Mega Man Xtreme 2 is a gameboy color mashup of several of the Mega Man X games.
Unlike the original run, this one uses Zero throughout, which while being a significant strategy difference, is only about 2 seconds faster. Because of emulation differences though the run is overall quite a bit longer, as accurate lag emulation adds up to many frames.
The trade off here is between the time it takes to buy and equip X's weapons, as well as switching back and forth to X, and the inefficiency of using Zero to fight bosses and navigate the latter stages of the game.
Here is a basic list of the tradeoffs:
+5 seconds : buy weapons for X
+5 seconds : equip and switch back and forth between X and zero
+1 seconds : load in animation for X
+7 seconds : Launch Octopus Stage
+3 seconds : 1 second for each boss taking 8 hits instead of 7
-1 seconds : improved movement in Blast Hornet stage (would be closer to 2 seconds but some frames lost to inefficient movement)
-2 seconds : defeat Sigma 1 boss in 6 hits instead of 8
I only found the last improvement in Blast Hornet's stage recently, and without that, all it would have been too close to call considering there is also a lot of lag everywhere in this game. I also traded a few frames (~20) in the early stages to maximize income in order to make sure I had enough to buy Zero's saber 2 after the 8 bosses.
There are a couple of actual improvements in the early stages as well. Flame Mammoth is improved by ~1 second with a better route, and Neon tiger's stage is ~40 frames faster as Zero, so since you start as Zero there anyway, you can stay as him until reaching the end of the stage.
It's unfortunate that VBA didn't do a good job emulating lag, so comparing runs is pretty hard. As soon as you have more then 2 enemies on the screen you have to start adding frames compared to the original. Also VBA counted all loading frames as one frame, so ~600 frames is added just due to counting loading time correctly.
This run though has the benefit of being console verified, so at least the lag is known accurate:
Overall I'm happy that this run actually worked out the way I hoped. While it's still slightly improvable, particularly with a real understanding of RNG, I am at least confident that the Zero route is the optimal choice.
New SRAM verificaiton movie: http://tasvideos.org/userfiles/info/68657981202423899
GoddessMaria: First, I'd like to say that I wasn't expecting this submission; it was a nice surprise. There is no mistake that this has been optimized further in gameplay on more accurate emulation and shows off more variance in action than its predecessor. Good work on the improvements!
Accepting as an improvement to the previous movie.
Also added the "Extreme mode" branch label, as the previous movie is labelled as such.