1965813 - Add the possibility to decode jpeg cmyk image coming from pdfs

Status: RESOLVED FIXED

(Core :: Audio/Video: Web Codecs, enhancement)

Description

[Calixte Denizet (:calixte)]

Attachment: invitation.pdf

Jpeg cmyk images in pdfs have inverted channels.
There are no specific markers which could help to guess if the channels are inverted or not: the only reason why they're inverted is because the image has been embedded in a pdf.
The attached pdf contains an few cmyk jpeg images. Right now we're falling back on pure js jpeg decoder and then we have to apply a cmyk color profile on the pixels in order to get the RGBA images: it's very slow.

Comment 1

[Calixte Denizet (:calixte)]

Attachment: Bug 1965813 - Add the possibility to decode cmyk jpeg images coming from pdf files r=aosmond

Comment 2

[Calixte Denizet (:calixte)]

I don't know if this is the right approach: maybe adding an option in the ImageDecoderInit dictionary would be better.

Comment 3

[Calixte Denizet (:calixte)]

A try auto:
https://treeherder.mozilla.org/jobs?repo=try&revision=0a0306a61d0545c309b0da8969f8615927446799

Comment 4

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/48a758441a03
Add the possibility to decode cmyk jpeg images coming from pdf files r=aosmond,necko-reviewers,valentin

Comment 5

[Cristina Horotan [:chorotan]]

https://hg.mozilla.org/mozilla-central/rev/48a758441a03