Help
Information on content playback.
Machine Requirements
For playback at the high-quality video resolution 7680x3840 (8k), your machine must have a capable enough GPU.
Tested Systems
System A
- OS: Windows 11
- GPU: RTX 2060
- CPU: AMD Ryzen 5 3600 6-Core / Intel-i7
- Memory: 16.0 GB
System B
- OS: Windows 10
- GPU: RTX 3080 10GB 2.0 PCIe 4.0 Gigabyte
- CPU: AMD Ryzen 7 5800X AM4
- Memory: 32 GB DDR4-3200
Audiovisual Spatial Alignment
4th-order Ambionsics .WAV files have a -90° spatial rotation in Yaw. This is to spatially align audio with Unity's projection mapping of 360° images using skyboxes.
Read below for explination.
Playback Compatability
Many 360° video players have limitations in either the maximum video resolution, or spatial audio compatability. There are few plug-and-play tools available that will be compatible with both higher-order Ambisonics above 3rd order, and high-resolution video files above 4K.
To render higher-order Ambisonics, many VST plug-ins are available that can be used to decode and render Ambisoncs audio channels independently of the video. See the SPARTA spatial audio real-time applications, or IEM Plug-in suite for more information. To render 8K video, the Unity game engine was used in conjuction with the VideoPlayer component.
Using Unity
Unity can be used to send real-time head tracking information to incoming OSC ports of any of the spatial audio plug-ins above, to rotate the Ambisonics audio with respect to head movements. The main draw-back of this approach is that the unwraping of a 360° image to a Skybox in Unity will cause a +90° shift in Y rotation. As the audio is rendered independently, it is unaware of this shift resulting in audio visual spatial misalignment of 90°.
Current Implementation
The current solution provided by default is a -90° rotational shift in the 4th-order Ambisonics audio files. When used in conjuction with Unity and the Unity VideoPlayer component, audio and video will be already correctly alligned. The 1st-order Ambisonics contained in the YouTube videos do not have this spatial rotation.
To use the 4th-order Ambisoncis with no spatial rotation, re-render the audio files using a Ambisonics scene rotator (see above plug-ins) with a +90° rotation in Yaw.
Creating Degradations
Many studies wish to use databases to investigate the effect of changing paramaters, often representative of the processing or transmission chain, on objective or subjective quality measures. Here's a few examples of how to create degaraded versions of the provided database for both audio and video.
Video
Examples of creating video degradations using FFmpeg
Reducing the bit-rate
ffmpeg -i Input_file.mkv -preset slow -c:v hevc -x265-params "pass=1" -b:v 25000k Output_file_Pass1.mp4
ffmpeg -i Input_file.mkv -preset slow -c:v hevc -x265-params "pass=2" -b:v 25000k Output_file_Pass2.mp4
Reducing the framerate
ffmpeg -i Input_file_60fps.mp4 c:v libx265 -crf 18 -r 30 Output_file_30fps.mp4
Reducing the resolution
ffmpeg -i Input_file_7680x3840.mkv -c:v libx265 -crf 1 -vf scale=3840x1920 Output_file_3840x1920.mkv