Quest for Glory II: Trial by Fire

Edit history:

SDAVerification: 2018-01-02 11:28:50 am

2017-11-07 09:19:30 am

Game Page: http://speeddemosarchive.com/QuestForGlory2.html

Quest for Glory II: Trial by Fire (pc) (pc) [Any %] [Single Segment] [New Game+] [Character: Fighter]

Decision: Reject

Reason: Good execution and general route, but there is a big potential time save by manipulating a global timer during a waiting sequence that was not used

https://queue.speeddemosarchive.com/verificationfiles/2113/

This run will be available for a month. After that these link(s) will no longer work.

Thread title:

SDAVerification

2017-11-07 09:19:31 am

Run Information

Quest for Glory II: Trial by Fire (pc) (pc) [Any %] [Single Segment] [New Game+] [Character: Fighter]

Verification Files

http://v.speeddemosarchive.com/QuestForGlory2-20171101/

Please refer to the Verification Guidelines before posting.

Please post your opinions about the run and be certain to conclude your post with a verdict (Accept/Reject). If you wish to remain anonymous, you can also send a pm with your reply to 'sdaverification' (please state clearly in that case which run you have verified). This is not a contest where the majority wins - Each verification will be judged on its content.

LotBlind

2017-11-16 10:35:24 am

Are you aware that you can use memory watching during the run if you do it in segments? You just check it right after saving or when time has stopped. Of course this is in SS. Other than that, did you look around carefully for anything that happens in Raseir only on even or odd seconds? Or at any fixed time at all? Is the timer that's used to determine whether you get or lose 26 seconds global or does it reset at some point? Even if it's global, does it help you to start a stopwatch earlier in the run or does it increment in a way that's difficult to follow? Even if you can't find any way to follow the timer aside from cheating by looking at the RAM directly, it seems to me you have a 50-50 chance of cutting short the time during the first Raseir wait like this:

1) Just pass through the door.
2) After exactly two [in-game] seconds, pass back. Or buffer it like it seems the TAS is doing.
3) If you happened to start on an even second and managed to stay consistent with your timing, surely the time will continue to increment until it's time to sleep.

If you can't possibly do this the whole way through, how much time would you save if you got it right like four times and then waited? I see that even the TAS takes more than two minutes here so those 26 seconds each go can't be in real-time. Still, even if you can't do it the TAS way, maybe there's some second best solution.

Other than this, is it for sure no other inputs could be any faster, such as "wish teleport" or "wish tele"?

Are you aware that you can use memory watching during the run if you do it in segments? You just check it right after saving or when time has stopped. Of course this is in SS. Other than that, did you look around carefully for anything that happens in Raseir only on even or odd seconds? Or at any fixed time at all? Is the timer that's used to determine whether you get or lose 26 seconds global or does it reset at some point? Even if it's global, does it help you to start a stopwatch earlier in the run or does it increment in a way that's difficult to follow? Even if you can't find any way to follow the timer aside from cheating by looking at the RAM directly, it seems to me you have a 50-50 chance of cutting short the time during the first Raseir wait like this:

1) Just pass through the door.
2) After exactly two [in-game] seconds, pass back. Or buffer it like it seems the TAS is doing.
3) If you happened to start on an even second and managed to stay consistent with your timing, surely the time will continue to increment until it's time to sleep.

If you can't possibly do this the whole way through, how much time would you save if you got it right like four times and then waited? I see that even the TAS takes more than two minutes here so those 26 seconds each go can't be in real-time. Still, even if you can't do it the TAS way, maybe there's some second best solution.

Other than this, is it for sure no other inputs could be any faster, such as "wish teleport" or "wish tele"?

mrprmiller

Edit history:

mrprmiller: 2017-11-18 05:18:22 pm

mrprmiller: 2017-11-18 06:51:32 am

2017-11-18 06:50:54 am

Everyday is puppies and sunshine...

Quote from LotBlind:

Are you aware that you can use memory watching during the run if you do it in segments? You just check it right after saving or when time has stopped.

I had no idea. I thought that would have been considered a tool-assist, but I can understand how that would literally be impossible to stop in segmented runs.

Quote from LotBlind:

Other than that, did you look around carefully for anything that happens in Raseir only on even or odd seconds?

The issue is that not much happens in Raseir. c-square was pretty thorough in his assessment: the even-odd counter has no effect in the hallway, only on transitions from one screen to another. The only things you can actually do in Raseir on the first day are go out and fight in the desert (the desert walk from town actually don't use any time all, frustratingly enough), explore the town (with the same 26-second advances being important), order drinks at the bar (tested), and steal the oil from the back table. What we did find is that you can actually reverse time, Superman style, if the in-game day/night speed is set to it's lowest.

Quote from LotBlind:

Or at any fixed time at all? Is the timer that's used to determine whether you get or lose 26 seconds global or does it reset at some point? Even if it's global, does it help you to start a stopwatch earlier in the run or does it increment in a way that's difficult to follow?

The timer that determines whether you lose or gain literally is a formula attached to transitions testing against the "in-game" clock. In pseudocode: TimeOfDay=Even, TimeOfDay+26. If TimeOfDay=Odd, TimeOfDay-26. When you set the in-game timer speed, what happens is this: normal speed is every second is a second. The second setting moves the clock forward +1, then +2 alternating. The fastest setting is +1, +3 alternating back and forth.

Quote from LotBlind:

Even if you can't find any way to follow the timer aside from cheating by looking at the RAM directly, it seems to me you have a 50-50 chance of cutting short the time during the first Raseir wait like this:

1) Just pass through the door.
2) After exactly two [in-game] seconds, pass back. Or buffer it like it seems the TAS is doing.
3) If you happened to start on an even second and managed to stay consistent with your timing, surely the time will continue to increment until it's time to sleep.

A buffer could be created... but that buffer could screw me over. The time advancement that happens after talking to Ferrari could serve the purpose you're describing for getting a solid bearing on even-odd, there's no way to know if I am buffering on an odd or even increment, especially in the inconsistency that occurs with load-times on occasion. The DOSBox cycles make some transitions take longer and make movement slow at times for no apparent reason for the first couple of steps at time. It's hard to notice when watching a TAS or Any%, but when you do a 100% run it's a little more noticeable. Those delays do not necessarily translate to clock delays. It'd be like hitting a field-goal on a constantly moving goal-post.

Quote from LotBlind:

If you can't possibly do this the whole way through, how much time would you save if you got it right like four times and then waited? I see that even the TAS takes more than two minutes here so those 26 seconds each go can't be in real-time.

You can save up to 90-seconds with TAS precision on frame-perfect movement, according to c-square. You're correct. Those 26 seconds are in-game day/night clock seconds. Because the game polls the keyboard 1/7 frames to see if a button is pressed, MAYBE 60 seconds could be saved if everything went right. The game does poll the mouse for input every frame, so the TAS is using the mouse for every movement, but he can also zip the mouse to where he needs to at frame-by-frame perfect input, whereas a human being needs to move the mouse over those frames making that 1/7 frame input a problem. That 1/7 frame problem would also cause problems for finding the buffer.

I always felt like something was happening when I started typing commands and it missed the first letter in my parser. I now understand it's because I hit that first letter on the 1/7 frame. Annoying, but vindicating.

Quote from LotBlind:

Other than this, is it for sure no other inputs could be any faster, such as "wish teleport" or "wish tele"?

You're right on this. CTRL+E enters "escape" which does just that; you first for an escape. This was something I found out after the TAS came out. :/ hehe c-square held that little trick back on me. Also, "y" can be used for "yes" and "n" can be used for "no". "strip" can be used in place of "give clothes". There are a few extra keystrokes here and there that could probably shave about 10 seconds or so off.

[quote="LotBlind"]Are you aware that you can use memory watching during the run if you do it in segments? You just check it right after saving or when time has stopped.[/quote]
I had no idea.  I thought that would have been considered a tool-assist, but I can understand how that would literally be impossible to stop in segmented runs.
[quote="LotBlind"]Other than that, did you look around carefully for anything that happens in Raseir only on even or odd seconds?[/quote]
The issue is that not much happens in Raseir.  c-square was pretty thorough in his assessment: the even-odd counter has no effect in the hallway, only on transitions from one screen to another.  The only things you can actually do in Raseir on the first day are go out and fight in the desert (the desert walk from town actually don't use any time all, frustratingly enough), explore the town (with the same 26-second advances being important), order drinks at the bar (tested), and steal the oil from the back table.  What we did find is that you can actually reverse time, Superman style, if the in-game day/night speed is set to it's lowest.

[quote="LotBlind"]Or at any fixed time at all? Is the timer that's used to determine whether you get or lose 26 seconds global or does it reset at some point? Even if it's global, does it help you to start a stopwatch earlier in the run or does it increment in a way that's difficult to follow?[/quote]
The timer that determines whether you lose or gain literally is a formula attached to transitions testing against the "in-game" clock.  In pseudocode: TimeOfDay=Even, TimeOfDay+26.  If TimeOfDay=Odd, TimeOfDay-26.  When you set the in-game timer speed, what happens is this: normal speed is every second is a second.  The second setting moves the clock forward +1, then +2 alternating.  The fastest setting is +1, +3 alternating back and forth.

[quote="LotBlind"]Even if you can't find any way to follow the timer aside from cheating by looking at the RAM directly, it seems to me you have a 50-50 chance of cutting short the time during the first Raseir wait like this:

1) Just pass through the door.
2) After exactly two [in-game] seconds, pass back. Or buffer it like it seems the TAS is doing.
3) If you happened to start on an even second and managed to stay consistent with your timing, surely the time will continue to increment until it's time to sleep.[/quote]
A buffer could be created... but that buffer could screw me over.  The time advancement that happens after talking to Ferrari could serve the purpose you're describing for getting a solid bearing on even-odd, there's no way to know if I am buffering on an odd or even increment, especially in the inconsistency that occurs with load-times on occasion.  The DOSBox cycles make some transitions take longer and make movement slow at times for no apparent reason for the first couple of steps at time.  It's hard to notice when watching a TAS or Any%, but when you do a 100% run it's a little more noticeable.  Those delays do not necessarily translate to clock delays.  It'd be like hitting a field-goal on a constantly moving goal-post.

[quote="LotBlind"]If you can't possibly do this the whole way through, how much time would you save if you got it right like four times and then waited? I see that even the TAS takes more than two minutes here so those 26 seconds each go can't be in real-time.[/quote]
You can save up to 90-seconds with TAS precision on frame-perfect movement, according to c-square.  You're correct.  Those 26 seconds are in-game day/night clock seconds.  Because the game polls the keyboard 1/7 frames to see if a button is pressed, MAYBE 60 seconds could be saved if everything went right.  The game does poll the mouse for input every frame, so the TAS is using the mouse for every movement, but he can also zip the mouse to where he needs to at frame-by-frame perfect input, whereas a human being needs to move the mouse over those frames making that 1/7 frame input a problem.  That 1/7 frame problem would also cause problems for finding the buffer.

I always felt like something was happening when I started typing commands and it missed the first letter in my parser.  I now understand it's because I hit that first letter on the 1/7 frame.  Annoying, but vindicating.

[quote="LotBlind"]Other than this, is it for sure no other inputs could be any faster, such as "wish teleport" or "wish tele"?[/quote]
You're right on this.  CTRL+E enters "escape" which does just that; you first for an escape.  This was something I found out after the TAS came out.  :/  hehe c-square held that little trick back on me.  Also, "y" can be used for "yes" and "n" can be used for "no".  "strip" can be used in place of "give clothes".  There are a few extra keystrokes here and there that could probably shave about 10 seconds or so off.

mrprmiller

Edit history:

mrprmiller: 2017-11-18 05:15:25 pm

2017-11-18 04:02:15 pm

Everyday is puppies and sunshine...

Quote from LotBlind:

Other than this, is it for sure no other inputs could be any faster, such as "wish teleport" or "wish tele"?

I should also answer that the preposition "for" is needed for the other options. Forgot to mention that. I can "search pass" rather than "search for pass(age)," but apparently, whoever programmed the djinni decided that "wish teleport" was too unspecific.

LotBlind

2017-11-19 08:01:15 am

Nothing's cheating so long as you're not writing into memory or directly monitoring it during recording attempts/segments.

Quote:

whoever programmed the djinni decided that "wish teleport" was too unspecific.

Guy was a selective grammar nazi. He never finished grammar nazi school nor did he earn the respect of der Führer die Grammatik.

So how much real-time does each 26 seconds correspond to again? If you're on the fastest setting, it's 13 right (the average of 1 and 3 being 2)? If so, this is still a pretty major time save opportunity which is difficult to lose in the part that comes before AFAICS. Did I count that right? It just doesn't seem to differ from any of the typical kinds of RNG manipulation found in other runs. There isn't any RNG other than it is there (other than inputs getting lost)? A lot of the run is waiting around as well so it's pretty relaxed in that way. Am I missing something?

LotBlind

2017-11-30 11:09:26 am

So I'm thinking you were left looking for ways to incorporate some more RNG manipulation? Based on everything you said, I don't know see how it's not worth it if it's a potential 13, 26, 39 etc. seconds off. Even with 50-50 odds, you'd still get 52 seconds off every 8th attempt which seems way more significant than execution or (as I got it, unless passing the elementals by is really dangerous) the rest of the RNG.

So unless something else is said
reject

SDAVerification

2018-01-02 11:28:51 am

Decision posted.