This web page was created to accompany the webinar, Fundamentals of H.264 Encoding, produced on September 4, 2013. It includes sample files assessing the quality of output produced using the Baseline, Main and High profiles.
The problem on the table is this; if you produce a set of adaptive files for mobile distribution, you must use the Baseline and Main profiles for compatibility with older iDevices. However, computers and OTT devices can play videos encoded using the High profile. Is the qualitative difference between files encoded using the Baseline and Main profiles, and those encoded using the High profile, sufficient to warrant producing two separate sets of files?
To add a real world element to the analysis, I’ve included Apple’s recommendations from TN 2224. I encoded one set of files using the recommended parameters (with some minor data rate adjustments), the other using the High profile at the same data rates, and provide four comparisons below. I encoded the files using Elemental’s Cloud Encoder and I’m making them available via Sorenson 360, which doesn’t re-encode the files like most OVPs. There are no still image comparisons, as I’ve used in the past, but it should be easy enough to open two browser windows and compare the files side by side.
Table 1. Encoding Recommendations from Apple Tech Note TN2224.
416×234@200kbps Baseline Profile
416x234@200kbps High Profile
480x270@400kbps Baseline Profile
480x270@400kbps High Profile
640x360@600kbps Baseline Profile
640x360@600kbps High Profile
960x540@1400kbps Baseline Profile (click to full screen to compare)
960x540@1400kbps High Profile (click to full screen to compare)
While you can see minor differences between the files, they don't appear to be as dramatic as one would expect. From where I sit, the difference doesn't warrant two separate encodes, one for computers and OTT, the other for mobile. However, your results may vary depending upon the encoding tool that you use, your encoding parameters and your source footage. My strongest recommendation is that you don't assume that separate files are required for optimal quality; run similar comparisons using your encoding parameters and encoder, and representative source footage, and see what it looks like to you.
Note that I performed a similar analysis here. The encoding parameters were different, and there are still images for comparison purposes, but the comments and debate provides the real value.