DICe calibration with one camera_RMSE

[Yatingou]

Hi guys, it would be great if you could help me with this issue. Many thanks in advance for your time and investment.

I am trying to calabriton my DICe with the symmetric dot grid with special corner dots.

There are some querstions from my seite:

1. The RMSE is greater than 0.5 (the recommended value). What I have shown here is the best calibration results with RMSE 0.82. When I calibrate with other pattern, the RMSE value is much larger almost 2.5.
2. After calibration, I try to analyze my object. But there is a bigger difference between the DICe results and the real values. The real deformation in y-axis in my case is 1mm, but the DICe result is 2mm.
3. where can I find the cal.log file in the output folders?

DICe version: DICe_setup_v2.0-beta.16-win32-x64
Camera.6960 3904 Pixel
Inspected object: 50mm 50mm
Targeted deformations: 1mm

And the .xml it is in the attachment.
DICe-Output.zip
There is the link of figures for calibration:
https://drive.google.com/drive/folders/1-AK0pFf6tqTYE1GkEMbz17EXMHSRWXLw?usp=sharing

I have tried many way but unfortunately not successful. I will be great helpful if someone can give me some suggestions and inspirations.
:) If you need more information, please let me know. Thank you!!!

[dicengine]

Would it be possible to post one of your calibration images to this issue so I could check that you have the right parameters? (You have the link to your drive folder above, but it's all .zip files which I don't want to download)

Also, I would use a more recent version of DICe. Here's the latest: https://github.com/dicengine/dice/releases/tag/v3.0-beta.8. This will have better calibration routines.

[Yatingou]

Hi, thanks for the reply.
Here is the photo.

If you want photo series, I reload them on the google driven without .zip
https://drive.google.com/drive/folders/1-AK0pFf6tqTYE1GkEMbz17EXMHSRWXLw

I have also tried the new version. I have one more question about the new version. Where can I add the desired tracking point?
I mean this funktion>> add live plot pt

have a nice day! :)

[dicengine]

Looking at you cal images, it seems you only have images with the calibration target lying on the table. To get a good cal you need images with the calibration target tilted at different angles (in and out of the plane of the table).

For the live plot points, once you select the button you have highlighted above, you just click the location in the image where you want the live plot.

[Yatingou]

Thanks for the advice. :)
I try to make the camera on the camera stand, keep it fixed. And the cal. target on a plate and make the in and out of plane. But unfortunately it donsent make the results better.

• Is it affected by the distance between the target and the camera? The distance should be constant?
• I guess, depth of field and focal length have an effect on calibration.
  -Should I make the black dot bigger?

-another question beyond cal.____ There is no large surface on my object. It is a cross pattern with small area, aprrix. 1.2mm*50mm. Is DICe suitable for it?
or should I use the same small cal. pattern for calibration?

and in the new version, i cant find the button for 'add live plot pt'. Colud you please help me.

Have a nice day!

[dicengine]

Making the dots bigger would help since the calibration depends on the center of the dots which gets averaged from the pixels that the dot covers. The distance shouldn't be a big deal, but not being able to angle the cal target towards and away from the camera will likely limit the quality of your calibration. The size of the sample is not really a problem, but making sure that the speckle pattern is at least roughly 3 pixels for the dark and white patches will lead to better results.