New Windows App does not have ability to render ASS subtitles at video resolution instead of display

[cubatilles 3]

Hello,

I've switched from Emby Theater to Emby for Windows since some weeks ago. In general the app works ok, but I've figured I'm not able to use anymore a custom mpv.conf on the new app.

Previously, in Theater, I was using a custom mpv.conf and set the render to gpu (instead of gpu-next) because in gpu mode there was a mpv parameter, called "blend-subtitles", which I had setup to "blend-subtitles=video". With this parameter, ASS subtitles are rendered to video's native resolution despite of your display's resolution. This is important for ASS anime subtitles, because if they are rendered at display resolution (4K in my case), some effects (like blending on font borders, for example) are rendered incorrectly (basically blending effects are less visible). To avoid this problem mpv.conf has this parameter, "blend-subtitles=video", which forces subtitles to be rendered to the video resolution, whichever your display resolution is. This ensures ASS subtitles and its effects are correctly rendered as they are meant to.

Is there any way to support this behavior in new Emby for Windows app? And, in case it's not supported... Any other workaround? Any plans to support it? I don't care if custom mpv.conf is not available anymore, but would like an option to render subtitles at video resolution.

Thanks for your support.

 

[softworkz 4534]

Hi,

it is not quite clear to me what you're up to.

The subtitles are sized to the video content area, not to the app window's size:

 

 

[pwhodges 1833]

Obviously positioning and movement information is scaled from the video pixels to the display pixels at some stage.  The suggestion is that for some reason blending effects are not so scaled - which frankly I find hard to believe (the fact that MPV might have a parameter related to this surely simply points to its own deficiency, or maybe it's just too clever for its own good).

At this stage I'd want to see some ASS code that demonstrates the issue and images showing the differences in the two MPV renderings and the Emby one.

Paul

[softworkz 4534]

8 hours ago, pwhodges said:

Obviously positioning and movement information is scaled from the video pixels to the display pixels at some stage.  The suggestion is that for some reason blending effects are not so scaled - which frankly I find hard to believe (the fact that MPV might have a parameter related to this surely simply points to its own deficiency, or maybe it's just too clever for its own good).

The setting controls at which stage of processing the subtitles are applied. AFAIK, "blend-subtitles=video" causes the ass subtitles to be "burnt-in" to the video before it is going to be rendered on screen while otherwise, the subtitles are blended via GPU composition during rendering without ever changing the video frames.
The latter approach is much more efficient because it doesn't need to perform processing on each video frame which is very costly in terms of performance.

I'm not aware of "blending issues", the one difference I'm aware of is that with "blend-subtitles=video", the subtitles will be subject to the same color corrections and tone mapping like the video itself which isn't the case normally (and not necessarily desired as well).

9 hours ago, pwhodges said:

At this stage I'd want to see some ASS code that demonstrates the issue and images showing the differences in the two MPV renderings

Yes, that would be interesting.

9 hours ago, pwhodges said:

the two MPV renderings and the Emby one.

The latter is the same as one of the two.