Experienced Forum User, Published Author, Experienced player
(751)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
Just to contradict myself, I managed to gain 8 more frames in SYZ3 (WIP). But in any event, I will submit it as soon as I find a suitable screenshot.
Edit: There, it is done.
Experienced Forum User, Published Author, Experienced player
(751)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
Improved SYZ1 by 3 frames and LZ1 by 1 frame.
Full game WIP
I think this will be the final WIP; I managed to improve the attempted SYZ3 shortcut by 2 frames (check here), but hacking a savestate to get full spindash speed with a single rev (gaining 5 frames) shows that it cannot be done.
Experienced Forum User, Published Author, Experienced player
(751)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
Just a quick note: being able to boot a Linux Live CD without no problems does not rule out hardware failure. I have fixed one laptop (battery malfunction) and one desktop (harddisk failure) both of which showed problems on XP but worked flawlessly with an Ubuntu Live CD (the latter wouldn't even boot in XP anymore, but lasted until the harddrive died in Linux, which was much more than enough to back up everything without data loss).
With that in mind, I have a few suggestions:
1) If it is a virus, malware or some form of spyware, any testing done while XP is running is compromised and may give false results; it would be ideal to test it with XP offline. One way to do it is to use the Ultimate Boot CD for Windows to boot and scan your computer. It requires a XP CD, which it uses to build a preinstalled environment with a variety of virus/malware scanners which you can boot to. Just to be safe, you will want to make the CD in another computer.
2) Sysinternals has a lot of cool utilities, two of which deserve special mention:
(a) Process Explorer. There are some forms of CPU usage that don't show up in standard Task Manager; in particular, telling Process Explorer to show the "Context Switch Delta" column and watching the "DPCs" metaprocess can give a clue to hardware malfunctions that can slow the computer down without showing up elsewhere (the laptop I mentioned had a problem with the battery that manifested in this way);
(b) Rootkit Revealer. The way it is designed to work makes it virtually impossible for any rootkit to hide/compromise the program -- the rootkit programmers would have to know a lot more about the inner workings of the NT kernel than they care about.
Experienced Forum User, Published Author, Experienced player
(751)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
I managed to reduce SY1 time by 3 frames, but I am stuck trying to make some improvements to SY3. I am this close to a major improvement -- but sadly, it seems to be impossible... which is kind of frustrating.
I mean: just look at how close I get... I recon that some 5-10 frames ought to do it, but the number of re-records of this movie shows just how much different things I have tried on this level... since I started trying to improve this part in the beginning.
Experienced Forum User, Published Author, Experienced player
(751)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
DemonStrate wrote:
What is the requirement of a hill or angle to be able to jump off of it to gain forward momentum?
Ground speed, slope gradient and how (and if) the slope gradient is changing. The optimal jumps to make in slopes are the ones that not only actually increase your speed, but that leave you with (a) a positive vertical speed (going down) or (b) the ones that leave you with a really high negative vertical speed (going up) and end by hitting the ceiling before speed reaches the drag range (upward speed in the 0 to -1000 range). With Tails in particular, any slope jump that allows you to start flight at or above the screen's top edge before drag kicks in is also optimal, in that your vertical speed will be set to zero when you start flight in these cases and prevent further loss to drag (SLZ2 is the perfect example).
Jumping also makes a slight (1-frame) pause during which position is not changed. This must be considered when determining if a slope jump is worth it.
DemonStrate wrote:
There are a few levels where there is a hill, and I didn't know if there was a reason you don't jump forward off of it.
The most likely reason is that I would lose most of the speed gained (and then some) because of drag. The second most likely reason is that the speed gain is not enough to reduce the time further (because of the 1-frame cost or because the distance is too short or both), and the jump would not have any entertainment value.
Of note:
DemonStrate wrote:
Like in Marble Zone 3, at the end of it before the boss.
Jumping from that slope loses most of the benefits to drag, even if I jump at the last frame with slope 16 (I actually have two jump one frame before that or the loss will be even worse). It also prevents me from making that jump at the end; that jump is made at high speed in the change from slope -8 to slope 8 (possible due to speed and subpixel position). That jump grants me the vertical speed boost of slope -8 with the forward speed boost of slope 8; the time it takes for the vertical speed to reach the drag range makes me suffer from only 7 frames of drag before it becomes irrelevant (as the screen stops scrolling because I reach the boss area).
DemonStrate wrote:
Or in Spring Yard 2, at then end of it.
In this case, the speed gain nets exactly the frame lost due to the jump and nothing else.
DemonStrate wrote:
Edit: Also, as of not the latest WIP, but the one before it, mine desyncs at, I believe, Labyrinth Zone 3. It still does this in the latest also.
Nineko beat me to the punch (the server was under heavy load when I wrote the original reply), but: it is very likely that you are using the version v2.1 of the hack; I have checked, and the movie desynchs in Labyrinth 3 with that version. The movie is meant to be played with v2.1.1, a bugfix release that was released shortly after v2.1.
Experienced Forum User, Published Author, Experienced player
(751)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
Pu7o wrote:
You see, while getting in the wall as Sonic or even Knuckles is somewhat difficult for the casual user, as Tails it's ridiculously easy to just fly in there. So I added the wall to prevent that.
I figured as much. For what is worth, there are a couple of walls in GHZ3 (bottom route) that Tails can fly into too; but none of them are as easy or obvious (or useful) to get into as this one in GHZ2.
On another note, I managed to reduce Scrap Brain 2 time by another 2 frames. The WIP is here.
Experienced Forum User, Published Author, Experienced player
(751)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
Taking the top route at the end is two frames slower then the present route; entering the wall at the beginning is not possible because there are more walls in the rom to prevent just that, it seems.
Having the key input video would definitely be better than that YouTube video (as it would be possible to look frame-by-frame at what he does), but it seems to me that he spindashes at the beginning, jumps right before the rock, activate glide to drop to it, spindash again and glide to the end. That or he spindashes, jumps over the rock and activates glide mid-jump. I don't think he goes through the rock at all.
Experienced Forum User, Published Author, Experienced player
(751)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
bkDJ wrote:
Great work! 2 things though.
1) In labyrinth act 3, when you hit the switch to move the waterfall wall, you spindash on it, hit a wall, fall, and spindash again. Wouldn't it be faster to just walk off the switch and spindash only once since you have to fall anyway and the wall negates any good speed you could get from the first?
I tried both ways; but the speed I get from the spindash makes me start to fall slightly earlier than if I were to run off the edge, so that I hit the ground slightly earleir, and I already hit the ground at zero horizontal velocity because I bump into the wall, so that I can spindash slightly earlier (it is tricky to slow down exactly to zero horizontal speed in the air and still land facing the correct direction). All put together, it is seven frames faster to spindash from the top of the switch and I end up slightly ahead in terms of position, which may effectivelly add another couple of frames or so.
bkDJ wrote:
2) probably a hack glitch or something I pressed during the end credits but tails jumped off the scrap brain gravity spinning wheel the wrong way and was killed by lightning :D
I hadn't watched it this far yet; after you pointed that out, I did and the same happens even if you press nothing at all. Probably due to the differences between Sonic and Tails.
Experienced Forum User, Published Author, Experienced player
(751)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
I already managed to reduce it to 0:08:45 (11 frames) by turning slightly earlier and optimizing the in-wall movement; I will check again about the flying, it might be possible to optimize it a bit.
Experienced Forum User, Published Author, Experienced player
(751)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
upthorn wrote:
Scrap Brain 3 looks inoptimal. You want to get the moving platform unloaded from memory as soon as possible at the start, because when you approach it from the right, it loads in its final position.
It also unloads (and teleports to the end position) when you get far enough to the left. In fact, in the current WIP, if I turn back a few frames earlier (I haven't counted them exactly), the moving platform will not even be there, having vanished completely. I have tried doing it with your suggestion, but the best I have got so far is 9 frames slower than the WIP.
upthorn wrote:
(incidentally, the factor that determines which piston he will appear in next is how many frames you were in contact with his hitbox on the previous pass)
That is good to know; I will see what I can do with it.
Experienced Forum User, Published Author, Experienced player
(751)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
Managed to improve SY2 (2 frames), SY3 (18 frames) and SL2 (16 frames). SL2 took me quite by surprise, I was expecting at most a 2 frame improvement...
WIP here.
Experienced Forum User, Published Author, Experienced player
(751)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
Hm. I reduced SY3 by 8 frames, so I will have to resynch the whole Labyrinth zones again; I will do it tomorrow (and hope Microstorage is back by then).
Edit: improved SY3 by another 9 frames, being a total of 17 frames faster now.
Experienced Forum User, Published Author, Experienced player
(751)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
upthorn wrote:
Mainly it was that long flight that looked dubious. Though, frame-by-frameing it, it looks a bit less dubious. I still wonder if you could have started the flight a few frames later, so there would be less speed loss, though.
From memory, I wanted to say "no"; I decided to check, and indeed it was "no". Delaying by 4 frames means hitting the fish; delaying by 1-3 frames means that I can't boost over the bug at the end of the flight without losing speed, and have either to make a suboptimal boost or jump over it (or use the glitch bottom route, which is slightly faster...). In either case, the new route makes this moot, as the frames that might be gained this way will not add up to the 22 frames gained by the new route.
I would upload the combined WIP with all improvements, but Microstorage seems to be down for the count...
Experienced Forum User, Published Author, Experienced player
(751)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
Hm. Right after posting the last message, I tried a bottom route again and came up 22 frames faster. I am still polishing it, and will post the WIP later.
Edit: 24 frames faster through the bottom route, for a time of 0:15:15. WIP. The "entrance" into the second bottom route (frames 3695-3712) I had actually found in the v2 ROM, but the diffeences in flight made it useless there as far as speed is concerned (Tails would hit the right side of the waterfall and lose forward momentum).
Experienced Forum User, Published Author, Experienced player
(751)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
Thanks. Can you give an example in GHZ2? Because everything else I have tried resulted in it being one frame or more slower (or, at the very best, matching the current time).
Experienced Forum User, Published Author, Experienced player
(751)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
Nicely done. Somehow, I just knew it would have an improved boss, but the quick stop midway for spindash was a pleasant surprise; I had actually tried doing that in the past, but I didn't try walking back to stay on the sprite for the spindash and just assumed it would waste time.
I combined both improvements and resynched the movie while I wait for what you have in Labyrinth 2; here is the combined WIP that finishes the game. I also updated the times at the earlier post.
Experienced Forum User, Published Author, Experienced player
(751)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
Not content in letting you have all the fun, I had to improve it further :-)
Here is a 3-frame improvement on your SY2 improvement; those frames where extremely hard to find! A 1-frame imprecision meant that at most 1 frame could be gained, if any at all.
I will wait for your SY3 attempt before trying to resynch the video, as lag differences make Labyrinth zones a pain to resynch.
Edit: Another frame gained: WIP here. This brings down the SY2 (in-game) time a total of exactly 2.5 seconds lower than the time I had.
Experienced Forum User, Published Author, Experienced player
(751)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
Wow. 146-frame improvement! Nice work! I now can't wait to see what you have in store for SY3...
If you don't mind, I'll hex edit the changes in and credit you with this improvement.
Experienced Forum User, Published Author, Experienced player
(751)
Joined: 9/29/2008
Posts: 964
Location: 🇫🇷 France
So, in the end I did change something else: I managed to reduce MZ2 by 11 frames. After fixing chain desynchs due to lag differences, I got a movie that finished the game: Full game WIP. The extra lag happens mostly after the timer has stopped; for example, slow animals in capsules and level loading times. In fact, the Labyrinth zones (except for 3) had slightly less in-game lag.
