TIFF and LibTiff Mail List Archive


2003.05.12 17:12 "Photoshop layers, Lab colorspace", by Stephan Assmus
2003.05.12 18:38 "Re: Photoshop layers, Lab colorspace", by Chris Cox
2003.05.12 23:39 "Re: Photoshop layers, Lab colorspace", by Stephan Assmus
2003.05.13 07:45 "Re: Photoshop layers, Lab colorspace", by Kai-Uwe Behrmann

2003.05.13 07:45 "Re: Photoshop layers, Lab colorspace", by Kai-Uwe Behrmann

Am 12.05.03, 19:12 +0200 schrieb Stephan Assmus:

I'm new on the TIFF list. I don't know if any FAQs exist, and am just hoping that I don't bore anyone with my questions. I have written a graphics application for the BeOS, which is not yet published. TIFF would support saving any information I would like to remember in my image files. However, the main reason I'm interested in TIFF is that Photoshop can now save layer information in TIFF files. Unfortunately, I have not figured out, how it does that. My first guess, after reading a bit through the libtiff documentation, is that it maybe uses multiple TIFF directories, one for each layer. This seems not the case though. Is there any information available on how Photoshop saves layers in TIFF files?

Fine to hear about Your choice to do support multilayerd tiffs. Once I could test photoshop (6) it did not save layers. I dont know if I was mislead by an invisible option to set or Photoshops lacks this anyway (was ps6 - maybe this changed). The layerd tiffs I know are coming from the ImageMagick package and from tiffcp of libtiff. Since some months it is possible to do the same with filmgimp/cinepaint <>. I am not aware of any guis under win/mac.

Photoshop stores some information about visibility, over all transparency and some more of it's capabilities in an tag called TIFFTAG_PHOTOSHOP 34377. You get the documentation only while installing Photoshop with the SDK. Hope these tags get included somedays in the libtiff.

My second question is about the Lab colorspace. I have read much about it, but am having difficulties implementing the correct conversion routines for RGB to Lab and back. I have used the matrix constants and algorithms described in the Color Space Conversions by Adrian Ford and Alan Roberts, however, converting some RGB color to Lab and back again produces some slightly more red result. And my routines are slow. Reading about in the TIFF spec, I was surprised to find some paragraphs on Lab, and this got my curious again, since it says that conversion can be fast using LUTs. Can someone point me to some resources that would be helpful? Or even provide code?

I used sometime ago lcms for converting and displaying CIELab 's in filmgimp. The patch is available but I did not yet continue. Give it a try - GNU. The speed is reasonable <>.


- Good informations depends on our courage to bear them.               -
- Desinformation may become very inconvenient sometimes.               -
- more?             @         <>               -