Joined: 1/24/2018
Posts: 304
Location: Stafford, NY
Not really a new discussion topic, but more to announce my current TAS attempt at the A2600 port of the arcade game, which is more than 28 seconds faster than GameBreaker64's woefully unoptimized second submission for the game.
Userfile: http://tasvideos.org/userfiles/info/59062312797578783
Video Encode:
Link to video
This was performed on BizHawk 2.3.2 using the default game start and settings, with the final input being on frame 7,273 (2:01.37), a full 1,713 frames (or 28.59 seconds) faster than GameBreaker64! Note that despite the amount of savings I have found here, I do not believe this effort is submission-worthy yet because A2600 Jungle Hunt appears to mimic the original arcade game in the difficulty increasing up to the fourth loop.
While continuing my efforts to get through the full four loops of the game for a proper submission, I will invite anyone to give their feedback and possible improvements to the current 1st-loop userfile in this topic.
--------------------------------------------
Individual Section Comments:
* Deadly Forest: After finishing this section, I was surprised to find it was actually ~32 frames slower than GameBreaker64's time! I tried introducing deliberate delays at various points, but nothing seemed to allow me to catch an earlier vine cycle down the line. I then tried playing back GameBreaker64's inputs back to see what was going on, but that resulted in a desync after a few jumps. When I tried resyncing the inputs, I found the exact same jump pattern I had already come up with, so I'm convinced the loss in this section is the result of changes to AtariHawk in 2.3.2 (either a slower scrolling speed or a different vine pattern than the one GameBreaker had in 2.3.1), but I would like someone to prove me wrong.
* Reptile River: As feos pointed out in the discussion topic for GameBreaker64's second submission, it is much faster to kill all the crocs rather than letting some go past you. I was able to improve this section by roughly the same amount feos had.
* Boulder Field: Standard "run left and jump as needed" affair, but it turns out the collision detection on the larger boulders isn't great - when in mid-air, you can clip anything from your leading foot downward through the boulder without triggering a death! This means stopping to duck boulders is completely unnecessary - all you need is a correctly timed jump.
* Cannibal Camp: After jumping each cannibal and continuing left, the game will normally stop you against an invisible wall briefly before unlocking the screen scrolling to let you continue on. Constantly jumping after clearing the cannibal allows bypassing this invisible wall (or shortening your time behind it), allowing you to get closer to the left side of the screen before the controls lock during the screen scroll. (This works for all three transitions - I think this is the same as the "double jump" strat Spikestuff found, but that only was applied for one of the transitions.)
-------------------------
Potential Improvements:
* If someone found a way to match GameBreaker's strat for Deadly Forest in 2.3.2, then the ~32 frames lost there could be recovered.
* Skipping the transitions completely in Cannibal Camp could lead to a sub-2:00 time - I can get close with jumps, but it seems the controls completely lock as soon as the next sprite in the sequence appears on screen.
* In theory, you could jump right up to the cannibals in Cannibal Camp to save time, but I believe it would require extensive manipulation of their behavior to be useful (usually the cannibal just runs right into you if you try). In the absence of any evidence for a potential benefit, I will leave that to someone else who believes such a benefit can potentially be found.
c-square wrote:
Yes, standard runs are needed and very appreciated here too
Dylon Stejakoski wrote:
Me and the boys starting over our games of choice for the infinityieth time in a row because of just-found optimizations
^ Why I don't have any submissions despite being on the forums for years now...
Thanks to Spikestuff, I'm working on the side to this game but only onthe seconddifficulty so I'll see on your version if i can improve it. I'll give updates.
Joined: 1/24/2018
Posts: 304
Location: Stafford, NY
Quick update on the project status:
I just tested my current 1st loop submission in the new BizHawk release (2.3.3), and just like I found with GameBreaker's submission in 2.3.2 I got a desync after a few jumps in Deadly Forest. This means I'm reasonably confident emulation differences are indeed the cause of the loss in that section.
c-square wrote:
Yes, standard runs are needed and very appreciated here too
Dylon Stejakoski wrote:
Me and the boys starting over our games of choice for the infinityieth time in a row because of just-found optimizations
^ Why I don't have any submissions despite being on the forums for years now...
