Would [BODY_APPEARANCE_MODIFIER:BROADNESS:75:75:75:75:75:75:125]
hold a different meaning than
[BODY_APPEARANCE_MODIFIER:BROADNESS:3:3:3:3:3:3:5] (a simplified proportion)?
The scale here is in percent, so it's not just a proportion. In the second example, you will have always extremely thin creatures.
This is not true. After extensive testing and debugging with a personal mod, I discovered that not only are the numbers relative, the position matters. I was trying to determine which direction LENGTH and THICKNESS scaled (ie, is 0 long or short?) and, assuming the same thing you did above, I set a test to 5:5:5:5:5:5:10. Result?
Nothing. In other words, average.
After another hour or so spent testing, I was able to determine that the 4th/center value sets the average value and all the others are relative to that value. So, an entry of 5:5:5:100:5:5:5 will return "very short" 6/7 and nothing at all 1/7 of the time, while an entry of 5:5:5:5:5:5:100 will return nothing at all 6/7 the time and "extremely long" 1/7 the time.
I also discovered an oddity with the TL_COLOR_MODIFIER tag. It's supposed to use a weighted average, but it doesn't appear to use the
numbers for weights, but rather the
position in the list. Items closer to the front of the list are orders of magnitude more likely to appear than items at the end of the list, even if the item at the front has a weight of 1 and the item at the rear has a weight of 10000. The more items there are in the list, the more obvious the difference becomes. Get a long enough list and you'll never, ever see the colors in the second half of the list.
If I had to guess, the TL_COLOR_MODIFIER thing is a bug. I'd guess it's totaling the weights cumulatively; in other words, the first item in the list gets its weight added to the weight of every other item in the list, whereas the final item only gets its own weight.