AWARE SYSTEMS
TIFF and LibTiff Mail List Archive

Thread

2003.12.03 06:20 "[Tiff] Double Precision TIFF images", by Chris Cox
[...]
2004.01.23 18:55 "Re: [Tiff] How to convert MINISBLACK to PHOTOMETRIC_RGB", by Frank Warmerdam
2004.01.23 19:15 "Re: [Tiff] How to convert MINISBLACK to PHOTOMETRIC_RGB", by Joris
2004.01.23 19:19 "Re: [Tiff] How to convert MINISBLACK to PHOTOMETRIC_RGB", by Bob Friesenhahn
2004.01.23 20:12 "Re: [Tiff] How to convert MINISBLACK to PHOTOMETRIC_RGB", by Chris Cox
[...]

2004.01.23 20:12 "Re: [Tiff] How to convert MINISBLACK to PHOTOMETRIC_RGB", by Chris Cox

Frank Warmerdam wrote:

>"sample" == "band" == "channel" == "component" ==?

Thanks Frank. Now I know I finally did understand it correctly. Am I also

correct in thinking that the only proper way ignorant software (as in

>>software that doesn't know about how the image came about or what it is

intended to contain) can correctly interpret this 3-channel image as a

>grayscale image with 2 extra channels that have no specified

application/meaning? If ignorant software were to have to render this image,

the correct thing to do is to render the first channel as grayscale, and

>>discard the others as far as rendering is concerned, is that right? I

realize this may also be trivial questions, but this thread is very

confusing.

Joris,

I am by no means an authority on the TIFF specification. My interpretation of the file in question is that a sensible application should render the first band in greyscale and ignore the 2nd and 3rd since they are extra samples with no meaning attached.

Or load the 2nd and 3rd channels as alpha (not transparency) channels -- as Photoshop does.

It would be incorrect to interpret this image as RGB.

Chris