A user recently reported that they felt our subtitle sizing was too large for text-based SSA subtitles.
Here's a comparison between MC and MPC-HC playing the same clip at 1080p (resized by 50% so it fits better here):
I don't think there's really a "right" answer on this one, but let me explain our subtitle sizing logic and we can discuss if there are any improvements that could be made.
The sizing works like this:
1) Use 9.5% of the height of the video for a normal text subtitle (this does not include black bars, unless the bars are baked into the video)
2) If the user sets the size setting, scale that 9.5% by the user setting
3) If the subtitle file uses varying sizes, pick the median size and scale all the titles relative to that
As an example of #3, the video above uses these sizes:
20, 28, 30, 40, 40, 50, 50, 50, 50
These might be point sizes or something, but they don't mean too much to us since we're rendering at varying sizes and always have pixel-perfect rendering (as opposed to MPC-HC that scales the subtitles with the video, at least that I can tell).
So we pick 40 as the normal mark (that's the median) and scale a 50 (what is shown in the screenshot) by 1.25.
The result is that the subtitle above uses 11.875% of the height of the video for each character (9.5% * 100% user setting * 125% for SSA specifier).
Is there anything that should be changed in that logic? Should the 9.5% thing change as the video changes size? Should it consider width as well as height? Any other tips?
Thanks for any help.