http://recursive-design.com/blog/2012/07/28/exif-orientation-handling-is-a-ghetto and I concur. CG added EXIF support in bug 19688 for imageDocuments and for page <img> via the page WebKitShouldRespectImageOrientation setting. We should add this EXIF support to the chromium port for imageDocument only, with the page setting to permit testing (leave the setting off-by-default) since auto-orientation of page <img> tags should be opt-in (author controlled) ... [css3-images] Honoring EXIF orientation from images http://lists.w3.org/Archives/Public/www-style/2011Dec/0001.html [css3-images] Processing model for transformations (image() vs. image-orientation) http://lists.w3.org/Archives/Public/www-style/2011Dec/0002.html http://dev.w3.org/csswg/css4-images/#image-orientation img { image-orientation: from-image } So in this bug, be aware of the CSS image-orientation work bug 89052 and how the platform code needs updating to examine an image renderer->style()->imageOrientation() at draw time, see bug 91566 and bug 92330, let's not break them.
The image-only exif rotation bits are done in bug 100164 and bug 100179 (and http://trac.webkit.org/changeset/132260). From what I can tell, this should not collide with the bugs linked above.
Bug 100200 will clash with one of davidbarr's changes, but in a minor way (the code it adds will need to grow 2 additional lines to deal with the passed-in orientation he's adding.)
This is mostly done.
http://crbug.com/56845 marked fixed, all done for Chrome M24.