Bug 138657 – Video as image

• Status: NEW
• Component: Layout and Rendering

[Antti Koivisto]

Support real video formats wherever images are allowed (<img>, CSS backgrounds, etc.) Semantics like animated GIF.

[Antti Koivisto]

Created attachment 241422 [details]
wip patch

[Eric Carlson]

Comment on attachment 241422 [details]
wip patch

View in context: https://bugs.webkit.org/attachment.cgi?id=241422&action=review

Nice!

> Source/WebCore/platform/graphics/VideoImage.cpp:84
> +    if (m_player->readyState() >= MediaPlayer::HaveEnoughData)
> +        m_frameTimer.startOneShot(16_ms);
> +
> +    if (m_player->duration() > MediaTime::zeroTime() && m_player->currentTime() >= m_player->duration()) {
> +        fprintf(stderr, "seeking %d\n", m_player->readyState());
> +        m_player->seek(MediaTime::zeroTime());
> +    }

It seems like we should seek to the next frame if possible instead of using a hard coded interval. How about adding MediaPlayer::step(int numberOfFrames) - and have the default MediaPlayerPrivate implementation seek 16_ms * numberOfFrames so it works like this does until ports implement stepping logic?

[Antti Koivisto]

> It seems like we should seek to the next frame if possible instead of using
> a hard coded interval. How about adding MediaPlayer::step(int
> numberOfFrames) - and have the default MediaPlayerPrivate implementation
> seek 16_ms * numberOfFrames so it works like this does until ports implement
> stepping logic?

Makes sense. I was actually trying to find a "new frame available" callback so the media player could drive the paints. MediaPlayerClient::mediaPlayerRepaint sounded like it but it didn't seem to actually get called. The timer was more of a hack.

[Antti Koivisto]

Created attachment 241448 [details]
wip, rebased and with actual mime type handling

[Antti Koivisto]

Created attachment 245080 [details]
attempt to use media source extensions