superxavix: some research on the bitmap layer, slight visual improvement where it gets used + general XaviX refactoring and improvements + 7 new sets #13074
Conversation
…ent where it gets used
|
the title of the tmy_thom set could probably be improved now the title screen renders more correctly |
|
The Japanese title is Soreike Thomas - Sodor Tou no Nakamatachi (それいけトーマス ソドー島のなかまたち). The English title (Thomas & Friends on the Island of Sodor) should be noted too, though. |
…e memory bypass (as some later SuperXaviX games will need it)
|
there are some baby steps in here regarding a general XaviX cleanup, with a view to moving the SoC logic into a core, rather than the driver at some point. There will likely be a performance hit to doing this, but it feels like the right time to start doing it. |
|
While there are definitely still some issues, being able to map RAM into external space, and adding the bitmap plotter allows the Super TV-PC units to start drawing to their bitmap layer, which is important as they decompress data there. Some of the data that gets copied is corrupt though, I suspect due to bad XaviX2000+ opcodes, although there could also be a data conversion feature I'm missing.
This will eventually lead to a cleaner implementation of the XaviX memory handling, even if it'll be a few PRs beyond this one before we're fully there. |
|
The bitmap plotter was further improved, might not be perfect, but the different width data can now be written which fixes things like the filling in the Hello Kitty paint demo in attract |
|
Outside of that, this set of changes gives improvements to various other SuperXaviX games using the bitmaps. |
Definitely some issues with the text rendering still. It's readable but seems to be missing pixels on the thinner lines, and the parentheses in the quiz answers look messed up. Also, are the colours supposed to be so desaturated? |
|
It's interlaced, and probably meant to be twice the x resolution, but until
I understand the crtc stuff this is the best I can get, I'll need to pixel
double the standard xavix output to get the higher red bitmaps superxavix
supports. One step at a time.
…On Fri, 13 Dec 2024, 22:05 Justin Kerk, ***@***.***> wrote:
The bitmap plotter was further improved, might not be perfect, but the
different width data can now be written which fixes things like the filling
in the Hello Kitty paint demo in attract
Definitely some issues with the text rendering still. It's readable but
seems to be missing pixels on the thinner lines, and there seem to be extra
pixels on the parentheses in the answers making them look doubled up. E.g.
I think the top left answer should say "本（ほん）"
Also, are the colours supposed to be so desaturated?
—
Reply to this email directly, view it on GitHub
<#13074 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/BBR6GZJN4YXYNECZJJCVSUL2FNK2LAVCNFSM6AAAAABTPZXRJWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKNBSGQ2TMMRYGY>
.
You are receiving this because you authored the thread.Message ID:
***@***.***>
|
|
the current changes improve the visuals on Choro Q, although it still crashes shortly after this (it doesn't crash if you return 0 from the math unit, although then the cars don't move either) |
|
there do still seem to be some plotter issues, but it is a bit better with the full horizontal resolution for the bitmap layer |
|
FWIW, a future PR will likely move the XaviX chips to be SoC/CPU implementations, but that will require significantly more refactor and is out of the scope for this. The work done here is basically tweaks and improvements based on new knowledge gathered from recent dumps. |
|
I've added a better summary of the changes too |
|
The Anpanman game is the first SuperXaviX one we know to use the extended secondary bus feature which appears to be usable by the sprites and bitmap layer at least. We have a few others with extended ROMs, but it isn't clear if they're doing the same, or have banking, because those don't boot far enough yet. Unfortunately implementing this requires some extra checks in places, which slows things down, but the current line of thought is "make it work" while re-optimizing it will be looked into later. It also uses the bitmap plotter to plot to main ram instead of extra RAM, which will need some work (for the graph in one scene) There seem to be some opcode / logic bugs with a few of the mini-games too, but in this case I'd consider that a positive because the more 'not quite right logic' places we find, the better when it comes to debugging it.
|
mamehaze
changed the title
|
Doraemon Moving! Oekaki (Japan) [TeamEurope, David Haywood] likely need better translations at least, maybe the other Anpanman mat too |
|
Disney needs a better name |
|
Disney Characters Oto! Iro! Don-Don! Miracle Parade (ディズニーキャラクターズ オト！イロ！トン・トン！ミラクルパレード) |
mamehaze
changed the title
|
thanks, replaced that one. I've left the "Let's TV Play" part because it's on the box even if it isn't on the screen https://muuseo.com/henly/items/561 are the others OK? Doraemon Moving! Oekaki (Japan) [TeamEurope, David Haywood] |
|
Correction: the transliteration should be Ton-Ton, not Don-Don. |
…ks corrupt in places, add 2 workaround to boot other sets so that graphic features can be better tested
new NOT WORKING machines
Anpanman Kazoku De Ikunou Mat DX (Japan) [TeamEurope, David Haywood]
Let's TV Play Dragon Ball Z Battle Experience Kamehameha 2 Ossu Ome Goku Tenkaichi Budokai (Japan) [TeamEurope, David Haywood]
Doraemon Moving! Oekaki (Japan) [TeamEurope, David Haywood]
Anpanman Pyon-Pyon Ikunou Mat (Japan) [TeamEurope, David Haywood]
Doraemon anywhere - Japan travel game DX experience! Where is the Dragon Grand Prix! (Japan) [TeamEurope, David Haywood]
Let's! TV Play Futari wa PreCure MaxHeart Dance on the mat Let's go to MaxHeart (Japan) [TeamEurope, David Haywood]
Let's! TV Play Disney Characters Oto! Iro! Don-Don! Miracle Parade [TeamEurope, David Haywood]