Big error after calibration #10594
Comments
|
Hi @Obrepal Could you confirm a couple of points about the target image that you were using, please?
https://play.google.com/store/apps/details?id=com.intel.realsenseviewer17613&gl=GB
You could also try using a tool built in to the RealSense Viewer to reset the camera to its factory-new default calibration to see whether this provides the desired image quality / accuracy. Instructions for doing so can be found at #10182 (comment) |
|
Yes, I downloaded app via google store (Dynamic Target tool version 1.0) Ok, should I print it or it can be displayed on monitor? |
|
The calibration bar image provided by the Android app should be displayed on the screen of the Android device and not printed. The camera should be pointed at the image on the Android screen. Aside from using the Dynamic Calibration tool in a location with good lighting, there is not much else that can be done to improve its performance further. The Tare calibration tool offers more flexibility, as it has a Get button to custom-adjust the depth calibration for the distance of the camera from the observed target image before clicking the 'Calibrate' button to begin calibration.
|
|
Yeah, in Dynamic I used screen of phone.
Ok, I will try it and let you know.
Thanks :)
śr., 15 cze 2022, 13:11 użytkownik MartyG-RealSense <
***@***.***> napisał:
… The calibration bar image provided by the Android app should be displayed
on the screen of the Android device and not printed. The camera should be
pointed at the image on the Android screen.
Aside from using the Dynamic Calibration tool in a location with good
lighting, there is not much else that can be done to improve its
performance further.
The Tare calibration tool offers more flexibility, as it has a *Get*
button to custom-adjust the depth calibration for the distance of the
camera from the observed surface before clicking the 'Calibrate' button to
begin calibration.
[image: image]
<https://user-images.githubusercontent.com/41145062/173812872-2144da61-de9c-4e8b-9aea-dfa8eb9cdd6c.png>
—
Reply to this email directly, view it on GitHub
<#10594 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ALRKFXBUZ2K3QSEPKFGLI7TVPG27PANCNFSM5YYQSCQA>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
|
Hi @Obrepal Do you require further assistance with this case, please? Thanks! |
|
Hi @MartyG-RealSense I think it was improved results are not as good as expected but it will do it. Above are three sensors and RMSE |
|
Comparing the estimated RMSE curve for D435 on Intel's official graph from their camera tuning guide to your own graph, they seem to be similar at the 4000 mm distance. Intel's graph is clearly marked as 60 mm RMSE at 4000, whilst your own curve looks to be around 60 at that distance if a ruler is put over the chart at the halfway point between 0 and 100.
|
|
I am using 640x480 insteas of 848x480 thus the difference
wt., 21 cze 2022, 14:17 użytkownik MartyG-RealSense <
***@***.***> napisał:
… Comparing the estimated RMSE curve for D435 on Intel's official graph from
their camera tuning guide to your own graph, they seem to be similar at the
4000 mm distance. Intel's graph is clearly marked as 60 mm RMSE at 4000,
whilst your own curve looks to be around 60 at that distance if a ruler is
put over the chart at the halfway poit between 0 and 100.
[image: image]
<https://github.com/assets/41145062/a4c94546-b925-414f-908c-38728c90112c>
—
Reply to this email directly, view it on GitHub
<#10594 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ALRKFXDYAHW7J47VDF7BH73VQGXFHANCNFSM5YYQSCQA>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
|
In general, as resolution is increased the depth accuracy increases, and accuracy decreases as resolution is reduced. The optimal depth accuracy resolution for the D435i camera model is 848x480. I do not believe that there will be a huge reduction in depth accuracy at 640x480 though. RMS error may reduce if the camera's IR emitter and its dot pattern projection are disabled. If the scene that the camera is observing is well lit then the camera can use the ambient light in the scene to analyze surfaces for depth detail instead of analyzing the dot pattern. |
|
Hi @Obrepal Do you require further assistance with this case, please? Thanks! |
|
Case closed due to no further comments received. |
RS was calibrated with on Chip and dynamic tool (with RGB). I used my phone Redmi Note 8T.
.
Result in chip was 0.00 which is the best result one can obtain thanks to pattern provided here (https://dev.intelrealsense.com/docs/self-calibration-for-depth-cameras). During the dynamic part, I didn't change distance between pattern and camera plane.
Also, phone was kept straight and whole scene was covered (15 pictures were spread).
I tried it multiple times and each of the results is horrible. Below I put comparison of d435 and d435i (1000MM).
As far as I know, RMS should be close to 9 MM (according to Intel).
The text was updated successfully, but these errors were encountered: