How did you make your profile picture?

[ndavd]

Hi, I really like how your profile picture looks. Is this how you made it?

[AndrewRadev]

Unfortunately, no, I kind of wish I had, because at least I'd know how to repeat it 😅.

The photo was rendered on this Apple 2 emulator: https://www.scullinsteel.com/apple2/. A friend of mine gave me a .DSK file with it, which I rendered on the emulator and then downloaded. Where did he get the photo? He ran it through this program: http://www.appleoldies.ca/bmp2dhr/. The problem is, the linux download link is broken, and I can't find it in the arch user repository. I think maybe I tried setting it up for windows at some point, but couldn't get it to work. And this friend of mine no longer has it installed either :).

So, uh, if you feel like trying the process out, good luck :D. Let me know how it goes. There's also this website, but I think it produces a somewhat different effect: http://c64.superdefault.com/

[ndavd]

Oh wow didn't expect such an interesting answer! I shall investigate then :D

[ndavd]

Some quick updates... I found this website that talks about that program http://appleoldies.ca/cc65/docs/dhgr/MonochromeDitheringBmp2DHROverview.htm which reads... Some of Bmp2DHR’s color distance program code was adapted from a routine in Joel Yliluoma's arbitrary-palette positional dithering algorithm: http://bisqwit.iki.fi/story/howto/dither/jy/

[ndavd]

Been a while 😄, Just remembered about this and decided to give it another shot...

So, with the help of Web Archive was able to access the Linux downloading page https://web.archive.org/web/20170704165407/http://hoop-la.ca/apple2/appleoldies/bmp2dhr/.

Unfortunately there's no snapshot of the Makefile for Linux, but at least there's the source code. I was able to build the b2d binary but it keeps telling me that my .bmp is in the wrong format... Any change you're still interested on this, lol?

[AndrewRadev]

Hm, it's tricky. I compiled and ran the code in 32-bit mode, I had to include a missing header, and I'm getting the same thing. I added some debug statements for the parsed structures:

% gcc -m32 -o b2d b2d.c && ./b2d ../bmp/lwizard.bmp
Palette 5: tohgr NTSC DHGR Colors
Preview Palette 5: tohgr NTSC DHGR Colors
bfi.bfType:      BM
bfi.bfSize:      0
bfi.bfReserved1: 0
bfi.bfReserved2: 54
bfi.bfOffBits:   2621440
bmi.biSize:          4587520
bmi.biWidth:         3604480
bmi.biHeight:        65536
bmi.biPlanes:        24
bmi.biBitCount:      0
bmi.biCompression:   764149760
bmi.biSizeImage:     247726080
bmi.biXPelsPerMeter: 247726080
bmi.biYPelsPerMeter: 0
bmi.biClrUsed:       0
bmi.biClrImportant:  4278190080
../bmp/lwizard.bmp is in the wrong format!

These numbers are somewhat off, but I can't figure out how. The width and height of this image are 70x55:

../bmp/lwizard.bmp: PC bitmap, Windows 3.x format, 70 x 55 x 24, image size 11660, resolution 3780 x 3780 px/m, cbSize 11714, bits offset 54

The hex code for 70 is 46, if you look for that hex codes in the output of xxd ../bmp/lwizard.bmp | head:

00000000: 424d c22d 0000 0000 0000 3600 0000 2800  BM.-......6...(.
00000010: 0000 4600 0000 3700 0000 0100 1800 0000  ..F...7.........
00000020: 0000 8c2d 0000 c40e 0000 c40e 0000 0000  ...-............
00000030: 0000 0000 0000 00ff ff00 ffff 00ff ff00  ................
00000040: ffff 00ff ff00 ffff 00ff ff00 ffff 00ff  ................
00000050: ff00 ffff 00ff ff00 ffff 00ff ff00 ffff  ................
00000060: 00ff ff00 ffff 00ff ff00 ffff 00ff ff00  ................
00000070: ffff 00ff ff00 ffff 00ff ff00 ffff 00ff  ................
00000080: ff00 ffff 00ff ff00 ffff 00ff ff00 ffff  ................
00000090: 00ff ff00 ffff 00ff ff00 ffff 00ff ff00  ................

There's a number 4600 and if I tweak the long/short types, I can get out 70 and 55, but it's a bit of a mess. Like, I can change the top 3 types of one of the structs from ulong to ushort and:

Palette 5: tohgr NTSC DHGR Colors
Preview Palette 5: tohgr NTSC DHGR Colors
bfi.bfType:      BM
bfi.bfSize:      0
bfi.bfReserved1: 0
bfi.bfReserved2: 54
bfi.bfOffBits:   2621440
bmi.biSize:          0
bmi.biWidth:         70
bmi.biHeight:        0
bmi.biPlanes:        55
bmi.biBitCount:      0
bmi.biCompression:   24
bmi.biSizeImage:     764149760
bmi.biXPelsPerMeter: 247726080
bmi.biYPelsPerMeter: 247726080
bmi.biClrUsed:       0
bmi.biClrImportant:  0

So 70 and 55 are there, but they're just sort of in the wrong places.

C is not exactly my specialty, though I admit it's kind of interesting to try to figure out how to fit the data... Right now, I don't have the time, but I might in a few weeks.

[AndrewRadev]

@ndavd Alright, managed to get it compiling: https://git.sr.ht/~andrewradev/bmp2dhr

The relevant commit is this: https://git.sr.ht/~andrewradev/bmp2dhr/commit/14e0058193274e11d096da55bb9ced05fc0a89d5

The following command-line produces a monochrome image with dithering, the v creates a preview BMP image:

./b2d BMP/PROFILE/austin_7.bmp v mono

Note that the output directory has to be in all-caps, otherwise the program fails. This generates:

Preview	In Apple II emulator

Annoyingly, the dithering style is not quite the same as my original profile image, but the only thing I can say is that you could experiment with different "dither" and "pallette" command-lines. The program is pretty poorly documented, though, I'll tell you that. Here's one example of a working command-line and its output:

$ ./b2d BMP/PROFILE/austin_7.bmp v hgr D4 P9 
Palette 9: Legacy Win32 Colors
Preview Palette 5: tohgr NTSC DHGR Colors
Color Dithered Output:
Dither = 4 - Atkinson
Preview file BMP/PROFILE/austin_7_Preview.bmp created!
BMP/PROFILE/AUSTIN_7C.BIN created!

So, the P9 is pallette 9, the D4 is for dither 4, and v generates a preview BMP:

From this point on, I'd say try to compile it locally and experiment with the options. Let me know if you reach some interesting results :)

[ndavd]

Those are some amazing results, will give it a try today and let you know!