When captions are played in full screen mode, they are about half the size we would expect. It looks like we are configuring the text track representation's graphics context to scale by the device scale factor, but later we apply that device scale factor again when generating the final image we pass to the full screen compositing step.
<rdar://problem/16507482>
Created attachment 230685 [details] Patch
Comment on attachment 230685 [details] Patch r=me
Committed r168192: <http://trac.webkit.org/changeset/168192>
Re-opened since this is blocked by bug 132498
Created attachment 230726 [details] Patch
Comment on attachment 230726 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=230726&action=review > Source/WebCore/platform/graphics/avfoundation/objc/MediaPlayerPrivateAVFoundationObjC.mm:1605 > + if (representationLayer == m_textTrackRepresentationLayer) { > + if (m_videoFullscreenLayer && m_textTrackRepresentationLayer) { > + // Keep our text track layer in sync with the video layer at all times > + CGRect textFrame = m_videoLayer ? [m_videoLayer videoRect] : CGRectMake(0, 0, m_videoFullscreenFrame.width(), m_videoFullscreenFrame.height()); > + [m_textTrackRepresentationLayer setFrame:textFrame]; > + } This is good, but it will miss when the video changes sizes during a single cue display. (I.e., changing rotation while a cue is up will not cause a re-layout.) Brent and I discussed over IRC and alternate possibility for fixing this.
Created attachment 230738 [details] Patch
Comment on attachment 230738 [details] Patch Nice! r=me
Comment on attachment 230738 [details] Patch Clearing flags on attachment: 230738 Committed r168245: <http://trac.webkit.org/changeset/168245>
All reviewed patches have been landed. Closing bug.