3 replies to this topic

[Synthpopalooza]

Apologies in advance if I am treading old ground ...

I was doing some experiments in my ICE CIN editor, on my Atari 1200XL, and came up with something strange.

Firstly, I was doing a character dithering experiment in the Super PCIN mode (this is Graphics 10 mixed with Graphics 12). When mixing these modes, the playfield registers 708-711 are used in both 10 and 12, So, I tried plotting a checkerboard pattern using the same two pairs of colors. The pictures below in Altirra illustrate this:



In this example I blended PF1 and PF2 on alternate VBI frames. In the altirra screenshot I got a solid color field, however when I did this on my Atari 1200XL, I noticed the pixels had a slight variation in brightness. Further experimentation showed that Pf0-PF3 in Graphics 10 is 1/2 step brighter than the PF0-PF3 in Graphics 12.

I also did the same thing in Super MIN (Graphics 9 + Graphics 12) mode:



In this case I blended PF0 in Graphics 12 with the accompanying shade in Graphics 9. Again in Altirra this created a solid color field, but on the 1200XL I noticed that the Gr. 9 pixel was brighter. Further experimentation showed that this pixel was 1/4 step brighter than the Graphics 12 pixel (1/2 of a Graphics 9 luminance step!)

Questions: Can anyone with a real Atari verify this? And would the results be the same if we did this in bitmap mode (Graphics 15+10 or Graphics 15+9) by alternating modes each scanline? Or is this unique to the 1200XL?

One caveat: if you blend 15 with 10, you will also need to adjust HSCROL by one color clock on the GR. 15 lines so that the pixels will line up with the Gr. 10 pixels (the HIP bug).

[Rybags]

Is the "Graphics 12" in normal or GTIA mode setting via PRIOR ?

GTIA doesn't follow the Antic concept of "graphics modes" - all it knows is that a stream of commands comes down the AN0-AN2 bus lines and it only does one of 6 possible things with it during the active display:

3 * GTIA modes, normal multicolour graphics, hires graphics, and the "reassigned palette" you get if you switch from a GTIA back to hires partway through a scanline, where PF0 becomes background and bitpairs 01/10/11 are mapped to PF1-PF3.

Also, you shouldn't judge a pixel's brightness if it's only being shown every 2nd frame, a pixel alternating between luma 6 and luma 0 will obviously appear dimmer than one fixed at 6.

[Synthpopalooza]

The graphics 12 setting is normal, amd it switches with a graphics 0 screen with the gtia settings. What happens is, 0 to 6 (12 to 10) is brighter than 6 to 0.

Edited by Synthpopalooza, Fri Nov 11, 2011 6:07 PM.

[Rybags]

GTIA luma output is digital, then it's buffered and turned into analog externally.

So really, inside GTIA there's no way of generating 2^4 = 16 different values.