TIRs are almost always preferred by me these days, so I’ll agree. Also, I did some very quick and dirty layout measurements, and a ~32mm LEP and 12* ~13mm LED optic with 5mm emitter hole could yield ~170kcd and 12k lumens with the W2 running 50A total (Vf would be ~3.2V). The same LEP with the bigger optic should yield ~750kcd.
I don’t know if the driver (resistance wise) can deliver that much more amperage. The prototype on your cells is only pushing ~20A it appears with an emitter Vf of ~3.25V.
I estimate then about 32A total, 4 per emitter. Vf around 3.4V. This would suggest a circuit resistance of 20mOhm assuming each vtc5a is about 15mOhm. Not bad, but maybe could be improved some. Are springs bypassed? What FET (s) does this driver use?
According to these assumptions, 8 * 50.2 should put out about 17k otf and 48A
This looks like an awesome light but with two separate light sources, I would expect to control them using two separate switches. Going through multiple click hoops to get what you want seems really cumbersome. The Fenix LR40R got it right in this regard.
There’s a video of the prototype’s UI in the second post, if you missed it.
I have their D1, which uses their own hold-to-turn-off UI. The driver is good but I’m not keen on the UI. I also have their tint-ramping FW4X, which I didn’t really like the UI of as it had a separate config mode to change the channels. I was concerned that they’d use the FW4X’s UI here but they haven’t.
Here's the Thor Pro prototype's turbo output visualised against the Lumintop GT70, which has similar lumen and candela numbers but is a few times the size and weight: