Armory Sponsor
Posted: 9/2/2026 10:34:12 PM EDT
[Last Edit: lazyengineer][Edited]
|
I did some load development work today, for some plinking ammo. I set up a bunch of different targets, of different grab-bag styles and tips, and had a great time shooting for a few hours to see how my progressive loaded ammo would go. I then took a photo of each target, and it either was a target with it's own 1 MOA grids on it, or just laid one near it, and took my picture of each target. I dumped about 20 quick smartphone pictures into the AI chat, with instructions to analyze for group size, Mean radius, and zero off-set of impact, and make a table listing the round, the gun, the velocity, etc. AI cranked for almost 2 minutes an image, doing its assessments. . It actually worked surprisingly well and it did a good job descerning whats what in the many images, better than I was expecting. Today's testing was noisy and messy, with limited variables. Using surplus American Reloading components and factory seconds (so not exactly precision stuff here). I told it to give me a report summarizing the results and any recommended loads from this, as well as to graphically show the best tested rounds of each time, their distance to 1000 ft-lb and distance to Transonic, and well as gong-range usage input. Here's the report, which actually agree's with my own observations, and a pretty good presentation. I liked it.  |
|
Then I asked it to make a summary of the phone-target image dump, with instructions on how I wanted it to be summarized and to look. with each lot# called out, and the SFAR AND the RPR on each target of each lot, with what sight settings for each, it would use, to get the group centered. It cranked non-stop for 7 minutes, and came up with this, after one go-around of clarifying instructions, which it cranked for another 5 minutes. So starting from images that looked like this:  with filenames: Target 20 - RPR - 5shots - 65-197e - at0.3R0.3 2696-7 140 SST.jpg It took a dump of those, and made this: |
Armory Sponsor