Re: [dev] Looking for simple, alpha supporting image format

From: FRIGN <>
Date: Tue, 15 Jul 2014 20:12:09 +0200

On Tue, 15 Jul 2014 19:55:16 +0200
Markus Teich <> wrote:

> Why do you overwrite the blanks before checking if they are blanks? If the file
> format specifies these characters as blanks, they should be checked for.
> Otherwise the format specification should not say „blank“ but „any one byte“.

There obviously is a little confusion about blanks (0x00) and
ASCII-blanks. Replacing the "spaces" with \0 is just a trick and has
nothing to do with the data-represenation.

> ASCII only uses 7bit and since you probably want to use this one extra bit as
> well, I would at least consider the data portion to be binary and not a string.

Yup, my mistake. The basic idea fortunately isn't affected by that.

> I also think it is a well defined format. If someone want's to transfer an image
> over the network, he can easily use an arbitrary compression algorithm,
> therefore we do not need one inside the format and also get rid of different
> pixel formats (peek at the possibilities of DirectX if you want to throw up).
> One problem could be if someone needs more than 8 bit per color channel, but I
> think 8 bit are enough.

I just imagine how well this format would work with gzip, especially
for instance when an image doesn't have an alpha-channel and the
gzip-algorithm can substitute each "fourth 0".



Received on Tue Jul 15 2014 - 20:12:09 CEST

This archive was generated by hypermail 2.3.0 : Tue Jul 15 2014 - 20:12:11 CEST