Smoothing a curve

[IanMaz]

Hi Math Gurus,

I was wondering if anyone knew how to smooth out data in a curve.

Here is what the data looks like on a graph as it is.  kinda lumpy.  it should look more like one on the right.
Note they're different data sets.  But the effect is that it should not have chunky abrupt transitions.
I have attached a text file with the data for the left curve.

These are calibration aim files for an imaging device.  They have 26 steps x3 for red, green and blue.
The end steps spread naturally in the process so thats normal.

Searching google, I cant find a good way (or any way) of doing this in FileMaker.

Any ideas?
 

Thanks!

Data txt file link https://wormhole.app/o4DxmQ#FvptNf39LaTzr-XsK0m9ng

 

aim.txt

Edited Wednesday at 03:51 PM by IanMaz

[comment]

Are we talking about Filemaker's native chart tool?

If yes, there is an option to style the chart as curve-fitting, which may serve your purpose.

 

--
P.S. Your file cannot be downloaded ATM - see:

 

[IanMaz]

Hi,

I actually need the data changed to reflect the smoothing.  Its used in a calibration file that a printer will try to follow.

The curve view is a web viewer using google charts.  For this, it was a lot better than the built in charting.

Heres the data txt file https://wormhole.app/o4DxmQ#FvptNf39LaTzr-XsK0m9ng

[comment]

16 minutes ago, IanMaz said:

I actually need the data changed to reflect the smoothing.

You would need to specify exactly how to change the data. I believe we have already been at this point:
https://fmforums.com/topic/106749-linear-number-list/#findComment-482406

BTW, if Google Charts does what you need, why not simply use it instead?

 

 

 

Edited Wednesday at 04:08 PM by comment

[IanMaz]

Thanks, I never saw that last post.  I will try to find a mathematician.

The chart is just a way to visualize it.  Curve fitting or filtering doesn’t alter the actual data.
I need the data to be smoothed out so I can generate a new aim file.

The data in the case of these aim files is derived by trial and error (as far as I know) by the film manufacturer.  I got this from them for a different machine so it's not ideal.
The machine prints out a test strip, process and read that in, then its software calculates corrections.  It all depends on the aim file being a good one.

From looking at the all of the curves, it's loosely based on an exaggerated H&D curve.  H&D curves are Log10 (if im reading that right).
Where contrast builds slowly to allow better control in the lower and mid regions then builds faster in the areas that are very hard to control.
Usually you jam step 26 up to DMAX to guarantee a good black and let it float to whatever the process is feeling that day.  The DMIN and mid tones needs the most control to stay neutral.