I have this exact same problem. I'll share a little more of what I've discovered.
My machine is a Samsung NP700Z5C-S01US running ML 10.8.2, similar to the one described in the above post. After trying to inject each of the different ig-platform-id's, I noticed that only 0x01660003 and 0x01660004 produce any image on my internal display, indicating that the display was connected to port 5. However, while both id's enable QE/CI, they produce markedly different results.
When the id 0x01660003 is injected, the appears either red-tinted (most of the time) or blue-tinted (some of the time). Either way, the display is stretched out so that only the left half is actually shown. A picture of this can be seen
here.
However, when the id 0x01660004 is injected, the display appears to be the correct resolution, but an incorrect refresh rate and odd colors, as in banditos93's image. There is constant flickering and lines scanning downwards and the color spectrum appears almost inverted, although not quite.
Looking through AppleIntelFramebufferCapri.kext, I noticed the difference in the connector tables for the id's 0x01660003 and 0x01660004 for the LVDS connection:
Code:
0x01660003: 05 03 00 00 02 00 00 00 30 [COLOR=#ff0000]00[/COLOR] 00 00
0x01660004: 05 03 00 00 02 00 00 00 30 [COLOR=#ff0000]02[/COLOR] 00 00
Fiddling around with that value, I found that if I take that byte modulus 3, I get the following results:
- 00: (Mostly) Red-tinted, stretched to only show left half of the screen.
- 01: More often blue-tinted, stretched to only show left half of the screen.
- 02: Odd, close to inverted color spectrum, shows entire screen but flickers (vertical refresh rate seems to be off).
For example, setting that byte to 00, 03, and 06 seemed to produce the same, mostly red-tinted and stretched display, while setting the byte to 02, 05, or 08 seemed to produce the same, inverted and flickering display. Changing the other values besides the port in the framebuffer appear to only change how the display shows up in About this Mac > Displays.
Besides trying the above changes, I've also verified that the EDID that is picked up for the display is the same in both Windows and Mountain Lion. Also, interestingly enough, if I don't inject an EFI string and just use shared 64mb memory (no graphics card), the internal display shows up perfectly with no problems, but without QE/CI since no hardware acceleration is used. It appears that this display problem is an issue with the connector tables in AppleIntelFramebufferCapri.kext, but I'm not quite sure what exactly the problem is.
Can anyone shine any light on this problem?