Inquiry About Missing "Frame Color Calibration Module" in dv-gui (SDK 1.6.2, Ubuntu 20.04)

[Chentong Xu]

While using dv-gui with a DAVIS346 Color device, I encountered an issue regarding the Frame Color Calibration Module. Specifically, I am unable to find this module under the structure section in dv-gui.

Additionally, I have searched through the dv-runtime source code on GitLab, but could not locate any reference to the Frame Color Calibration Module there either.

 The specific issue is illustrated in the screenshot below.  

I would like to ask if the Frame Color Calibration Module has been deprecated. If so, how can I perform Frame Color calibration with DAVIS346 Color?

For reference, I was consulting the following documentation:
https://docs.inivation.com/software/dv/modules/built-in-modules/frame-color-calibration.html

Thank you very much for your time and assistance.

Best regards,

Xu Chentong

[Luca Longinotti]

Hello, that was a rare case of the documentation getting ahead of the releases. :-) The newly released DV 1.7.0 includes the frame color calibration module.

Have a great day!

Get Outlook for Android

---

From: dv-u...@googlegroups.com <dv-u...@googlegroups.com> on behalf of Chentong Xu <330057...@gmail.com>
Sent: Wednesday, July 9, 2025 12:53:31 PM
To: dv-users <dv-u...@googlegroups.com>
Subject: [dv-users] Inquiry About Missing "Frame Color Calibration Module" in dv-gui (SDK 1.6.2, Ubuntu 20.04)

 

--
You received this message because you are subscribed to the Google Groups "dv-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dv-users+u...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/dv-users/652c1391-73bc-4ce9-8045-4e90f0c33296n%40googlegroups.com.

[Luca Longinotti]

Hello, the frame color calibration module requires OpenCV 4.5, which means it's only available on Ubuntu 22.04 Jammy and 24.04 Noble.

Ubuntu 20.04 Focal only has OpenCV 4.2 available and cannot support it, sorry this wasn't communicated clearly.

The maximum value of the reprojection error depends on your application, for precise 3D triangulation it's usually <0.2 pixels, for simple undistortion <1 is fine.

It can be a bit annoying to calibrate event cameras, it all depends on the data quality of the reconstructed frames, and often requires a lot of retries.

-- 

Luca Longinotti (llongi)

Senior Software Engineer
iniVation AG - https://inivation.com/
A SynSense Group company

-----Original Message-----

Subject: Private message regarding: [dv-users] Inquiry About Missing "Frame Color Calibration Module" in dv-gui (SDK 1.6.2, Ubuntu 20.04)

Date: 07/20/2025 05:59:41 AM

Thank you very much for your response. I have updated dv-gui and dv-runtime to version 1.7.0 on Ubuntu, but the Frame Color Calibration module is still not available. However, I found the Frame Color Calibration module in the Windows version of dv-gui. Does this mean that it hasn’t been updated on Ubuntu?

I also have another question: when calibrating DAVIS346, the default value of Max Reprojection Error is set to 1, but the actual computed value is always greater than 1, sometimes even exceeding 2. What is the maximum value of Max Reprojection Error that I can set to ensure that my experiment does not produce significant deviations?

Thank you again for your time and effort.

Kind regards,

Chentong Xu

[Chentong Xu]

Thank you very much for your response. I encountered a new issue during my experiments. At night, due to low lighting conditions, I set the exposure time to 600,000 microseconds. However, I noticed some unexpected stripes appearing in both the APS images and the event frames as highlighted by cyan circles, which do not belong to the actual scene, as shown in the attached figure.

                                                                                      

I would like to ask: could this problem be related to the rolling shutter effect? If so, how can it be resolved?

Thank you for your time and effort.

Kind regards,

Chentong Xu

[Luca Longinotti]

Hello, I don't believe this to be caused by rolling shutter effect. First of all the DAVIS346 works in global shutter mode by default, and if there was any kind of readout artifact it would be more regular, present over the whole image, and in straight lines, not curved like this.

It's more likely to be some kind of lens reflection. Or an actual real light that's too faint for the eye to see.

But the fact it appears on events too is strange, depending on how often you see it there.

Coupling between frame capture and events is possible during the frame readout phases, which are very short (~10-15ms) before and after exposure.

If you look at recorded data, are the lines in the events always there or only for some milliseconds, then 600ms they're gone, then back and so on?

If you see this stimulus all the time in the events, then it really looks like this is a real stimulus.

Hope this helps, have a nice day!