Skip to main content
R
RedyCode
Inspiring
January 5, 2022
Answered

ExtendScript: Find fastest point on a graph

  • January 5, 2022
  • 1 reply
  • 257 views

Hey everyone, I've been searching for days and I can't find a solution:

How can I find the fastest point of a graph via scripting? ( so in the speed graph it's the highest peak )


I have a working solution ( The script goes into a for loop and it gets the value of every frames between the 2 keyframes and then it finds the difference between every value to find the biggest difference, if that makes sens ) but it doesn't work perfectly with everything—especially when it comes to the CUSTOM_VALUES or some path—so I want to find a better solution, maybe a solution working with the easing ( like the influence and speed )

Thanks in advance,
Redy

This topic has been closed for replies.
Correct answer Mylenium

You need to create a differential and measure the length of the resulting vector if simple thresholding and differentiation doesn't provide distinct results. That and of course you have to view this at a much larger scale. Such calculations based on only two immediate keyframes rarely provide stable results. Ideally you would have a convolution/ pyramid and iterate through different time spans, then average and smooth the results. That would also allow to extrapolate results if no proper solution is found between two adjacent frames. In your case you likely would also need to figure in the vectors of the original keyframes, which such a solution could also take care of by dynamically roving beyond the ends of your segment. Anyway, you will have to make this more complex to get good results.

 

Mylenium

1 reply

Mylenium
MyleniumCorrect answer
Legend
January 6, 2022

You need to create a differential and measure the length of the resulting vector if simple thresholding and differentiation doesn't provide distinct results. That and of course you have to view this at a much larger scale. Such calculations based on only two immediate keyframes rarely provide stable results. Ideally you would have a convolution/ pyramid and iterate through different time spans, then average and smooth the results. That would also allow to extrapolate results if no proper solution is found between two adjacent frames. In your case you likely would also need to figure in the vectors of the original keyframes, which such a solution could also take care of by dynamically roving beyond the ends of your segment. Anyway, you will have to make this more complex to get good results.

 

Mylenium

R
RedyCodeAuthor
Inspiring
January 7, 2022

Thanks for the answer, I really appreciate it. I found an easier workaround tho, pretty hard to explain so if someone really needs it you can ask me

Terms & ConditionsAccessibility statement
Using the community Terms of use Privacy Cookie preferences Do not sell or share my personal information ad choicesAdChoices