do the graphics update, add the correct id (only your device id, delete the others)into the info.plist for geforce, nvdaresman, nvda40hal, then use marvin utility for patch the opengl frameworks. Forget about the leonvidia kext, you want 10.5.2... don´t you?

I've got ToH 9.2.0 and graphics update from Rev1 (currently Rev2) DVD. The original GraphicsUpdate1 always panics the kernel. Without NVinject/wahtever I have failsafe mode, with it - native res, no QE/CI. With kexts from "leo_vid_kexts.zip" (unfortunately I don't know their origin) - black rectangles instead of menus, render bugs in Dashboard and mostly the mess in other places that use OpenGL.

I might haven missed something but is your device id definatly in geforce.kext and nvdanv40hal.kext?
If I miss my device id in one of these I get to change res but no QE CI

Leo_vid..._kexts are a 10.5 version and they are not compatible with the new opengl frameworks (10.5.2 graphics update) and they are causing all of your graphics problems (balck....). You can try this, download this, extract and move the "patches" folder to the root do the graphics update again, boot with -s and type

mount -uw /
cd /patches
./replace.sh
./patcher cpuid.txt
./patcher cpuid-graphics.txt
reboot

edit the info.plist for your kext, nvdaresman, nvda40hal, geforce, and add only your device id, delete the others, copy into the extensions folder nvinjectgo open a terminal and type

sudo -s
(password)

cd /system/library/extensions

chmod -R 755 * && chown -R root:wheel *
cd ..
rm extensions.mkext
kextcache -a i386 -e
reboot

boot with -f -v

That's what I was thinking lol

Okay, I've installed GraphicsUpdate (and got a system freeze neary instantly after setup).
Rebooted in single mode, applied replace.sh, patcher cpuid.txt and patchr graphics-cpuid.txt (you seem to miss that, but this is the one cpuid that patches OpenGL).
Rebooted just to see kernel panic (GeForce.kext/NVDAResman.kext)
Rebooted in single mode, restored my Gf* and NV* stuff (from Rev2).
Rebooted, fixed IDs in NVDAHAL40, NVDAResman, GeForce and NVinject (noticed the side-effect of replace.sh - completely dead BT mouse).
Rebooted to see OSX in the state I've started from (not counting dead mouse, *sniff*).

Other ideas?

btw, my board is 10de:0398

Why not put your device id's into the gf and nv from the upgrade, instead of replacing them?

Is that going to fix kernel panic?... ok, I'll try now (the truth is that I didn't remember ID, and lspci doesn't work in single mode).

good luck don't know if it will work but I reckon its worth a go.