Kai-Uwe, You're using the outdated address of the LibTiff mailing list. That no longer works. Use [EMAIL PROTECTED] instead, if you're a subscriber at this time. If not, subscribe first, over at http://lists.maptools.org/mailman/listinfo/tiff.
I should add though that I tend to nag badly when people start including more parties in CC lists in the midst of a conversation. That's not just because of my personal character flaws, it's also because it is hard for the new recipients to know what is going on. We should at least try to make it easy for recipients to decode our messages. So instead of just adding stuff to CC, and a whole lot of overquoting at the bottom, as if answers usually precede questions, it is much more friendly and efficient to build a new message that explains what has been discussed before when you add recipients. Thus, I've sent this reply to old channels only. Kai-Uwe Behrmann wrote: > While in CinePaint I use similiar encodings to TIFFPM6.pdf for > 8/16-bit 8 and 9 photometric tags, there is room to improve the float > encoding. > > The ICC suggested float ranges are oriented on 8-bit math. They are > not very useable for real imagery and may cause a lot of confusion. I > would go as far to say the traditional value range of 0.0 -> 1.0 for > RGB should apply to all colours equally if CIE*Lab, CMYK or RGB. I'm not sure what you mean by this... Range 0.0->1.0 for CIE Lab? Then you would have to at least specify some scaling, and I'm not sure where to do this, depending on your application and context. > The CinePaint code is some years old and I did not look at further > specification efforts. So maybe something has allready cleared. > > How does photoshop cs2 store floats? I dont have any examples to tell > my own. What is the reference? Assuming you mean float Lab in TIFF... That should be pretty straightforward, though I doubt very much any mainstream reader is anticipating it. Essentially, you'd have to merely specify the correct SampleFormat (see http://www.awaresystems.be/imaging/tiff/tifftags/sampleformat.html), correct bitdepth, and there you go. As to the range, since there is no need for neither specification of any scaling, unscaled range should be stored. Meaning, I'd expect normal Lab values L to be between 0.0 and 100.0, for example. As to 'range' of a and b, we probably agree there is no such thing, and so does the TIFF specification (Section 23, first topic, 'the theoretically unbounded a* and b* ranges'). One of the major benefits of floating point Lab storage would be precicely this: it enables lossless storage of 'totally wild' ranges. Just make sure to apply no additional scaling before storage, meaning a and b will usually be in the range -128.0 to 128.0. As to the Photometric of such imagery, I think Photometric 8 should be used in this case. Same scheme applies to store floating RGB in TIFF, for example, and I have actually seen some images that do this. I think it is used in some GeoTIFF related branches. Range in this case is indeed the 'traditional' 0.0 to 1.0. But be aware, while this would be a perfectly valid TIFF, I don't expect many readers to be able to cope with it. As to Photoshop CS2 floating point storage, I don't know, I don't have it and haven't seen any specifications of what it does, except for one, defining 16bit floating point storage. That one isn't yet published through Adobe's official channels, so I don't think it is actually a 'specification', yet. Plus, it seems that it is of no use to you I think as 16bits for a float does not yield lots of precision. Drop me a line if you want me to send it to you, though. I've also seen testimages accompanying this not-yet-specification. If I remember correctly, Photometric was RGB, and ranges were 0.0 to 1.0, which is compatible with what is said earlier. Is this all at all useful, or am I misunderstanding you completely? Joris Van Damme [EMAIL PROTECTED] http://www.awaresystems.be/ Download your free TIFF tag viewer for windows here: http://www.awaresystems.be/imaging/tiff/astifftagviewer.html ------------------------------------------------------- SF.Net email is Sponsored by the Better Software Conference & EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf _______________________________________________ Lcms-user mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/lcms-user
