Yay! I'm glad you tried out some new spindash spots and the upper path on EH1, excellent redo so far. Also, I'm glad you discovered the "Steganos hump hopping" on CP1, awesome improvement.
I do have a nitpick on EH2. When you take the vertical spring, you should press left for a little while, so that you can start moving to the right before you actually clear the wall that bounds you on the right.
Thanks greatly for your help and analysis, Truncated. I just got 0"24 with a frame to spare. This probably doesn't mean I will finish the whole video any time soon, but the first level will be awesome when I finish the whole run some time in 2012!
Link to the new first level: http://www.geocities.com/xebranick/Sonic1Act1byxebra.zip
If you are working on another movie, take all the advice people are giving you to heart. This site is by the community, for the community. If the community isn't happy with your product, then you have no one to sell it to.
Make it your top priority, on every level, to beat by a significant margin the times listed at the Sonic Center. They are real. Having done speed runs for a few of the S2 levels, all of the S1 levels, and all of S3 competition mode, I know how the games work, I know human limitations, and I know how rerecording can be exploited to circumvent these limitations. In almost every case you should be executing impossible maneuvers to improve upon the human runs.
- Queitust, I don't believe your conjecture about CP2 is true. Look for a better path.
- Kudos on the CP1 time, I always knew the 0"22 I acheived on a console was beatable.
- Shoot for 0"18 on EH1.
- Chao got 0"36 on EH2.
- Shoot for at most 0"48 on DE. I've done 0"49 on a console and know it can be improved.
Unfortunately those are the only levels I'm familiar with, so that's the only advice I can give.
I for one would like to see tracks 1 through 5 done once, as well as a demonstration video on a homemade track showing those "orbital speeds" and interesting glitches you talked about.
In vid 1 you jump trough the box 8 frames earlier than in vid 2. Any particular reason for this? I see no immediate benefit.
No reason.
Jumping over the stone and up the ledge later in vid 2 gains almost one frame in air. (Why?)
I believe it is because the ground is not level, and jumping from different spots launches you at different speeds.
Looking back at your older video (acts 1-10), it has the same behaviour as vid 1 in and when exiting the pipes. Whatever you did there, do it again.
I never bothered to optimize v2 after exiting the pipes because I couldn't find a path to exit the S-pipes appropriately.
I really appreciate your detailed analysis, and you've confirmed my fears that there's probably not much to be done. One last thing I mean to try is approaching the S-pipes only one frame ahead. Perhaps it will result in more bizarre behavior that will catapult me ahead of even v1!
If that doesn't work ... well, I'll give up. 4-5 frames from 24 seconds is a huge improvement over my acts 1-10 video, at least. I should probably go through the other acts and optimize them in a similar manner, but just the thought makes me shudder.
I finally fixed my comp! A month or two late I know ...
Anyways, I've uploaded a zip file with two GHZ1 runs: http://www.geocities.com/xebranick/GHZ1WIP.zip . (You may need to copy the link to a new window.)
Version 1 of the run is my fastest completion of GHZ1 so far, it's 4 (or 5) frames away from 0:00:24 (depending on whether the clock increments on the frame you finish the level.)
Version 2 of the run is two frames faster upon entering the pipes (after the loop) due to a single frame being pared from two different maneuvers earlier in the level. Mystery of mysteries, this run is slower upon exiting the pipes.
I think the goal of any collaborators (if anyone even wants to work on this movie with me =/ ) should be to figure out what in the heck is going on inside the pipes, and whether there are any more time saving maneuvers to be had.
P.S. Lezard, where are you?
It's undoubtedly safe to assume that overclocking a console would have unpredictable side effects. People that claim merely reducing lag couldn't possibly have an effect on gameplay need to recall certain jumps in Q3 that can only be made at ~75 fps or better ...
Furthermore, if LocalH is correct in that overclocking a NES changes the rate at which sound is played, then it's probably not desirable for making entertaining movies. The allure of tool assisted superplay movies is that they are familiar, yet amazing. SMB with music played a quarter octave too high isn't familiar, it's retarded.
No, you had it right the first time. An observer orbiting the Earth will note that our surface clocks are running more slowly than his.
A good way to remember this that doesn't directly involve any complex reasoning about relativity, time dilation, and frames of reference is to think of an extremal example, namely what a clock passing through the "surface" of a black hole would look like to an orbiting observer, if there were a way to see it. (Specifically that the image of the clock is frozen in time.)
Changing oil is not difficult on most cars.
First check your owner's manual to see how much oil you need, go buy that many quarts of oil and a new filter (if you know the make and model of your car, any auto parts store will be able to tell you what kind of filter you need.)
Now get something to hold the oil (a wide and deep pan with enough volume to hold the contents of a full sized milk jug will do) and crawl under the front of your car.
You should be able to find a large tank with a bolt coming out of it. Unscrewing this bolt will cause the oil to start to drain, so make sure you aren't directly under it when you do this. Try to get all of the oil into your pan. Once all the oil is drained out, put it in a sealable container, you can dispose of it at almost any gas station.
Screw the bolt back in, and locate the oil filter in your car. You know what it looks like since you already bought a new one, and most of the time they are located in an easily reached place on the side of your engine. Some filters require "filter wrenches" to unscrew but a lot of the time you can do it with your hands. You might want to ask the dude at the auto parts store about this when you are buying the new filter. There will probably be a small amount of oil in the old filter, try not to let it get everywhere since oil stains concrete pretty well. Screw in the new filter, and don't make it too tight.
Pop the hood of your car, find the oil cap, take it off, pour in the new oil with a funnel so you don't spill it everywhere all over your engine. Check the oil level with a clean dipstick when you are done to make sure you put in enough. Done!
For a chess program it's enough to code some basic principles on how to evaluate the position and then implement some minimax or other similar bruteforce algorithm to search moves forward.
Terribly inaccurate. Don't forget opening theory. Disable a computer's opening book (which is centuries of human knowledge and experience, distilled and compressed into a form easily indexed by a computer) and even the best computer in the world can be defeated by relatively weak players. I don't have a good estimate of how weak computers become without opening theory, but I personally have seen a 2200 FIDE stomp all over Fritz in a standard game on a powerful computer in such a situation.
The point here is this: even the best computer chess programs don't just rely on basic principles and brute force algorithms. First they are built upon a human-supplied foundation -- opening theory -- that computer programs are as of yet incapable of spontaneously generating. Furthermore, programs are only just now starting to approach "super" grandmaster strength because of innovative strategic algorithms that attempt to imitate things like intuition through machine learning.
Despite the fact that Kasparov performs poorly against computers, the general consensus is that they still aren't anywhere near as good as humans at chess. Peter Svidler has no trouble with any of the top programs, and in general, when a strong grandmaster loses to a computer program, it's usually because he has committed a serious blunder -- a side effect of fatigue, sickness, or just having a bad day which are all unfortunately human qualities computers will probably never learn to emulate.
If anyone is seriously considering doing this run, I have quite detailed hand drawn maps of both overworlds (with all towns and moongate coordinates indicated) and every dungeon (with all doors, chests, fountains, etc. indicated).
Yeah, I took my gaming seriously when I was a kid. Anyways, if anyone thinks those would be helpful, let me know and I'll scan them.
I plan to begin work on this video again in the near future. The past week, instead of studying for finals, I spent a good deal of time on the first act of the first zone -- that is, Green Hill Zone Act 1. Hero Chaos Chao asked me once if 24 seconds was possible on that map, and I declared it impossible. My public video is already perfect, right?
Well, I had to see for myself. I have performed a frame by frame analysis of every instance of jumping and rolling throughout the entire level. My original time for the level was 00:37:53, my current best time is 00:37:09. The clock changes to 25 seconds on 00:37:04 (on the file I'm working on now. I believe it changes 1 frame earlier on my public video, which is a mystery to me.)
I've since found a way to shave 2 more frames off the first half of the map, and to maintain the 2 frame lead against my best time all the way until I enter the pipes after the loop. For some reason, the "faster" path enters the pipes ahead, but exits them behind. Because of this odd inconsistency, the only thing that remains for me to do on GH1 is try all the different double jumps on the loop within a window of 5 or so frames to see if one of them inexplicably results in me exiting the pipes ahead of my current best.
Anyways, the reason I'm telling you all this is that I'm sick and tired of this level, but don't want to give up on 24 seconds just yet. Critiquing and finding mistakes in your own work is always hard. Does anyone want to work together with me on this level, and possibly the whole video, to help me find new strategies and pinpoint the most minor errors?
Josh: I haven't seen the NG2 vid, but if there is a valid use to executing an impossible maneuver, then it is acceptable. For example, ultrafast shooting in Metroid and Megaman, or spinning really fast in Zelda when it is required to hit multiple targets, these are all valid uses of impossible maneuvers.
Read over my previous posts. You will find I take exception only to "needlessly executing a humanly impossible maneuver."
