MSX HDMI mutlimedia card

페이지 15/56
8 | 9 | 10 | 11 | 12 | 13 | 14 | | 16 | 17 | 18 | 19 | 20

By giuseve

Paladin (744)

giuseve의 아바타

21-06-2014, 19:31

Any news about this project??

By maxis

Champion (512)

maxis의 아바타

22-06-2014, 01:15

giuseve wrote:

Any news about this project??

DId you see the last demo on youtube?

9918 functions are fully implemented and tested.

TEXT2 mode video - next week.

By ARTRAG

Enlighted (6493)

ARTRAG의 아바타

22-06-2014, 10:10

Maxis can I suggest you add an optional feature for sprites?
I would use the color byte in the sat as index to a color profile (keep the EC bit, 127 profiles will be sufficient).
why ? To allow patched versions of msx1 games with multicolor sprites

By maxis

Champion (512)

maxis의 아바타

22-06-2014, 11:31

ARTRAG wrote:

Maxis can I suggest you add an optional feature for sprites?

Hello, ARTRAG. No problems at all, this byte is always fetched regardless the sprite mode (16 bit data bus). I have a different sprite fetch algorithm. It is single pass, instead of two passes on the original 9918/38 and all the data from the SAT is available internally. And there is 90% of time available for the additional memory access algorithms.
Since you are talking about the HW upgrade why not to do the full 4bpp or even 8bpp color sprites w/o the color clash. The 4-th SAT byte could just have the index (as you suggest its use) to the alternative sprite pattern table not in the 9918/9938 address space. For the 256 color sprites we will have to reserve 16x16x32 for the start, which is not a lot...
IMHO it would be great enhancing the existing games with the new graphics and the least coding effort. Another interesting option would be adding more sprites per line and enhancing the quantity of the sprite table entries (say 64 instead of 32). Since Z80 has no interrupt precision and no throughput either the most logical way IMHO is to work out the sprite HW. Also I would suggest to add 3 more chunky bitplanes with 8bpp and dedicated palette per plane, to make the smooth parallax scrolling in the existing games (talking about MSX2 games now). These bit planes could be also out of the VDP address space. This way the original game GFX engine will be intact.
The main advantage here is the memory. Out of 64MBytes, almost all of it could be dedicated to the graphics.

Let's discuss all of this. I'm looking forward to implement the HW architecture for the provided FW support.

Currently I'm looking at the SMS ---> Genesis VDP series. IMHO this is the most logical and absolutely brilliant extension of the 9918 line. We can add this functionality too or some of it.

By ARTRAG

Enlighted (6493)

ARTRAG의 아바타

22-06-2014, 11:51

my proposal: use as index for full bitmap sprite definitions the 12 bits stemming from frame and color in the sat.
For 8x8 sprites you need to store 4096 bitmap frames
For 16x16 sprites you need to store 1024 bitmap frames
All games that do not change the spt on fly would be ennanched without changing a single byte of their code
Just preload alternative sprites, enable the super sprite mode and run the standard game

For games changing the spt pointer on fly you could allow multiple banks of ennanched sprites swapped with value of register 5 (or 6 I do not remember which changes spt address).

For games updating the spt on fly one should decode the actual sprite definition and color.
It would be too much (32x8+4 = 260bits, so maybe one should use some hash....)

By hit9918

Prophet (2899)

hit9918의 아바타

22-06-2014, 15:48

@maxis,
"9918 functions are fully implemented and tested" - the problem is that Nemesis etc dont really use the 9918.
ARTRAG just had a case, the toshiba VDP is not compatible on mask registers. Have you seen his uridium, it does scroll scroll Smile

The lower maskbits in colorregister, are they like 9918 or 9118?
Have you tried MSX demo "waves"? It does timed splitscreen things.

By hit9918

Prophet (2899)

hit9918의 아바타

22-06-2014, 16:18

Another idea for great VDP enhancement:
The lovely 9929 has dark gray color and further colorfest. With PAL linecolor mixing. If fpga has RGB/VGA port, it should be able to do it there, too.
Well, "enhancement", again I am rather worried about missing features of the original VDP Wink

By maxis

Champion (512)

maxis의 아바타

22-06-2014, 17:46

ARTRAG wrote:

my proposal:

Interesting. So, then some sort of a sequencer per sprite will be employed to provide the looping for the ever changing sprites. Do you have any particular titles in mind to modify first?

By maxis

Champion (512)

maxis의 아바타

22-06-2014, 18:20

hit9918 wrote:

@maxis,

The lower maskbits in colorregister, are they like 9918 or 9118?
Have you tried MSX demo "waves"? It does timed splitscreen things.

So far I've tried Konami and Ponica titles only and some synthetic tests. Achieving compatibility is not a problem, really. Already fall badly on masking. For example, in Text2 bit 2 of the color base is not always 1 when manipulating with cx5m2 bios extension.
So, I'll use "waves" demo for the compatibly regression test. Can you suggest any other program? I also would try Uridium game, if I would have that.
One more thing: since Procyon FPGA model is a completely indigenous design made from scratch, it bears all the errors of my misinterpretation of the spec and lack of knowledge of the real 9918 implementation. But the control of the design is in my hands, so easy to change and fix on the spot. 9918 spec functionality is achieved, but haven't tested yet the unspecified features. Also have only 9938 and 9958 machines.

By maxis

Champion (512)

maxis의 아바타

22-06-2014, 20:00

hit9918 wrote:

Another idea for great VDP enhancement:
The lovely 9929 has dark gray color and further colorfest. With PAL linecolor mixing.
Well, "enhancement", again I am rather worried about missing features of the original VDP Wink

PAL color mix for 9929 will be emulated too. I'm just wondering what game was developed originally for this color set?
Don't worry, all the hidden features of the original 9918 will be implemented. Smile Apart from HANS TMS9918 archive are there any documents summarizing the undocumented features/findings?

페이지 15/56
8 | 9 | 10 | 11 | 12 | 13 | 14 | | 16 | 17 | 18 | 19 | 20