foobar opened the ticket and wrote:

OpenTTD r23401
Windows 7 32bit

I'm using fire cycle colours as a base for cargo specific recolouring using custom colour translation tables. This works fine when using the DOS palette for the NewGRF. However, when using the WIN palette (no other changes), the colour indices don't match. For example if I draw colour E9, I have remap colour EC in the colour translation table to actually have the E9-coloured pixels recoloured by the translation table ingame.
(With the DOS palette I can draw E9, map E9 and everything is well).
From a test it appeared that the fire cycle is shifted by 4 positions and the block cycle by 3 positions. The other cycles don't have this problem.

For this I assume that positions in the colour table for the fire cycle are the same in DOS and WIN. indices may indeed differ depending on what colours you want in the endresult. If this is a wrong assumption, there may be a bug in NML instead. To make things absolutely clear: the problem also occurs when using colours of which the and indices are exactly the same in DOS and WIN palette (fire cycle is in the same position, as well as for instance the light yellows for a grain colour translation).

Please find attached two test NewGRFs, one in DOS palette, one in WIN palette. Build a random train (available from around 1985) and attach a couple of Hoppers until the test pattern appears. The test pattern is fully visible from the vehicle properties window. With the DOS grf the upper and lower lines of the test pattern are identical. With the WIN grf, the patterns for the fire and block cycle are shifted. A comparison of what happens is provided in the attached image.

The NML colour translation table used is pasted below:

recolour_cycl = reserve_sprites(1);
replace(recolour_cycl) {
recolour_sprite {
//block
0xE3: 0x90;
0xE4: 0x91;
0xE5: 0x92;
0xE6: 0x93;
0xE7: 0x94;
//fire
0xE8: 0x90;
0xE9: 0x91;
0xEA: 0x92;
0xEB: 0x93;
0xEC: 0x94;
0xED: 0x95;
0xEE: 0x96;
//red
0xEF: 0x90;
0xF0: 0x91;
//yellow
0xF1: 0x90;
0xF2: 0x91;
0xF3: 0x92;
0xF4: 0x93;
//water
0xF5: 0x60;
0xF6: 0x61;
0xF7: 0x62;
0xF8: 0x63;
0xF9: 0x64;
0xFA: 0x65;
0xFB: 0x66;
0xFC: 0x67;
0xFD: 0xA2;
0xFE: 0xA3;
}
}

Attachments

• test_pattern.png (4.00 KiB)
• dutchtrainsDOS.grf (88.31 KiB)
• dutchtrainsWIN.grf (88.31 KiB)

Reported version: trunk
Operating system: Windows

This issue was imported from FlySpray: https://bugs.openttd.org/task/4868