Home Page Forums MAME MAME SVN builds - new driver flood

I'm on kind of a roll, so... more FM Towns love! This time it's the turn for half-size sprites, which CRI used a lot to simulate the sprite scaling in Sega's arcade games, but also show up in surprising places...

Well, why not go all the way in and make sprite rotation work so the After Burner games at least look half-decent?

(the resolution is weird but it's easy to see the improvement)







The sky still looks wrong in AB1, though...

Now for something less glamorous, but no less important... I got the keyboard type flags working so now it behaves like a real FM Towns JIS keyboard and not a weird mix of everything.

Sadly, the kanji dictionary only works on MS-DOS (something related to unimplemented I/O ports, I think) but at least it's possible to type actual Japanese in some way.

Nice :-) Although you usually don't see "dekiru" written in Kanji ;-)

I still hope to find an actual FM Towns for sale in Japan at some point...

Status for MyBrain 3000 & clones driver shape up prpject, DOS 1.25 finally boots asking for date & time!

Mybrain 3000 keyboard is working now but the prompt is in the wrong corner! But hey, we got A: prompt!

r09

any chance you could look at tnzs on the towns?

https://s9.postimg.org/54zcdrrnj/0003.png

there seems to be an off-by-1 error once you start scrolling the screen and it goes past what I assume is a page boundary.

I suspect it's probably a simple maths bug when wrapping or something similar, but I imagine it has the potential to be causing less obvious bugs elsewhere too.

It's always been like this in MAME, but real hardware videos don't seem to show the glitch, so it might be a good one to look at.

I might be wrong, but I suspect that bug is also related to debut2 and secre2 (and maybe others) looking completely messed up. So far I haven't found the cause, though.

Okay, either I'm missing something obvious (which is probably the right answer, since I'm not exactly very experienced in all these emulation things) or TNZS does things with the VRAM that I don't quite understand.

Basically, the background is rendered as a 16-bit bitmap layer (sprites are only used for the characters and foreground elements), and uses CRTC register 17 to scroll horizontally. From what I can see, the game writes directly to the high VRAM area (0x80000000). So if we look at the point in VRAM where the "split" happens, we see this:



Which is... exactly what's showing up on screen. It corresponds to the black-to-white transition in this:



I'm guessing the bitmap drawing routines are missing something... but since the only documentation I have is in a language I can barely read and the only other open source emulator (FM-bochs) doesn't even run the game, I'm a bit stumped, to be honest.

Have you tried it on UNZ to see if it happens? It's obviously not conclusive since we know they have some per-game hacks to fix things that happened on hardware, but it's better than nothing