search

mist-devel / fpgagen

SEGA Genesis/Megadrive core, running on the MIST
8 stars 6 forks source link

Some counter rework (again) #11

Closed gyurco closed 6 years ago

gyurco commented 6 years ago

Now the H_CNT - HV_HCNT schizophrenic behaviour is eliminated. The counter values are taken from real measurements by Nemesis in the spritesmind.net forum. There's a small difference in H40, but shouldn't matter. Unfortunately rendering is not timed, and would be very hard to achieve with the current shared VRAM-SDRAM architecture.

harbaum commented 6 years ago

I wonder if the entire stage 1 sprite processing should only be done once per whole frame. The VDP docs seem to say that the sprite attribute table is only evaluated once before rendering starts and changes during display phase don't have any effect and you cannot e.g. increase the number of visible sprites by altering the sprite table halfway during screen display.

Then the objinfo buffer needs to be extended to be able to hold all information from the sprite attribute table.