there is a HUGE difference in making a canister light that works, and engineering one.
for example. Several big names in lighting are using off-the-shelf parts and putting them into a waterproof housing. Fairly easy if you have a mill and know where to buy the bits and pieces. You will get about 60% efficiency doing that because the boards aren't heat sinked properly, the emitters aren't appropriate for the in-water environment etc. Those will be "good enough" but you have to go way overboard to get a comparable amount of light because of the efficiency. The cost then comes against battery size and heat.
You can properly engineer it and get over 90% efficiency doing it, but there is a ton of engineering involved and you aren't going to save any money.
You can see here a comparison of a light that was designed using off the shelf parts, against one that was fully engineered. Wattage is really a stupid unit for LED's because it doesn't mean anything like it did for HID, but we'll use it. You can see below the LM 12, Hollis 25, and LD-15. The LD-15 is on high and consuming about 15w of power, the LM and Hollis didn't have settings, but are consuming 12w and 25w. You can see that the light output of the LM12 is nowhere near that of the LD15 despite the power only being 20% more. It's easily double the light output. They claim 1200lumen out of that vs 1500 lumen from the LD-15, but the difference is that the LD-15 is better than 90% efficient so that much more of the light is coming out and being useful. The Hollis 25w looks to be a similar amount of light to the LD15, but is consuming 1.7x as much power to do it.
If you're going to try to DIY your own light head, you are going to optimistically get the efficiency seen in that LM12 and Hollis25 which means your battery has to be that much bigger to get realistic burn times and it's all because of the optics and heat sinking. That's not bad if you have your own machining equipment and really want to DIY it, but you aren't going to save any money if you really want a high end light. The best way to start DIYing it IMO is to start with a mag light head, machine the battery end down and put a cable on the end of it. Your beam angle is going to be wide, but that's where I'd start. Otherwise lights from companies like ANO are super cheap and you aren't going to be able to DIY it cheaper than that
Beam Comparison
