Many web meeting services have a recording functionality. Most recordings are provided as MP4 files with the video encoded in H.264 because that offers the most universal compatibility. However, it also needs a lot of disk space. H.264 has a successor, H.265, which only requires half the space for the same visual quality. This post shows how to use StaxRip, a free tool, to re-encode H.264 video into H.265 quickly by making use of GPU hardware encoding.
- Download and extract StaxRip. I used the current stable version 1.7 x64
- Start StaxRip
- When opening the first video file, StaxRip may ask you to install AviSynth. Do so by clicking Install AviSynth+.
- Click x264 and choose one of the following depending on your GPU vendor: NVIDIA H.265, Intel H.265 or AMD H.265
- Click MKV and select MP4 (mp4box) instead
- Click the Opus entry next to the first audio stream field and select copy/mux
- Click the Opus entry next to the first audio stream field and select no audio
The result should look like this:
- Right-click Source > Open > File Batch and select the files you want to convert
- Click Next to start the conversion
- The output files are placed in the same directory as the input files with the extension _new
The Nvidia GTX 1060 GPU in my desktop PC encoded H.265 at the impressive rate of approximately 420 frames per second (full HD, 1920×1080).
The Intel HD Graphics 620 (Core i7-7500U) in my laptop only reached about 113 frames per second for the same content. Still impressive, but a lot less fast.
An interesting difference between the two GPUs: while the Nvidia encode used the GPU’s dedicated video encoding engine, the Intel encode used the GPU’s generic 3D engine.
Another noteworthy difference: the file generated by the Intel encode was 38% smaller than the file generated by the Nvidia encode.
The original videos of a four-day training recorded with Skype had a size of 7.6 GB. Converted to H.265 the size was reduced to 2.4 GB, which amounts to 68% savings!
If you are interested in monitoring your GPU’s performance and find out how its various engines are used, take a look at our uberAgent product. During the Nvidia encoding, for example, the GPU’s video encoding engine was nearly at 100% load and its generic compute engine at approximately 20%: