Username:
B
I
U
S
"
url
img
#
code
sup
sub
font
size
color
smiley
embarassed
thumbsup
happy
Huh?
Angry
Roll Eyes
Undecided
Lips Sealed
Kiss
Cry
Grin
Wink
Tongue
Shocked
Cheesy
Smiley
Sad
page  <- 12345678910 <- 1 ... 10
--
--
List results:
Search options:
Use \ before commas in usernames
Edit history:
tigrou: 2016-06-23 02:06:00 am
tigrou: 2016-06-21 09:48:55 am
tigrou: 2016-06-21 09:48:05 am
tigrou: 2016-06-21 09:46:05 am
tigrou: 2016-06-21 09:45:34 am
tigrou: 2016-06-21 09:41:54 am
tigrou: 2016-06-21 09:41:38 am
About lag : I took a look at the game internals and here is what I found out :

- The game internal loop (which take care of updating life engine, input, physics, rendering, ...) is trying to target 70fps (and not 60fps as I thought). If it ran faster than this, the game will slow down to match that speed.
You can measure those FPS by using FRAPS and the technique I explained before (there is a free version on the official website). or this :
1) start CheatEngine.
2) Raise number of cycles to a high value (eg : 35000 cycles).
3) Search "70" (2 bytes value).
4) Lower cycles to 200.
5) Search "0" (2 bytes value)
6) Repeat 2-5 until one value only is found : this is the internal fps counter used by the game itself (very precise)

- As in most games, there is some internal mechanism that make sure player will run/rotate at same speed, regardless fps.
You can try it yourself easily : lower the number of cycles (Eg : 1000) check time needed to rotate OR to move player from on point to another : should be pretty similar regardless CPU cycles.
I am not sure how accurate it is. It might that implementation is not perfect and player still ran slightly slower when CPU is not fast enough.

Anyway, there is some actions which are not covered by this mechanism and which will make you lose some time :
- Switching between cameras.
- Switching between floors.

When one of those events occurs the game just freeze and all time spent is lost.
Try yourself with low cpu cycles and room viewer linked : start running from one place to another, when the camera is switching the player stop moving for a while and resume from where is was before the freeze.
The potential move that could have been done during the freeze is not recovered.
It seems playing some sounds also slow down the game (i'm not 100% sure about this)

One reason why room/camera switch is so slow is because all ressources are compressed inside PAK files. When camera is switching, it is like the game has to unzip an image from a zip file each time.
The game has very little memory available (640K) so once a ressource had been used, it is thrown away (and reloaded later if needed).

Also : what definitely slowdown the game (without impacting player speed) is the rendering. Once you have 2-3 enemies on the screen (like in the dining room) the FPS will go down.


Shut up and go back to another earth
I'll keep an eye on the subject but saddly, for me the choice is clear. What made me not use DOS internal reccorder was precisely the fact that the run I reccorded in wich there was the gradual lag was kind of cheated. At this time it was the route of my first 11:54 route. As I said, this lag appeared because I touched a ghost while leaving the ballroom. Imagine that to this moment until the end of the run (almost during 6 minutes), the game decelerated gradually. It was indeed annoying, but it was also a really good way to anticipate everything easily because the game became so slow that it was a piece of cake to be accurate with my moves. And in the final video nothing appeared, like the game had normal speed all along. For me it was exactly like cheat as the "slowmotion" helps a lot to play well whileit is totally invisible in the video.
tigrou: CheatEngine crashes on my compy. If that value was linked to the viewer, would it tell me how long the game has been lagging for? You never answered my suggestion about creating a lag meter in-game. Is it possible?

So I can't quite tell based on your answer - do you think NHG is losing a considerable amount of time compared to  a video recorded on DB?

NHG: Here's another little variation on the alternative boss exit I showed you earlier. You run through to R5 which causes you to fall down one level. Then jump back into R6 near the corner of the trigger while activating the hover. This means you don't need a separate hover later. The advantage is you're only falling down one level instead of 2.  I don't see the alternative boss exit even mentioned in the table you were supposed to fill out.

Also this idea. If you start rotating towards the right (I presume you'll want to approach the E5 trigger from the W) right before you clip through and fall, you'll get a quick turn as you land and are sent into the trigger straight away. You won't have the hover yet though.

lag: but this time you'll have no ghosts... but yes a single-segment run is different from segmented. I forgot about that. If you wanted to, you could ask in the tech support about it (unless tigrou can help) - maybe there's some way to change DosBox' settings so it doesn't lag up, or maybe we can change it to MAX FPS to correct things, as is done with some games. If you do that now, you'll know what to do with 3.
Edit history:
NHG: 2016-06-21 01:26:44 pm
Shut up and go back to another earth
Alternative boss exit: Wait,  Are we supposed to do this after defeating Pregzt ? Because I don't even have to time it to see it's a lot longer, even in term of distance traveled/actions realised, not to mention it asks the viewer for the last blind part. I mean, it's just a more complicated path who activates the jump-trick sooner, but it does not give any advantage to activate it now or during boulder opening. "you're falling down one level instead of 2", but with the classic way there's no need to fall at all Oo I don't see the advantage here. Or maybe you were comparing to the first alternative boss exit ? I kinda forgot to put it into the board but I don't think I would have done it anyway: no matter the category or the "general path" you use for the run, it's longer. If I had to write every possible path even when they don't worth it in any case I indeed forgot to mention a lot of things ^^'

Idea: I tested it before doing the last speedrun, I did not keep it because it gives sligtly the same time (basically you don't have the boulder opening time + hook activation but you have a OOB + 2 levels falls + jump trick activation a little less optimized) but it's really easy to lose time by bad clipping.

Lag: Ghosts were just an exemple, if it cancels big lags, there's no reason it does not cancel all the "mini-lags". That's exactly why SDA does not accept emulators (and so Dosbox neither some years ago), because despite the fact that you can modify a lot of things in it (wich is against the idea to play in "normal conditions"), an emulated game never works exactly as it does on its original support. In any case and if I read SDA rules about Dosbox runs, we MUST keep the settings
given by the GOG version as it is considered as an "offical dosbox bundled release" (https://forum.speeddemosarchive.com/post/virtualization_software_and_dosbox_allowed_for_old_games.html). Concerning the cancelling of lag it in indeed a good question, maybe SDA'll consider those legit, but I don't. For me, having a final video slightly different that what you actually experimented during the play and who has consequences on the final time is not okay.
NHG: Okay, it's slower but could have been worth a brief mention in the table just so we can see someone's tested it.

The other idea: okay I might have to test it more for segmented then if it's roughly the same.

Lag: Well it doesn't matter at all if everyone gets basically the same amount (i.e. if it's only a factor of the CPU cycles). I don't really notice the lag while recording short videos, and most of my segments are going to be so short it doesn't matter. I guess there's no real problem other than if recording introduces external lag for you (I'm super lost and confused about this stuff) then you lost some time to that and we can't necessarily compare the run times directly. But that's fine isn't it?
Shut up and go back to another earth
The other idea: I guess if you manage to do it with a perfect execution (meaning you directly have a good angle for clipping) it is possible to save a little time, but frankly I'm not even sure.

Lag: Oh, of course it is fine ^^ I mean technically my opinion on the subject doesn't even matter, and unless you're triggering the whole mansion in a row I think the final time won't make a huge difference with or without lag so no matter what, comparing both times'll give an "accurate" idea anyway. But be careful about this whole "settings modification" thing, it could make the run refused for submission.
Edit history:
tigrou: 2016-06-23 02:09:14 am
tigrou: 2016-06-22 03:20:57 pm
tigrou: 2016-06-22 03:20:05 pm
tigrou: 2016-06-22 03:17:14 pm
tigrou: 2016-06-22 03:14:43 pm
tigrou: 2016-06-22 03:08:35 pm
tigrou: 2016-06-22 03:08:02 pm
tigrou: 2016-06-22 03:07:35 pm
tigrou: 2016-06-22 03:06:27 pm
tigrou: 2016-06-22 03:06:11 pm
tigrou: 2016-06-22 02:49:10 pm
tigrou: 2016-06-22 02:48:48 pm
tigrou: 2016-06-22 02:30:15 pm
tigrou: 2016-06-22 02:23:16 pm
tigrou: 2016-06-22 02:22:48 pm
tigrou: 2016-06-22 02:22:17 pm
tigrou: 2016-06-22 02:22:03 pm
tigrou: 2016-06-22 02:08:13 pm
tigrou: 2016-06-22 02:07:39 pm
@LotBlind : I think your computer has some RAM issues (because what CheatEngine does is to scan memory). Did you tried ArtMoney, as I suggested ?

Lag meter : what would you use this for ? How would it help to improve speedrunning with the GOG version ?

Even it is theoretically possible (eg : by comparing total number of frames executed vs expected number), the lag reported (eg : in milliseconds) would be wrong because it would take in consideration the whole lag, while the game as some internal mechanisms to counter some of them (eg : player rotate and move at same speed regardless FPS). What is required is to measure time lost to perform some specific actions (eg : switching camera/floor). This cannot be done in the viewer (it does not know what piece of code AITD is executing at a given moment).

If you want to measure that yourself, what you can do is to record the game and then use a video editing program to measure how much time has elapsed (eg : to switch camera). Multiply this by the number of camera switches in NHG run and you will a good estimation of how much time has been lost in the whole run. As I explained before, on "cycles = max" or "core = dynamic" those actions are almost instantaneous (thus you can consider lost time is near zero).

About "DOSBOX cancelling it's own lag" : I am not sure to understand what you mean guys. Does it means that when recording the game using DOSBOX, if there is any lag because of the recording it won't appear in the video ? (while it would be noticeable if recording was done using an external program).

Quote:
NHG : an emulated game never works exactly as it does on its original support

While DOSBOX is far from perfect, there are some special emulators (used in TAS) which are almost 1-1 with the original hardware (regarding timings and such). They are so accurate that if you send the inputs to a real machine (using some electronic device) the TAS run will not fail (most tricks used in TAS require very precise timings (at frame level)). Here is an example.
Edit history:
LotBlind: 2016-06-23 05:57:10 am
LotBlind: 2016-06-23 05:26:07 am
NHG: yup, not about to touch settings... I think tigrou just suggested it for testing.

BTW could you quickly test to see if the same DosBox recording lag happens with 2 and 3 as well?

EDIT: Did you know there's some differences in the floppy and CD versions of A2? There's at least this thing. Did you ever try the floppy one just in case?

tigrou: Okay so it looks like (as per NHG's tests earlier) the game takes a different amount of time when you enter the menu depending on where you're at at the time. Presumably this has to do with 1-2 things such as maybe the number of items in the menu, and when you close it, how much stuff is in the room around you because that seems to get completely re-loaded. So by changing where you change actions or use items (the oil mainly) you might be able to save a little bit of lag time. If there was a lag meter, I could fairly easily test where the best spots are and route my run accordingly. This could also be used for AitD 2 and 3 for further optimizations. I'm going to try to do it with or without such an aid but it would make it much faster to complete the run if it was available. But just think about it and tell me if it's too much work or whatever. That's absolutely fine.

Oh wait, are you saying we can't directly measure that stuff in the viewer? I don't mean lag in the sense of "running" lag, only loading times. How about do you think my theory is right - it's simply the count of objects, triggers and colliders in the room that determines how long it will load for? Could we have the viewer count those for me then? That would also save a bit of time.

What NHG meant about canceling lag, is that if you play the game on DosBox while recording, the longer you've played the bigger the lag gets until you're playing in slo-mo. However, this does not show in the video. When recording using FRAPS, I just asked IsraeliRD and he doesn't think that could cause extra lag in most cases unless your computer is slow. We think that the extra lag while recording really is only caused by the recording process itself. Israeli hasn't heard of anyone else getting it so maybe it's an inherent problem with DosBox and AitD. I'm actually going to ask the DB forums about it now...
Edit history:
NHG: 2016-06-23 08:57:08 am
NHG: 2016-06-23 07:23:53 am
NHG: 2016-06-23 07:21:20 am
Shut up and go back to another earth
I am going to try some records with DOSbox with all of the three games, to see if the thing still happens. Even if the physic gives the feeling is floating Carnby is floating, I think the sequels are not as much subject to lag to the first one. I'm going to test it now. In any case, maybe the gradual lag effect only happened with my previous computer. After all I never re-tested it since then.

Alone2: Yes, there are different versions of the game, I think there's one more where the whole Grace part is missing. Or maybe I'm wrong and there's just this version with a "shorter" Grace part. As far as I know it's the only difference. There's also a reported glitch about the sword allowing to pass through doors that I never saw or did by myself, Tigrou and I were wondering if it's because it's only possible in the floppy version. I did not test it though.

Edit: Oddly enough, It does not take the video, only the sound. DO I hav to install a pluggin/pilot to make it work correctly ? =o... Oh forget that it works.

Edit2: I think you can relax about the whole lag thing. I tested the internal reccorder and I can't see any difference with external recorder. Even ghost triggering doesn't do anything so at this point we can conclude that the problem came to my last shitty computer. It's a really good new because I found my old post in the forum where I talked about: it was the "quake effect" who produced the gradual lag. If it's not really a problem to avoid ghosts, it is impossible to avoid the quake at the end when you kill Pregzt... Anyway, my old PC probably had problem to make DOSbox works properly. Even if the internal reccoder "cleans" the video from external lag as I made the experience some years ago, we can assume that any actual computer wont lag because of such a game.
Passing through doors? Whoa! You gotta look into that later.

Thanks for looking into the recording thing! I'm getting packed right now and will be back in a few days...
Very minor little note, something I hadn't realized before: it IS possible to save/load without losing your running state, same conditions apply as with quick turns, i.e. hold down forwards before saving and hold it as you're reloading. It's really good for segmented runs ofc. Segment anywhere right about.

Another minor new discovery: it's actually fairly easy to get the game to stop drawing its graphics (everything except for the text in item pick-up prompts) by just hold down keys while leaving the save menu. I could attach a video of this but there's not much to see. Tongue I couldn't instantly see any altered behaviour otherwise, not even less lag so...

If you're holding the oil lamp, or are in some other state where your BUTTON action is null (i.e. space doesn't do anything), you can use space to perform snappy turns everywhere! The effect is reset everytime the view changes though.

I'm posting this as a note that this game is still being studied by myself: I've finished reading through all the scripts a while back and tigrou has been making awesome improvements in the Room Viewer steadily along the way. I don't think I've found anything (apart from maybe the lantern trick) that would make the existing run virtually any faster but I've still to do some last tests before getting into actual route testing finally (hoping to get there within a few months but it depends on my health). Afterwards I'll edit all my notes into the Guide but unless someone else really wants to start running one of these games right now, not yet.

PS: From now on, let's all record the Room Viewer output on follow player with all our runs for any of these games - makes for a better viewing experience. You should be able to extract just the game footage afterwards for the submission. Ofc you can't have the Viewer showing while you run.
Edit history:
Djeez: 2024-02-14 07:54:57 pm
Djeez: 2024-02-08 11:52:48 am
Djeez: 2024-02-08 11:22:16 am
Djeez: 2024-02-08 10:53:42 am
Djeez: 2024-02-08 10:52:53 am
Djeez: 2024-02-08 10:52:14 am
I guess it's time to revive this thread. LotBlind, you have done a lot of work on this game, there is no doubt about that. You've been the main driving force behind the evolution of Any%, and just in general organizing knowledge for this game, and I have a lot of respect for that.

However, this is shameful behavior from you.  https://www.speedrun.com/aitd1/forums/4hivy

It has been 3 years now, and I guess I had never noticed it, but here we are. Removing all the sugarcoat from your post, this is basically what you're saying:

1) Hi guys, I found a new strat!
2) However, I will not share it! Instead, you will have to contact me privately, and prove to me that you can execute to NHG's level (a feat no one has ever been able to do in 10+ years).
3) Once judged worthy, you may commence grinding. Once you get a run that meets SDA standard, you shall be allowed to post it on SDA, and I might post my segmented runs as well.

I'm sorry, but this is absurd. Something similar has happened in FF7, and no one thought it was done in good faith there either.
After 3 years of nothing since this post, this is strat hoarding.



First of all, strat hoarding in this era of speedrunning is extremely poorly regarded, EVEN when it's done in good fun.

Second of all, this game's already barely active. Do you think going into the forums, only to see a strat is being hoarded for over 3 years now, is going to make people want to pick up the game? Seriously, you are essentially holding Any% hostage.

Thirdly, the part where you say ''You would have to show you have potential for executing on the same level as NHG of course.'' is one of the worst thing i've ever read, in my now 15 years of speedrunning. Just WOW. So potential runners have to prove to you they're good enough, and THEN you feed them the strat? Absurd. Asking potential new runners that they need to reach a certain skill level, for you to unlock the strat to them, is INSANE. Doubly insane when said skill level must match NHG, someone who we, the public, have only seen play in near-perfect runs for a total of less than 1 hour of footage across the entire trilogy.

Finally, the part where the potential runner has to ''agree to submit the run to SDA'' shows what your true motives are. You have obviously been the last person truely holding down the fort for SDA, and it is respectable. However, your passion for SDA is clouding everything else. I mean you go on reddit to argue in favor of SDA rulings to this day, which I won't comment on, but is worth pointing out in this instance.

One final thing. I remember starting to speedrun Lands of Lore in 2009 and feeling stifled by the ALREADY draconian rules of this website. Please do not continue the time-long SDA tradition of stifling speedrunning growth all the way into the 2020s.
I'm going to reply to you in a PM first in case there's some misunderstandings here. Then one of us can copy a summary of that conversation here (to avoid creating a lot of fluffy text here). If you're still not happy after what I've told you there, I'll reply to this in this public thread too.
Edit history:
Djeez: 2024-02-14 09:51:09 pm
Djeez: 2024-02-14 09:38:42 pm
Djeez: 2024-02-14 09:11:56 pm
Djeez: 2024-02-14 09:11:11 pm
Djeez: 2024-02-14 09:10:47 pm
Djeez: 2024-02-14 09:10:27 pm
Djeez: 2024-02-14 09:09:49 pm
Yeah I did not want to sully this thread either, but oh well.

After reading LotBlind's private message to me, I have contacted another SDA administrator. If you are going to be the face of this website, it is unacceptable that you hold such elitist views towards speedrunning. It is quite frankly ridiculous to have to read things as such in 2024:



(excerpts from LotBlind's DM to me, posted for the record)

"I guess it matters to those who want to do casual runs. I haven't really ever been into or thought much about such efforts, so it's just not at the top of my mind. I don't think what they're doing is particularly useful in the sense of contributing something new to the theory. And if they don't care about that, what does it even matter what strats you're using, if it's just to challenge yourself?"

So many strats have been found by people of all skill levels, in so many games. This is just asinine and extremely gatekeepy to read. Deciding who's ''worthy'' of what strat is also crazy.


"It's not just a matter of sacrificing time, but rather taking a risk as well. So many people over the years have approached me and ended up not pulling through with almost any work done at all."

I've spent hundreds of hours helping people across multiple games, and I know a ton of people who have done the same. Not once did I think or worry how long/serious they were going to take the game. Get out of here with your ''risk''.


"I'm already putting in all this effort, so am I morally obliged to go further out of my way to appease people I don't really see as doing anything useful at all from my point of view? Why exactly? It probably wouldn't have amounted to anything, in my experience, which I have a fair amount of by now. Myself, tigrou and NHG are very exceptional people."

You didn't have any moral obligation. You could've kept everything to yourself and kept silent. But you decided to publicly announce you had new strategies. At that point, you have obligations.
Otherwise, you are dangling a carrot in front of people's face, and asking them to jump through a hoop to access.

Let the record show that I am not even outraged because I want to know the strategies that badly. I haven't even gotten around to check the ''new'' ones from like 5 years ago yet. I'm just appalled by the principle of the situation here, which is antithesis to everything I've known the speedrunning community to be about.