Auf Reddit hat jemand einen Test gemacht der die Unterschiede von Accuracy bei ARs zeigt:
Record a video of me firing 28 magazines of an AR, have some program chop the video into images. Have a second program analyse the images to track the location of the reticle. Use a third program to filter the data and display graphs.
Setup a macro to fire an AK-M so that every magazine and reload takes the same amount of time. Record that clip with Shadowplay. Import the video into MATLAB2019a and run a downloaded tracking script by Mike Claffey. Edit the script so that it would actually work (that took way longer than expected, lol). Use Python3 to filter and display the results.
Here's a GIF that shows the result of the tracking script. Note that it also tracks unrelated objects, those are filtered by Python3.
Here's a picture of all the graphs I plotted.
First row plots all the locations of the tracked reticle. Second row shows the normal distribution in x and y direction. The third row shows the exponential distribution of the radius from the center. Left columnhas a lower percentage accuracy than the right column. Accuracy percentage is determined from the in-game accuracy bar.
Higher accuracy has a significantly compacter spread. How much this impacts gameplay, I don't know. I did not give that much thought (yet). To me it would be logical to quantify spread in travel distance to target accuracy. However how to do just that is not yet clear to me.