Hi jake_nolan_3933, thanks so much for the kind words, I really appreciate it!
On the limitations you mentioned: that was the first real hurdle. At the start I set everything up with idle GIFs and actions shown/hidden via triggers so it all lived in one flow. But the handoff from the idle loop to the attack sequence was never truly clean - there was always a small, noticeable hiccup. In practice I couldn’t reliably align the last frame of the loop with the first frame of the action; between GIF decoding, visibility toggles, and timeline timing, a micro-break would appear and break continuity.
The fix: inspired by Darkest Dungeon, I added a quick camera push on the action so the cut is “masked” by the transition and the eye reads it as a natural move. Technically, I kept the idle as lightweight GIFs and prerendered the attacks as MP4s; I hook them up with precise cue points, then pull back to the wide shot and resume the idle. This gave me smooth motion and consistent timing.
The trade-off is complexity: moving to video meant multiplying clips and slides, especially to handle variants when Guard is active. With GIF-only I could have kept the structure lighter and added more combinations (status effects, extra physical/magic attacks, specials), but I couldn’t consistently eliminate that tiny break between loop and action. So I prefer the reliability of prerenders with a camera push and cue points, which gave me a smoother, more readable result across the board.