mendevi 1.3.2

Video encoding and decoding measures.

Useful links: Binary Installers | Source Repository | Online Documentation |

Description

This Python module performs energy and metrics measurements on video, for encoding and decoding. It also provides several detailed dataset and a visualisation tool that generates complex matplotlib figures.

It manages the following parameters:

1. It supports the libx264, libopenh264, libx265, libvpx-vp9, libaom-av1, libsvtav1, librav1e and vvc cpu encoders.

2. It supports the h264_nvenc, hevc_nvenc, av1_nvenc and *_vaapi gpu encoders.

3. Distortions are measured using the lpips, psnr, ssim, vif and vmaf metrics.

4. Complexity are measured using the rms_sobel and rms_time_diff metrics.

5. Encoding efforts are fast, medium and slow.

6. It takes care about the colorspaces (range, transfer and primaries).

7. Iterate over different effort, encoder, mode, quality, threads, fps, resolution and pix_fmt.

8. Energy measurements are catched with RAPL and an external wattmeter on grid'5000.

9. Get the cpu, gpu, ram and temperature activity.

10. Get a full environment context, including hardware and software version.

11. It support the mode (constant bitrate) cbr and (constant quality) vbr.

12. Ability to modify ffmpeg commands on the fly to perform specific tests.

13. It take care to transfer files to RAM if possible to avoid biases related to storage space access.

14. Provides a guide to compile ffmpeg with all optimizations in order to compare encoders/decoders at their limits.

Pipeline

This is the pipeline used for measurements:

Example of result

Example of rate distortion curve:

mendevi plot mendevi.db -x bitrate -y psnr -y ssim -wx profile -c encoder

Example of energy per encoder:

mendevi plot mendevi.db -x quality -y energy -wx profile -wy mode -c encoder -m effort

Alternatives

1. The GREEM video encoding measurement tool.

2. The MVCD database also includes video encoding and decoding energy measurements.

3. The COCONUT database also includes video decoding measurements.

4. The SEED and VEED dataset offers a comprehensive LCA and GPU measurements.

5. The CTC videos and Big Buck Bunny are used for the tests.