This topic is for the purpose of discussing #1769: Wyster's N64 GoldenEye 007 in 21:52.18

I enjoyed the run and it was well played, I espeacially like how you made a complete mockery of the Control objective when protecting Natalya. There is one problem though/ a couple of problems though 1. You used antids Its not like you used an outdated emu, or a different one from what most people would choose, the main problem is that its not approved and it has problems creating AVIs i.e. there corrupted. Plus I don't see how it was neccesary to use it because you didn't even require a reset. Also there seems to be some horrible unnatural lag in places where regular Mupen doesn't have it, such as throughout most of the Cradle level. 2. You choosed to do the easiest level You may choose to bring up that Duke_64 plays on the easiest setting, but the problem here is that 00A setting makes the game harder and more interesting as well as complicate the game with extra and stricter objectives. Many of the levels on agent difficulty can be completed just by running to the end of the level e.g. Dam, Streets and Depot. I congratulate you for your efforts though and hope that you'll choose to obselete Tempest's run someday. I'll give this one a 'meh'

I voted yes. The easiest difficulty was chosen, which means the completion time is as fast as it could be. The playing was also very optimal. The only thing I didn't like was the delays at the pause menu in Frigate. I know that this was done for manipulation purposes, but are you sure there weren't other, more entertaining ways to manipulate the hostages?

Maybe you guys don't see the difference, but difficulty means (in most cases) less objectives thus different strats to completing them asap. imo each difficulty would deserve it's own run, but I'm pretty sure it's too strict for that here, but that doesn't mean we can't do it for our own community, so yeah, yes'd for the sake of pwning all of this game, 2/3rd done if not counting any side projects such as turbo/dark ltk.

Well the reason i didn't do 00A was mainly cause Rising Tempest just finished his run and it's incredibly optimal already, i probably wouldn't be able to improve it, also i wanted this run to be as varied as possible, so choosing a diffrent difficulty seemed like the best way to do that. And lastly, it's always interresting(IMO) seeing a game completed in fastest possible time, which is exactly what this run does. I guess you could call it an any% run while Tempest's is 100%(if aztec and egypt is included). And for frigate, sure it's possible to do manipulation with angles and stuff but there's several problems making it practicly impossible to do: The randomness is decided exactly when the host takers die meaning you'll be in the middle of the pause when this happens, and you can't do angle stuff while in pause(can pause later too but that would loose alot of time. Secondly and most importantly, you need to manipulate both hostages at once, and since every host has like 6 diffrent routes that's 6*6=36 diffrent possibilies and getting the best possibility only happens like once every 100-200 frames. I figured the only possible way to do that kind of manipulation was by waiting out the pause, which also proved effective but not very esthetical. And about the mupen thing, I suppose because your text is overlined that it's not a problem anymore? Cause i've had poeple make avi's using that version. And i'm sorry for using that version but i didn't know there were any other diffrences other than the playback thing and it was theh only version i had configured when i started the run and i'm a bit lazy to configure the other version.

do you think it would be possible to get 33 on cradle? you had a very low decimal there (34.05 or so).

Fun stuff, Henrik. I voted yes, I think this would fall into the "any %" category (guess that means when (if?) I get around to improving my current one I had better include Aztec/Egypt). NFQ, 33 secs on Cradle just can't be done- think of it this way, the time in the level doesn't matter at all until Trevelyan activates the console (provided you enter the hut after him optimally). From there on, there's like seven seconds of gameplay where your speed actually -matters-, and as can be seen watching both mine and Henrik's rendition of the level there's really no way to get any more frames out of that (barring some freaky fall through the ceiling glitch or an extra grenade for a boost which still might not be enough).

Optimized, and beats the game as fast as possible, right? Yes vote then.

I guess I don't mind so much the easier difficulty, but really I enjoyed RT's run more because of the extra objectives he had to complete. The 00A run is a 9/10 in entertainment for me, and this is more 7/8, TBH. Although your control was really good, but too many of the stages were not as fun compared to the 00A run because of less objectives. I also don't as easily see two Goldeneye runs on the sight, especially since many members don't hold a double standard (i.e. of course we can have 50+ categories for Super Metroid and Castlevania AoS, but a N64 game? no way!) So I think you did a great job, I give a yes and I'd be very willing to encode the stages individually for youtube/the-elite.net and such, but I'm going to be realistic and say I don't see this easily becoming published for the site.

That was a lot more different than I thought it'd be. Yes vote. Do you have a list of your level completion times?

Sure: Dam 0:54(+1) Facility 0:43(-2) Runway 0:23(+1) Surface 1 1:01(-2) Bunker 1 0:17(0) Silo 0:58(-5) Frigate 0:22(-1) Surface 2 0:49(0) Bunker 2 0:24(0) Statue 2:17(-2) Archives 0:16(0) Streets 1:14(+1) Depot 0:26(0) Train 0:50(-11) Jungle 0:50(-2) Control 3:53(-6) Caverns 1:01(-1) Cradle 0:34(0) THe (x) how far behind/ahead of the console world record i am. For Cradle if it's like you say NFQ that it's 34.05 then 33 might actually be a possibility, but it'd require a diffrent strat or a more optimal Trev in beginning(i think mine was pretty bad, not sure though).

Why should any of those times be slower than the world record for unassisted play on the same difficulty level?

The levels with the + times, is that when you went into the pause menu or something?

For Cradle if it's like you say NFQ that it's 34.05 then 33 might actually be a possibility, but it'd require a diffrent strat or a more optimal Trev in beginning(i think mine was pretty bad, not sure though).

check for yourself with the in-level frame counter (009654A0). your cradle is 2043 frames. 2043/60=34.05. if you could shave off only 4 frames, it would be 33. you also had a very low decimal on streets even though you didn't use GL.

I didn't even know there was an in-level frame counter. Anyways, 4 frames should be possible, but i don't really know where it could be saved. Ok so explanations for the times that are slower than current world record: Dam - I have no idea how to manipulate the double gates. Runway - Because of the sky i chose not to throw grenades upwards which would've lead to more boosts and therefor wr. Streets - I tried but manipulation the grenade launcher dude was impossible, also i chose not to use lookdown in this run since so many poeple around here didn't like it, meaning more lag and worse time.

any chance of seeing a test avi of this? No mupen on mac :(

I didn't even know there was an in-level frame counter. Anyways, 4 frames should be possible, but i don't really know where it could be saved. Ok so explanations for the times that are slower than current world record: Dam - I have no idea how to manipulate the double gates. Runway - Because of the sky i chose not to throw grenades upwards which would've lead to more boosts and therefor wr. Streets - I tried but manipulation the grenade launcher dude was impossible, also i chose not to use lookdown in this run since so many poeple around here didn't like it, meaning more lag and worse time.

nfq mentioned in the 007 thread that he found the memory address for the in-game time. Also, the reasons you stated for why you got slower times equates to fail. TASes shouldn't be beaten by currently existing real time WR's. Dam: you should have PM'd RT on how he manipulated the double gate. Or, if the Speedrun WR did it, you could have did what they did. Don't make a TAS of a competitive game without knowing how to beat currently existing records. Runway: you made a conscious decision not to look up to use grenade boosts, simply because of the sky. Bad move. Streets: lookdown is not supposed to affect lag as much as on console, according to RT's tests. If you should have used lookdown to turn corners better or something, maybe that could have been done. Also, just because you can't luck manipulate the grenade launcher at first doesn't mean it's impossible. Does the WR speedurn do it? if so, it's possible. Try to vary what frame you end the cinematic, that helped luck manip for facility in the 00A run. No offense, but that post makes me regret voting yes.

Okay i gotta make a few things clear: RT didn't manipulate gate as far as i know, with his gate he probably wouldn't have gotten 53 either. For Runway, sure i could've used lookup, but as things seem there wouldn't have been an avi for several months. And finally streets, Lookdown does matter in TAS, i proved that in depot level. And the thing that made GL manipulation impossible in streets was that there was no real point where his position was determined. Even with same frame of ending cinema he was at diffrent positions, and neither i or simon sternis could find any real determination point. Besides, RT's run was not so optimal either, just look at frigate, it's ridiculous. If the 00 Agent console world records were as good as the Agent ones he wouldn't have tied the wr. Anyways, NFQ's posts got me thinking and i decided to try a redo of cradle, and i actually managed to get 33! So i'm not exactly sure what to do now, i suppose i'll have to replace the submitted m64 with the new one or something?

Anyways, 4 frames should be possible, but i don't really know where it could be saved.

complex 3D games have so many random factors that you can pretty easily save a couple of frames by doing random stuff. for example, your cradle was about a half second faster than jack's, even though you didn't do anything different. check it out: cradle 33 + save state

Watched it, and I'm sorry to say it's just not as entertaining as the 00Agent run. Without the extra objectives, it just seems like a lot of running, and since that running isn't optimized on several levels, I have to vote no. The runway has a lot of lookdown without damage boosting, while the dam doesn't use lookdown when it should be. I also don't think this game deserves more than one run, since the differences between the difficulties is not very great, and a run completing the game as fast as possible with all the objectives is inherently more entertaining than completing the game with 30% of the objectives. Hope this makes sense.

nfq, get aim and be online some :p

Just thought i'd add the improved ending in case someone wants to see it: http://download.yousendit.com/FEC100243ADF8955 So anyways, some last comments on the run: You say the run is unoptimized because it doesn't beat all wr's, But what if you add some perspective, the-elite has the most maxed wr's in any game(maybe mario kart or something is better), so overall even by not beating the world records the run is still way more optimized then let's say Diddy kong racing run which actually got published. Of course i realize you want all world records to be beaten but as i said, the difficulty in doing that is very relative to which game your TAS'ing and how the game works. Also on the "bad choice to not use lookup in runway" comment, i might as well add that i loose time by not using lookdown as well, I could've saved a second on at least 2 levels by doing that. So i think some guidelines to exactly what should be used and not should be written before someone attempts an new run(which i doubt will ever happen). So in the end, i think the run is as good as it will ever get. Even if someone attempts an improvement i'm still not so sure it will be faster than this one. So the questions remains, Does goldeneye need another speedrun and is this run entertaining enough? Oh and one last thing, I'd really love to see a list of the exact times on each level. Especially Depot which should be extremely low??

But what if you add some perspective, the-elite has the most maxed wr's in any game(maybe mario kart or something is better), so overall even by not beating the world records the run is still way more optimized then let's say Diddy kong racing run which actually got published. Of course i realize you want all world records to be beaten but as i said, the difficulty in doing that is very relative to which game your TAS'ing and how the game works.

Mario Kart 64 is I'd agree the other heavily tight WR community. However, do know a few things: -After 10 months, my MK64 run is still not done with the 2nd course. 3 months have been sunk into my Mischief Makers TAS, and I'm once again back at the beginning. Perfection takes time, especially when you choose a highly competed game like 007. -It IS stated in the rules that you should not be beaten by realtime speedrun records. There are very few exceptions, usually ones that have to do with timing differences or emulator problems or what have you. -WR's are beatable. One of the tightest fast lap for Mario Raceway in MK64, which is arguably the least improvable fast lap in the game, was beaten by Fried by 0.20". So it is possible. Granted, 007 only has a second timer and not microseconds, but you still should not end up with a time slower than a speedrun.

Oh and one last thing, I'd really love to see a list of the exact times on each level. Especially Depot which should be extremely low??

  54.8 dam
  43.6 facility
  23.3 runway
1:01.9 surface 1
  17.8 bunker 1
  58.3 silo
  22.6 frigate
  49.3 surface 2
  24.7 bunker 2
2:17.9 statue
  16.5 archives
1:14.0 streets
  26.2 depot
  50.6 train
  50.7 jungle
3:53.8 control
1:01.5 caverns
  34.0 cradle

17:12

i wrote down these when i watched the run, but i rounded up to the nearest tenth. you can check more correctly with the memory watching tool.