clueless wrote:
I'd like to animate background tiles and sprites (for water / shoreline effects like in Crystalis (MMC3 btw)). Since right now I'm restricting myself to MMC1, I've opted to use non-switching CHR-RAM. So I can animate very few tiles at a time, but its workable.
Yeah, the figure for NTSC is 10 tiles + OAM DMA.
Quote:
I want to have 8-way background scrolling. I'm willing to live with scroll artifacts on the edges if I must, but like most people, would prefer to avoid them.
Scrolling glitches are 100% avoidable in two cases:
- You have an IRQ counter. In this case, use vertical mirroring and use a scanline IRQ to skip over the status bar. Crystalis does this.
- You aren't playing samples on channel 5, and your status bar is made of sprites (as in Ikari Warriors or Contra or SMB2 or Mega Man series or Rollerball). In this case, use vertical mirroring and DPCM Split.
Quote:
I need a ~2 row status bar. I've put mine at the top and use sprite-0 hits to switch nametables. I'm ok with this, but if there is a better way (like, say, an IRQ), then I'd like that instead.
If you don't play samples on channel 5, check out
my DPCM Split demo that (ab)uses one of the NES's built-in IRQs. Cut off roughly the top 40 and bottom 40 lines, and you have a GBA-esque widescreen that looks good in HDTVs' zoom mode. Bonus: if you play it right, you get more vblank time to copy more tiles.
Quote:
ps- Tepples / Memblers, How would your theoretical mapper detect that the PPU is trying to render tile $FF? What is it looking for at the wire level?
In much the same way that MMC2 and MMC4 detect drawing tiles $FD and $FE. It'd use CHR /A12 AND A11 AND A10 AND A9 AND A8 AND A7 AND A6 AND A5 AND A4 to pull IRQ low, and a write to an acknowledge register to put it back high. But it's not as necessary now that the DPCM Split technique is known.
I smell split as soon as I know what to call the new topic.