Joined: 10/12/2011
Posts: 6435
Location: The land down under.
Chef Stef wrote:
Jump+spin+triangle didn't work there because the loading screen cancels it if you're in range.
Sorry. Got my notes wrong.
Something like this:
http://tasvideos.org/userfiles/download/35342506771892646
Is what I should've provided/mentioned instead.
(PAL Version, Hawk 1.11.6. Should sync on 1.11.8.2)
Bit lazy on the input I provided.
Chef Stef wrote:
Edit: Oh, and what do you mean about sliding on the edges?
In the running from Boulder/Bear sections.
You're able to slide on the little lip to get distance better, but I don't recall if it's a benefit or not.
It is less entertaining than no-GO but...
Chef Stef wrote:
I can do non GO-abuse if you prefer, that's why I posted the WIP so early :D
Just keep GOing on with it... kills himself.
WebNations/Sabih wrote:
+fsvgm777 never censoring anything.
Disables Comments and Ratings for the YouTube account.Something better for yourself and also others.
Joined: 5/23/2006
Posts: 361
Location: Washington, United States
Interesting. I think that only works on PAL - holding right in NTSC isn't enough to escape the load trigger. You have to actually slide to get far enough away.
I experimented a bit on PAL and here are some hard numbers on the speed differences between versions:
-- NTSC: 9 frames of sliding at 18.43 speed, 12 frames of spinning at 12.29 speed = 14.92 speed per frame
-- PAL: 8 frames of sliding at 22.53 speed, 9 frames of spinning at 15.02 speed = 18.55 speed per frame
So movement in PAL is about 25% faster. Crash also jumps higher and falls faster. Sounds like a winner to me. I'll start a PAL version and see how it goes.
Opinions seem split on game over abuse vs. no game over abuse. I'm a fan of fastest-time-at-all-costs as a concept but no-death would show off more of the game. I can defer the decision until after level 1 so if anyone else feels strongly please chime in.
Joined: 5/23/2006
Posts: 361
Location: Washington, United States
After messing with the PAL version for a weekend I've decided to continue the run on the NTSC version. Once you take into account the framerate differences (50 vs. 60 FPS) the speed increase in PAL is much more modest than I suggested in my earlier post. The increased fall speed gets in the way of tricks like sliding across pits, and overall it's harder to achieve the same level of precision in PAL due to differences in how movement angling works. Don't get me wrong, I think a properly-optimized PAL version would ultimately be faster, I just don't see as huge a benefit to a version change as I thought at first. Also, I want to make sure I don't kill my motivation on this TAS in the first place! So with all that said, onward to level 4 - I'll post another video once I've made some progress.
Joined: 10/12/2011
Posts: 6435
Location: The land down under.
I'm sorry, but I'm going to have to post it like this.
You've created one of the most bullshit posts and excuses on why you're using the NTSC version and continuing forward.
Not knowing on whether or not using the faster version which is the PAL version is a sheepish excuse to completely avoiding it.
Chef Stef wrote:
Also, I want to make sure I don't kill my motivation on this TAS in the first place! So with all that said, onward to level 4
No, you're not moving on.
As I've pointed out to you.
Your movement is lacking quite a bit of wiggling which will save you a lot of time in the long run.
WebNations/Sabih wrote:
+fsvgm777 never censoring anything.
Disables Comments and Ratings for the YouTube account.Something better for yourself and also others.
-- NTSC: 9 frames of sliding at 18.43 speed, 12 frames of spinning at 12.29 speed = 14.92 speed per frame
-- PAL: 8 frames of sliding at 22.53 speed, 9 frames of spinning at 15.02 speed = 18.55 speed per frame
Because I'm curious, I'm just going to convert these to seconds.
NTSC:
9/30ths = 0.3 seconds of sliding at 18.43 speed.
12/30ths = 0.4 seconds of spinning at 12.29 speed.
9*18.43 + 12*12.29 = 313.35 distance per slide spin.
313.35 distance per 0.7 seconds = 447.6428571428571... distance per second. Call it 447.6ish due to assuming 30 FPS instead of the actual 29.97.
PAL:
8/25ths = 0.32 seconds of sliding at 22.53 speed.
9/25ths = 0.36 seconds of spinning at 15.02 speed.
8*22.53 + 9*15.02 = 315.42 distance per slide spin.
315.42 distance per 0.68 seconds = 463.852941... distance per second.
Comparative rate:
PAL is 3.62% faster assuming all slides and spins are on flat ground and run for their full duration. However, what really jumps out to me here is that PAL slides are 2% faster and cover 8.66% more distance than NTSC slides. Noting that slides downhill keep the slide speed while slides uphill will be followed by a slidespin, constant slidespinning is very clearly a decent bit faster on PAL.
Missing speed variables:
Walking, air speed. (If PAL is faster here, load times or lag would be the only way NTSC would be better overall... and it's known that NTSC is laggier?)
How zigzagging affects this. (RTA runs primarily use PAL... but I honestly don't know how TAS level zigzag would affect this)
Joined: 5/23/2006
Posts: 361
Location: Washington, United States
Sheesh, I guess I've offended you somehow, Spikestuff. I never said PAL was slower, I'm saying I found its particulars more of a headache to deal with and decided the speed-up wasn't worth it. You're right - I didn't think hard about the version choice at the beginning, I started with NTSC because it's the version I play and it's where I did all my sandbox testing prior to starting the run. I thought I was making a reasonable decision sticking with NTSC but it sounds like it's a bigger deal to you than I thought.
I will clarify that I don't like the idea of using PAL on the basis of speed alone, especially when it's on the order of a few % rather than 25+%. If there are actual tricks that can only be done in the PAL version I think the argument is much stronger. Otherwise I'd argue the PAL version is at the very least discouraged under TASVideo's version rules.
I'd also like to single out this point:
As I've pointed out to you.
Your movement is lacking quite a bit of wiggling which will save you a lot of time in the long run.
I brought this up before but didn't get an answer. Where exactly have I failed to do this? I'm doing this on every single slide and I'm pretty sure I'm doing it optimally. If you're convinced that's not the case, can you please demonstrate with a movie so I can improve my run? I'm not doing this TAS in a vacuum, I've done a lot of research on Crash's movement behaviors to make sure I'm doing things correctly. There's a reason it's taken me 25k rerecords to do two levels.
Edit:
I have no name wrote:
Missing speed variables:
Walking, air speed. (If PAL is faster here, load times or lag would be the only way NTSC would be better overall... and it's known that NTSC is laggier?)
How zigzagging affects this. (RTA runs primarily use PAL... but I honestly don't know how TAS level zigzag would affect this)
Thanks for slicing the numbers. The ones I posted were with zigzagging and analog sticks, so I believe those are maxed out (assuming I didn't miss something).
Walk speed is 7.51 in NTSC, 9.56 in PAL.
Air speed is 9.56 in NTSC, 13.65 in PAL (both including wiggling).
All of these numbers assume that you're travelling optimally in a given direction. Travelling off-axis from the level, for example, will reduce Crash's overall speed to varying degrees.
Of course, there are special cases like sliding on ice or wall-clipping where the speed values will differ too. I've also noticed that the game compensates for lag frames sometimes by increasing the slide or spin distance by 1.5x for a frame. I imagine this is the same in both versions.
Walk speed is 7.51 in NTSC, 9.56 in PAL.
Air speed is 9.56 in NTSC, 13.65 in PAL (both including wiggling).
PAL walking speed is 6% faster than NTSC walking speed. Then again, you're not walking very much.
PAL air speed is 18.9% faster than NTSC air speed. There's... a lot of jumps.
Thus, between the faster slidespins (not by as much as I expected) and the faster air movement, PAL is definitely the better version.
Also worth noting: Crash jumps slightly higher on PAL. I believe NTSC gains on the jetboard sections in the water, but that's it. I'm not the most knowledgeable about Crash games, but I've picked up some information from being around those who are.
Another version difference is that the jetpack moves faster on PAL: this allows getting through the 2 hit scientists in only one attack. It's possible to completely skip the jetpack in at least one version both version of the game, though I don't know if both jetpack levels are doable without the jetpack.
Lastly, TASVideos version guidelines. Yes, they would discourage PAL over NTSC, but that's less because NTSC is always preferred and more because NTSC is almost always faster. NTSC-J vs. NTSC-U comes down to which is faster, which is usually J for dialogue reasons, though U can be used as a speed/entertainment tradeoff. For Crash 2, I'd say PAL has the entertainment edge, but it definitely wins for speed.[/s]
Joined: 5/23/2006
Posts: 361
Location: Washington, United States
My reading of the TASVideos rule is that speed differences can be attributed to inexact (improper) conversion to PAL format, and therefore the PAL format is not preferred for cases where speed is the only difference. But you do make some good points. I focused on slidespin gains being only marginally better in PAL but overlooked the larger jump speed difference, which will probably add up to a lot over a full run. Another argument is that a game-over-abuse run is focusing on speed over all else already and should therefore use the PAL version for the extra time savings.
Let me give the PAL version another shot - If I can at least match my results from the NTSC version (trick-wise, not time-wise, we all agree PAL is faster) I'll feel better about switching.
No one is going to reject a Crash 2 TAS for using PAL, that's for sure. I would even consider it more likely that a judge would hold it against a TASer to use NTSC in this case.
Current project: Gex 3 any%
Paused: Gex 64 any%
There are no N64 emulators. Just SM64 emulators with hacky support for all the other games.
Joined: 5/23/2006
Posts: 361
Location: Washington, United States
I'd prefer 100% too but it seemed daunting to follow up on pirohiko's amazing WIP. I figured I'd start with a shorter run first.
Anyway, thanks all for getting me to try the PAL version again. Level 1:
Link to video
Bizhawk movie file
In-level time for NTSC was 2380 frames, and in PAL it's 1880 frames (equivalent to 2256 NTSC frames). So the PAL version of level 1 is at least 2 seconds or 5% faster than NTSC.
One nice thing in PAL was that the extra speed let me just barely clear the mud pits in one slide, versus the NTSC version where you catch the edge and have to jump out. I was able to slide past all three in the level. On the flip side, there was one (regular) pit that's slide-able in NTSC but wasn't slide-able in the PAL run. The NTSC run relies on lag frame abuse to cover more distance, but the PAL version doesn't have lag frames there.
Another interesting note is that Crash's movement is more finicky in the PAL version. But in practice there's more leeway to do stuff and still keep top speed. For example, NTSC's top sliding speed of 18.43 is only possible when zig-zagging directly along the level axis, whereas in PAL you can go maybe 10 degrees off-axis and still keep the optimal 22.53 speed.
Joined: 5/23/2006
Posts: 361
Location: Washington, United States
Warepire: I'm considering just doing both at once since there's so much overlap. If that's doable I'll be slightly prioritizing the no-deaths run, since having full-level runs as a baseline will help plan where game overs actually save time in the GO-abuse run. Though I suppose we could just figure that out with enough testing beforehand.
Joined: 5/23/2006
Posts: 361
Location: Washington, United States
Finished level 5:
Link to videoBizhawk movie file
Running the numbers against the NTSC version, level 5 is 2.8 seconds or about 11% faster in PAL. But that includes a lot of time from faster loading during the deaths at the end, so if we exclude that, PAL is 1.5 seconds or about 5% faster.
I also started the no-game-over-abuse run and finished level 2. (Level 1 is identical to the GO-abuse run.)
Link to videoBizhawk movie file
Out of curiosity, how many time do you save by GO-abuse level 5 ? Because with your fast pace, it isn't obvious that it's faster to do so on this level.
Joined: 5/23/2006
Posts: 361
Location: Washington, United States
Hmm, that's actually a good point. Pirohiko's 100% run finishes the level and regains control in the warp room in about 32 seconds. PAL any% will be faster by a few seconds.
My run regains control in the warp room 29.5 seconds after collecting the crystal... and there's still another ~2 seconds that have to be spent turning in the crystal later on. So at first glance it looks close and probably in favor of just finishing the level normally.
This is the sort of thing I was hoping to work out with the no-death run for comparison but I didn't expect it to be a possibility so early. Prioritizing the no-death run seems like a good idea now so we can properly route-plan the GO-abuse warp room.
I am shocked that after all of these years, there is no 100% run yet in the PSX section of Tasvideos on Crash 2! One of the greatest games of all time is sorely lacking in tas videos! :'( I don't want to bother you guys with this, but if you don't work on/and finish 100% Tas's for all Crash Bandicoot games, the planet Earth will implode within one year! ;)
I'm posting this here because the Playstation subforum is dead.
Joined: 10/12/2011
Posts: 6435
Location: The land down under.
Did you try not hanging on the walls during Hang Eight to see the time you'd get or did you just do that?
Whilst going across the sides you seem to lose for a few frames when rubbing across it.
Also what's with the 45 degree angle of the board during it?
I know it's wallboosting, but you seem to slow down for a tiny bit.
Question on The Pits would spinning the second/third bird benefit in any speed? As you're able to get a boost off of the birds.
WebNations/Sabih wrote:
+fsvgm777 never censoring anything.
Disables Comments and Ratings for the YouTube account.Something better for yourself and also others.
Joined: 5/23/2006
Posts: 361
Location: Washington, United States
I tried a bunch of strats in Hang Eight and this was the fastest one I found. The angle was necessary to get the full wallboosting speedups. There are a few sections where the level geometry sticks out, causing slowdown for a few frames, but it's outweighed by the extra speed. I generally minimized those by angling the stick away for a few frames.
For The Pits, the maximum boost from spinning birds is only marginally faster than the average spin-slide speed, and since it delays the next possible spin it's pretty situational. The double-bird sequence was the only place I was able to save time with it, and that was only because it let me glitch through the second bird. But I'll double-check the other places too.
Edit: I confirmed that boosting is slower due to the set-up involved to make it work. The speed improvement gets cancelled out by either having to jump higher over the bird (to have enough time for a second spin), or having to wait frames for the spin to finish after the boost is over (the boost only lasts 6 frames). In an RTA the trick probably saves time because it's actually a good bit faster than Crash's normal speed, but in the TAS we get much better speed from abusing analog input, so the boost isn't as meaningful.
