Re: [css3-color] ICC implementation

>
> Chris Murphy wrote:
>
>> And then also that there has been no implementation of gamma  
>> correction, and still there is no implementation of gamma (only)  
>> correction, Section 3.1.1 which remains in the spec?
>
> Do you mean gamma correction in general, or a CSS specification for  
> it?  You can't do transparency or anti-aliasing properly without it  
> (although I suspect that a lot of image replacements are not  
> properly gamma corrected!).

I web browsers do not apply the specifed gamma correction as stated in  
Section 3.1.1. Images do still look lighter on the Macintosh than they  
do on Windows, as to be expected since the Mac OS has a different tone  
reproduction curve. If a correction curve were applied, it would  
compel those images to display darker, and if everything worked out  
just right the resulting image TRC would be the same as it is on  
Windows which has no correction.

This correction isn't occurring. Web browsers on the Mac don't do it.  
It's faster to just use ColorSync anyway, which they also do not use  
for untagged content, not even Safari.


> I would be interested to know how many authors actually understand  
> gamma correction.  It is very common on amateur and in house  
> designed web sites to find images with gamma 1.0, yet the authors  
> seem to be oblivious as to how dark they render).

An image with no encoding ("gamma 1.0") or linear encoding would look  
completely washed out. I think it would be obvious to anyone with  
vision that there is a serious problem so I'm not sure how big of a  
problem this really could be unless people just aren't paying attention.



>  It is also not that uncommon to find ones with the Mac gamma,  
> although they are more likely to come from a design agency.  To some  
> extent this is exacerbated because it seems that some digital  
> cameras generated gamma 1.0 but label it as sRGB.

I know if no camera that does this. Digital cameras all do have linear  
response to light, and record it that way. The raw data is processed  
in the camera, tone mapped where it has a tone reproduction curve  
defined by either the sRGB curve or gamma 2.2, and is then JPEG  
compressed and tagged with EXIF data as to its color space.

Off hand I can't think of a single camera that is embedding ICC  
profiles in images.

And here is another area where W3C should take some leadership and  
that is whether or not web browsers should honor EXIF color space  
data. I think it should, and I think at least mozilla is looking into  
honoring the EXIF color space tag.


> (The problem being that image sensors generally approximate gamma  
> 1.0 devices, so need gamma correcting for either sRGB use or to  
> achieve equal perceptual brightness steps between digitisation  
> levels.  sRGB approximates CRT behaviour; the Mac approximates equal  
> perceptual brightness).

This is already done by the time the JPEG is written out. I don't  
understand at all the suggestion that this is not being done and that  
images are linear encoded

Raw images are linear encoded, which are processed with software  
designed to do so, and their main function tone mapping.

>
> Images with bad gamma appear in PDF as well.

What does bad gamma mean?

Chris Murphy

Received on Friday, 11 July 2008 18:03:37 UTC