protected]
Please reach out if you need support or would like to collaborate on a project.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent