nesdev.org

Very good question. To be honest, those tend to be used mostly for large empty areas. But there's another case, when you have multiple pages of tiles in order to animate (please correct me here if I'm mistaken), you tend to have lots of duplicated tiles for the static stuff. This allows a group to r...

Well there's always the NesEncoder part of my scheme. Without the Huffman coding, SMB.chr is around 5800 bytes instead of 5200 but that's still fairly good starting from 8192 bytes.

tokumaru wrote:My decoder was very slow too, I favored code size rather than speed. I believe these complex schemes will always be significantly slower, and that's why I'm pursuing something "lighter".

Other than RLE? What sort of characteristics are you after? (random access etc)

Thanks tepples. I implemented the Huffman decoding and.. man is it ever slow. I should have listened to Tokumarus warning. :) It currently takes (guessing) about a second to decompress a bank full of tiles, that's almost useless IMHO. De-HuffmanEncode: http://code.google.com/p/artwork-compression-ki...

Ah. I think I'll leave that alone for now, I replaced the ASL with "CMP #$80 / BCC" because I do need A unaffected. It's not a hotspot and this, I think, is easier to read. Thanks though. Very thankful for any hints you guys can provide. :) One thing I'm wondering, and I think I haven't se...

I actually got Nestopia to compile for Linux, and my test failed spectacularly. It turned out to be the BIT #$80, easily replaced with a shift and BCC. It now seems fine. Interestingly Nintendulator didn't fail the BIT #$80 version, looked identical to FCEU. I think I'll stick to Nestopia now that I...

The two things you mention bother me as well. Column mode is pretty nasty, I had plans to see if I could have kept it row-oriented and bunched the unique columns toward one end of each byte instead. Decoding would be more ALU-oriented, and while taking 8 bytes like a raw pattern it might skew the va...

Ah, thanks. I'll see if I can run those, using Linux so I tend to prefer native emulators. In this case finding a workaround is easy, I can roll bit 7 into carry instead, but I really want to be compatible. If you're interested, here are my efforts for now: http://code.google.com/p/artwork-compressi...

I've so far ported NesDecoder and have it decompressing the SMB CHR correctly. Next thing up is the HuffmanDecoder. I'm getting the hang of assembly on this thing, but had a couple of headaches. E.g. confusion over which instructions affect which flags, and it seems the various 6502 docs disagree ov...

Fair enough. I think it's probably not that bad, and might take a stab at it. Would you mind at all if I used your decoder/test source code as a starting point? I am versed in assembly, just not 6502, and can probably learn the platform as I go.

Tepples, that's cool. :) Can't really claim it was something that I carefully planned. Was mostly figuring out a way to make the Huffman decoding cheaper, and it turned out to work well. Tokumaru, if someone were to pick up writing a 6502 version that would be awesome. I'd have to learn the platform...

First, I'm a reader and admirer of the people of this forum. While I'm not a NES developer myself (I do some DS) I'm tickled by the intricacies of old console hardware. Reverse engineering, and pushing the limits of these old consoles is fascinating to my nerdy brain. Especially Tokumaru's implement...