[vlc-devel] [vlc-commits] display: convert picture in vout_display_Prepare()

[Rémi Denis-Courmont]

Le maanantaina 14. tammikuuta 2019, 13.04.12 EET Steve Lhomme a écrit :
> The API didn't have two pictures but it was still possible to hack

Hypothetical hacks are not a valid design reference.

Before this change, the interface required filtering pictures sequentially on 
the one hand, and preparing then displaying pictures sequentially on the other 
hand.

Now, it requires filtering, then preparing then displaying pictures 
sequentially. However, since the interface was never reentrant, you could not 
filter in parallel. And since the display pool was always small, you could not 
interleave either. So in effect, the requirement was already the same.

> Now that vout_display_Prepare() returns a picture it's trickier or we'd
> have to tell the prepare which eye(s) the prepare is for.

Filtering already returned a picture. Nothing new here. And I don't think 
passing separate pictures for separate channel is a good idea anyway.

Not that it matters. Comsumer stereoscopic video is dead. I just checked the 
largest electronic shop in town, and it was almost all about 4K and HDR. The 
only "3D" items I could find were replacement RC's.

-- 
Реми Дёни-Курмон
http://www.remlab.net/