So…maybe give up on sharp edges and try to reflect towards the centre? 
Then that would basically be a wavien collar, and you would not end up with a round die projection like the luminus LEDs.
This could be interesting. The mechanism would be more complex.
Here’s a D4 shorty next to an Olight S-Mini, S1, and S10. IIRC the On The Road light is a clone of the S10. I’d like to see a BLF S-Mini though…
The Olight S-series clips (especially the S-Mini) have been some of the best I’ve ever seen. I didn’t think a clip-on clip could be good before trying these, but they work really well. I also hadn’t ever cared for Olights in general, until someone gave me one and I found it was actually pretty nice.
I’d love to see a BLF version of the S-Mini with an extra-small FET+1 driver (or even just the +1, or perhaps a 2+1, or similar) and the latest tiny85 BLF firmware.
Or perhaps a 14500 version of the FW3A.
Or the CNQG brass AA with Nichia 219c, 1x7135 driver, Crescendo, and a clip.
On all of the above, I’m thinking high CRI at 4500K or 4750K. Maximum output doesn’t need to be very high; it’s more about the size, shape, quality, beam, and interface. Specs would probably not appear impressive on paper, since the goal is to make a really nice EDC. Preferably mod-friendly too, but the traits with priority are things which don’t really change much over time so it likely wouldn’t need modding to remain relevant.
Got it
Like the small size of the Olight but so hate the UI and it shitting off without warning on low voltage and not being able to turn it on again. A BLF firmware would solve my issues with them for sure.
The FW3A could be a good starting point for going smaller.
If Fritz wants a downscale to 18350 or even 16340 tube would bring it down to really small.
In that pic the D4 looks big compared to the left one, but the 18650 tube shows they are all small to tiny 
Or of course a logical next after Q4/Q2 which are about the illusive AA 14500 driver
Made big in the Q4, then can be done smaller in a tiny lights capable of running AAA.
Would still be longer then the D4 short i think
I understand why they did it that way (CR123 support), but it still bothers me. I want LVP and battcheck, but the Olights don’t do that and instead rely on the cell’s protection circuit to avoid damage. 
I’d say I’m aiming for something about the size of a finger, but I don’t know how big your fingers are.
For a 14500 light, the host I’d like to use is 76mm x 18mm. It might need to be slightly longer to allow for protected 14500 cells.
For a 16340 light, I’d ideally like to make it just as small as the S-Mini, 55mm x 21mm.
The D4 shorty is relatively large (wide) at 63mm x 28mm.
Personally, I like 18350 more than 14500 and 16340. Same size ballpark, but significantly more power and capacity. Actually looking at the fatty Emisar D4 shorty and Peak Eiger Ultra AAA I find the size difference too small to justify the smaller light. Though it does feel better in the mouth.
I would love to see a good 18350 light smaller than D4 though. Especially a triple.
So…for me there’s only one step-down from 18650, it’s 18350. Down from it, I wouldn’t get anything larger than 10280.
I prefer 18350 also, significantly. TK doesn’t use the brightest settings often though, so the lower power and capacity is a fair trade off for the smaller size. Ah, choices…
Since a lot of flashlight manufacturers are making pocket clips that are clipped to the flashlight body with an open ring (like a clamp or a jaw (don’t know the appropriate word)), a groove is needed to accomodate it. To keep the wall with enough thickness behind this groove, the whole body wall is generally thicker than the one of flashlights that have a pocket clip with a complete ring or one that is screwed to the tailcap.
The reylight pineapple’s pocket clip is the kind which is the best for me.
Really looking forward for this as the BLF team next project.
It makes me think just to call Olight and ask for a NW ramping version for us
But calling doesnt seem to work all that good, tried getting Maglite to do Old Lumens version but did not get passed some slick “have a nice day” secretaries.
OP updated, thanks ISO!
For an e-switch light: could you have a one-shot “disable LVP” (or “use primary cell”) configuration option with the setting stored in volatile memory? LVP would be on by default, but if you put a CR123 in, you could use the configuration option to turn LVP off until power was lost at the next cell change. At that point, LVP would automatically reactivate until you used the configuration option again.
The tradeoff is that you’d have to set the option every single time you put a CR123 in / used tailcap lockout / turned power off with a series switch. That might still be acceptable to people who normally use Li-Ion but would like an emergency CR123 capability, though; things would automatically work as expected when using a Li-Ion cell.
For a series-switched light: you’d have to save the LVP configuration in non-volatile memory, so the user would have to take responsibility for reactivating it when using a cell that needed LVP. I imagine some people would still be fine with that, especially if they normally used one type of cell or the other, so they could just set it and forget it, but there is admittedly a risk of ruining a Li-Ion if they forget to turn LVP back on after using a CR123.
Yes, there are several decent ways firmware could support both 16340 and CR123 without over-discharging a 16340. Unfortunately, Olight doesn’t even try.
OTOH, the attiny85 I’ve been using lately has unfortunately high power overhead during use (~5 mA), so it makes moon mode rather inefficient. Perhaps a more moon-friendly MCU also makes it harder to add fancy features. I’ve been meaning to experiment with the tiny85’s settings to see if I can find a way to reduce its operating power level, but I suspect the only options might require reducing PWM frequency to visible speeds… which is arguably worse than an inefficient moon mode.
It would be great to see a AAA/10440 or AA/14500 truly pocketable BLF light with a good U/I and high cri LED. Preferably (tail) e-switch with lighted switch or very low drain moon mode to be used as night beacon.
Being able to blast in DD on Li-ion and be versatile to use on alkalines when not in a situation to get a recharge (but lower output) would be the bees knees. With Narsil or something alike would make it a priceless light to me, and I think for a lot of people.
I modded a Thrunite Ti head with MTN-12DD driver with guppydrv and XPL HI allmost 2 years ago and legoed it to the tailsection of a bare aluminium Lumintop Tool. For a 10440 it’s relatively long and it could do without the hard “click” from the button, but love it a 1000 times over a twisty.
It gets by far the most use of all my lights. Capacity is rarely an issue for daily (fairly short) use. It gets charged at least every week, but small size with big power always comes at a prize. I rarely depleted it though, 14500 would be more than enough for its use.
It is unnoticeable in a (skinny) jeans or a suit trousers pocket, length here is less of an issue than width. I wear it every day and therefore use it multiple times a day.
It’s great, but via BLF we could come up with something even better. How nice to have a light like this available to everybody!
A DQG sized AA light would be comparable in size with more capacity. Any thicker and I won’t EDC it in my pants and therefore hardly use it. I have a very nice modded 18650 DQG, but it gets way less use, since its too thick to carry in my pocket on most occasions.
The BLF-348 is nice, but I stopped using “dumb” flashlights without batt. check, programmable modes etc. IMHO the simple three or one mode flashlights are obsolete.
I do agree with TK that there is a void to fill in the small section. I by far use my 10440 lights the most. And in direct drive it is still very bright, especially for its size.
With the BLF A6, Q8 and GT there is a lot of throw and lumens in nice builds with good U/I out there.
I would love to see this in a truly pocketable BLF version!
ST, Short Thrower. But quad series cells,like QT more
And a check if I can use another hoster for images:
be prepared for my wicked sketch skills 
QT, same overal size as Q8, but longer head
QT set, head, a short tube and a looong tube
Lego party:
I really love this idea! The QT is exactly what I have wanted to see for a long time. All of the other options are cool also. I would be on this interest list in a second.
Really like the concept. If I knew I could expect ST in 6 months I would skip MF-02 and wait for ST instead.
Wouldn’t extension tubes be a better idea than long ones? It would enable for example 3s lights.
Future idea, for a light that will be thread-incompatible anyway:
Make a battery tube that is male threaded on one side and female on the other, so one can just use another regular tube as an extension.
It is very hard to align things nicely so lots of the double tube lights have different designs for the two tubes and I haven’t seen such light that looked good.
It also would increase the OD or create a wider rim.
The QT with long tube and 18350 in series could run a XHP35HI with rge GT driver (using 8 that will be a pain to charge but still
The Miller, I like your lego set very much but it seems to imply a driver which accepts either 1S or 2S. I’m not sure what the practical options are there but it rules out using the Q8 in its current form with the long tube.
Also, could the long tube not be an extension instead? Then you could buy 3 and build a 4S4P monster light Probably too crazy an idea for mass production though.
Awesome idea, I’d want to be on this interest list!