TASVideos

Posted: 8/11/2007 11:18 PM

moozooh

Experienced Forum User, Senior Moderator

Joined: 8/4/2005
Posts: 5789
Location: Away

Sorry, but there already is an improvement to Genisto's movie published here: http://tasvideos.org/830M.html And it's 16 seconds faster than yours, too, due to using two players, and other reasons.

Posted: 8/11/2007 12:58 PM

moozooh

Experienced Forum User, Senior Moderator

Joined: 8/4/2005
Posts: 5789
Location: Away

There are a few people who really make useless comments, generally with silly reasons or plain wrong reasons

Posted: 8/11/2007 12:43 PM

moozooh

Experienced Forum User, Senior Moderator

Joined: 8/4/2005
Posts: 5789
Location: Away

Titus Kwok wrote:

There are tool-assisted runs of Doom, aren't there? I would think Hexen wouldn't be much of a stretch if there are.

Right, except they're done on PC version of Doom, not an [poorly] emulated version of Doom. The engine was specifically customized for allowing TAS conditions in it, which isn't the case with both N64 and PC versions of Hexen.

Posted: 8/10/2007 10:50 PM

moozooh

Experienced Forum User, Senior Moderator

Joined: 8/4/2005
Posts: 5789
Location: Away

DK64_MASTER wrote:

He was referring to the other rating system where there was no such warning.

Except it's supposed to be even more strict there: "Do not rate a movie after having watched it only once. Watch it at least twice, and the second time concentrate on the rating categories given above."

Posted: 8/10/2007 7:47 PM

moozooh

Experienced Forum User, Senior Moderator

Joined: 8/4/2005
Posts: 5789
Location: Away

Saturn wrote:

I'm pretty sure I still optimized my Ceres rooms to the max, because when even JXQ, hero and especially moozooh (who is known for extremely crazy frame optimization) didn't manage to beat them "with" using the memory watcher for all possible stuff, then this says much, I think.

Hey, I haven't even tried the escape yet. So far I'm only in the second room. And you forgot those 3 frames you have lost in the first pre-ridley room. ;) Also, it feels like you're comparing apples and oranges, since we weren't able to monitor randomness via RAM watch, either. We are under the same conditions, and so far the only thing that prevents us from having perfect randomness is willingness to test every delay combination (there can be thousands of them, all very similar). And since hero and JXQ were going for realtime frame count, it wasn't even among their goals. I and evilchen are pretty much the only ones except you who strive for better ingame frame count/escape timer.

Posted: 8/10/2007 7:05 PM

moozooh

Experienced Forum User, Senior Moderator

Joined: 8/4/2005
Posts: 5789
Location: Away

I liked the camera work this time. Excellent job, guys.

Posted: 8/10/2007 5:03 PM

moozooh

Experienced Forum User, Senior Moderator

Joined: 8/4/2005
Posts: 5789
Location: Away

Chamale wrote:

In Mario Bros, can 1 player pick up the other one? That could make for interesting moves.

You can't pick anything in MB/SMB. The only thing you can do is bumping player 2 just as any other enemy by punching the platform they're standing on.

Posted: 8/9/2007 11:36 AM

moozooh

Experienced Forum User, Senior Moderator

Joined: 8/4/2005
Posts: 5789
Location: Away

Still not long enough to top the list. :(

Posted: 8/9/2007 11:34 AM

moozooh

Experienced Forum User, Senior Moderator

Joined: 8/4/2005
Posts: 5789
Location: Away

Snes9x 1.51 has the same feature, too. (TOO BAD IT DESYNCS, BOOHOO.)

Posted: 8/9/2007 11:32 AM

moozooh

Experienced Forum User, Senior Moderator

Joined: 8/4/2005
Posts: 5789
Location: Away

I think you guys haven't even looked at that Get Well card. Transparent hint: look at it.

Posted: 8/9/2007 11:27 AM

moozooh

Experienced Forum User, Senior Moderator

Joined: 8/4/2005
Posts: 5789
Location: Away

Saturn wrote:

Nope, it wouldn't be of any benefit for me at least. But anyway, here is how I did it in detail: I optimized the subpixel position for every door, and I manipulated luck for perfect steam by delaying some frames in the intro scene and at Ridley. The combination of this was the key. There is nothing more about it.

Hmm? Weren't you asking me how to optimize the subpixel positions (and what did they mean at all) a couple months back? It doesn't seem like it would help anywhere compared to, say, JXQ's run, though: I've tried to optimize the pre-Ridley part with absolute precision recently (had both better subpixel position and speed at every door, much thanks to acceleration storage tech), and was held back by 2-frame door opening cycles, jumps and other events like that, impeding (or rather, nullifying) sure frame gains. I'm pretty sure the way back will show similar results; the only place where further subpixel optimization might gain a frame is at the jump in 3rd room, making it one frame earlier if possible. I will look into it when I return home. Otherwise I'm pretty sure it's more about randomness than anything else. On a related note, have you determined any patterns in escape counter behaviour that might be useful/manipulable somehow? Because the results I get are kinda weird in the least; for example, in the test I was talking just above I left Ridley's chamber (frame 2412, IIRC) at 59'86 while having fully matched JXQ's times at entering all the rooms (he leaves at 59'87). Seems like this logic-defiant timer can be affected by literally everything, not just timing of pressing start at the intro cutscene, entering the rooms and taking a hit from Ridley. What the hell.

Posted: 8/8/2007 2:59 PM

moozooh

Experienced Forum User, Senior Moderator

Joined: 8/4/2005
Posts: 5789
Location: Away

Truncated wrote:

moozooh: I am curious how you find the subpixel adress. Do you look on adresses near the whole pixel adress, or is it by the usual way of searching? Perhaps this calls for a new topic...

Actually, it depends on the game. In game like Super Metroid, this can be done by, for instance, placing Samus under conditions that would lower her overall speed to less than 1 px/f and searching for increasing address each time the position doesn't go over the next pixel boundary (and decreasing each time it does). Games like Metroid Fusion track the character's position already in subpixels (quarterpixels in case with MF and MZM). While I was searching for horizontal position, a handful of 5-digit addresses emerged, with one of them incrementing by 320 every frame of movement. Judging from other addresses, I quite quickly found that Simon's speed was 1.25 px/f, and 320 is 1.25*256, which led me to believe it's tracking subpixel position. So as you can see, Castlevania's method of tracking subpixels is something inbetween the above two methods, which makes it very convenient to use. And for some games (especially NES, which has far lower overall amount of addresses), opening the RAM table and looking at it is indeed the easiest way. Subpixel position for NES Metroid was found this way.

Posted: 8/8/2007 2:03 AM

moozooh

Experienced Forum User, Senior Moderator

Joined: 8/4/2005
Posts: 5789
Location: Away

Here are some useful addresses for you guys: 7E003A,2u — ingame frame counter (excellent for measuring time between different emulators); 7E054A,2u — X pixel (center); 7E0549,2u — X subpixel (1 unit = 1/256 of a pixel); 7E1F9C,2u — Y pixel (head); 7E054E,2u — Y pixel (center?); 7E1F9A,2u — Y pixel (feet); 7E13F6,1u — boss energy (not very useful as it simply duplicates the red bars). For AngerFist's convenience, I've uploaded a memory watch pack with the updated addresses (just unzip it anywhere and run alongside Snes9x). Also, could anyone tell me why does this happen and if it's going to be useful anywhere? I found this glitch on console many years ago, but never heard of any useful application since then.

Posted: 8/7/2007 11:33 PM

moozooh

Experienced Forum User, Senior Moderator

Joined: 8/4/2005
Posts: 5789
Location: Away

TGPrimus wrote:

Sliding was annoying, but i can understand that it saves time.

It only saves time in one part in the entire run (though I'm not sure if it was due to sliding at all), stages 4-7 are not improved even though they were slid through.

Posted: 8/7/2007 8:43 PM

moozooh

Experienced Forum User, Senior Moderator

Joined: 8/4/2005
Posts: 5789
Location: Away

Hmm, apparently, I was slightly disappointed. While you did improve the published movie, you did it by a very small amount. However, the exact reason of my disappointment was your stylistic choices. Sliding got really old halfway through the movie (and it didn't really help anything, as evident from your improvement list), and post-boss actions were all the same, while Scumtron's were all different and very amusing. I can't help but give it a meh vote, sorry. :\

Posted: 8/7/2007 6:23 PM

moozooh

Experienced Forum User, Senior Moderator

Joined: 8/4/2005
Posts: 5789
Location: Away

wazkatango wrote:

I already knew about the whole "fast forward your movie to the end, and then start recording again", but what if you have a movie that is hours long. ( Example: Earthbound) Someone ought to make a utility to load movies at a certain time and allow re-recording without having to watch it every time. ( Unless I'm not looking hard enough again and it has already been done.

That utility is called a savestate. Before you close your emulator, do a savestate. After you open it again, load it in read-only mode to continue watching, and in read+write to continue recording. Loading it otherwise is impossible because a movie doesn't contain information for each frame, and thus have to be "reproduced" up to the needed point.

Posted: 8/7/2007 5:06 PM

moozooh

Experienced Forum User, Senior Moderator

Joined: 8/4/2005
Posts: 5789
Location: Away

Well, I know there are, but I'm not sure it would justify using a password. Playing it under default conditions makes it more easily comparable to conventional play, which for some people is easier to comprehend and thus enjoy. On another note, will it make the movie faster or not?

Posted: 8/7/2007 4:33 PM

moozooh

Experienced Forum User, Senior Moderator

Joined: 8/4/2005
Posts: 5789
Location: Away

Nah, I don't think it's necessary. For the record, it's the same thing as when you pass the game and start again without resetting.

Posted: 8/6/2007 9:40 PM

moozooh

Experienced Forum User, Senior Moderator

Joined: 8/4/2005
Posts: 5789
Location: Away

What I'm very surprised about is the amount of turns he does in mid-air. It looks like it's done without rerecords, either.

Posted: 8/6/2007 9:36 PM

moozooh

Experienced Forum User, Senior Moderator

Joined: 8/4/2005
Posts: 5789
Location: Away

I think this game would benefit from speed/entertainment tradeoffs, like most fighting/brawling game do (btw, it really reminds me of DBAA). Is there some attack that can potentially make you run faster than while simply running?

Posted: 8/6/2007 7:16 PM
(Edited: 8/6/2007 7:17 PM)

moozooh

Experienced Forum User, Senior Moderator

Joined: 8/4/2005
Posts: 5789
Location: Away

klmz wrote:

Then just play my movie in 6400% speed, isn't it 00:00:09?

And if you play the published movie at 6400%, it's 00:08'47. The point, however, is that you failed to improve upon the published movie. Even if you had just taken all the input from it and applied it to an NTSC game, it would have been 9:22. Yours is 9:48, 26 seconds slower than what's supposed to be optimal. How do you explain the difference?

primorial#soup wrote:

I think that's exactly the point here. One should not and cannot disregard the difference in playback speed. The current movie is PAL (E) and was recorded in NTSC (J U). Had it been recorded in PAL mode, as it should have been, it would have had a length of 11:15.

Movies between NTSC and PAL-unoptimized game versions have to be compared under same conditions. Otherwise, one could have taken the NTSC version of the game, run it in 11:14 and claim it's faster than the current movie, which is actually ridiculous.

Posted: 8/6/2007 6:50 PM

moozooh

Experienced Forum User, Senior Moderator

Joined: 8/4/2005
Posts: 5789
Location: Away

klmz wrote:

Just open the Blue Shadow ROM with FCEU 0.98.16 and load the published movie, then look at the "Length: 00:11:15". Isn't it longer than 00:09:48?

It's longer just because it plays slower. If you discard the difference in playback speed, it's 9:22.

Posted: 8/6/2007 3:46 PM

moozooh

Experienced Forum User, Senior Moderator

Joined: 8/4/2005
Posts: 5789
Location: Away

Amazing. :D I would love to see a full run.

Posted: 8/5/2007 3:13 AM

moozooh

Experienced Forum User, Senior Moderator

Joined: 8/4/2005
Posts: 5789
Location: Away

My favorites, in order of the time I spent listening to them: Touhou 07—9.5 (all PC) — GODLIKE; Cave Story (PC) — ALSO GODLIKE; Killer Instinct (SNES; came with OST CD); DKC 1&2 (SNES); Super Castlevania IV (SNES); Jets'n'Guns Gold (PC); Earthworm Jim SE (Sega CD); Earthworm Jim 2 (PS1); Quake (PC) — absolute 100% match for the game's atmosphere; Rise of the Robots (SNES) — shitty game but the music is excellent; Diablo 1&2 (PC) — rich and fitting; Chrono Trigger (SNES); Final Fantasy VI (SNES); Sonic 3 & Knuckles (Genesis); + everything else I forgot.

Posted: 8/4/2007 8:38 PM

moozooh

Experienced Forum User, Senior Moderator

Joined: 8/4/2005
Posts: 5789
Location: Away

What's wrong with GoogleVideo which has nearly no limitations, far less restrictions than YouTube, and doesn't require installing anything?

Posts for moozooh