The ghosting of sprites is actually documented by TI themselves to some degree…
See the programmer’s guide section 8.4.2: “What this mode does is effectively shrink the Graphics II Mode Color and Pattern Table down from Hex 1800 bytes to Hex 800 bytes. [...] When in this mode 32 sprites can no longer be used. If you try to put more than eight sprites on the screen at once, they will start to duplicate themselves on the screen.”
Surprised this isn’t emulated by openMSX…?
Another way to increase the number of sprites beyond 32 is by using a screen split which changes the attribute table address mid-screen.
Since when are sprites limited to 32 per screen? I thought the more important limitation is # sprites per scanline.
No reason you can't change Sprite Attribute Table (or sprite patterns) several times per frame. Not to mention software sprites although that takes more effort programming wise.
And yes collision detection is a useful feature. But I'd think that still applies even if you're messing with SAT or sprite patterns halfway into a frame. That would just put some restrictions on what exactly is or isn't detected.
It is the well known and documented cloning effect, as Gauw said
It is unstable and can vary from chip to chip even with the temperature
So it is hard to get something useful with it
Bleah! Do not expect nothing that is really useful from the crappy TMS VDP.
Even some undocumented modes that have some usefulness are plagued by some others side-effects that make them unusable as ARTRAG said
