automatic gun stat generator

~~jtyotJOTJIPAEFVJ~~ · 03-25-2018, 02:21 PM

I made a calculator that can convert real-life gun stats into a GURPS weapon stat block using MACHINE LEARNING™ and INTERPOLATION®.[1]

You put in bullet mass, muzzle velocity, bullet diameter and weapon mass and get out an approximation of damage, recoil, ST requirement and bulk that tries to match similar weapons using real-life performance numbers of guns listed in high tech.

It seems to work fairly well for guns ranging from pistols to high-caliber rifles and even gives somewhat useful numbers for cannons. This calculator can at least serve as a useful baseline for an 'average' weapon with the stats you put in, which you can then tune to your liking.

Some examples of guns that do have GURPS stats but are not present in the 'training' dataset in the calculator:

VSS Vintorez (~290 m/s muzzle velocity, 9 mm bullet diameter, 16.8 g bullet mass and ~6 lb weapon weight) gives following statline:
2d+2 pi, rcl 3, st 9t, bulk -4
which is close enough to the statline found in here: http://www.sjgames.com/pyramid/sample.html?id=1719
(2d(2) pi with AP ammo, which is an exact match for damage)

2A42 30mm (960 m/s, 30mm, 385 g, 254 lb) gives following statline:
(7d+1)x5 pi++, rcl 4, ST 30 (M), bulk -12
Sounds very similar to the stats available here:http://www.sjgames.com/pyramid/sample.html?id=2234, most importantly the KE damage works out to 127 average given by my calculator, versus 105 in the pyramid article, which is still close enough.

You can find the calculator on google sheets through this link: https://docs.google.com/spreadsheets...it?usp=sharing

For clarity, it does contain stats of a bunch of firearms from High Tech, but I removed weapon names and all labels from the datasheet from this version so it's pretty much impossible to tell what the unlabeled numbers mean unless you already own High Tech.

[1] I use a simple kNN classificator to find a couple of similar statlines from the input data and use regular old linear regression to interpolate given stats from those close matches.

03-25-2018, 02:21 PM	#1
~~jtyotJOTJIPAEFVJ~~ Banned Join Date: Mar 2018	automatic gun stat generator I made a calculator that can convert real-life gun stats into a GURPS weapon stat block using MACHINE LEARNING™ and INTERPOLATION®.[1] You put in bullet mass, muzzle velocity, bullet diameter and weapon mass and get out an approximation of damage, recoil, ST requirement and bulk that tries to match similar weapons using real-life performance numbers of guns listed in high tech. It seems to work fairly well for guns ranging from pistols to high-caliber rifles and even gives somewhat useful numbers for cannons. This calculator can at least serve as a useful baseline for an 'average' weapon with the stats you put in, which you can then tune to your liking. Some examples of guns that do have GURPS stats but are not present in the 'training' dataset in the calculator: VSS Vintorez (~290 m/s muzzle velocity, 9 mm bullet diameter, 16.8 g bullet mass and ~6 lb weapon weight) gives following statline: 2d+2 pi, rcl 3, st 9t, bulk -4 which is close enough to the statline found in here: http://www.sjgames.com/pyramid/sample.html?id=1719 (2d(2) pi with AP ammo, which is an exact match for damage) 2A42 30mm (960 m/s, 30mm, 385 g, 254 lb) gives following statline: (7d+1)x5 pi++, rcl 4, ST 30 (M), bulk -12 Sounds very similar to the stats available here:http://www.sjgames.com/pyramid/sample.html?id=2234, most importantly the KE damage works out to 127 average given by my calculator, versus 105 in the pyramid article, which is still close enough. You can find the calculator on google sheets through this link: https://docs.google.com/spreadsheets...it?usp=sharing For clarity, it does contain stats of a bunch of firearms from High Tech, but I removed weapon names and all labels from the datasheet from this version so it's pretty much impossible to tell what the unlabeled numbers mean unless you already own High Tech. [1] I use a simple kNN classificator to find a couple of similar statlines from the input data and use regular old linear regression to interpolate given stats from those close matches.