MechWarrior 3 - August 2, 2022

Edit history:

SDAVerification: 2022-08-02 03:01:26 pm

2022-07-20 12:39:20 am

Game Page: http://speeddemosarchive.com/MechWarrior3.html

MechWarrior 3 (pc) (pc) [Any %] [Segmented]

Decision: Accept

Congratulations to 'Falconer Gray'!

Thread title:

SDAVerification

2022-07-20 12:39:21 am

Run Information

MechWarrior 3 (pc) (pc) [Any %] [Segmented]

Verification Files

http://v.speeddemosarchive.com/MechWarrior3-20220719/Mechwarrior%203%2030.44%20Segmented%20Speedrun.mp4

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.

ImEliteGlitches1608

2022-07-21 02:08:17 am

A/V: mostly good; however the audio was a little quiet for me, even on max (but that could partly be due to my bad hearing). Visuals were good, apart from some anomalies, e.g. at 0:20, 3:54, and 15:10 - which appear to show the taskbar. I'm unsure as to the severity of this, but it may break the clean video rule.

Cheating: none

Gameplay: No complaints - solid, as to be expected of segmented runs. Comments are well done too, certainly helped my comprehension of the game (having never played it before).

Accept

Falconer Gray

2022-07-21 05:10:01 am

Quote from ImEliteGlitches1608:

A/V: mostly good; however the audio was a little quiet for me, even on max (but that could partly be due to my bad hearing). Visuals were good, apart from some anomalies, e.g. at 0:20, 3:54, and 15:10 - which appear to show the taskbar. I'm unsure as to the severity of this, but it may break the clean video rule.

Cheating: none

Gameplay: No complaints - solid, as to be expected of segmented runs. Comments are well done too, certainly helped my comprehension of the game (having never played it before).

Accept

Thanks for the feedback. Especially regarding the comments. I kept thinking I was writing way too much, but I wanted someone who had never seen the game before to feel comfortable with what they were watching.

Regarding the video comments, that seems to come up on the transition between menu and gameplay. From talk in the old thread on this site, people were suggesting that the menu and game run in different resolutions. That may have something to do with it.
I'm can't think of anything I can do to eliminate this when recording, but I suppose if it's an issue those frames could potentially be edited out? Unfortunately I wouldn't know how to do that properly.

[quote="ImEliteGlitches1608"]A/V: mostly good; however the audio was a little quiet for me, even on max (but that could partly be due to my bad hearing). Visuals were good, apart from some anomalies, e.g. at 0:20, 3:54, and 15:10 - which appear to show the taskbar. I'm unsure as to the severity of this, but it may break the clean video rule.

Cheating: none

Gameplay: No complaints - solid, as to be expected of segmented runs. Comments are well done too, certainly helped my comprehension of the game (having never played it before).

Accept[/quote]

Thanks for the feedback. Especially regarding the comments. I kept thinking I was writing way too much, but I wanted someone who had never seen the game before to feel comfortable with what they were watching.

Regarding the video comments, that seems to come up on the transition between menu and gameplay. From talk in the old thread on this site, people were suggesting that the menu and game run in different resolutions. That may have something to do with it. 
I'm can't think of anything I can do to eliminate this when recording, but I suppose if it's an issue those frames could potentially be edited out? Unfortunately I wouldn't know how to do that properly.

ImEliteGlitches1608

Edit history:

ImEliteGlitches1608: 2022-07-21 05:12:56 am

ImEliteGlitches1608: 2022-07-21 05:12:17 am

2022-07-21 05:12:07 am

Quote from Falconer Gray:

I'm can't think of anything I can do to eliminate this when recording, but I suppose if it's an issue those frames could potentially be edited out? Unfortunately I wouldn't know how to do that properly.

Out of curiosity, what are you using to record?

Falconer Gray

2022-07-21 05:50:06 am

Quote from ImEliteGlitches1608:

Out of curiosity, what are you using to record?

Recorded in OBS.
When figuring out how to make this work, I trialled a couple of different recording programs. I had tried to use the 'game capture' feature (that captures directly from the game itself, rather than just whatever is on the screen) but wasn't able to make that work in any of the programs I tried.

Mechwarrior 3 is from that weird time right after DOS and has major problems with running on current systems. It takes a bit of effort to get running properly and doesn't play nicely with modern things!

LotBlind

2022-07-21 09:35:09 am

I'll give proper feedback later but the showing the desktop/taskbar thing is a common issue when games change resolutions. (You guys should see Deus Ex Invisible War! That game literally has to reboot itself to start a new level.) So anyway we've accepted that kind of thing in the past so probably not an issue here either.

PROX

2022-07-26 05:05:56 am

No watermarks, gameplay was pretty clean. Shame about OP4M1. OP2M2 was really good with all the headshots. OP4M3 was really nice as well with the elemental handling. Overall, incredibly solid run. My only gripe is how quiet the audio is, but they will definitely fix that if they accept it.

Overall, I don't see a reason as to why they wouldn't accept it this time. The watermark earlier made some sense, but I really don't think they have an excuse not to this time lol

Falconer Gray

2022-07-26 06:27:52 pm

Thanks Prox.

The audio annoys me as well - that's actually the real volume that I get when playing. All volume settings are maxed, no idea of what causes it to be so low.

Your comment about OP4M1, is there something up with the run, or is it more just about the nature of the mission itself? Figuring out the right strat balance was an interesting challenge. I can get there sooner with a faster mech, but then can't get as close to the train because I wouldn't be able to avoid the enemy mechs. Going heavier means I wouldn't have to worry about them, but then I wouldn't reach the train in time.
I trialled a bunch of different ways to approach this before landing on the end result. I enjoyed doing it. Funny how much work goes into literal seconds of gameplay - but that's speedrunning!

PROX

2022-07-27 12:21:19 am

Quote from Falconer Gray:

Thanks Prox.

The audio annoys me as well - that's actually the real volume that I get when playing. All volume settings are maxed, no idea of what causes it to be so low.

Your comment about OP4M1, is there something up with the run, or is it more just about the nature of the mission itself? Figuring out the right strat balance was an interesting challenge. I can get there sooner with a faster mech, but then can't get as close to the train because I wouldn't be able to avoid the enemy mechs. Going heavier means I wouldn't have to worry about them, but then I wouldn't reach the train in time.
I trialled a bunch of different ways to approach this before landing on the end result. I enjoyed doing it. Funny how much work goes into literal seconds of gameplay - but that's speedrunning!

When you fell off the mountain before the second relay station. Everything else was solid for that mission. When you fell off the mountain, was it intentional?

Falconer Gray

2022-07-27 05:03:23 pm

Quote from PROX:

When you fell off the mountain before the second relay station. Everything else was solid for that mission. When you fell off the mountain, was it intentional?

Ah yep, I know what you're referring to. It wasn't intentional, but it didn't cost any time. As it happened I realised it meant I couldn't receive fire from the mechs on the right, so it's something I might actually look at using intentionally in later runs.

Worn_Traveler

2022-08-02 07:55:21 am

Formerly known as Skullboy

A/V is fine. No cheating detected.

As this run is apparently segmented and on the same difficulty as the current SDA run, I've watched the currently published run and provided mission times to compare and contrast between the two runs.

SDA Published Run Run in Verification
O1M1: 00:01:50 O1M1: 00:01:41
O1M2: 00:02:13 O1M2: 00:01:16
O1M3: 00:03:26 O1M3: 00:03:22
O1M4: 00:01:55 O1M4: 00:00:40
O2M1: 00:02:14 O2M1: 00:02:01
O2M2: 00:02:59 O2M2: 00:01:29
O2M3: 00:00:53 O2M3: 00:00:42
O2M4: 00:01:24 O2M4: 00:00:36
O3M1: 00:03:00 O3M1: 00:02:56
O3M2: 00:02:21 O3M2: 00:01:27
O3M3: 00:03:06 O3M3: 00:02:21
O3M4: 00:01:53 O3M4: 00:01:02
O3M5: 00:01:13 O3M5: 00:01:00
O3M6: 00:01:47 O3M6: 00:01:02
O4M1: 00:03:31 O4M1: 00:03:09
O4M2: 00:01:13 O4M2: 00:00:35
O4M3: 00:03:27 O4M3: 00:02:26
O4M4: 00:01:25 O4M4: 00:00:44
O4M5: 00:02:02 O4M5: 00:01:32
O4M6: 00:01:01 O4M6: 00:00:42
Total: 40+ minutes Total: 30+ minutes

This run in verification crushes the previous run by using better strategies and being more aggressive.

I think the current run is listed as 42:53... I wonder if the time spent on the salvage menu (which the submitted run either skipped or didn't record) gave the extra time there.

Regardless, this run should obsolete the current one barring any problems

Accept

A/V is fine. No cheating detected.

As this run is apparently segmented and on the same difficulty as the current SDA run, I've watched the currently published run and provided mission times to compare and contrast between the two runs.

SDA Published Run	Run in Verification
O1M1: 00:01:50	O1M1: 00:01:41
O1M2: 00:02:13	O1M2: 00:01:16
O1M3: 00:03:26	O1M3: 00:03:22
O1M4: 00:01:55	O1M4: 00:00:40
O2M1: 00:02:14	O2M1: 00:02:01
O2M2: 00:02:59	O2M2: 00:01:29
O2M3: 00:00:53	O2M3: 00:00:42
O2M4: 00:01:24	O2M4: 00:00:36
O3M1: 00:03:00	O3M1: 00:02:56
O3M2: 00:02:21	O3M2: 00:01:27
O3M3: 00:03:06	O3M3: 00:02:21 
O3M4: 00:01:53	O3M4: 00:01:02
O3M5: 00:01:13	O3M5: 00:01:00
O3M6: 00:01:47	O3M6: 00:01:02
O4M1: 00:03:31	O4M1: 00:03:09
O4M2: 00:01:13 	O4M2: 00:00:35
O4M3: 00:03:27	O4M3: 00:02:26
O4M4: 00:01:25	O4M4: 00:00:44
O4M5: 00:02:02	O4M5: 00:01:32
O4M6: 00:01:01	O4M6: 00:00:42
Total: 40+ minutes	Total: 30+ minutes

This run in verification crushes the previous run by using better strategies and being more aggressive.

I think the current run is listed as 42:53... I wonder if the time spent on the salvage menu (which the submitted run either skipped or didn't record) gave the extra time there.

Regardless, this run should obsolete the current one barring any problems

Accept

SDAVerification

2022-08-02 03:01:27 pm

Decision posted.

LotBlind

2022-08-03 01:16:34 pm

The decision here isn't going to change based on my feedback but looks like this was closed a little early this time. Thanks for everyone who verified it including Worn Traveler with the time comparisons. Maybe you could even work that table into your comments so everyone who saw the old run knows how much time you saved in each mission?

A/V: Was there any talk about the framerate? Briefly glancing at forums I can see this game benefits from it being limited? And that's why it's a steady 25?

I see you're playing on "hard" difficulty. Since it's not the default setting (assuming), I just edited it into the run entry so it'll show up on the page like the old run had it. Are the other settings in the options menu that you show the defaults or did you for any reason deviate from those? Might want to mention that you did/didn't in the comments. And maybe someone would like to know how the difficulty affects how the game works if you wanted to address that too.

You're only runner I can remember using "we" to describe the action in the notes. Very professional. The only step that's even left to take along that path is to start a strategy guide *wink wink*. Tried to make the template as easy to use as could possibly be without codework being involved.

Also before I forget (I may have pointed you at this already), there's this challenge that PROX found of interest so maybe it'll amuse you as well.

C1M4: Wow, that firefly is like a velociraptor.

C2M1: Can't believe you can dupe the enemy like that with nav points! Nice jump shots.

C3M2: "One in a million shot!" Did someone just perform... the Holdo Maneuver?

"As the story goes, there are other friendly mechs on the planet in a similar situation to us. One of them is carrying artillery weapons, and is happy to fire off a few shots when we call for it." That's a pretty neat piece of world building... or world destroying I guess. And the glitch where the mech bulldozes the building on its own is interesting too.

C3M6: So nowhere else you could have used the trick where you set a nav point and enemies attack it to destroy some building faster? I suppose it's usually more than one building that needs destroying in the same area. Did you see if other missions have the same restart trick where some enemy mech can bulldoze a target down if you just set the nav point next to it?

C4M1: I suppose this could be topped by launching the missiles before the thing has even spawned. If you went to some location on the hills next to the tracks (with a lighter mech maybe) and used the time you have to set up a dumbfire shot from a longer distance, could you in principle kill the train a second faster?

C4M5: I think your comments cut off here after where it says "Getting stuck on the terrain".

Yeah so at any rate, even if there was something in what I said that could improve on some mission, this is already more than up-to-par, so let me spell it loud and clear

*ACCEPT*

The decision here isn't going to change based on my feedback but looks like this was closed a little early this time. Thanks for everyone who verified it including Worn Traveler with the time comparisons. Maybe you could even work that table into your comments so everyone who saw the old run knows how much time you saved in each mission?

A/V: Was there any talk about the framerate? Briefly glancing at forums I can see this game benefits from it being limited? And that's why it's a steady 25?

I see you're playing on "hard" difficulty. Since it's not the default setting (assuming), I just edited it into the run entry so it'll show up on the page like the old run had it. Are the other settings in the options menu that you show the defaults or did you for any reason deviate from those? Might want to mention that you did/didn't in the comments. And maybe someone would like to know how the difficulty affects how the game works if you wanted to address that too.

You're only runner I can remember using "we" to describe the action in the notes. Very professional. The only step that's even left to take along that path is to start a [url=https://kb.speeddemosarchive.com/Category:Games]strategy guide[/url] *wink wink*. Tried to make the template as easy to use as could possibly be without codework being involved.

Also before I forget (I may have pointed you at this already), there's [url=https://forum.speeddemosarchive.com/post/challenge_for_rts_runners_especially_fastest_controller_down_13.html]this challenge[/url] that PROX found of interest so maybe it'll amuse you as well.

C1M4: Wow, that firefly is like a velociraptor.

C2M1: Can't believe you can dupe the enemy like that with nav points! Nice jump shots.

C3M2: "One in a million shot!" Did someone just perform... the Holdo Maneuver?

"As the story goes, there are other friendly mechs on the planet in a similar situation to us. One of them is carrying artillery weapons, and is happy to fire off a few shots when we call for it." That's a pretty neat piece of world building... or world destroying I guess. And the glitch where the mech bulldozes the building on its own is interesting too.

C3M6: So nowhere else you could have used the trick where you set a nav point and enemies attack it to destroy some building faster? I suppose it's usually more than one building that needs destroying in the same area. Did you see if other missions have the same restart trick where some enemy mech can bulldoze a target down if you just set the nav point next to it?

C4M1: I suppose this could be topped by launching the missiles before the thing has even spawned. If you went to some location on the hills next to the tracks (with a lighter mech maybe) and used the time you have to set up a dumbfire shot from a longer distance, could you in principle kill the train a second faster?

C4M5: I think your comments cut off here after where it says "Getting stuck on the terrain".

Yeah so at any rate, even if there was something in what I said that could improve on some mission, this is already more than up-to-par, so let me spell it loud and clear

*ACCEPT*

Falconer Gray

2022-08-04 03:48:25 am

Thank you all for the comments, and for supporting speedrunning in this way.

Quote from LotBlind:

The decision here isn't going to change based on my feedback but looks like this was closed a little early this time. Thanks for everyone who verified it including Worn Traveler with the time comparisons. Maybe you could even work that table into your comments so everyone who saw the old run knows how much time you saved in each mission?

A/V: Was there any talk about the framerate? Briefly glancing at forums I can see this game benefits from it being limited? And that's why it's a steady 25?

I see you're playing on "hard" difficulty. Since it's not the default setting (assuming), I just edited it into the run entry so it'll show up on the page like the old run had it. Are the other settings in the options menu that you show the defaults or did you for any reason deviate from those? Might want to mention that you did/didn't in the comments. And maybe someone would like to know how the difficulty affects how the game works if you wanted to address that too.

You're only runner I can remember using "we" to describe the action in the notes. Very professional. The only step that's even left to take along that path is to start a strategy guide *wink wink*. Tried to make the template as easy to use as could possibly be without codework being involved.

Also before I forget (I may have pointed you at this already), there's this challenge that PROX found of interest so maybe it'll amuse you as well.

C1M4: Wow, that firefly is like a velociraptor.

C2M1: Can't believe you can dupe the enemy like that with nav points! Nice jump shots.

C3M2: "One in a million shot!" Did someone just perform... the Holdo Maneuver?

"As the story goes, there are other friendly mechs on the planet in a similar situation to us. One of them is carrying artillery weapons, and is happy to fire off a few shots when we call for it." That's a pretty neat piece of world building... or world destroying I guess. And the glitch where the mech bulldozes the building on its own is interesting too.

C3M6: So nowhere else you could have used the trick where you set a nav point and enemies attack it to destroy some building faster? I suppose it's usually more than one building that needs destroying in the same area. Did you see if other missions have the same restart trick where some enemy mech can bulldoze a target down if you just set the nav point next to it?

C4M1: I suppose this could be topped by launching the missiles before the thing has even spawned. If you went to some location on the hills next to the tracks (with a lighter mech maybe) and used the time you have to set up a dumbfire shot from a longer distance, could you in principle kill the train a second faster?

C4M5: I think your comments cut off here after where it says "Getting stuck on the terrain".

Yeah so at any rate, even if there was something in what I said that could improve on some mission, this is already more than up-to-par, so let me spell it loud and clear

*ACCEPT*

Great feedback LB.

The framerate is important, in that the physics engine is tied to the framerate. One of the issues people have with modern systems, is things going completely haywire when running at a higher framerate, such as mechs and vehicles flying through the air as if they were weightless. Good old 1990's programming I guess.

My writing style is based on wanting to bring the reader along for the journey, hoping to make it somewhat immersive. Might also be some crossover from my day job - I author documents for a living (procedures, guidelines, FAQs, etc).

The "One in a million shot" is just a random line of dialog, which then gets a chuckle from me as I have had it said multiple times in the one mission.

There's a lot of details about how the game works that still isn't understood. Using the enemy AI to attack targets seems straightforward, but then there are also instances where they just don't do it at all. For example, in O4M4, the two mechs guarding the final objective turn to face the target, but they never attack it.
The scenario in O3M6 is something of a perfect storm, where everything falls into place to make this strategy viable. It's a small, single target rather than multiple buildings, and the enemy mech is in the perfect location.
Off the top of my head, I can't think of another mission where this trick would be quicker than the player doing it themselves. At least, not in any reliable way. I've tried to use the AI (both friendly and enemy) more in other missions (such as O4M2) but it never worked out for me.

The restart trick is hard to pin down, as every single mission in this run involved restarts, so it's difficult to say where exactly the differences might occur. I realised what was happening in O4M2, as destruction of the buildings is early enough in the mission that there normally will have been no need to restart at that point. My previous strategy on this mission was to use lasers, until by chance I noticed a friendly mech do a lot of damage with their missiles. This then led to me trialling a missile loadout of my own, which worked to great effect. However, when I then reached this mission on this current speedrun and made my first attempt at it, this loadout was very ineffective. Puzzled, I restarted the mission and everything worked as expected. From there, I realised something was going on and was able to replicate this by launching the mission from the main menu and then replaying from the in-game menu.

I was using the same trick in O3M4 without realising it was happening. It was only after the above discovery that I was able to understand it better. The building destruction in O3M4 happens because an enemy mech spawns inside one of the target buildings. When it activates, it moves forward and bumps into the structure. This collision damage is what destroys the objective for us, but only if the game has been restarted. I knew the AI was destroying the building, but I had thought it was due to the nav point trick. Once the restart trick was found I was then able to understand that's actually what was happening.

Dumbfiring at the train before it spawns would be possible in theory, but the shot would be extremely complex to calculate and pull off.
My strategy was to get as close to the train as possible, so as to reduce the flight time of the missiles.
The train spawns in at 3:00, my missiles hit at 3:02. So there is the possibility of a timesave by closing the gap between those two points, but pulling it off would be extremely tight.

But of course, to all those points - never say never, I hope that someone is able to improve on these times one day.

I've updated the commentary file to reflect the points above.

Cheers all!

Thank you all for the comments, and for supporting speedrunning in this way.

[quote="LotBlind"]The decision here isn't going to change based on my feedback but looks like this was closed a little early this time. Thanks for everyone who verified it including Worn Traveler with the time comparisons. Maybe you could even work that table into your comments so everyone who saw the old run knows how much time you saved in each mission?

A/V: Was there any talk about the framerate? Briefly glancing at forums I can see this game benefits from it being limited? And that's why it's a steady 25?

I see you're playing on "hard" difficulty. Since it's not the default setting (assuming), I just edited it into the run entry so it'll show up on the page like the old run had it. Are the other settings in the options menu that you show the defaults or did you for any reason deviate from those? Might want to mention that you did/didn't in the comments. And maybe someone would like to know how the difficulty affects how the game works if you wanted to address that too.

You're only runner I can remember using "we" to describe the action in the notes. Very professional. The only step that's even left to take along that path is to start a [url=https://kb.speeddemosarchive.com/Category:Games]strategy guide[/url] *wink wink*. Tried to make the template as easy to use as could possibly be without codework being involved.

Also before I forget (I may have pointed you at this already), there's [url=https://forum.speeddemosarchive.com/post/challenge_for_rts_runners_especially_fastest_controller_down_13.html]this challenge[/url] that PROX found of interest so maybe it'll amuse you as well.

C1M4: Wow, that firefly is like a velociraptor.

C2M1: Can't believe you can dupe the enemy like that with nav points! Nice jump shots.

C3M2: "One in a million shot!" Did someone just perform... the Holdo Maneuver?

"As the story goes, there are other friendly mechs on the planet in a similar situation to us. One of them is carrying artillery weapons, and is happy to fire off a few shots when we call for it." That's a pretty neat piece of world building... or world destroying I guess. And the glitch where the mech bulldozes the building on its own is interesting too.

C3M6: So nowhere else you could have used the trick where you set a nav point and enemies attack it to destroy some building faster? I suppose it's usually more than one building that needs destroying in the same area. Did you see if other missions have the same restart trick where some enemy mech can bulldoze a target down if you just set the nav point next to it?

C4M1: I suppose this could be topped by launching the missiles before the thing has even spawned. If you went to some location on the hills next to the tracks (with a lighter mech maybe) and used the time you have to set up a dumbfire shot from a longer distance, could you in principle kill the train a second faster?

C4M5: I think your comments cut off here after where it says "Getting stuck on the terrain".

Yeah so at any rate, even if there was something in what I said that could improve on some mission, this is already more than up-to-par, so let me spell it loud and clear

*ACCEPT*[/quote]

Great feedback LB.

The framerate is important, in that the physics engine is tied to the framerate. One of the issues people have with modern systems, is things going completely haywire when running at a higher framerate, such as mechs and vehicles flying through the air as if they were weightless. Good old 1990's programming I guess.

My writing style is based on wanting to bring the reader along for the journey, hoping to make it somewhat immersive. Might also be some crossover from my day job - I author documents for a living (procedures, guidelines, FAQs, etc).

The "One in a million shot" is just a random line of dialog, which then gets a chuckle from me as I have had it said multiple times in the one mission.

There's a lot of details about how the game works that still isn't understood. Using the enemy AI to attack targets seems straightforward, but then there are also instances where they just don't do it at all. For example, in O4M4, the two mechs guarding the final objective turn to face the target, but they never attack it. 
The scenario in O3M6 is something of a perfect storm, where everything falls into place to make this strategy viable. It's a small, single target rather than multiple buildings, and the enemy mech is in the perfect location. 
Off the top of my head, I can't think of another mission where this trick would be quicker than the player doing it themselves. At least, not in any reliable way. I've tried to use the AI (both friendly and enemy) more in other missions (such as O4M2) but it never worked out for me.

The restart trick is hard to pin down, as every single mission in this run involved restarts, so it's difficult to say where exactly the differences might occur. I realised what was happening in O4M2, as destruction of the buildings is early enough in the mission that there normally will have been no need to restart at that point. My previous strategy on this mission was to use lasers, until by chance I noticed a friendly mech do a lot of damage with their missiles. This then led to me trialling a missile loadout of my own, which worked to great effect. However, when I then reached this mission on this current speedrun and made my first attempt at it, this loadout was very ineffective. Puzzled, I restarted the mission and everything worked as expected. From there, I realised something was going on and was able to replicate this by launching the mission from the main menu and then replaying from the in-game menu.

I was using the same trick in O3M4 without realising it was happening. It was only after the above discovery that I was able to understand it better. The building destruction in O3M4 happens because an enemy mech spawns inside one of the target buildings. When it activates, it moves forward and bumps into the structure. This collision damage is what destroys the objective for us, but only if the game has been restarted. I knew the AI was destroying the building, but I had thought it was due to the nav point trick. Once the restart trick was found I was then able to understand that's actually what was happening.

Dumbfiring at the train before it spawns would be possible in theory, but the shot would be extremely complex to calculate and pull off. 
My strategy was to get as close to the train as possible, so as to reduce the flight time of the missiles. 
The train spawns in at 3:00, my missiles hit at 3:02. So there is the possibility of a timesave by closing the gap between those two points, but pulling it off would be extremely tight.

But of course, to all those points - never say never, I hope that someone is able to improve on these times one day.

I've updated the commentary file to reflect the points above.

Cheers all!

LotBlind

2022-08-05 09:13:41 am

Framerate: I figured that might be the case here. Same as MechWarrior 2 then.

One in a million: maybe the run as a whole is that?

Quote:

For example, in O4M4, the two mechs guarding the final objective turn to face the target, but they never attack it.

Whenever you have something like that, typically it's because the AI is not in its normal state at the time so it doesn't charge out or otherwise reveal itself before the designers intended. Meanwhile, making sure you can't easily blindside those mechs by just approaching from behind might be why they're still turning to face the closest target. That's my first intuition. If so, might wanna check what the trigger exactly is to make them activate. Probably just proximity or line of sight or specific trigger boxes. Maybe after that they CAN be gotten to attack the nav point target? I'm sure this is all too easy to have been overlooked here though.

Mission restart: Weird weird weird. Needless to say, the restart code is slightly different from the just-start code. Or something.

Framerate: I figured that might be the case here. Same as MechWarrior 2 then.

One in a million: maybe the run as a whole is that?

[quote]For example, in O4M4, the two mechs guarding the final objective turn to face the target, but they never attack it. [/quote]

Whenever you have something like that, typically it's because the AI is not in its normal state at the time so it doesn't charge out or otherwise reveal itself before the designers intended. Meanwhile, making sure you can't easily blindside those mechs by just approaching from behind might be why they're still turning to face the closest target. That's my first intuition. If so, might wanna check what the trigger exactly is to make them activate. Probably just proximity or line of sight or specific trigger boxes. Maybe after that they CAN be gotten to attack the nav point target? I'm sure this is all too easy to have been overlooked here though.

Mission restart: Weird weird weird. Needless to say, the restart code is slightly different from the just-start code. Or something.

Falconer Gray

2022-08-06 05:03:15 pm

Quote from LotBlind:

Whenever you have something like that, typically it's because the AI is not in its normal state at the time so it doesn't charge out or otherwise reveal itself before the designers intended. Meanwhile, making sure you can't easily blindside those mechs by just approaching from behind might be why they're still turning to face the closest target. That's my first intuition. If so, might wanna check what the trigger exactly is to make them activate. Probably just proximity or line of sight or specific trigger boxes. Maybe after that they CAN be gotten to attack the nav point target? I'm sure this is all too easy to have been overlooked here though.

Mission restart: Weird weird weird. Needless to say, the restart code is slightly different from the just-start code. Or something.

The mechs in O4M4 will stay put even after I've been attacking them, as long as the nav point is closer to them than I am. They activate by proximity (I should say activate rather than spawn, all the mechs loaded into the game when the mission starts, but they are dormant until triggered) - sometimes depending on speed or path taken I can see them turning around to face the nav point as I'm jumping up the structure. They have line of sight on me, but turn to the nav instead and sit there doing nothing. Very strange.

As for the restart, weird indeed. I don't like not actually knowing what's happening, but everything to do with that is basically guesswork an assumption on my end!
The game carries *some* information over when restarting a mission. If explosions (from missiles, etc) have left craters in the ground, those craters will stay there for each time you restart a mission. I've had speedruns where I've had to back out to the menu because after a high number of restarts, the ground is so bumpy from craters that it impacts my movement.

But there are obviously other things that are supposed to reset. Buildings, enemy mechs etc. If they're destroyed and you replay the mission, they're all reset properly.
So we can assume that the damage thing is from something carrying over or changing that wasn't supposed to.
I don't know why it uses this method - maybe so that the load times are shorter? That is a noticeable difference (when playing a mission for the first time, it takes a while to load everything up. When replaying, you can jump into the mission much quicker).

Another detail is that the game seems to carry over certain things from the last frame when restarting. When restarting a mission, if there was previously smoke or other particle effects on the screen, they'll still be there for a moment. Missile plumes, are another. You can restart a mission and for a moment appear to see missiles, which then disappear.
It happens with sound as well, where I've restarted missions and had a laser firing sound play immediately when no shot was fired. Sometimes even in the menu, which is weird.

Here are some videos I made to demonstrate this to the MW3 discord. They show off the village destruction in O4M2, the fortress in O3M4, and then the carry over explosion in O4M5.

[quote="LotBlind"]Whenever you have something like that, typically it's because the AI is not in its normal state at the time so it doesn't charge out or otherwise reveal itself before the designers intended. Meanwhile, making sure you can't easily blindside those mechs by just approaching from behind might be why they're still turning to face the closest target. That's my first intuition. If so, might wanna check what the trigger exactly is to make them activate. Probably just proximity or line of sight or specific trigger boxes. Maybe after that they CAN be gotten to attack the nav point target? I'm sure this is all too easy to have been overlooked here though.

Mission restart: Weird weird weird. Needless to say, the restart code is slightly different from the just-start code. Or something.

[/quote]

The mechs in O4M4 will stay put even after I've been attacking them, as long as the nav point is closer to them than I am. They activate by proximity (I should say activate rather than spawn, all the mechs loaded into the game when the mission starts, but they are dormant until triggered) - sometimes depending on speed or path taken I can see them turning around to face the nav point as I'm jumping up the structure. They have line of sight on me, but turn to the nav instead and sit there doing nothing. Very strange.

As for the restart, weird indeed. I don't like not actually knowing what's happening, but everything to do with that is basically guesswork an assumption on my end!
The game carries *some* information over when restarting a mission. If explosions (from missiles, etc) have left craters in the ground, those craters will stay there for each time you restart a mission. I've had speedruns where I've had to back out to the menu because after a high number of restarts, the ground is so bumpy from craters that it impacts my movement.

But there are obviously other things that are supposed to reset. Buildings, enemy mechs etc. If they're destroyed and you replay the mission, they're all reset properly. 
So we can assume that the damage thing is from something carrying over or changing that wasn't supposed to. 
I don't know why it uses this method - maybe so that the load times are shorter? That is a noticeable difference (when playing a mission for the first time, it takes a while to load everything up. When replaying, you can jump into the mission much quicker).

Another detail is that the game seems to carry over certain things from the last frame when restarting. When restarting a mission, if there was previously smoke or other particle effects on the screen, they'll still be there for a moment. Missile plumes, are another. You can restart a mission and for a moment appear to see missiles, which then disappear. 
It happens with sound as well, where I've restarted missions and had a laser firing sound play immediately when no shot was fired. Sometimes even in the menu, which is weird.

Here are some videos I made to demonstrate this to the MW3 discord. They show off the village destruction in O4M2, the fortress in O3M4, and then the carry over explosion in O4M5.

https://youtu.be/v4Q7fG1PNSg
https://youtu.be/xMDf2RwTR8E

LotBlind

2022-08-10 01:03:46 pm

Craters: For what it's worth, the much shorter load times sounds like a good theory as to why they would have left it like that. They just figured it wasn't enough of a difference in casual play. I've noticed that whenever a game has carryover from mission to mission, or e.g. an RPG has the carryover of all the items, doors, enemies etc. in their old positions whenever you revisit the same area, the amount of data to store in save files grows very very fast. Divine Divinity is one such game. This game also seems to have a carryover as evidenced by how you destroy stuff ahead of time in some missions. I don't know if there's some link there as I've lost my train of thought. But yeah, loading times are always a concern.

As for the splash damage glitch. I'm thinking something along the lines "the buildings have specific hitboxes that aren't the same as the shapes that they're built out of; after a restart, these hitboxes are abandoned for whatever reason in favor of a default hitbox". The DOS game I'm studying myself has had so many inscrutable things about its behavior of the kind I can't see myself having had any chances of ever getting to the bottom of without the guy I'm working with (a coder) having studied it on the code level that my confidence in my theorycrafting prowess has probably net dropped since the beginning. One relatively common theme in the glitches I've seen is stuff reverting to a default behavior from a more specific one that was set by a piece of code that's no longer being executed.

In case my guess about the hitboxes being affected was right, you might be able to see a difference if you shoot at the buildings with non-explosive weapons too. Or if the explosions happen further away from them. Maybe the game tracks damage dealt per-polygon instead of using hitboxes though? Seems like the kinda game that does that. Otherwise maybe the buildings go from not hurting each other into damaging each other as they explode (maybe there's a value that determines the damage dealt to surrounding things). Maybe they lose their ability to block the explosion for some reason.

Or are you saying the LRM splash damage is what's changed? In that case do you see an effect on other targets as well? If it is the splash damage that's increasing, you should be able to see that in a RAM watch utility if you knew what the initial damage was. In theory. Somehow.

Did you try bringing the kind of behavior on generic DOS forums as well just in case there's a wiz kid in the house?

BTW: Using the replay bug to your advantage might prompt a "NG+" category? Can you fire off lots of missiles at targets and make them blow up at the start of the mission?

Craters: For what it's worth, the much shorter load times sounds like a good theory as to why they would have left it like that. They just figured it wasn't enough of a difference in casual play. I've noticed that whenever a game has carryover from mission to mission, or e.g. an RPG has the carryover of all the items, doors, enemies etc. in their old positions whenever you revisit the same area, the amount of data to store in save files grows very very fast. Divine Divinity is one such game. This game also seems to have a carryover as evidenced by how you destroy stuff ahead of time in some missions. I don't know if there's some link there as I've lost my train of thought. But yeah, loading times are always a concern.

As for the splash damage glitch. I'm thinking something along the lines "the buildings have specific hitboxes that aren't the same as the shapes that they're built out of; after a restart, these hitboxes are abandoned for whatever reason in favor of a default hitbox". The DOS game I'm studying myself has had so many inscrutable things about its behavior of the kind I can't see myself having had any chances of ever getting to the bottom of without the guy I'm working with (a coder) having studied it on the code level that my confidence in my theorycrafting prowess has probably net dropped since the beginning. One relatively common theme in the glitches I've seen is stuff reverting to a default behavior from a more specific one that was set by a piece of code that's no longer being executed.

In case my guess about the hitboxes being affected was right, you might be able to see a difference if you shoot at the buildings with non-explosive weapons too. Or if the explosions happen further away from them. Maybe the game tracks damage dealt per-polygon instead of using hitboxes though? Seems like the kinda game that does that. Otherwise maybe the buildings go from not hurting each other into damaging each other as they explode (maybe there's a value that determines the damage dealt to surrounding things). Maybe they lose their ability to block the explosion for some reason.

Or are you saying the LRM splash damage is what's changed? In that case do you see an effect on other targets as well? If it is the splash damage that's increasing, you should be able to see that in a RAM watch utility if you knew what the initial damage was. In theory. Somehow.

Did you try bringing the kind of behavior on generic DOS forums as well just in case there's a wiz kid in the house?

BTW: Using the replay bug to your advantage might prompt a "NG+" category? Can you fire off lots of missiles at targets and make them blow up at the start of the mission?

Falconer Gray

2022-08-12 02:30:06 am

To clarify, the missiles aren't actually there on the restart. It's like a "ghost" frame or something, in that you see them as they were when the mission was restarted, but then they aren't actually there and they disappear immediately. So the game does wipe them clean on the restart, but does part of that during the first frame of the new mission? Something like that. All guessing though.

The splash damage is a suggestion of mine because it doesn't seem as if the hit points change. It still takes the same amount of shots from lasers to take out the buildings, but the explosive damage from the missiles has a much greater effect. I don't know any forums focusing on this era of games, but will keep an eye out. The MW3 discord didn't come up with anything concrete.

BTW I also did some further testing with exploiting the AI through the use of nav points and learnt a little more. When ordering the friendly AI to a nav point (or point on the ground), the enemy AI recognises this as a target. However, if that point is something that has already been designated as being on the enemy team, the enemy AI will not fire on it. This is why the two final mechs in O4M4 do not attack the objective they are guarding. The nav point and the objective are the same. When I order lancemates to the objective, the enemy AI then recognise it as a target and then turn to face it. But they do not fire on it, presumably because they are programmed not to attack their own team.

Where this differs from O3M6 (where we do use the enemy AI to destroy an objective) is that the nav point in that mission is NOT directly on top of the target structure, but is close enough that the splash damage from the missiles can destroy it. (That target has very low HP, whereas the target in O4M4 is much higher).

To clarify, the missiles aren't actually there on the restart. It's like a "ghost" frame or something, in that you see them as they were when the mission was restarted, but then they aren't actually there and they disappear immediately. So the game does wipe them clean on the restart, but does part of that during the first frame of the new mission? Something like that. All guessing though.

The splash damage is a suggestion of mine because it doesn't seem as if the hit points change. It still takes the same amount of shots from lasers to take out the buildings, but the explosive damage from the missiles has a much greater effect. I don't know any forums focusing on this era of games, but will keep an eye out. The MW3 discord didn't come up with anything concrete.

BTW I also did some further testing with exploiting the AI through the use of nav points and learnt a little more. When ordering the friendly AI to a nav point (or point on the ground), the enemy AI recognises this as a target. However, if that point is something that has already been designated as being on the enemy team, the enemy AI will not fire on it. This is why the two final mechs in O4M4 do not attack the objective they are guarding. The nav point and the objective are the same. When I order lancemates to the objective, the enemy AI then recognise it as a target and then turn to face it. But they do not fire on it, presumably because they are programmed not to attack their own team.

Where this differs from O3M6 (where we do use the enemy AI to destroy an objective) is that the nav point in that mission is NOT directly on top of the target structure, but is close enough that the splash damage from the missiles can destroy it. (That target has very low HP, whereas the target in O4M4 is much higher).

LotBlind

2022-08-12 02:11:07 pm

Right so the missiles that are left over from the last attempt can't damage anything? I thought maybe in the second video above they actually hit that target.

DOS forums: The only one I know of off the top of my head might be the DOSBox forum since there may be technically skilled people there.

Nav point targeting: And you can't send the AI to an arbitrary point near the target but not exactly on top of it? Maybe one of their lasers or something could still hit it even if their shots mostly miss.

Falconer Gray

2022-08-13 01:35:44 am

Sorry, I should have explained that second video better. There are no missiles in that one. When a mech falls into the lava, it overheats and explodes. The explosion is large enough to destroy objectives and other mechs. What is happening is that the mission is being reset after the explosion is triggered but before it actually happens. So when the mission restarts, the explosion then happens in the new mission.

Re the nav points, that would be the way to make it happen and I did trial exactly that, but it's far too time consuming. Those mechs aren't carrying enough splash damage weapons for it to be a quick destruction, and setting the nav point on the exact right point is too specific than just using the preset nav point and doing as per the existing run. Also given that they don't activate until the player is close, i don't think it would be any faster. But who knows in the future!

LotBlind

2022-08-13 12:41:14 pm

So does the explosion that happens in the second video actually impacting the mission or is it just cosmetic? Can you abuse things like that?

Okay, I think my armchair wisdom is running dry here... but good luck whatever runs you chase after next.

Falconer Gray

2022-08-13 06:20:06 pm

Yes it does impact mission objectives. The messages that appear at the bottom of the screen after the explosion show that objectives are being completed.
I absolutely was trying to abuse it for this run, but couldn't figure out a way to make it quicker than just doing the mission normally.

In that mission, we need destroy 4 towers, 3 mechs and 12 elementals. I was able to use the explosion to destroy about a third of the required objectives, leaving the rest to be destroyed normally. I was calculating the run time for this as being the mission time it takes to jump into the lava PLUS the mission time on the replay to destroy everything else. Adding these times together was actually taking longer than just doing it all in one go, as I wasn't able to be sure where the remaining mechs and elementals would be. It was a lot harder to predict and plan for.

Having said all that, I do think there's a possibility for future speedrunning use. If you could find a way to have it destroy more of the objectives, or be more consistent in the outcome, then it could be useful.

Yes it does impact mission objectives. The messages that appear at the bottom of the screen after the explosion show that objectives are being completed. 
I absolutely was trying to abuse it for this run, but couldn't figure out a way to make it quicker than just doing the mission normally.

In that mission, we need destroy 4 towers, 3 mechs and 12 elementals. I was able to use the explosion to destroy about a third of the required objectives, leaving the rest to be destroyed normally. I was calculating the run time for this as being the mission time it takes to jump into the lava PLUS the mission time on the replay to destroy everything else. Adding these times together was actually taking longer than just doing it all in one go, as I wasn't able to be sure where the remaining mechs and elementals would be. It was a lot harder to predict and plan for.

Having said all that, I do think there's a possibility for future speedrunning use. If you could find a way to have it destroy more of the objectives, or be more consistent in the outcome, then it could be useful.

LotBlind

Edit history:

LotBlind: 2022-08-15 07:04:23 am

LotBlind: 2022-08-15 07:02:09 am

2022-08-15 07:01:51 am

Actually, that's not the way such (segmented) runs are timed! You don't include the first part where you set it up, only the time for the try where you finish the mission. This has many precedents and the reason is simple: it's not your fault if the game isn't perfectly coded. You're starting the mission and completing it, fair and square. The same argument as for using any glitchy exploits really. Funny enough, the game I mentioned earlier, Divine Divinity, has exactly this kind of exploit available to make the New Game start in a weird state (where the player e.g. is in a different location AIR). When you do such setups, it's good to include the video of the setup as part of the submission so everyone can see you're not cheating.

In fact, this makes it a bit confusing how this should be done... on the one hand you're already "exploiting" replaying missions in the sense that there's stuff that doesn't happen, as you've demonstrated, without first resetting it once. So it's already "NG+" in that sense. On the other hand those are exploits that are literally impossible to avoid so you're not really setting them up specifically. So there might be an argument for leaving this category the way it is and if you do anything more, such as trigger explosions at the end of the previous try, to speed it up more, that could then be how we define the NG+ category. Do you see such a definition leading into problems where it's not quite clear if some runner did or didn't gain something extra from the restart that doesn't usually happen?

Yeah, I can see planning and executing that kind of run would be a bigger challenge.

Actually, that's not the way such (segmented) runs are timed! You don't include the first part where you set it up, only the time for the try where you finish the mission. This has many precedents and the reason is simple: it's not your fault if the game isn't perfectly coded. You're starting the mission and completing it, fair and square. The same argument as for using any glitchy exploits really. Funny enough, the game I mentioned earlier, Divine Divinity, has exactly this kind of exploit available to make the New Game start in a weird state (where the player e.g. is in a different location AIR). When you do such setups, it's good to include the video of the setup as part of the submission so everyone can see you're not cheating.

In fact, this makes it a bit confusing how this should be done... on the one hand you're already "exploiting" replaying missions in the sense that there's stuff that doesn't happen, as you've demonstrated, without first resetting it once. So it's already "NG+" in that sense. On the other hand those are exploits that are literally impossible to avoid so you're not really setting them up specifically. So there might be an argument for leaving this category the way it is and if you do anything more, such as trigger explosions at the end of the previous try, to speed it up more, that could then be how we define the NG+ category. Do you see such a definition leading into problems where it's not quite clear if some runner did or didn't gain something extra from the restart that doesn't usually happen?

Yeah, I can see planning and executing that kind of run would be a bigger challenge.

Falconer Gray

Edit history:

Falconer Gray: 2022-08-16 07:39:44 pm

2022-08-16 07:38:10 pm

I think would be difficulty in determining what actually has or has not happened in a run. There's so much we don't understand about MW3 at the moment. There are some conversations going on at the discord about things like weapon fall off, but it's all speculation. We have observed that *sometimes* a weapon will do a little bit less damage than normal, but we don't know why.

There is also script relating to damage at max range, which we also don't quite understand fully. I'm told this script has a percentage value assigned to it, and the best guess that has been made is that this represents a percentage chance for the weapon to not do its full damage, but we still don't know exactly at what point. One suggestion has been that the percentage covers both range and damage fall. So if "damage at max range" was 75%, then being beyond 75% of the max range, there is only 75% chance of the weapon doing full damage. Again, all just speculation.

Where this relates to speedrunning and the replay feature I've exploiting, is that it then becomes hard to know what is actually a result of the replay bug and what isn't. I'm currently back working on single segment runs again and can give some examples of what I mean.

In O2M4, my segmented run sees the mech factory destroyed with a single shot of 12 ER Small Lasers. I've been unable to recreate this in a single segment run. But, I don't know for sure if this is because of the replay bug.
This is because achieving that result (destroy the factory with 12 ER Smalls) wasn't just a case of replaying the mission, I had to run that over and over and over to finally score the kill. It was extremely inconsistent and I can't make heads or tails of exactly why. I also had trouble with finding the right range, which is why you see me run right up to the factory when destroying it, which should be well inside the range of those lasers.
So maybe there is an intentional element of RNG involved and with luck, I would be able to score the same result on a single segment run, first attempt. But I don't know for sure.

There is also the fact that other missions may be impacted by this bug, but that I haven't been taking advantage of it. For example, maybe replaying a mission makes the building objective easier to destroy, but my lancemate (who destroys that building) is using enough firepower that it doesn't make a difference. This hypothetical example is something that could be taken advantage of by a different strategy, but we wouldn't recognise that until it actually happens.

I'll also say that I'm not at all experienced in speedrunning and don't have an understanding of how things like this are viewed and any prior examples that might set precedence. These are just my observations as I see them. What an interesting topic!

I think would be difficulty in determining what actually has or has not happened in a run. There's so much we don't understand about MW3 at the moment. There are some conversations going on at the discord about things like weapon fall off, but it's all speculation. We have observed that *sometimes* a weapon will do a little bit less damage than normal, but we don't know why.

There is also script relating to damage at max range, which we also don't quite understand fully. I'm told this script has a percentage value assigned to it, and the best guess that has been made is that this represents a percentage chance for the weapon to not do its full damage, but we still don't know exactly at what point. One suggestion has been that the percentage covers both range and damage fall. So if "damage at max range" was 75%, then being beyond 75% of the max range, there is only 75% chance of the weapon doing full damage. Again, all just speculation.

Where this relates to speedrunning and the replay feature I've exploiting, is that it then becomes hard to know what is actually a result of the replay bug and what isn't. I'm currently back working on single segment runs again and can give some examples of what I mean.

In O2M4, my segmented run sees the mech factory destroyed with a single shot of 12 ER Small Lasers. I've been unable to recreate this in a single segment run. But, I don't know for sure if this is because of the replay bug. 
This is because achieving that result (destroy the factory with 12 ER Smalls) wasn't just a case of replaying the mission, I had to run that over and over and over to finally score the kill. It was extremely inconsistent and I can't make heads or tails of exactly why. I also had trouble with finding the right range, which is why you see me run right up to the factory when destroying it, which should be well inside the range of those lasers. 
So maybe there is an intentional element of RNG involved and with luck, I would be able to score the same result on a single segment run, first attempt. But I don't know for sure.

There is also the fact that other missions may be impacted by this bug, but that I haven't been taking advantage of it. For example, maybe replaying a mission makes the building objective easier to destroy, but my lancemate (who destroys that building) is using enough firepower that it doesn't make a difference. This hypothetical example is something that could be taken advantage of by a different strategy, but we wouldn't recognise that until it actually happens.

I'll also say that I'm not at all experienced in speedrunning and don't have an understanding of how things like this are viewed and any prior examples that might set precedence. These are just my observations as I see them. What an interesting topic!