AWARE [SYSTEMS] Imaging expertise for the Delphi developer
AWare Systems, Imaging expertise for the Delphi developer, Home TIFF and LibTiff Mailing List Archive

LibTiff Mailing List

TIFF and LibTiff Mailing List Archive
May 2004

Previous Thread
Next Thread

Previous by Thread
Next by Thread

Previous by Date
Next by Date

Contact

The TIFF Mailing List Homepage
This list is run by Frank Warmerdam
Archive maintained by AWare Systems



Valid HTML 4.01!



Thread

2004.05.20 05:35 "Extra samples handling", by <mayankg@iiitb.ac.in>
2004.05.20 23:42 "Re: Extra samples handling", by Chris Cox
2004.05.21 03:35 "Re: Extra samples handling", by <mayankg@iiitb.ac.in>
2004.05.21 19:22 "Re: Extra samples handling", by Chris Cox

2004.05.20 23:42 "Re: Extra samples handling", by Chris Cox

At 11:05 AM +0000 5/20/04, mayankg@iiitb.ac.in wrote:
> Hi,
>
> Please confirm if i am correct with the extra samples handling in TIFF images.
>
> If TIFFTAG_EXTRASAMPLES has:
>
> Type as 1(associated alpha): then the color values have to be taken as it is.
> also the number of extra channels will always be one only.

No, associated alpha means that it is transparency, and images will 
have to be un-premultiplied.

> Type as 2(unassociated alpha): then all the color values have to be multiplied
> by the alpha channel value.

No - unassociated means that it is NOT transparency.
And in no case would you multiply values when reading a TIFF --  the 
values in the TIFF file are premultiplied only if they have an 
associated alpha (transparency) channel.

> Question 1: Are multiple unassociated alpha channels possible? If 
> yes, then how
> are the final color values calculated?

Yes - unassociated alpha channels don't contribute anything to the 
color or transparency.

> Type as 0(unspecified data): then the extra channels have to be 
> simply ignored.

No, they can be loaded (as in Photoshop).

> Question 2: Are combinations of Type 2 and Type 0 possible for a particular
> TIFF image?

Yes, of course.
You could even have all 3 types.

> If yes, do we need to ignore the alpha channel/multilply the color
> values with the vaues of the color channel accordingly?

See above.

Chris