Archive: Color BIT Mapper


28th April 2003 11:42 UTC

Color BIT Mapper
Well as i said (see ColorMapper2) i had bee doing some researches and wrote a prog that does 2D Colormapping referring to a picture as map.

Unfortunately, the whole thing is written in crappy Delphi and therefore cant be used as ape (it couldnt be either if it was C-code bacause i used objects). But anyway, the mapping algorithm works, i even added some more channel options.

I'll post the prog and the sources here as soon as i uploaded it to my webspace - already mailed it to unconed, so he may be faster.

I'd say lets work on this and get it running :D


2nd May 2003 12:43 UTC

Uploaded ZIP
Here you go!
Just check it out (sorry, a bit slow) and tell me what you think about that. Maybe one of you can gimme a hand in rewriting this in C as an APE.


2nd May 2003 15:51 UTC

This slooooooooooooooooooooooooowww. Try not updating the progressbars everytime, that could speed it up.

And well, APEs like this definitely need to be written in MMX :).


2nd May 2003 20:12 UTC

Well i had em off in the first place, but then i saw no progress for a long time, and it wasnt really faster. This is only to visualize that here is at least something happening.

The main fault is that i used the cavas of a TImage object, as well as a temporary TBitmap. Those objects havent been designed for fast operations.

But anyway, thats what i wrote in my mail. took you a pretty long time to respond.

[EDIT]

Originally posted by Magic.X
*me senses a need of a "UnConeD explains use of MMX" thread

2nd May 2003 20:57 UTC

Actually I wrote a short blurb about MMX in a mail to shreyas, but considering it didn't help him at all, it's probably badly written.
Ask him.


3rd May 2003 11:27 UTC

here you go:

and UnconeD, your mail did help me a lot...I could rip of picture II and make Dynamic Clear Screen.

well, seriously,
I used some parts (for multiply and XOR blend and maximum blend) of picture II for my APE, (is it ok?)


5th May 2003 08:50 UTC

Ok, that reveals the meaning of those mysterious sequences in the picture 2 code.

To smit up, the MMX code is able to split a 32bit RGB val up into its 4 channels and do the same opertaions to all.

Gonna do researches on that.

Is it possible to use 3Dnow and ISSE command sets? Those have been developped later and should contain also usable things.