371 replies to this topic

[ckh]

I've developed a Windows program called Foucault Unmasked that analyzes unmasked Foucault Test images.  This method results in more objective and accurate mirror profiles than can be obtained by eye using a Couder mask or Everest pins. The number of zones is limited only by memory space.

 

The program automatically crops the images to the edges of the mirror and finds the precise radius of null zones (to sub-pixel accuracy).  Note that automatic cropping requires a dark background. Pre-cropped images can be used, but the Crop checkbox must be unchecked in that case.

 

Results can be saved in the format used by Figure XP or by Sixtests so that those programs can be used for subsequent analysis the zone data.  A native CSV format is also provided that can be loaded into Excel.

 

For convenience, the program includes a native surface analysis (with surface plot) based on ATM Mirror Mathematics by Jim Burrows (click the Show Results... button).  This has not been extensively tested so it is best to compare the results with those produced by Figure XP or Sixtests.

 

In testing the program, I had access to images from a limited number of mirrors.  I used my own 6" F8 images and images of a 450mm F3.56 mirror kindly provided by the Mark Cowan.  My own results on Mark's images nearly match those he obtained using a manual flip-diff image processing method to find null zones.

 

In testing I also generated 40 Diffract images of a simulated perfect parabolic mirror. These images were analyzed yielding an RMS of 2 nm and Strehl ratio of 0.998. The difference from perfect is largely due to the simulated diffraction effects near the edge.

 

The Help button in upper left hand corner of the main window provides an Overview of the software and instructions for use. You will need Microsoft .NET version 4 or later. 

 

To install:

I have placed the windows installer file "Setup.msi" on the file sharing site "ge.tt". To download the file click on this link http://ge.tt/6wDoeNn2.  A large grey window will appear (after the ads) containing "Download Setup.msi". Click on "Download" and then run the downloaded Setup.msi file (just double click on the file) to install Foucault Unmasked. (The file size is about 1 MB).

 

The program is added to the startup menu as "Foucault Unmasked".

 

Let me know about your experience or if you have any problems, questions or suggestions.

 

Here are some screenshots showing the analysis of Mark Cowan's images:

 

 

 

 

 

Carl

[mark cowan]

Carl, looking good! 

 

I'll have to give it a spin.  

 

Post the URL out to the Strehlafane collection when you think you can release it (unless I said that already, but still do it).

 

https://stellafane.o...t/software.html

Edited by mark cowan, 18 November 2017 - 09:50 PM.

[kingjamez]

Very nice. I look forward to giving this a try. 

-Jim

[Pinbout]

 

...Strehlafane...

Edited by Pinbout, 20 November 2017 - 09:57 AM.

[ckh]

Missed it.   

 

Hope somebody tries it out.  There have been a few downloads. Looking forward to feedback from you two, Jim and Mark.

 

Going back to polishing and testing my own mirror now.  Been messing with this software for a long time while my mirror just sits on it's lap.

 

Carl

[Pinbout]

 

while my mirror just sits on it's lap.

https://www.cloudyni...flat/?p=8223970

[ckh]

There have been 10 downloads of the software but no feedback yet.  Has anyone installed the software and run it?

 

Carl

[JohnH]

Just going to fix my tester to use on my 12.5" f4.5

[kingjamez]

I've got to get my tester back up and running too. Got the stand fixed yesterday, so 1/2 way there.

 

Jim

[ckh]

Thanks guys.  Let me know how it works out.

 

Carl

[sixela]

I've advertised the existence on a French and Dutch forum.

[jeffg]

Carl:

 

The program installed easily.  I note that it works on Win10 , but not XP--probably due to the NET 4.1 requirement.

 

Selecting "Overview" in "Help" generates an Unhandled Exception error.  Hitting continue generates an empty dialog box.

The "About" and "Analysis Methods" tabs work as expected.

 

Since the overview doesn't seem to be working, I am unsure of what images and other data (I assume some sort of zonal measurements) I need to drive the program.  Otherwise, it looks impressive.

 

Jeff.

[ckh]

In addition to the mirror parameters, all you need is a set of images (use Select Images... to load them) and the KE offset for each image.  I'll look into the Overview issue.

 

Thanks,

Carl

[ckh]

Not sure why the Overview was not included, but a rebuild fixed the problem.  I have posted the updated installer at:

 

http://ge.tt/1sN65Xn2

 

Just download Setup.msi and run it by double-clicking.  The Overview (on the Help menu) explains how to use the program in detail. 

 

If you have already installed the program, select the Repair option (default) when installing.

 

Carl

[mark cowan]

Hmm, on the display, even though I put in a conic of -.514, I get a surface plot referenced to -1.000... 

 

Other than that quite nice.  The numbers look to be in close agreement to the last test I ran and analyzed manually.

[ckh]

Mark,

 

With your older images (quite close to parabolic), I set the conic to -.514 and got this plot which shows about 2 waves overcorrected.  Can you provide more details about the problem you saw?

 

 

 

Carl

[mark cowan]

Oh never mind - I didn't load the offsets correctly. 

After taking care of that it's pretty close to what I got, at least at that rough stage of the work.

[ckh]

That happened to me numerous times during testing.  I suppose it could give a warning if the offsets result in more than 2 waves from the conic.

 

Carl

[mark cowan]

Better to highlight it at some point so the user gets reminded, I suspect.  I code the offset into the filenames so if it were me I'd just pull it from that. 

[lethoncbon]

Thanks a lot for sharing this

 

I would love to test it, but the download link doesn't seem to work anymore.

[ckh]

I updated the program as noted in post #14 (the Overview on the Help menu was missing). However I just tested both links and both worked.  This is the latest version.

 

http://ge.tt/1sN65Xn2

 

Did you try this link?  What problem did you encounter downloading?

 

Carl

[lethoncbon]

Now it's working  

I was trying from work earlier, guess that's where the problem was coming from (the grey window wasn't showing at all ...)

[Dorin Blajan]

Hello Carl and all,

I tried your program on my first mirror that I made a couple of years ago. It is a 150 mm diameter mirror of 1015 mm focal length (f/6.77).
I took several images, 0.25 mm apart. I used a webcam so the images are not the best quality but the program finds the null zones quite nicely.

But...
I took two series of images, taking care that the images are shifted a little bit along the optical axes between the series. I mean the images in the two series are not taken exactly at the same KE position so the null zones are slightly different.
Analyzing the two series I expected to get identical or very similar results but they are wildly different. That means that if I measure the mirror many times I'll get very different results. This should not happen. Here you can see the results from the two series:

For each series I also exported the results to FigureXP and its analysis agrees with yours (in FigureXP the mirror is slightly better but the difference is not significant). So the problem comes from finding the nulls zones not from the analysis.
It would be interesting if somebody could try this and see if he gets the same inconsistent results. Or am I missing something here?

I really want to thank you for your effort and for sharing it. I would be very interested to use your program if it gives consistent results.

 

Dorin

[ckh]

Hi Dorin,

 

Can you post both sets of images?  Also how much did you shift the offset in the second set?

 

Possibly Mark or someone else can at least check some of them using the diff/flip method.  

 

I'd certainly like to find out what's going on.  I presume that nothing in the Foucault setup was changed except the initial offset.

 

Thanks,

Carl

[Dorin Blajan]

I don't know the offset between the series, I just moved the tester a little bit along the optical axis. This is all that was changed between the two series of images.
I'll post the images soon.

Thanks for your help!