now that's a fun discussion here.. first of all, as it was mentioned, the light isn't really a great light, but you did well!
the plants look healthy and they flowered, which is what's most important.
to the discussion below: saying that the cxm32 is "better" than the cxb3590 is a bit of a strange statement, as it first has to be defined what means "better"? do you mean the light output? do you mean value/price?
it should be understood that both, the luminus cxm32, as well as the 3590's are a range of LEDs, with various multiple specifications and generations (well.. the cxb is the second generation of the cxa), and if i have the chance to get a gen1 cxm32 at CRI70 or a cxb3590 as a DB bin, then i go for the cxb every time.. (yes, unlike what many believe, CD is NOT the top bin for the cxb3590, it's DB)
i do however fully agree with Nobody01 that at this point, i would not invest too much in COBs anymore, but go the quantum board route with either lm301b's or lm301h's (cost a bit more, for a very slight difference)..
Having said that I do have and still use a cxb3590, but i also run 2 qb's with lm301b & 660nm reds, and i have 3 more quantum boards on the way (with lm301h's, 660nm and a single 430nm and a single 730nm diode.. whether they make a difference i will test)
coming back: the plant eventually doesn't really care about what lamp she's under, as long as she get's sufficient DLI (daily light integral) (yes, it's a bit simplified, as spectrum does play a role), which for cannabis would be a minimum of 20 mol a day, or for best yield around 40 mols a day. more brings only diminishing returns with much higher cost of electricity, and then you'd also have to add extra co2 for the plant to be able to actually use that higher intensity..
How these 20 (min)/40 (best) mol per day are split up, depends on the light-cycle.. if you're running 18 hours through veg and bloom (as an auto), then this would attribute to a minimum of 308 micro-mol (umol) per second, or 617 umol/s in the best scenario for highest yield per power-ratio. this is basically the often-seen PPFD value that growers look at when comparing and testing grow-lights..
If you have a photoperiodic plant, then these 20 mol/40 mol would have to be divided up by 12 hours instead of 18, and then you'd be around 462 (minimum) umol/s or 925 umol/sec (optimal, but requires co2 to be added).
Next, obviously this needs to be spread out evenly over the canopy, and not just in some hotspots.
That's one of the main reasons, why i prefer the quantum boards, as they pair the high efficiency with a good direct spreading.. and looking at China, these boards don't have to be too expensive either. Kingbrite uses original Samsung diodes, and you can even get added Cree's (for the 660nm). They were a bit slow in responding to me, so i eventually ordered from Meijiu which are the second trustable manufacturer, and i have to say that they were a charm to work with so far.