nesdev.org

Hello,everyone!
I have a problem on my emulator. My emulator use loopy's The skinny on NES scrolling
The graphics generate glitches, some games at scanline 0 , and some demos at every visible scanline.(scanline scope is -1 ~ 260, -1 represents pre-renderline, 0 ~ 239 represents visible scanlines, 240 represents post-renderline, 241 ~ 260 represents vblank stuff)

Here are some pictures








What's wrong?

I guess the reasion maybe $2005 write

who can help me?

1. Expecting a reply within 1 hour of your initial post is unreasonable - some of us only check this forum once per day, often even less frequently.

2. Do not expect older demo ROMs to work correctly, since most of them were designed to work on emulators and were never verified to work correctly on real hardware. It's entirely possible that those glitches are supposed to appear. Similarly, always check this list before trying to "fix" graphical glitches in a particular game.

3. It looks like you're using an outdated version of scanline.nes (the 2nd-last picture) - use this one instead.

Some games (mostly early ones) set the vertical scroll to -1 in order to compensate for the 1 scanline delay in sprite coordinates. This may cause attribute table data to be rendered as name table data in the first scanline of the screen, which most of the time will look like garbage. Since this isn't visible in most TVs, the problem was never detected by the developers. You should check each game (in other emulators and possibly a real console though video capture) to verify whether it's supposed to have those glitches before trying to fix them.

CMCWavy and Soccer look correct on Nintendulator, so the roms probably aren't the problem here. Try posting a screenshot of Slalom, that game uses both negative scroll and lots of horizontal scroll changes.

Dwedit wrote:CMCWavy and Soccer look correct on Nintendulator, so the roms probably aren't the problem here. Try posting a screenshot of Slalom, that game uses both negative scroll and lots of horizontal scroll changes.

Slalom has glitches too. The same problem occured in F1 Race

Somehow, I always think the reason is that I didn't properly deal with first write to $2005 to change coarse X(Tile Number) and find x.
Do you agree with me?

tokumaru wrote:Some games (mostly early ones) set the vertical scroll to -1 in order to compensate for the 1 scanline delay in sprite coordinates. This may cause attribute table data to be rendered as name table data in the first scanline of the screen, which most of the time will look like garbage. Since this isn't visible in most TVs, the problem was never detected by the developers. You should check each game (in other emulators and possibly a real console though video capture) to verify whether it's supposed to have those glitches before trying to fix them.

I will disassemble the ROM to verify it.

Quietust wrote:1. Expecting a reply within 1 hour of your initial post is unreasonable - some of us only check this forum once per day, often even less frequently.

2. Do not expect older demo ROMs to work correctly, since most of them were designed to work on emulators and were never verified to work correctly on real hardware. It's entirely possible that those glitches are supposed to appear. Similarly, always check this list before trying to "fix" graphical glitches in a particular game.

3. It looks like you're using an outdated version of scanline.nes (the 2nd-last picture) - use this one instead.

1. You are right, but I am a bit hurry.

2. Yeah. The page I accessed frequently. It's very useful to me. Simultaneously, Tricky-to-emulate games also important.

3. I am trying to understand your scanline.asm's every instruction

Looks like a timing error to me.

WedNESday wrote:Looks like a timing error to me.

Maybe, but I couldn't find the actual reason.
I guess sprite hit should be OK, because Battletoads could be run properly.
Do you know any test ROM for accurate timing?

Boolean wrote:Do you know any test ROM for accurate timing?

CPOW once posted a link to the collection of test ROMs he had accumulated when I was having trouble with my CPU/PPU, and they helped a lot: https://github.com/christopherpow/nes-test-roms

Some of these are timing-related test ROMs. Maybe they will prove to be useful.

ArsonIzer wrote:

Boolean wrote:Do you know any test ROM for accurate timing?

CPOW once posted a link to the collection of test ROMs he had accumulated when I was having trouble with my CPU/PPU, and they helped a lot: https://github.com/christopherpow/nes-test-roms

Some of these are timing-related test ROMs. Maybe they will prove to be useful.

Yeah!
Those ROMs were downloaded by me last year, and I did some tests.
Now, I think I should test ROMs about timing carefully.

I test a rom called apocalypse.nes.
Nametable window is ok, but FC87 main window is error.
It seems that nametable switch not properly.
I need to check The skinny on NES scrolling implementation in my emulator.

The demo apocalypse downloaded from https://github.com/christopherpow/nes-test-roms

I find a reason about horizon shaking. The reason is that my cpu's crossing page implementation
is wrong, and I no longer judge crossing page the video is better than ever.

for example
boing.png:
Judging crossing page left is shaking and right is shaking too.
Don't judge crossing page right is shaking only.
scanline.png:
Judging crossing page left is shaking and right is shaking too.
Don't judge crossing page neither left nor right is shaking.