yet another graphics converter

Page 2/2
1 |

By sd_snatcher

Prophet (3551)

sd_snatcher's picture

30-01-2010, 03:03

Hi Artrag!

First I have to thank you for this excellent tool! Smile

About the file format, The-Gimp has the following options when saving the TGA file:

- With or without RLE compression (checkbox)
- Origin: "Bottom left" or "Top Left" (combo box)

Maybe it's not even necessary to code the support for the other variations, but only add a readme file explaining the TGA requirements for the converter: No RLE compression and origin = "Bottom Left".

The algorithm of your tool outputs impressive results. It seems very possible to adapt it for screen-12 conversion too. What do you think about this? OTOH, screen-10 conversion, as I was talking with DamageX seems much more complicated. The mixed YJK+RGB colorsystem itself is a great challenge, but also the rest of the requirements I listed to allow the generated images to be used on gaming.

Last, I hope I'm not asking too much, but could you please generate Mac OS-X binaries of your tool too? Smile


Enlighted (6862)

ARTRAG's picture

30-01-2010, 08:57

hi sd_snatcher

about a Mac OS-X version, I do not have that sytem but
if you have any C compiler, try yourself to compile my sources
with gcc i do

GCC scr2floyd.c -oscr2floyd.exe -O3 -s

any other C compiler should work

sadly i do not have skilld to support other input formats, so I'll add your remarks on the format to teh downoload page

About extending the code to other video modes (screen 10 or 12), I'd like to integrete it with palette optimization first (fulls creen 4)
but for luck of time and lazyness it is not a priority atm

By sd_snatcher

Prophet (3551)

sd_snatcher's picture

30-01-2010, 19:55


What do you think about using the Spatial Color Quantization algorithm? It's said to yield the best results for this kind of very restricted color conversion as required by 8-bit systems.

By DamageX

Master (217)

DamageX's picture

10-02-2010, 11:01


I made some progress, adding the Floyd-Steinberg dithering (option 2) and some other features. MSX1 conversion continues to be inferior to ARTRAG's program because there is no brute-force search of color combinations. I made a utility to simulate screen 10/12 but there is no use of RGB colors so far. I spent a lot of time on the automatic palette combining routine but was not very successful :(

@sd_snatcher do you have a pic demonstrating color bleed in screen 10 that can be improved with RGB colors? As for the spatial color quantization, I can't say much about it. The code looks complicated and the link to the paper was broken.

Page 2/2
1 |