The Metal squeeze: How is it possible? Let's do the math!

[Ontwikseltsaar]

Today I was looking at the final squeeze in the optimal path for The Metal. This score has, to date, only been achieved once, by ukog-monkey. This last squeeze is particularly ridiculous so I wanted to know how much leeway there is to hit it.

Basically, what I did was this:

1. I used GHTCP to extract the expert chart for The Metal from the PC version of GH3.
   
2. I exported the chart, took the raw data and prettified it to be human-readable.
   
3. I calculated the timestamps of the last moment the frontend squeeze can be hit as well as the earliest moment the backend squeeze can be hit.
   
4. Based on the frontend activation, I calculated the timestamp of the moment when SP ends.
   
5. If (SP end time >= backend time), the squeeze is possible.

A crucial question here is: What is the exact timing window in GH3?
The only answer to this I could find by just googling the question (which included results from Score Hero) is this Reddit comment, which states the timing window for GH3 is 100ms. For the purposes of my calculations, I took a timing window of 100ms on either side of the note (so that would be a timing window of 200ms).

I have a Google sheet that contains the calculations with an editable timing window. I would share it here, but I believe there is a policy on Score Hero against publicizing the charts. I'll try to summarize the sheet:

• A measure consists of 4 beats. Every beat has a fret bar. Around the time of activation and the time when SP ends, the time that elapses between fret bars is 535ms or 536ms. This data can be found in the chart files (which is nice, because I expected that I'd have to do much more conversions between beats and time!).
  
• Initially, I calculated the amount of SP saved by activating 100ms late by taking the ratio 100/536 (= 0.187 beats of SP saved) and calculating how much time that would save when SP ends, but since the BPM at the beginning and end of SP is almost exactly the same, I realized that the ultimate squeeze (hitting the first note 100ms late) would save about 100ms worth of star power.
  
• Hitting the first note 100ms late and hitting the last note 100ms early, I still end up being about 134ms short of being able to hit the squeeze, which means we'd need a timing window of about 167ms (on either side of the note!) to be able to hit this. I calculated that star power ends 151057.25ms into the song, and the earliest the squeeze can be hit is 151058ms into the song! Assuming some minor rounding errors, this means the squeeze would indeed be possible if both the frontend squeeze (with the activation) and the backend squeeze are hit with 1ms precision!

In the process, I made the following assumptions:

• Notes are always at the center of their timing windows. This means that you can hit a note equally as late as you can hit it early, and vice versa.
  
• The chart for the PC version of GH3 is the exact same as the chart for the Xbox 360 version.

Let me know if any of this is not the case.

I must have made a mistake somewhere in my thinking (or my calculations). I can think of the following explanations:

• Maybe there's a way to abuse the infinite frontend window of HOPOs so that this squeeze can be hit.
  
• Maybe the path is incorrect.
  
• Some other kind of bug or glitch which makes this score possible.

Thus concludeth my investigations. How is this squeeze possible?

After corrections, it seems that with some rounding errors this truly insane squeeze is indeed possible! Is it possible that the 60 FPS frame rate makes this easier? I'm not sure how GH3 handles the timing of the button input on console when the screen has a 60 FPS output.

[Ontwikseltsaar]

Today I was looking at the final squeeze in the optimal path for The Metal. This score has, to date, only been achieved once, by ukog-monkey. This last squeeze is particularly ridiculous so I wanted to know how much leeway there is to hit it.

Basically, what I did was this:

1. I used GHTCP to extract the expert chart for The Metal from the PC version of GH3.
   
2. I exported the chart, took the raw data and prettified it to be human-readable.
   
3. I calculated the timestamps of the last moment the frontend squeeze can be hit as well as the earliest moment the backend squeeze can be hit.
   
4. Based on the frontend activation, I calculated the timestamp of the moment when SP ends.
   
5. If (SP end time >= backend time), the squeeze is possible.

A crucial question here is: What is the exact timing window in GH3?
The only answer to this I could find by just googling the question (which included results from Score Hero) is this Reddit comment, which states the timing window for GH3 is 100ms. For the purposes of my calculations, I took a timing window of 100ms on either side of the note (so that would be a timing window of 200ms).

I have a Google sheet that contains the calculations with an editable timing window. I would share it here, but I believe there is a policy on Score Hero against publicizing the charts. I'll try to summarize the sheet:

• A measure consists of 4 beats. Every beat has a fret bar. Around the time of activation and the time when SP ends, the time that elapses between fret bars is 535ms or 536ms. This data can be found in the chart files (which is nice, because I expected that I'd have to do much more conversions between beats and time!).
  
• Initially, I calculated the amount of SP saved by activating 100ms late by taking the ratio 100/536 (= 0.187 beats of SP saved) and calculating how much time that would save when SP ends, but since the BPM at the beginning and end of SP is almost exactly the same, I realized that the ultimate squeeze (hitting the first note 100ms late) would save about 100ms worth of star power.
  
• Hitting the first note 100ms late and hitting the last note 100ms early, I still end up being about 134ms short of being able to hit the squeeze, which means we'd need a timing window of about 167ms (on either side of the note!) to be able to hit this. I calculated that star power ends 151057.25ms into the song, and the earliest the squeeze can be hit is 151058ms into the song! Assuming some minor rounding errors, this means the squeeze would indeed be possible if both the frontend squeeze (with the activation) and the backend squeeze are hit with 1ms precision!

In the process, I made the following assumptions:

• Notes are always at the center of their timing windows. This means that you can hit a note equally as late as you can hit it early, and vice versa.
  
• The chart for the PC version of GH3 is the exact same as the chart for the Xbox 360 version.

Let me know if any of this is not the case.

I must have made a mistake somewhere in my thinking (or my calculations). I can think of the following explanations:

• Maybe there's a way to abuse the infinite frontend window of HOPOs so that this squeeze can be hit.
  
• Maybe the path is incorrect.
  
• Some other kind of bug or glitch which makes this score possible.

Thus concludeth my investigations. How is this squeeze possible?

After corrections, it seems that with some rounding errors this truly insane squeeze is indeed possible! Is it possible that the 60 FPS frame rate makes this easier? I'm not sure how GH3 handles the timing of the button input on console when the screen has a 60 FPS output.

[UndeadFil13QC]

This has to have an extra layer of complexity due to the game running at 60 FPS. I have a feeling every note's hit window has to be on a specific frame range, and that would mean that every note has a window of 13-ish frames, the center of which depending on the video lag setting (visually speaking, not in relation with other notes).

Let's first point out that 16th notes in measure 67 are approximately 8 frames apart each, which makes it possible to make use of the entire back-end window for a maximal squeeze. For the back-end of the activation, this doesn't matter since you can hit a note as early as the window allows regardless of other notes being present.

Now, assuming that a perfect squeeze can save you a total of 12 frames (0.2s), and that the first and last note of the activation are 4 measures + 0.375 beats apart at approx. 112 BPM (first R of measure 67 to first B of measure 71), we need (0.375)*(60s)/(112) = 0.200892... seconds squeezed in, which probably rounds closest down to 12 frames (0.2s). Hence the squeeze would be possible, but probably frame perfect.

Please note that this analysis is based on assumptions of the width of hit windows and note placement in relation to the frame rate. These numbers might be incorrect but it seems pretty close to reality.

This kind of reasoning could potentially be used to figure out squeezes such as in CoP and PaJ, but it gets even weirder when whammy rates are involved.
_________________

ULTIMATE GOAL: FC Im The One on GH: Van Halen Expert Guitar
Solo 2A FC Count: 39 (Best runs: many -1s, three outro chokes and two 100% overstrums!)

Side goals: GH Expert Vocals Full Series FC [4/8] | Rock Band Expert Bass Full Series FC [8/9] | COVID Vaccine & Booster 100% FC [4/4]

47|64|30|70|41|86|49|48|85|46|93|42 (10|65|10)
Full Series Total: 701/702 (786/787)

[UndeadFil13QC]

This has to have an extra layer of complexity due to the game running at 60 FPS. I have a feeling every note's hit window has to be on a specific frame range, and that would mean that every note has a window of 13-ish frames, the center of which depending on the video lag setting (visually speaking, not in relation with other notes).

Let's first point out that 16th notes in measure 67 are approximately 8 frames apart each, which makes it possible to make use of the entire back-end window for a maximal squeeze. For the back-end of the activation, this doesn't matter since you can hit a note as early as the window allows regardless of other notes being present.

Now, assuming that a perfect squeeze can save you a total of 12 frames (0.2s), and that the first and last note of the activation are 4 measures + 0.375 beats apart at approx. 112 BPM (first R of measure 67 to first B of measure 71), we need (0.375)*(60s)/(112) = 0.200892... seconds squeezed in, which probably rounds closest down to 12 frames (0.2s). Hence the squeeze would be possible, but probably frame perfect.

Please note that this analysis is based on assumptions of the width of hit windows and note placement in relation to the frame rate. These numbers might be incorrect but it seems pretty close to reality.

This kind of reasoning could potentially be used to figure out squeezes such as in CoP and PaJ, but it gets even weirder when whammy rates are involved.
_________________

ULTIMATE GOAL: FC Im The One on GH: Van Halen Expert Guitar
Solo 2A FC Count: 39 (Best runs: many -1s, three outro chokes and two 100% overstrums!)

Side goals: GH Expert Vocals Full Series FC [4/8] | Rock Band Expert Bass Full Series FC [8/9] | COVID Vaccine & Booster 100% FC [4/4]

47|64|30|70|41|86|49|48|85|46|93|42 (10|65|10)
Full Series Total: 701/702 (786/787)