XP-G2 on the left, original XP-G on the right.
XP-G2 tint is 1A from Cutter, driven at 50mA vs the XP-G at 40mA, will take some better photos once I have a second p60 host handy. Taken after each had been running for a couple of minutes.
Boards = http://www.dealextreme.com/p/17mm-2800ma-5-mode-regulated-led-driver-circuit-board-for-flashlight-diy-dc-3-4-5v-127684?item=4 with 5 chips removed
Nice work manual man, looks like a very decent increase in output 
Any update on this ?
Oops yeah… This light did about 120-130K lux with 4xEnergizer Lithium primaries. Now take this measurement with a HUGE grain of salt, as we were using a brand new lux meter that Jason pulled fresh out of the box and our only reference for calibration was a 200K Olight SR95. So we took a lux measurement of the Olight beast and used that for the conversion factor.
In terms of throw distance, the XPG2 build was very close to the thrunite TN31. The beam patterns were very similar too which was kind of surprising considering the size difference between the dies. Both lights held a tight beam focus that opened up to a smooth flooding spot WAY out, at similar distances.
The XPG2 build was definitely more of a pencil-sharp beam than a fat XML throw beam from say an HD2010 or skayray STL-V2. FWIW my stanley fatmax and defiant spotlight DIY builds both measured 76-78K that night… so both of my best XML budget throwers were out-classed. I could easily see how the XPG2 build held onto its tight focus and spat out a tight/bright beam while my XML throwers flooded out at shorter distances.
I think the coolest thing about the build though was that its a TOTAL sleeper. It did nearly the same thing as the TN31 using 4AA cells and a plastic carrier, it also drew slightly (IIRC) less power from its 4 cells than the TN31 from its 3x18650. It was the smallest and lightest thrower of the night… by a significant amount too. Part of that credit goes to the light-weight AA lithium cells. The host surprisingly did not generate THAT much heat. It was definitely warm, but it wasn’t painfully / scalding hot, and a good percentage of the heat came from the cells as they dumped ~2.5-3A into the LED.
Oh color tint… that was a BIG surprise. IIRC the build used a 1D tint BIN and it was pale white with “maybe a very slight” pink / warm color. And when I say “very slight pink” I mean the kind of color thats ONLY visible when comparing againt other pale whites. By itself, without a color reference it would easily pass as a pale-paper white. DEFINITELY none of the green ugliness that I so dislike from the regular XPG emitters.
Ok, so how much are they in the group buy? 
Thanks very much for the update, kramer5150.
You know, I have only (1) light with an XP-G. An Olight T20. The color is really nice and makes objects more visible with less power than some…
Hope the color is similar.
Before you said the XP-G2 has a tighter hotspot, but it doesnt look like that of the beamshot.
Also we cannot really compare the XP-G2 at 50mA and the XP-G at 40mA, ok it is only 10mA less, but it is also 20% less !
Does also this mean the XP-G2 will have a bit less flood than the XP-G ? Those already are throwers, so for me it is good to have some flood with it though.
The beam-shots were hard to get right when the camera’s battery dies every 20 seconds and you only have 1 p60 host on hand so probably not the best for comparing but they give you a rough idea.
Despite the extra 20% current to the XPG2 it looked quite a bit brighter (30%?). I have since pulled the p60 apart to scavenge the board so can’t do any more comparisons however from what I remembered the hotspot was brighter right in the center and faded out to around the same size and brightness (as the XPG) further towards the edge of the hotspot.
To be honest I think flood levels will be the same as the 20% increase should replace any of the light being better focused.
Wonder how long it will take them to hit mainstream?, if I recall correctly it took about 4 months before the XML arrived in force (mind you the XML made a bigger splash then these should).