I already offered a great compromise for the segmented vs. unsegmented debate. For THIS site, we can have the one (or few) unsegmented run(s). These would probably be on Agent (for fastest time) and 00 Agent (for hardest *intended* difficulty.)
The unsegmented runs could be a different part of the-elite, if they allow it, of course. I'm not sure if any of them are even interested in doing tool-assisted runs, but if any of them are, they could work on the levels individually over there.
GoldenEye will be a hard game to perfect on a single-segment run. First of all, you'd already have to be a MASTER of the game to do the run adequately. Second of all, there are probably tons of tricks/glitches/strats which are possible on an emulator, but have never been done or are too risky to pull off in real life. This is why I suggest allowing a lot of time for the individual levels to be perfected before doing the single segment run.
I think that DLTK is a very valid strategy. I don't know what criteria I use to judge this, but, as an example... A swordless Zelda 1 run seems like a valid challenge. A "Secret to Everybody"less Zelda 1 run is... kinda dumb. Sure, it would require a different planning route and such, but why even do it?
DLTK is a very legitimate challenge. It is the hardest way to play the game without limiting any of your abilities. (They have weird things, like "no strafe" challenges also.)
Another reason to do DLTK is: it is an accepted gametype that has its own rankings in the die-hard GoldenEye community AND there are levels which haven't been beaten yet. We're not just going to be showing off the emulators capabilities for some arbitrary challenge; we're going to be showing what an emulator can do that no human has been able to do yet, even though they've tried.
Maybe a little bit of bias is coming in here, but... I think this game should be left to the GoldenEye fans. Someone who is just introduced to the game casually should not do this run. Someone who has not followed the-elite should not do this run. The accepted challenges by the GE community should not be disregarded because it'll be require too many video submitions.
So you're saying we should suddenly start doing segmented runs?
Is there a particular loyalty to unsegmented runs? If this is for entertainment purposes, yes, do it segmented. It will just be easier on everyone. If it's for, "omg what can actually be done on a console," then stay unsegmented.
But this is very different than old NES games. In an NES game, your action in the beginning of the game may affect something later in the game. (I seem to recall Birdo movements in Mario 2.) In GoldenEye, there are a few different level enemy movement patterns for certain levels. If these are chosen randomly, this can be luck manipulated by choosing to start the level at different frames. Simply, what you do in the Dam will not affect something in the Facility. If I'm wrong from a theoretical / technical / if-we-had-an-awesome-supercomputer aspect, fine, but practically, doing an unsegmented run won't be anymore impressive than the segmented one.
Um. Aren't we storing these movies in zips and torrents anyway? Have a series of emulator movies in the zip, and multiple avis in the torrent. The backend wouldn't need to be dramatically changed, and you could list the whole game under one entry. The hardest part would be adding the times.
But if one level is done better, the person would have to download that level individually. It would be silly to make them redownload all the other files along with the one new level. And if you release the bundle and the individual file seperately, that would be more to download. Say the old Facilitity is in the bundle, and the new Facility is released seperately. Someone will have to download the old Facility (even if they don't want to) to get the rest of the levels. If you placed the new Facility in the bundle, then someone would have to re-download all the levels they already have to get the one update. You could update the bundle AND allow for the individual file, to accomodate different conditions, but if that's the case, why not just release all the levels individually anyway?
Telling ya, the-elite got it right.
Yeah, that might work. See, that site only covers two games, while this site covers many games, so we try to limit movies to one per game, or sometimes two or three if there are other modes of play. The-Elite can afford to look at Goldeneye 007 in more depth.
Agreed. Even from the NES days of this site we allowed a few different videos, like Mario with warps and Mario without warps. However, GoldenEye has so many different challenges that it would be best discussed at the-elite. Not to take away any credit from this site or anything, just a friendly affiliation between two gaming sites.
If Bisqwit's site is to accept only a few videos, it would be the gametypes that were intended to be played (ie, Agent, Secret Agent, and 00 Agent.) - the rest were created from the imaginations of the fans. (This is similiar to Zelda. We would have the legit game rather than a challenge like the sworldess run. In fact, I just checked, and we don't even have a sworldless run.)
So, for this site, we would have probably two videos for GoldenEye: the fastest run, and the fastest hard run. Agent is definitely faster than 00 Agent, but 00 Agent is the hardest (intended) game type. Secret Agent gets left out.
LTK, DLTK, slappers, and individual level runs for all levels will be at the-elite. At this point in time, I would recommend we wait a long while for the best strats to be discovered with the aid of the-elite guys before doing the single segment run for this one.
Maybe the sheer number of movies would be too much for Bisqwit to handle. Perhaps we should talk with the-elite and see if they can start a "tool-assisted" part of their site for GoldenEye.
I really think this is a non-issue. There is no good reason to do a single-segment run. It would be the same length of time as all the single segments (of the same difficulty) combined, with some extra frames.
Furthermore, if a method of beating an individual level faster is discovered, the one level can be redone, rather than re-doing the entire single segment run.
It just makes good sense to do it in the style of the-elite.net
Hmm... To playback a video of Train on DLTK someone would already need the savestate. I guess we'll have to begin including savestates with the videos.
***********DLTK MEANS DARK LICENCE TO KILL. IT'S AN OPTION IN 007 MODE (A MODE AVAILABLE AFTER YOU BEAT ALL THE LEVELS ON 00A) WHERE EVERYTHING IS SET TO MAX. ENEMIES TAKE LIKE 10 HEADSHOTS TO KILL AND YOU DIE IN ONE HIT. **********
So help me God if I had an N64 controller hook-up to my computer I would have made a Train DLTK vid by now. I can only imagine how cool the tool-assisted vids of GoldenEye will be. Rather than having to aim at only one enemy at a time in DLTK, you would be able to stun many guys on the screen with perfectly-aimed headshots. Only four DLTK levels haven't been done by humans: Train, Streets, Cradle, and Silo. I'd recommend that these be the first GE tool-assisted accomplishments because 1) they've never been done by humans yet, and 2) DLTK isn't as much about speed as it is about style, so perfect planning and frame cutting aren't important (at least for the first drafts) so these should be easy videos to make.
The Silo and the Streets both have time-limits, which make them very hard for a human player, but (probably) less so for a tool-assisted run.
There should be several different categories for time-attacks, following the structure the guys at www.the-elite.net use. I'm not sure there would be much point for any single-segment runs, as they would just be a few seconds longer than the individual level runs combined.
The categories should be: Agent, Secret Agent, and 00 Agent, of course. Along with LTK and DLTK. Also turbo mode. These should be primary. These gametypes are all big at the-elite. Other runs could include enemy rockets, turbo mode + fast animation, and slapper only. Tons of different gametypes can be created, but most are dumb.
While it's not necessary, for stylistic effect, we could get 100% accuracy with 100% headshots. (And possibly more with guns that shoot through guys.) However, this would take a longer time.
***********DLTK MEANS DARK LICENCE TO KILL. IT'S AN OPTION IN 007 MODE (A MODE AVAILABLE AFTER YOU BEAT ALL THE LEVELS ON 00A) WHERE EVERYTHING IS SET TO MAX. ENEMIES TAKE LIKE 10 HEADSHOTS TO KILL AND YOU DIE IN ONE HIT. **********
Aha! So there was a reason for it ;) Anyhow. I'd still love to see something else but the floor, even though it might be slightly faster. I was actually amazed over that when I played the game, how similiar some areas are to the movie. For example the stairs in the second level when entering the last room where Trevalyan is. The stairs are the same!
Yeah, I never dug looking at the floor. I don't think the memorization of the levels without seeing them is really skill so much as die-hard fanboyism. I probably would have competed in human speed run competition of the game if it wasn't for the looking at the floor aspect.
I believe a GoldenEye master would be required to make runs of it. Simply, there are too many different level paths and strategues for just anyone to go in and make the run. Studying the world records, of course, would help, but I imagine with the possibilities of an emulator, many of the WR strategies will be obsoleted (for the emulator only, of course.)
So before worrying about speed in GoldenEye, I'd like just to go for style. All levels done on Dark LTK mode with all head shots.
Alright. Let's agree that it would make a TAS more efficient, but this tool could only be useful for short time frames, because after too many frames the sheer number of combinations would take too long to calculate.
Also, there are usually more factors to consider. Imagine that the top five coin-collecting methods involved Mario getting hit, but you need Mario to stay big for some reason. You would need to find more RAM addresses and put more parameters on your search.
This type of technology would be beneficial in bonus rounds of games where you're required to collect as many of something as possible in a given time.
Can you please explain to me what the method your code uses and how you know it is correct? Are you sure that all parantheses were considered also? Unary?
(I know that I didn't say unary operators were allowed in the first post because I thought they wouldn't make a difference. Now I think they may.)
I decided to make this a thread of its own, as I *imagine* the discussion will be long enough to derail the other thread.
The object of the game is to find the smallest positive whole number that cannot be represented with nine 9's using only the operations of addition, subtraction, multiplication, division, and parantheses. Sticking the numbers together, like 99 or 999,999 is not allowed. Also, the program should be as efficient as possible.
Between two 9's there are four operator combinations.
9+9, 9-9, 9*9, 9/9. This is 4^1.
Between three 9's would be 4*4 which is 4^2.
Between nine 9's there would be 4^8.
The total number of results is 65,536, and not all of them are whole numbers.
Here's my idea: 1) generate all possible numbers from any legal combination of operations, 2) sort the elements, 3) check that the value of any element n is 1 greater than the previous element. If the value is greater than 1, then we have found the smallest positive whole number that cannot be represented with nine 9's.
If anyone has found any mistakes I made, please correct me. If anyone can think of a more efficient (in terms of how computers work) way to do this, please tell me. If anyone is really good at math and can find a way to detect repeats based on some mathematical principle that I'm unaware of, rendering doing many calculations unnecessary, please please tell me.
I still have a very incomplete idea of how this works.
My intuition is telling me that there should be a limited number of locations to warp to, rather than being able to warp ANYWHERE on the overworld and underworld, because the flute only warps you to levels you've already beaten.
If we can go to the places up/down/left/right of the levels we've beaten, or any underworld shops/warps/secrets, we should mark all of these locations on the overworld map from GameFAQs. There shouldn't be that many.
Then, we would need to recalculate the shortest travel across the levels based on these new warp points. There exists the possibility that we would need to do the levels in a different order, even if it means walking more in the beginning, to make up more time with the glitch warps later.
Once again I'm onboard to help in the mental anguish of calculating this.
Funny. I was just thinking about this game and new glitches. I have an idea, but never tried it. If someone skilled would try it, that'd be nice.
Link gets boosted when hit by a bad guy or his flame. My idea is to get the flame close to a wall, and then have a baddie knock Link into the flame in a way such that he couldn't hit it normally. Then, with this boost, hopefully he will go through a wall. The purpose would be to exploit poor boundary detection. This may be used to get past locked doors, or maybe even the Old Man of Level 9. Of course, all of this is wishful thinking, and the glitch probably won't even work.
But this video was awesome. I'm not sure I quite understood it. Would it be possible to enter a dungeon room from the overworld? I saw the one time he entered Level 8, but that was the entrance. Also, he had already completed it. If he performed the glitch before having burned the bush, and he walked out of Level 8, do you think the bush would still be there?
Preferably, we could figure out a way to warp directly to Ganon's room. Even if not, I bet this glitch could cut a ton of time out of the run.