Tuning x264 and Per-Title Ladders Cuts Bandwidth 33% for a Fraction of the Encoding Cost
Optimizing x264 settings and per-title ladders
Testing x264 with FFmpeg on 13 clips across genres, the Streaming Learning Center found that switching from a fixed Netflix-style ladder to a tuned per-title ladder raised VMAF by 0.66 points while dropping average bandwidth from 4.9 to 3.27 Mbps, a 33% savings. Encoding time tripled, but the extra cost is recovered after just 186 hours of viewing, making optimization a clear win for all but the smallest shops.
As a codec researcher, you hate to spend weeks of testing only to reach the obvious answer; while there are multiple configuration options you can adjust for minor gain, the most effective adjustment you can make is to implement some form of per-title encoding.
- dylan604
It's been a long time since I've had to worry about encoding parameters, but even these settings feel like a set it and forget it compared to what we used to do in the past. Comparing encoding for streaming vs shiny round discs is so different. We would do scene by scene encodes with a minimum of 2-passes allowing for even more when necessary. We were strict on controlling the VBV but these formats had more bandwidth than streaming and were not susceptible to congestion and buffering issues. The video was even sent to QA to look for any kind of issues with the encoding. I doubt anybody looks at the best quality of the ladder let alone all of the variations before it reaches the users.
- netsharc
Gotta love an article about video compression where the top image is a blurry downscaled-and-then-upscaled image, that's not even clickable to zoom.
Yes, if you scroll down the image is there again, still a bit blurry (so much winning), but clickable to embiggen. Your CMS is giving everyone a terrible first impression!
- antisthenes
I'll save everyone the time:
You gain ~0.7 VMAF score (both above 90) and save 33% bandwidth at the cost of 3.2x longer encoding for (presumably) some set of 1080p clips used in testing.