Re: [Openexr-devel] Embedding ICC profiles in OpenEXR

[Geraldine Joffre]

On Mon, 21 Nov 2005 20:21:13 +0100 (CET), Kai-Uwe Behrmann wrote
> Geraldine,
> 
> a profile can be created on the fly out of the exr primaries and 
> back written from the same on the fly created ICC profile into the 
> OpenEXR header.
Yes, I know that. My point is that it requires additional code that would not 
be needed with 
embedded profiles.

> CinePaint obtained such a small layer recently based 
> on lcms. 
Yes, it is a very small layer and we have implemented it in our Photomatix 
software as well. But why 
impose it to all applications on the workflow that will process the OpenEXR 
file? With the 
mechanism of embedded profile, color specification would just need to be done 
once at file 
creation time. All subsequent color managed applications processing the file 
would not have to 
change the way they work to handle OpenEXR files.

> The same is possible with TIFF and I think as well with PNG.
Yes but both offer the option of embedding an ICC profile as well. I'm not 
asking for removing the 
current chormaticities attribute but for adding the ability to embed ICC 
profiles as well. This way, 
the OpenEXR format can be handled similarly as other image formats that color 
managed 
applications are supporting. 
 
> Only it seems not an good solution to me to open an second path for 
> describing colour characteristics in parallel if not realy needed.
It may not be needed for CG work but for a photography workflow it is needed. 

> Of course if OpenEXR could hold an API to convert from exr primaries 
> to ICC and accepting an ICC profile to convert to its native header 
> format, that would be great.
But it will not change the fact that ICC profiles can not be embedded in 
OpenEXR files. I believe the 
OpenEXR format won't gain wide acceptance among the photographer's community 
until it can 
support embedded profiles. 
 
> Saving of device characteristics for HDR devices is not covered by 
> the ICC standard. 
Why is it so? As far as I know, ICC specifications do not fordib a TRC with a 
gamma of 1? I don't see 
any reason why the colorimetric behaviour of an HDR camera or display could not 
be described by 
an ICC compliant matrix-shaper profile.

> Where there practical suggestions to hear at 
> Scottsdale regarding ICC and HDR device characterisation?
> 
> regards
> Kai-Uwe Behrmann
>                                 + development for color management 
>                                 + imaging / panoramas
>                                 + email: address@hidden
>                                 + http://www.behrmann.name
> 
> Am 21.11.05, 07:44 -0500 schrieb Geraldine Joffre:
> 
> > I've been to a conference on HDR-Photography and 3D-Visualisation in 
> > Hamburg last week. 
The 
> > audience was mainly professional photographers, so when Spheron talked 
> > about their HDR 
> > camera, they were asked about how the HDR files are color managed and 
> > whether they have 
> > ICC profiles (color management is very important for quality photography).
> > 
> > The answer was that the SheroCam HDR camera had been characterized but that 
> > no ICC profile 
> > is available for the HDR files produced. The only way to get an ICC profile 
> > is to shoot an LDR 
> > image so that the profile gets embedded in a 16-bit TIFF file. So, I 
> > intervened to explain that a 
> > newer version of the OpenEXR format makes it possible to define the 
> > chromaticities of the RGB 
> > primaries and white point, which means the camera just needs to populate 
> > the chromaticities 
> > attribute when writing the HDR file in order to get a 'color managed' HDRI.
> > 
> > However, from the discussions I had afterwards, I realized it might be more 
> > convenient for a 
> > photography workflow to have the chromaticities values available in the 
> > form of an ICC profile 
> > embedded in the OpenEXR file. The responsibility for creating and embedding 
> > the profile 
would 
> > be left to the application that produced the OpenEXR file -- the HDR camera 
> > or the software 
that 
> > created it from multiple exposures. The advantage is that the CMM of color 
> > managed software 
> > could directly handle OpenEXR files that have an embedded ICC profile. 
> > Otherwise, additional 
> > coding is necessary to extract and convert the chromaticities attribute for 
> > correct handling by 
> > the CMM. 
> > 
> > I think it could be a good idea to add the ability to embed ICC profiles in 
> > a future version of 
> > OpenEXR, the same way it can currently be done with the TIFF and JPEG 
> > formats.
> > 
> > Thanks.
> > 
> > Geraldine Joffre
> > 
> > --
> > HDR Imaging for Photography
> > http://www.hdrsoft.com