Over the last week I fleshed out an idea for re-working the Roman infantry trees and how to make the 14 or so available infantry types more interesting and valuable.
Basically, units would be upgraded based on their XP levels and not just barracks + tech. The new barracks would only build the lowest tier units (Hastati + Legionaries respectively). When Hastati hits level 4, they can be upped to Principes, and when those hit 7 they can be upped to Triarii.
Same thing for Legions, around XP level 5-6 they can be upped to vet legions and somewhere in the gold chevrons you can train one per legion to be the first cohort.
I haven't decided specifically how to handle Praetorians but my idea for now is converting them to General's guard units and allowing a similar upgrade to Eagle cohorts. 1-2 XP7-9 legionaries upgradable per Legio.
Some other changes that would be necessary are 1. slowing down the vet gains (beta 3 might fix this), 2. Changing the tech tree so you cant be post-Marian by turn 15, and 3. Modifying the starting Legios to account for this system.
Now I know this isn't the most historically accurate and that the current system is probably better for creating proper Legion builds if you so choose, but I think it would go a long way in making individual Legions feel truly valuable. Also, there's a serious problem when the game obsoletes units the same turn you unlock them. Whats the real point to building Legionaries when I can just mass Praetorians which are a flat upgrade?
I have run through the files with the PFM but unfortunately upgrading units seems tied directly to technologies and there doesn't appear to be a way to incorporate current unit XP into the system. Also, as it stands currently unit caps can only be handled faction wide and not per army, which also creates a problem.
My skills are very limited though and I figured I'd see if anyone here has any ideas or insight that may help me get further along.
For now, I'd rather this stay directed at the physical modding stuff and the about the idea or historical accuracy of the mod if preferable. Even if not specifically for this project I think making progress on this concept could be really good for the modding scene in general since the upgrade/retraining system has a lot of untapped potential.