GTA III 100%

2010-01-05 06:57:57 pm

Thanks mate. Smiley

And a guess for the final time, hmm.. I'm gonna go on the safe side here and say somewhere between 6:15 and 6:30. That's very rough though. I hope to be faster but it could very well be slower. It's lots of luck and randomness after all. But I think it'll turn out to be around there.

Warepire

2010-01-06 04:35:36 am

Heavy Metal Powered

Awesome segment, the landing in the first Unique jump was a little unlucky... but that doesn't really matter.

ridd3r.

2010-01-06 06:20:19 am

we have lift off

Really good segment, at half an hour long with lots of stuff to do I can see why it took you so long. As always your efficiency looks excellent and I couldn't see any major mistakes, slight misses with the sniper rampage but that's it really.

The only thing I noted was your decision to get 7 checkpoints then go off and do loads of out of the way stuff. Does that save a lot of time compared with finishing the race quicker (as you are obviously a lot faster than the computer A.I) and getting some of that other stuff when you have missions in those areas? Also when does the race finish, when you have the majority of the checkpoints or when all 15 have been obtained?

LORDZAPPY

2010-01-06 07:54:29 am

I've watched this segment and now I am suspicious of silamranza. I dunno I think he might be cheating because I saw some strange moves by him.

Neverdown

Edit history:

Neverdown: 2010-01-06 08:39:13 am

2010-01-06 08:09:12 am

I love YaBB 1G - SP1!

Don't feed the troll. Well actually do because he is funny in his own, special way.

edit: I just watched segment 9. It looks very great. Good job keeping up that quality of gameplay in quitte a long segment.

Silmaranza

2010-01-06 09:44:53 am

Thanks guys! Heh, I'm just laughing at Lord Zappy now, good stuff.

Quote from ridd3r.:

The only thing I noted was your decision to get 7 checkpoints then go off and do loads of out of the way stuff. Does that save a lot of time compared with finishing the race quicker (as you are obviously a lot faster than the computer A.I) and getting some of that other stuff when you have missions in those areas? Also when does the race finish, when you have the majority of the checkpoints or when all 15 have been obtained?

This very thing went through a lot of testing and experimenting. The answer is, this is the most optimal way. It's comparable to 'The Fuzz Ball' where I only pick up the minimum amount of girls. The idea is, I have to do all that stuff some time anyway, so the time it takes me to do those things is lost either way. I was surprised that CannibalK9 goes for 10 checkpoints in his run. The time I don't have to spend on the rest of the mission is pure gain, so this is the best way to do it. I figured this out pretty quickly, but the most testing went into finding out what could be done in that time and in what order. This route includes all very out of the way and tricky stuff that didn't really fit anywhere else and also usually comes out somewhat at the same time as the others finish the mission. The mission ends when all 15 checkpoints are obtained, and you win if you have the majority (I got 6 on a couple of attempts but still won).

And yeah I know about the one or two failed sniper shots and the ugly first unique jump, but expecting perfection is insane. I find this very acceptable. Smiley

Thanks guys! Heh, I'm just laughing at Lord Zappy now, good stuff.

[quote="ridd3r."]The only thing I noted was your decision to get 7 checkpoints then go off and do loads of out of the way stuff. Does that save a lot of time compared with finishing the race quicker (as you are obviously a lot faster than the computer A.I) and getting some of that other stuff when you have missions in those areas? Also when does the race finish, when you have the majority of the checkpoints or when all 15 have been obtained?
[/quote]

This very thing went through a lot of testing and experimenting. The answer is, this is the most optimal way. It's comparable to 'The Fuzz Ball' where I only pick up the minimum amount of girls. The idea is, I have to do all that stuff some time anyway, so the time it takes me to do those things is lost either way. I was surprised that CannibalK9 goes for 10 checkpoints in his run. The time I don't have to spend on the rest of the mission is pure gain, so this is the best way to do it. I figured this out pretty quickly, but the most testing went into finding out what could be done in that time and in what order. This route includes all very out of the way and tricky stuff that didn't really fit anywhere else and also usually comes out somewhat at the same time as the others finish the mission. The mission ends when all 15 checkpoints are obtained, and you win if you have the majority (I got 6 on a couple of attempts but still won).

And yeah I know about the one or two failed sniper shots and the ugly first unique jump, but expecting perfection is insane. I find this very acceptable. :)

ridd3r.

2010-01-06 09:54:17 am

we have lift off

My point was more that if you had stayed in the race until the end it would have finished quicker as you will be much faster than the computer at getting to all the checkpoints. So, the time you spend doing other stuff is not pure gain as maybe the race would have been over 30-60seconds sooner if you had stayed in the race. It's a question of finding the balance.

Wuv

2010-01-06 10:44:44 am

Quote from LORDZAPPY:

I've watched this segment and now I am suspicious of silamranza. I dunno I think he might be cheating because I saw some strange moves by him.

AHAHAHAHAHAHA XD

Anyways good job so far, looks like the world record is pretty much going to be yours.

Also r1dd3r, don't forget he ended the mission at Asuka's doorstep. How long does it take to drive from checkpoint 15 to there? At the moment, I don't think finishing the race manually is quicker.

Silmaranza

2010-01-06 11:00:03 am

Quote from ridd3r.:

My point was more that if you had stayed in the race until the end it would have finished quicker as you will be much faster than the computer at getting to all the checkpoints. So, the time you spend doing other stuff is not pure gain as maybe the race would have been over 30-60seconds sooner if you had stayed in the race. It's a question of finding the balance.

The race would be done sooner, but it would also cost me time to accomplish that, time that I now spend doing sidestuff that I otherwise would have to do some time else. Let's say that in my strategy the part of the race I do is 1 minute, the part they do is 3 minutes and the sidestuff also takes 3 minutes. That means the mission and all the sidestuff is accomplished in 4 minutes. Now let's say I do the entire race myself, finish it in 2 minutes but don't do any sidestuff, which will still take 3 minutes obviously. I have to do that some time else now, so in the end it takes me 5 minutes, hence I lose a minute. The time I spend doing other stuff is not pure gain, because that'll have to happen some time anyway, but the time I don't have to spend on the rest of the race is. There is no balance, because any time it would take me to complete the entire course will always be longer than getting just 7 checkpoints. Getting any more checkpoints is a complete waste of time. I see your point and the exact same thing crossed my mind when planning this, but I don't know how to explain it any better that this is indeed optimal.

[quote="ridd3r."]
My point was more that if you had stayed in the race until the end it would have finished quicker as you will be much faster than the computer at getting to all the checkpoints. So, the time you spend doing other stuff is not pure gain as maybe the race would have been over 30-60seconds sooner if you had stayed in the race. It's a question of finding the balance.
[/quote]

The race would be done sooner, but it would also cost me time to accomplish that, time that I now spend doing sidestuff that I otherwise would have to do some time else. Let's say that in my strategy the part of the race I do is 1 minute, the part they do is 3 minutes and the sidestuff also takes 3 minutes. That means the mission and all the sidestuff is accomplished in 4 minutes. Now let's say I do the entire race myself, finish it in 2 minutes but don't do any sidestuff, which will still take 3 minutes obviously. I have to do that some time else now, so in the end it takes me 5 minutes, hence I lose a minute. The time I spend doing other stuff is not pure gain, because that'll have to happen some time anyway, but the time I don't have to spend on the rest of the race is. There is no balance, because any time it would take me to complete the entire course will always be longer than getting just 7 checkpoints. Getting any more checkpoints is a complete waste of time. I see your point and the exact same thing crossed my mind when planning this, but I don't know how to explain it any better that this is indeed optimal.

ExplodingCabbage

2010-01-06 11:25:41 am

Sil, ridd3r's point is that the cost in time of getting the sidestuff that you get during the race might well be less if you could get it while passing through on later missions, instead of actively driving around for it. So there is - in theory, at least - a balance. For instance, if one of the hidden packages you get during Bling Bling Scramble takes 30 seconds to drive to in your run, but you park 5 seconds away from it in a mission later on and could grab it then anyway, then it could work out faster to not get that package during Bling Bling Scramble, and instead get more checkpoints to finish the race early and save getting the package to the later mission when it is more efficient to do so.

Whether this thinking could actually save time in practice depends upon whether any of the stuff you get during Bling Bling Scramble is sufficiently close to your route on any later missions.

Silmaranza

2010-01-06 11:43:09 am

Ah yes, you're very right there EC. In theory, there is a balance indeed. But all of the stuff I do are actually things that didn't fit in anywhere else. Like I said, lots of testing went into figuring out what to actually do with this time, and these are all things I never really come close to on other missions, or at least not in a convinient situation. For example I could get the package at the police station during 'Kanbu Bust-out', but it would be a pain with 2 stars. The unique jumps need a fast car and I want them at the beginning of a segment, this is the only time those two conditions can be met. The bomb needs to be placed now for obvious reasons and the package at the end of the dock is completely out of the way of everything. CannibalK9 gets this package on his first trip to Staunton Island I think, which takes much more time.

Ah well, I guess in the end you can never know what the actual most optimal possible route is for a game like this. Besides, that would probably be a near impossible one.

Ah yes, you're very right there EC. In theory, there is a balance indeed. But all of the stuff I do are actually things that didn't fit in anywhere else. Like I said, lots of testing went into figuring out what to actually do with this time, and these are all things I never really come close to on other missions, or at least not in a convinient situation. For example I could get the package at the police station during 'Kanbu Bust-out', but it would be a pain with 2 stars. The unique jumps need a fast car and I want them at the beginning of a segment, this is the only time those two conditions can be met. The bomb needs to be placed now for obvious reasons and the package at the end of the dock is completely out of the way of everything. CannibalK9 gets this package on his first trip to Staunton Island I think, which takes much more time.

Ah well, I guess in the end you can never know what the actual most optimal possible route is for a game like this. Besides, that would probably be a near impossible one.

Neverdown

2010-01-06 02:03:36 pm

I love YaBB 1G - SP1!

It would be funny to have a computer program figure out what the most optimal route is.

ZenicReverie

Edit history:

ZenicReverie: 2010-01-06 02:26:16 pm

2010-01-06 02:16:56 pm

Waiting hurts my soul...

Quote from Neverdown:

It would be funny to have a computer program figure out what the most optimal route is.

You'd have to collect the data by hand unless you meant for the program to figure that out too, but then it'd just be a matter of make points (nodes) for each thing you needed to do, and weight the path from one point to another with how much time it would take. It'd be like a traveling salesman problem. I think most people go through this by hand (or at least mentally) when they're make a speedrun.

The program would probably be able to just spit back the optimal route though. Of course, this is dependent on how reliable your original data (timing) was. If you weren't optimal in all test times, then the route might be flawed.

ExplodingCabbage

2010-01-06 03:11:38 pm

I think that just programming an algorithm to solve the problem in the first place would be as big a problem as the data gathering, Zenic, since it's vastly more complicated (in practice, not just in theory) than the travelling salesman problem.

Silmaranza

2010-01-06 03:56:58 pm

It wouldn't be possible. There isn't even an algorithm for the traveling salesman problem in the first place. It is kind of similar though.

ZenicReverie

2010-01-06 05:03:42 pm

Waiting hurts my soul...

Quote from ExplodingCabbage:

I think that just programming an algorithm to solve the problem in the first place would be as big a problem as the data gathering, Zenic, since it's vastly more complicated (in practice, not just in theory) than the travelling salesman problem.

I wouldn't say it's vastly more complicated. Basically each location you need to go to would be a node, and the path that takes you to the next node could be timed and weighted for a solution. Side missions/objectives/objects/collectibles would also be nodes. You'd only take those into account though for nodes that are close to another node. I agree though, doing it by hand would be faster than trying to figure out an algorithm for it.

I do stand by what I said about writing a program to play games optimally (gathering data), would be more difficult than writing a shortest path algorithm. I'm talking about for a single game, but it'd probably stand up when you generalize both of them.

Quote from Silmaranza:

It wouldn't be possible. There isn't even an algorithm for the traveling salesman problem in the first place. It is kind of similar though.

There's always the brute force style of checking all possibilities, but that seems to make the algorithm useless. For speedrunning though, taking only one weighted factor into account (time) makes it a little more possible. You could always go shortest point to shortest point, which could then be checked by hand.

[quote="ExplodingCabbage"]
I think that just programming an algorithm to solve the problem in the first place would be as big a problem as the data gathering, Zenic, since it's vastly more complicated (in practice, not just in theory) than the travelling salesman problem.
[/quote]
I wouldn't say it's vastly more complicated. Basically each location you need to go to would be a node, and the path that takes you to the next node could be timed and weighted for a solution. Side missions/objectives/objects/collectibles would also be nodes. You'd only take those into account though for nodes that are close to another node. I agree though, doing it by hand would be faster than trying to figure out an algorithm for it.

I do stand by what I said about writing a program to play games optimally (gathering data), would be more difficult than writing a shortest path algorithm. I'm talking about for a single game, but it'd probably stand up when you generalize both of them.

[quote="Silmaranza"]
It wouldn't be possible. There isn't even an algorithm for the traveling salesman problem in the first place. It is kind of similar though.
[/quote]
There's always the brute force style of checking all possibilities, but that seems to make the algorithm useless. For speedrunning though, taking only one weighted factor into account (time) makes it a little more possible. You could always go shortest point to shortest point, which could then be checked by hand.

ExplodingCabbage

2010-01-06 05:22:26 pm

Quote from ZenicReverie:

For speedrunning though, taking only one weighted factor into account (time) makes it a little more possible. You could always go shortest point to shortest point, which could then be checked by hand.

I have no idea what you mean by this, but at any rate you seem to be missing the fundamental point that in GTA3 mission objectives have to be travelled to in order, you can only have one mission or rampage active at once, and missions have to be unlocked in order, and that in practice the route is dictated by these points, not by some ideal of just taking the most optimal route through all the points on the map you ever have to visit in the run.

Quote:

I do stand by what I said about writing a program to play games optimally (gathering data), would be more difficult than writing a shortest path algorithm

Well, okay, but they're both practically impossible since neither problem can really be approached by any method other than brute force and there are too many possibilities for brute force to be practical.

[quote="ZenicReverie"]For speedrunning though, taking only one weighted factor into account (time) makes it a little more possible. You could always go shortest point to shortest point, which could then be checked by hand.
[/quote]

I have no idea what you mean by this, but at any rate you seem to be missing the fundamental point that in GTA3 mission objectives have to be travelled to in order, you can only have one mission or rampage active at once, and missions have to be unlocked in order, and that in practice the route is dictated by these points, not by some ideal of just taking the most optimal route through all the points on the map you ever have to visit in the run.

[quote]I do stand by what I said about writing a program to play games optimally (gathering data), would be more difficult than writing a shortest path algorithm[/quote]

Well, okay, but they're both practically impossible since neither problem can really be approached by any method other than brute force and there are too many possibilities for brute force to be practical.

ZenicReverie

2010-01-06 05:43:35 pm

Waiting hurts my soul...

Quote from ExplodingCabbage:

Quote from ZenicReverie:

For speedrunning though, taking only one weighted factor into account (time) makes it a little more possible. You could always go shortest point to shortest point, which could then be checked by hand.

I have no idea what you mean by this, but at any rate you seem to be missing the fundamental point that in GTA3 mission objectives have to be travelled to in order, you can only have one mission or rampage active at once, and missions have to be unlocked in order, and that in practice the route is dictated by these points, not by some ideal of just taking the most optimal route through all the points on the map you ever have to visit in the run.

I don't really know the game, but it seems like it'd be doable to list the game out as points you have to hit (sometimes in order, other times as modifiers to the weighted edges), and then just figure out what the best path is by minimizing the weight (time).

So, a node would be start of mission, end of mission, start of rampage, end of rampage, which need to be hit as start to end, but the next start could be any other mission available. Edges would be the time from start to end, but can be modified by adding intermediate nodes such as collectables, trick locations, whatever else, that can be done during any mission/rampage. I guess things get complicated when you can't start another mission until the current one ends and the ending of the mission is based on time rather than location or actions, but it's not impossible to map out.

Quote from ExplodingCabbage:

Quote from ZenicReverie:

I do stand by what I said about writing a program to play games optimally (gathering data), would be more difficult than writing a shortest path algorithm

Well, okay, but they're both practically impossible since neither problem can really be approached by any method other than brute force and there are too many possibilities for brute force to be practical.

It really hasn't been proven either way (possible/impossible). If someone could prove it then they'd be a millionaire.

[quote="ExplodingCabbage"]
[quote="ZenicReverie"]For speedrunning though, taking only one weighted factor into account (time) makes it a little more possible. You could always go shortest point to shortest point, which could then be checked by hand.
[/quote]

I have no idea what you mean by this, but at any rate you seem to be missing the fundamental point that in GTA3 mission objectives have to be travelled to in order, you can only have one mission or rampage active at once, and missions have to be unlocked in order, and that in practice the route is dictated by these points, not by some ideal of just taking the most optimal route through all the points on the map you ever have to visit in the run.
[/quote]

I don't really know the game, but it seems like it'd be doable to list the game out as points you have to hit (sometimes in order, other times as modifiers to the weighted edges), and then just figure out what the best path is by minimizing the weight (time).

So, a node would be start of mission, end of mission, start of rampage, end of rampage, which need to be hit as start to end, but the next start could be any other mission available. Edges would be the time from start to end, but can be modified by adding intermediate nodes such as collectables, trick locations, whatever else, that can be done during any mission/rampage. I guess things get complicated when you can't start another mission until the current one ends and the ending of the mission is based on time rather than location or actions, but it's not impossible to map out.

[quote="ExplodingCabbage"]
[quote="ZenicReverie"]
I do stand by what I said about writing a program to play games optimally (gathering data), would be more difficult than writing a shortest path algorithm[/quote]

Well, okay, but they're both practically impossible since neither problem can really be approached by any method other than brute force and there are too many possibilities for brute force to be practical.
[/quote]
It really hasn't been proven either way (possible/impossible). If someone could prove it then they'd be a [url=http://en.wikipedia.org/wiki/P_versus_NP]millionaire[/url].

Neverdown

2010-01-07 08:21:23 am

I love YaBB 1G - SP1!

Let's get back on-topic guys.

Paused

2010-01-08 12:05:37 pm

Quote from Silmaranza:

Segment 9 is finally done:

http://video.yahoo.com/watch/6719393/17452089

Great segment. Entertaining with several points where you ask, 'wait, whats he doing?' before seeing what you do next.

Well worth the wait. _{(Disclaimer:So long as we do not have to wait that long again..?)}

How are the rest of the remaing segments compared to this one, in terms of difficulty?

Silmaranza

2010-01-08 05:00:58 pm

Quote from Paused:

Great segment. Entertaining with several points where you ask, 'wait, whats he doing?' before seeing what you do next.

Thanks. Care to mention which points exactly? I like hearing feedback on routes and strategies. Smiley

Quote from Paused:

Well worth the wait. _{(Disclaimer:So long as we do not have to wait that long again..?)}

How are the rest of the remaing segments compared to this one, in terms of difficulty?

The segment I'm doing now isn't very hard so that shouldn't take too long. The one after that will be tough though. It features 'Evidence Dash' and Staunton Vigilante. That one and one further ahead with 'Bait', and a couple of challenging jumps and rampages, are the only two that I'd say are somewhat comparable to this last one in terms of difficulty. Although difficulty isn't really the right word. Nothing in this game is really difficult as in requiring skill, there's just a lot of things that require a shitload of luck.

Silmaranza

2010-03-16 03:39:22 am

*blows dust off thread*

Hello! Yes I'm still alive and going, be it somewhat slow. Sorry for the long wait again, but here's segment 10:

http://vimeo.com/10200353

This part used to be pretty easy until I found out that the unique jump from the carpark CAN be done after 'Multistorey Mayhem'. The jump has to be extremely accurate though, and I couldn't find a 100% reliable method to pull it off. About 90% of the attempts ended there.

After that there's nothing much to comment on really. Most strategies should be familiar if you've seen CannibalK9's run. The BF Injection spawns from 18:00 till 00:00 so it wouldn't have made a difference if I got to that point any faster because I'd have to wait for it. That was the last time restricted thing though I think for the rest of the run.

Smackdown is hard in a way that the spawnpoints of the dealers are completely random, so I have to decide on a route during the mission, and the BF Injection is hard to control and not very durable.

I'll try to get the next segment out a bit sooner!

*blows dust off thread*

Hello! Yes I'm still alive and going, be it somewhat slow. Sorry for the long wait again, but here's segment 10:

http://vimeo.com/10200353

This part used to be pretty easy until I found out that the unique jump from the carpark CAN be done after 'Multistorey Mayhem'. The jump has to be extremely accurate though, and I couldn't find a 100% reliable method to pull it off. About 90% of the attempts ended there.

After that there's nothing much to comment on really. Most strategies should be familiar if you've seen CannibalK9's run. The BF Injection spawns from 18:00 till 00:00 so it wouldn't have made a difference if I got to that point any faster because I'd have to wait for it. That was the last time restricted thing though I think for the rest of the run.

Smackdown is hard in a way that the spawnpoints of the dealers are completely random, so I have to decide on a route during the mission, and the BF Injection is hard to control and not very durable.

I'll try to get the next segment out a bit sooner!

Warepire

2010-03-16 08:41:14 am

Heavy Metal Powered

Great segment. A little unlucky with the fog during Smackdown though Undecided

Ghoti

2010-03-16 06:11:20 pm

Currently occupying the grey area.

I think you should know that Vimeo recently announced that they will no longer be hosting gameplay videos, hence the video has been taken down. You'll need to find a new host.

Silmaranza

2010-03-17 02:50:29 am

Oh great. >_> Alright then, does this one work?

http://sevenload.com/videos/3VqMjSr-GTA-III-100-speedrun-segment-10