Submission #5768: Alyosha's GG Tails' Skypatrol in 06:53.57

Game Gear
baseline
BizHawk 2.2.1
24782
59.9227510135505
8361
Unknown
Tails no Skypatrol (Japan)
Submitted by Alyosha on 12/27/2017 4:42:12 PM
Submission Comments
Tails' Sky Patrol is a short Game Gear game released exclusively in Japan in 1995. It has some pretty unique gameplay mechanics, requiring the player to use a special ring to interact with the environment as you traverse 5 side scrolling levels.
This is an improvement of 23 frames over the original run by Kumquat. That run was incredibly well optimized, and since the game is already pretty linear to begin with, the possibilities for improvements here were few and far between. I went over every non-trivial area of this run numerous times just to be able to find the small number of frame savings that I did.
I was able to improve this run due to a new clip technique I found on minecarts. It turns out that 2 slightly suboptimal clips on the minecarts are better then 1 totally optimal one (as in what Kumquat originally used.)
All levels were improved.
The breakdown of improvements per level are:
Training area 3
Rail Canyon 3
Ruin Wood 6
Metal Island 6
Dark Castle 3
Apparently I saved 2 other frames somewhere, but I can't really pinpoint where or how, lag is extremely random in this game, and since actual gameplay improvements occur in groups of 3, those are all I clearly account for.

Possible Improvements

Even though I was able to save some frames here, I can't really say I have any real understanding of how this game works. It is evidently quite complex, as several times I was able to gain a few pixels seemingly by random chance. I suspect it's possible to gain a few more frames here and there if someone were to VERY carefully study how everything is built. Or, just get really, really lucky.

Special Thanks

I have to thank Kumquat for the original run which I was very impressed by. Normally I would give co-authorship credit as well, but since I had to redo pretty much everything tht wasn't just holding right, I felt it was un-necessary.

ThunderAxe31: Judging.
ThunderAxe31: This submission appears to be 0.2 seconds slower than the current movie because the timing for this submission uses a more accurate framerate; this submission is actually faster if timing differences are not accounted for.
Accepting as improvement of the current movie.
fsvgm777: Processing.
Last Edited by fsvgm777 on 12/31/2017 3:02 PM
Page History Latest diff List referrers