this looks like it's going to be much better than the current run. i don't get why the latter doesn't do it on crazy hard. one of the advantages of doing so is that i know you'll have to go through the entire last boss. anyway, i'm looking forward to it.
Joined: 12/26/2006
Posts: 256
Location: United States of America
Bob A wrote:
i don't get why the latter doesn't do it on crazy hard.
Danny reckoned that taking damage would save time during things like boss fights. Also, his version three run was going to take damage to save time at the end-of-level bonus screens.
It doesn't seem to me that taking damage will add any significant time savings to the in-level time anywhere. And I'm going to submit this run in the "ignores delays caused by bonus effects" category to address the latter issue.
I'm glad so many people are interested -- I feel bad that progress is so slow. : (
Edit: Come to think of it, I would enjoy Dan's feedback in all of this if he's still out there...
I'm still out here. However, I don't have a computer and home so I can only check the forums when it's slow at work (as it is right now). I can't check the movie either, but I can echo Ouzo's thoughts:
Ouzo wrote:
Glad someone is finally improving this movie.
Maybe someday when I get a computer (probably a month or so) i'll be able to provide feedback (assuming you aren't done by then) but until that time comes, i'll just be lurking randomly... :P
But seriously, good luck with this and I hope it turns out well.
Joined: 12/26/2006
Posts: 256
Location: United States of America
Dan_ wrote:
I'm still out here. However, I don't have a computer and home so I can only check the forums when it's slow at work (as it is right now).
I guess that's what stopped you from completing your version three run. : )
Since its a safe bet that you don't have a Genesis emulator on your computer at work, would uploading the WIPs as video files at Google help you to be able to see them from work? I'll still put the .gmv files on Microstorage for everyone else.
Dan_ wrote:
But seriously, good luck with this and I hope it turns out well.
Thank you! It does look like it will take more than a month, given my available time and the work of doing a good TAS in general.
Joined: 12/26/2006
Posts: 256
Location: United States of America
moozooh wrote:
Is there any progress yet, btw? I'm itching to see what you will come up with to preserve your speed everywhere. :)
Bunny hopping is the biggest trick to preserve speed. I understand that this is true in many games: when the character jumps, whatever speed that is then present is preserved. Some loss of speed is inevitable due to friction, but bunny hopping after a rocket boost preseves high speed long enough until the booster is charged again.
I was also able to use one of the diagonal vines in the first level to bring my speed above rocket boosting speed. That's just cool. And when I needed to stop to fight the walker machines (stopping from a high speed in this game is usually difficult because "stopping" is really just accelerating the other way), I found (quite by accident) a convienant glitch where colliding with the hill the right way immediately sets my horizontal velocity to zero -- right where I needed to stop. That's also cool.
Boosting diagonally has less horizontal speed than boosting horizontally (which is logical, but the other players didn't seem to notice that). Also, when Sparkster is in his "falling" animation after a rocket boost (where he flaps his arms wildly), horizontal velocity is cut drastically, even if boosting forward, so I try to avoid this at all costs.
You'll also see me bunny hopping in areas where I don't use the rocket booster to gain speed. This is because, when walking, the horizontal speed switches between two values every frame. By jumping, I can preserve the fastest speed the whole time. (This is similar to a trick used in the Super Mario World run.) This is also why the first thing I do in the first level is jump. While the time saved would be counted in frames instead of seconds, it's still time saved nonetheless. : )
Another thing that will be helpful is simply the NTSC timing of the (U) ROM. The rocket booster value increases by one for each frame that an attack button is held down until it reaches its maximum, 63. This rule applies to all versions of the game. Since 63 frames in NTSC is faster than 63 frames in PAL, this means that I can take less time charging the rocket booster. This will definitely save some time in the first castle level, where Sparkster needs to boost a lot, and probably in other places as well.
Since Rocket Knight Adventures has a lot of bosses, optimal boss fighting strategy is a big part of the game. There are two kinds of bosses: hit-based and health-based. The first boss at the end of the first stage is a hit-based boss, a fact of which Dan and the other players were evidently unaware. This boss always takes (I think) six hits to go down, and there is a short invulnerablity period after each hit. Constantly swiping at the boss is pointless, and can actually waste time if the hits don't land on the first possible frame.
The best way to deal a lot of damage very fast for a health-based boss (and other enemies, like the trucks and walkers of level one) is to swipe the enemy as close as possible. This causes the game to register multiple hits in one swipe. Furthermore, when a health-based enemy is taking damage, it is usually possible to position Sparkster physically inside the enemy's collision box without harm. From this position, it is possible to deal very much damage very quickly, and by using frame advance, the next blow can be delivered before the collision box is reactivated. This strategy causes some functions of the game to lag, but it's not a problem because it is the fastest way to defeat the enemy, and other things (like the booster not recharging one unit per frame) can be easily compensated for.
moozooh wrote:
Is there any progress yet, btw?
Well, let me put it to you this way: I'm taking my sweet sweet time working on this run.
One thing I've seen a lot here is this: a person completes a TAS, submits it, then, once that run is published, submits an improvement. Then, once that run is published, they submit another improvement, and so on and so forth. As an example: how many times has Aqfaq submitted improvements to his Genesis Gods run?
There's nothing wrong or bad about this in any way whatsoever. I just intend to do my run somewhat differently. I'm a perfectionist. If I see a small way that my run can be improved, or even if there's a strategy that might save time and I haven't thoroughly tested it, then my run's not finished in my eyes. I want to try everything before declaring the run "finished".
Take a perfectly good example: I knew that, when fighting the trucks, the second truck doesn't even appear until the debris from the exploded first truck have settled down, and this happens a lot quicker if the first truck explodes in front of the tree. My first strategy had be explode the first truck, immediately boost to the far right side of the level, and defeat the second truck even before it fully appeared on the screen. I would then diagonal-boost over the wall. This sounds like a very logical, optimal strategy.
Guess what? Scrolling the screen just makes the second truck take longer to appear in the foreground. So my revised strategy eventually became to wait for the second truck, jump and attack it on the earliest possible frame, then immediately boost and bunny hop to the right. I can't remember exactly, but this saved something like two or three seconds of time. Anything that can shave two to three seconds off of a twenty-five second segment is definitely worth doing, plus I do the cool trick of jumping over the wall that the player's not supposed to be able to jump over. So everything is better that way. : )
I wrote before in response to Ouzo, Bob A, and TNSe that I felt bad that progress is slow. I retract that statement. My progress is going to be as slow as it needs to be for my run to be the best that it can be. That way, if someone does come along and obsolete the run, I get to be seriously shocked and surprised. : )
But you don't have to worry. I am absolutely NOT going to give up on this run. So even if I don't post anything for a while, I'm still working on it.
I'll release my next WIP when I complete level two. Unless I get pestered enough to do so sooner. : )
On a sidenote to the perfectionism comment, Rocket Knight Adventures is pretty straightforward compared to many other games, so it seems like I've found the perfect game of which to do my first TAS.
Anyway, that's enough out of me for today. : )
As an example: how many times has Aqfaq submitted improvements to his Genesis Gods run?
Only two times, so it's not a good example. :D There were better ones if you look closely at the most oftenly improved movies.
Deep Loner wrote:
I'm a perfectionist. If I see a small way that my run can be improved, or even if there's a strategy that might save time and I haven't thoroughly tested it, then my run's not finished in my eyes. I want to try everything before declaring the run "finished".
I myself fully approve of it, although there's a small caveat: if you accidentally discover a new generic trick or another route, or something like that applicable in the earlier part of the game, you'd likely have to start over. Many players try to avoid that either by doing and submitting almost-optimal runs in order to try an improvement later, or doing test runs. If such disrepancy hasn't been avoided in time, it compromises the quality of the whole run and the more of a perfectionist you are, the more daunting it can really be having to redo most parts of the run, especially if there's a lot of luck manipulation involved. I suppose you're aware of that, just wanted to make sure you realize that and are motivated enough to redo everything from scratch if needs be. :)
Deep Loner wrote:
Take a perfectly good example: I knew that, when fighting the trucks, the second truck doesn't even appear until the debris from the exploded first truck have settled down, and this happens a lot quicker if the first truck explodes in front of the tree.
Well… you sure DO know how to impress. There are not many players here ready to show such a thoroughness and indepth knowledge even of their favorite games, and that quality is very commendable.
Deep Loner wrote:
I wrote before in response to Ouzo, Bob A, and TNSe that I felt bad that progress is slow. I retract that statement. My progress is going to be as slow as it needs to be for my run to be the best that it can be. That way, if someone does come along and obsolete the run, I get to be seriously shocked and surprised. : )
But you don't have to worry. I am absolutely NOT going to give up on this run. So even if I don't post anything for a while, I'm still working on it.
Of course, please take your time to do everything as you want it to. It's not anything uncommon to wait a couple of weeks/months for an excellent run. :D
I still think you're overly cautious with your run being improved. I mean, you seem to already know much more than anyone here about this game, and will certainly learn much more things in the progress. There's hardly anyone able (let alone willing) to outperform you with such an attitude.
Deep Loner wrote:
On a sidenote to the perfectionism comment, Rocket Knight Adventures is pretty straightforward compared to many other games, so it seems like I've found the perfect game of which to do my first TAS.
Well, good luck to you in your endeavors. And I'm really curious if you have any specific thing about the current SMB3 run that can be improved, since it wouldn't be an easy (or even possible) task.
Warp wrote:
Edit: I think I understand now: It's my avatar, isn't it? It makes me look angry.
Joined: 11/18/2006
Posts: 2426
Location: Back where I belong
For this being your first game, you've obviously thought through the process an incredible amount, and I commend you for your perfectionism. I hope I can match the amount of patience you have in my current and future runs:). I'm also looking forward to any sort of WIPs, because this was one of my favorite games as a kid, and I was flabbergasted by Dan's run, so I have no doubt yours will floor me. Good luck and keep trucking along!
Joined: 12/26/2006
Posts: 256
Location: United States of America
moozooh wrote:
Only two times, so it's not a good example. :D
Touché. But you know what I mean. ; )
moozooh wrote:
Well… you sure DO know how to impress. There are not many players here ready to show such a thoroughness and indepth knowledge even of their favorite games, and that quality is very commendable.
I actually feel inferior to a lot of players on the site. I see a lot of players reverse engineering different things in the games by examining the ASM code. Bisqwit has a program that can calculate input based on the physics of Mega Man. Acmlm and FractalFusion were able to write a program to calculate the shortest games of Monopoly. I know there are more examples that I can't think of right now. ASM looks like gibberish to me; in fact, I barely know enough C++ to do anything useful.
Most of what I know about the game comes from just observing things in test runs and using the Gens memory search feature.
moozooh wrote:
I still think you're overly cautious with your run being improved. I mean, you seem to already know much more than anyone here about this game, and will certainly learn much more things in the progress. There's hardly anyone able (let alone willing) to outperform you with such an attitude.
I'm not against having my run improved (except under the circumstances we talked about before, but I'm not worried about that anymore) if someone else is able to do that. For example, I don't know of any super glitches in this game to exploit. If someone like Phil or nitsuja found a way to break the game somehow, my run would probably be toast. I just want my run to be as good as I know how to make it, so I can still be proud of my work even if it does get improved.
I used the Gods run as an (admittedly bad) example because the newest submission improves by only a little more than a second, and it's hard (for me anyway) to see the difference. Of course, by principle it should deserve to be published as an improvement, but this is the kind of optimization that I would want to make sure is in the run before I submit it.
moozooh wrote:
And I'm really curious if you have any specific thing about the current SMB3 run that can be improved, since it wouldn't be an easy (or even possible) task.
I don't. Partly it was a joke about my real-time playing skills. But I do intend to try, even if it's not easy or even possible, just because I want to sink my teeth into A Serious Big-Time TAS™ after I'm finished with this run. I know that I still have a lot to learn.
mmbossman wrote:
I'm also looking forward to any sort of WIPs, because this was one of my favorite games as a kid, and I was flabbergasted by Dan's run, so I have no doubt yours will floor me.
Joined: 11/18/2006
Posts: 2426
Location: Back where I belong
Deep Loner wrote:
Were you flabbergasted in a good or a bad way?
Oh definitely in a good way, but then again I was really new to the site and was still in awe of how someone could make a movie of a game and go as fast as he did. I haven't watched it in a couple months so I'm sure I could pick out some flaws now since I'm a little more experienced, but from what I remember, it was fun, fast paced, and entertaining. I'm sure yours will be no different.
Oh definitely in a good way, but then again I was really new to the site and was still in awe of how someone could make a movie of a game and go as fast as he did. I haven't watched it in a couple months so I'm sure I could pick out some flaws now since I'm a little more experienced, but from what I remember, it was fun, fast paced, and entertaining. I'm sure yours will be no different.
Really? I remember downloading that run when I first got into the site, and being very disappointed by how slow it seemed. Admittedly the first movie I saw was SprintGod's (now obsolete) S3K run, but Deep Loner's WIP thus far is living up to the expectations of speed before I watched the first run.
How fleeting are all human passions compared with the massive continuity of ducks.
Joined: 11/18/2006
Posts: 2426
Location: Back where I belong
Yeah, like I said, I was new to the TAS world, so probably any movie would have impressed me. You've inspired me to go back and watch it again for another review:)
EDIT: Ok, some thoughts after watching the first couple of levels again- definitely not at star struck, although I still think it's decently run, especially having played in on the console. However:
Level 1- Obviously improvable. Too much walking, not enough boosting, and some backtracking, which seems like it should be a sin in this game. And please, please, PLEASE do something more entertaining with the autoscrolling sections. Don't just hug the side of the screen waiting to move forward.
Level 2- I'm sure there is more entertainment to be had on the minecarts, especially during the overhead bomber phase. And I'm sure you can take the snake boss out in just two passes, looks like there was a lot of wasted time there.
Anyway, you get the picture. The biggest thing I noticed throughout the first couple levels is the poor setup for subsequent moves, which involved some waiting and some backwards motion, which makes me cringe just a tiny bit. And I'm sure that all the time added up could shave a significant amount off the clock, possibly as much as a few minutes through the whole run just from better movement planning.
Thanks Upthorn, for calling me on that.
Joined: 12/26/2006
Posts: 256
Location: United States of America
mmbossman wrote:
Level 1- Obviously improvable. Too much walking, not enough boosting, and some backtracking, which seems like it should be a sin in this game.
Level one will be improved substantially.
mmbossman wrote:
And please, please, PLEASE do something more entertaining with the autoscrolling sections. Don't just hug the side of the screen waiting to move forward.
I understand what you're saying. This game has a lot of autoscrolling portions, so for entertainment, I plan to work in the theme of Sparkster dancing to the music, especially on that later level where the player rides a platform and has to navigate up or down... that has some nice music to dance to.
I've noticed that dancing to music is really only effective with tool assistance.
mmbossman wrote:
Level 2- I'm sure there is more entertainment to be had on the minecarts, especially during the overhead bomber phase.
As a nice programming touch, the minecart can kill enemies even when Sparkster is high in the air. I'll probably show this off a few times. I might also dodge the enemies' attacks while dancing to the music -- this would be entertaining, and the enemies automatically die after a certain amount of time has passed anyway.
I'm not sure about the overhead bombing scene, because I don't know if it's possible to survive without killing the enemies. But I'll worry about that when I get there.
mmbossman wrote:
And I'm sure you can take the snake boss out in just two passes, looks like there was a lot of wasted time there.
Dan's unfinished version three run does this, so probably.
mmbossman wrote:
And I'm sure that all the time added up could shave a significant amount off the clock, possibly as much as a few minutes through the whole run just from better movement planning.
My goal for this movie is to complete the game in under thirty minutes, which Dan said would be very impressive. : )
I'm so happy to see so much interest in this run!
Joined: 12/26/2006
Posts: 256
Location: United States of America
Wow, people really are interested in this run!
I'm not going to release a WIP until level 2 is done, like I said before, just because too many WIPs become frivolous.
As for progress... well... be patient. My standards are very high. : )
If there are any suggestions are things you'd like to see in the run, feel free to mention those here.
Joined: 12/26/2006
Posts: 256
Location: United States of America
upthorn wrote:
How about a wip that completes level 1, and then don't do another one until you're through level 3?
Earlier in this thread I wrote:
I'll release my next WIP when I complete level two. Unless I get pestered enough to do so sooner. : )
OK. Sounds like a plan!
I will post one more thought while I'm here. I kind of see tool-assisted speedrunning as being kind of in its infancy (I know that's surprising to people who have been here a long time). For example, computer software development arose from just a bunch of guys who knew how to program to having standardized practices and procedures to the point that it can now be correctly called an engineering dicipline. From by brief experience trying (and ultimately failing) to make a level edit for Super Mario Brothers 3, the ROM hacking community hasn't quite reached that point yet; it's mostly a "find out for yourself" kind of thing. Perhaps the warez scene ("The Scene" -- thanks moozoh) has. I'm using my time with Rocket Knight Adventures to try to learn more about the actual process of making a tool-assisted speedrun, and how this can be abstracted into things like strategies, variations, sub- and micro-variations, and so on. Maybe someday I'll write a paper on it. : )
And anomolies. I encountered one sub-variation within a variation of a strategy that was a whole quarter-second (out of only a few seconds of gameplay) faster than all the other sub-variations (which were pretty consistant) for no apparant reason. I don't think this would have been discovered without a meticulous and thorough process for testing the variations. That's why I had mentioned that I would TAS Super Mario Brothers 3 after this. I want to learn more about the process of TASing -- and see how much (if at all) such a unified process could improve on an already tightly-optimized run.
But yes, once I solve all the variations for level one, I'll post it here.
I can't keep myself from mentioning how flattering it is to have some of the most well-respected and veteran TASVideos members so interested in my movie. It's not like just some guy named ILikeAnime4431. : )
Edit: also, one more thing that would help is if anybody knows any other runs I can compare mine to, for the purpose of developing metrics and stealing useful strategies and that kind of thing. : )
So far, this is what I have:
Joined: 11/18/2006
Posts: 2426
Location: Back where I belong
Very well thought out mini essay, I enjoyed reading it. I also like to know that I have tried and exhausted most possibilities with the games I run, and for me one of the more satisfying parts of what we do is finding those little programming errors or subvariations. For instance in my first level of my Turok run there is a large wall to climb. Most of the time, it takes around 10 seconds. But if you hit it juuuuust right, it takes about half that time. I don't know why it works, but when I found out about it I actually got excited.
It's because of the small variations available that I've entertained going through some of the older runs that have an abnormally low rerecord count and picking them up just to see if I can find anything strange. Plus, I very much like to have a template to work with too, I love to be able to count frames and see that my current run is x frames ahead at x % of game completion. So if I had any other runs to get you for your work, I would:)
I hope this continues to go well and good luck breaking this game down as methodically as possible!
Joined: 12/26/2006
Posts: 256
Location: United States of America
I do have to wonder how feasible this approach would be on something like, say, a four-hour RPG game. But maybe I'll try it out someday! ; )
Thanks for all the encouragement, guys!
I kind of see tool-assisted speedrunning as being kind of in its infancy.
I think I see what are you talking about. Actually, I wouldn't call it infancy anymore. The TAS scene was indeed in its infancy when all the runs were produced only by means of slowdown and rerecording (all the 2004 was like that). Guys like Phil & Genisto ruled supreme at that point, surprisinly being able to make very optimized runs without even such thing as frame counter.
Now that most TASers actively use memory watch, and are slowly moving on to using bots and ROM disassembling when/where possible, I don't think much more could be possibly done. So I guess that TASing as a discipline is already mature — it's just that not all the runners/runs are.
Deep Loner wrote:
also, one more thing that would help is if anybody knows any other runs I can compare mine to, for the purpose of developing metrics and stealing useful strategies and that kind of thing.
Nitsuja's Sparkster run is all I can think about except those you mentioned.
Warp wrote:
Edit: I think I understand now: It's my avatar, isn't it? It makes me look angry.
