Previous Page | Next Page
- by ugetab at 2:44 PM EST on December 22, 2007
- Try getting your hands on a newer rip. The one I've been using has 51 tracks, track 5 is dracula's castle, and track 9 is the ending.
- by marioman at 2:56 PM EST on December 22, 2007
- I was referring to Rush 'n' Attack. The ending theme for it is missing. Sorry that I was not more specific.
edited 2:57 PM EST December 22, 2007
- by JILost at 3:07 PM EST on December 22, 2007
- The rip I was using was your rip with the adjusted $4017, played using nezplug++ (v0.9.4.8 + 2 + 12). I don't think it's a bug in nezplug, though, because other games don't do it...but maybe it is. What player(s) would you suggest?
- by ugetab at 4:33 PM EST on December 22, 2007
- I've personally adjusted the 4017 writing in NotsoFatso(not official copies, but one on my site) to work like it does in FCEUXDSP. With this in place, I've had far less trouble with the feature, and developed it almost exclusively around making Mach Rider work correctly. 4017 issues are so common that I got tired of it after fixing 'A' from the NSF Pack. The NSF fixes involve initial writes and continuous writes to it. The player fix was just for continuous writes, which includes games such as SMB3. Without the fix in place, it's like not having any 4017 writes.
Rush'n'Attack had the double/triple issue of missing a bank of data, and the song init routine being too heavily modified to work correctly. I added the missing data, and wrote a small routine to use a lookup table, and added 100% original banks to it, and after that was done, all the trouble I was having with it suddenly disappeared.
- by marioman at 5:49 PM EST on December 22, 2007
- The Rush 'n' Attack rip sounds great. Thanks ugetab.
- by anewuser at 6:59 PM EST on December 22, 2007
- nezplug++ was updated to +2+16.1 (yeah, weird version numbers)
http://offgao.no-ip.org/program/nezplug++.html
- by marioman at 9:08 PM EST on December 22, 2007
- Thanks. It fixes the GBS issue that was introduced in version +15. It also adds a GBA Mode option for GBS files. Can't tell a difference though.
There still isn't a config dialog. Surely it will happen sooner or later.
- by JILost at 10:20 PM EST on December 23, 2007
- Listening to the rip in which you (ugetab) adjusted the 4017 on September 2, 2007 in the latest build of nezplug, the Dracula's castle music (which seems to be track 5) still exhibits the same problem: the first drum hit on the DPCM channel doesn't "pop" like the others. When the song loops, this is fixed. This same behavior is exhibited using foo_gep.dll from foobar2k, so it's clearly not coming from the player. If there's a newer rip around, please lead me to it.
- by ugetab at 1:31 AM EST on December 24, 2007
- Castlevania 2 needed a fix I've never had to use before, and didn't know about. Apparently, the requirements for some DPCM instruments includes a loud initialization sample in order for the 'kick' to kick in. In Castlevania 2, this comes with the first instrument of the first song played in the game. In FCEUXDSP, I managed this with a DPCM sample reading
00 00 00 00 00 00 00 00 00 00 00 00 00 00 FF FF
FF. It makes tracks pop when they start in the emulator, but the winamp plugins seem to handle it gracefully.
On another note, nezplug doesn't yet utilize 4017 writes correctly. Try listening to my updated Mach Rider rip in FCEUXDSP, track 4, about 9 seconds in. There will be a spot you'll need to remember. Now listen to track 4 in nezplug. The double-note about 9 seconds in should be, at worst, a barely audible blip of sound as it immediately changes to the second note in the blip. When fixed, track 1 must also sound correct 18 seconds in. The song will fade to silence if 4017 support is correct. If incorrect, it will blip at the end of the fade-out.
In order for 4017 support to be considered reasonably correct, both songs must exhibit the repaired behavior using the same coding.
I've implemented the FCEUXDSP coding into NotsoFatso, and that made it work with both tracks reasonably well. I've been trying to blindly implement it into nezplug, but the coding that needs to change is hard to find.
- by ugetab at 12:31 AM EST on December 26, 2007
- Well, this is annoying, but should be mentioned. It seems that one of the 2 square channels gets messed up by a pre-existing systemic error in NotsoFatso when the wave_square.h fix is in place. I'm pretty sure it was hiding behind the broken code I edited. Thus far, I've been unable to get enough usable data on the error to debug it.
The track that it made it's most noticable appearance in is the battleship track of Super Mario Bros. 3(track 21 in my copy). At about 0:10 seconds in, one of the notes plays for too long. I've determined it to be one of the 2 square channels, but what keeps it going, or what fails to make it stop is still a mystery to me.
I do believe the fix I added is valid, but it's also clear that it opens up an entirely new can of worms as far as fixes go. If you want the original NotsoFatso behavior for this NSF, you just need to look to the 'Config 3' tab, and change 'Force $4017 Write:' to 'Force $00'.
Hopefully, if I don't manage to get this fixed, the info will help someone else do it.
Previous Page | Next Page
Go to Page 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50
Search this thread
Show all threads
Reply to this thread:
HCS Forum Index
Halley's Comet Software
forum source