Thread by @GGnoRESPRiT

A few days ago I shared some random numbers without context related to bloom. Since I'm moving in less than a week and still need to finish getting ready, I realize it's probably not the best time to start working on a project, so I wanted to explain the numbers and share a bit.

Here's a link to the data I collected: docs.google.com/spreadsheets/d…

I'll explain a bit what I was trying to do and my early thoughts. This is really far from being conclusive but I figured I might as well share now.

I created a 9x9 grid of stand up targets in creative (not perfectly space because creative is a huge pain) and then positioned my character far away with a gun, such that my crosshair was centered(ish) on the middle target and the reticle was within the grid.

Since there's some recoil and the targets take time to reset (2 seconds min. I think), I made a simple macro that clicks every 3 seconds and let it run for ~100 shots.

I then went in creative and marked on my spreadsheet where each shot landed, skipping misses.

I only did this twice, with a leg. SCAR and a leg. revolver, and I also had the crosshairs positioned differently, so the data is kinda sparse, but here are my (as always, really rough) thoughts about how bloom works based on what I saw:

1. Bloom causes shots to land within a circle with a radius approx. the size of the distance between the center of the retical and the inside of the outer bars.

2. I believe this is distributed such that for any given point in this circle, the likelihood that your shot lands there decreases as the distance from the center increases. I think this is balanced such that all points of this distance would add up to the same probability. (1/2)

I'm not really a math person so there's probably a better way to articulate this, but basically this means that the distribution would look kind of like this if discretized (I think in actually it is continuous):

3. This one is an assumption: there is nothing more to the reticle besides this circle. Weapon type, rarity, distance don't affect anything as long as two reticles cover the same space on a target (e.g. a small reticle that contains an entire player up close = big from far).

Ok, finally, here's why I think this is actually worth figuring out: assuming that we can compute or test for these accuracies with some level of precision, we can develop heuristics for deciding when certain shots are worth taking or when to use certain shooting styles.

For example: let's say you see someone rotating by launching 100m away in the sky. Whether or not you should tap-fire or spray and track this person partially depends on how bloom works. Bloom may cause your hit rate to be too low, even with perfect tracking.

Your ability to track and your ability to tap-fire also affect this, but are presumably related skills, and if it turns out there are scenarios where spraying is almost never worth it, that's stuff that can be optimized, even if it's just a little bit.

I'm sure a lot of top players have already inadvertently optimized this just through feeling, learning over time when to tap-fire and when not to... but there's also the chance that the room for improvement makes a difference. But also, it's good to know how the game works.

Some things about this data: human error is a factor here, my cursor wasn't perfectly centered, for example. Also, my grid makes it kind of awkward to reason about circles, and I didn't do any geometry to see how a circle would overlap with each tile. Everything is ad-hoc here.