Strange lines in depth camera

Hi, I’m getting strange lines in my depth camera, is this normal? I see in other videos in youtube they have a clean image and no lines…

I think my Mini need calibration, but there is no calibration software available yet?

I’m getting very bad tracking and noisy results.

https://youtu.be/DhUEKsu8vHw

1 Like

The lines are abnormal and the scanned results as well, if the calibration don’t help you to improve your results please contact the Customer support via email showing them the results you experiencing, including the video link you posted.

Thank you for your reply. I did the recalibration, my first result was 0.08345 and the second time (with more care taken) it was 0.04325, so I saved it. It makes no difference however for the horizontal lines in the depth image. I notice these lines are a variation in exposure, so details passing by these lines disappear and appear again while passing. This maybe explains my bad tracking and inconsistent feature result. Is there anybody who sees similar lines on their depth camera?

See video:

That was a good calibration result , I see the object in preview are not affected by the difference of the exposure lines what is very good , I watched your video hoping to see the scanning process including meshing after the calibration .

It seems like the gain settings of one of your depth cam sensors is set to slightly higher value than the other, of course it should be not this way .

Please scan your object at 10 cm distance between MINI and the object , fuse it at 0.02 mm and mesh it at 6, denoiser 3 and closed holes . I would like to see the exact result with this settings, you don’t need to scan the whole object , just give me one rotation scan 360 frames at 0 degrees.
And post close up of the mesh image especially around ears-hair

If you use dual axis turntable use 35 speed, MINI don’t works well at slow speeds of the turntable, 30-35 is the desired speed .

1 Like

I checked both of my units , they have very slightly lines , one less than another but not like yours , the lines I have are regular lines coming from the glass covering the sensor and actually not affecting the scanning .
Screenshot 2022-09-16 161851

The result you should get is this when you scan the way I told you

I just watched your video , the surface is ok so the lines do now affect anything while you are scanning , but I find your result slightly too blurry, I see your bust is slightly different than my but still I don’t see the crispiness , more like a POP2 than MINI or like object scanned at 20 cm in place of 10cm

The algorithms changed a lot latelly regarding the point cloud , but in this case I should have less sharpness also. I tested on both device with the same result .

I see no problem with alignment here, beside the smoothness that you got on the details.
Your computer works perfect as well .

The only thing you can test more is scanning something that has smaller volume with more sharper details than the bust to be sure .

1 Like

Thank you so much for taking the time to help! Yes it seems to work ok. Somehow it is tracking much better than when I first got it, I think the calibration helped, see my first video, it looses tracking immediately when the Dual Axis stars turning, now it never looses tracking (after calibration).

So I’m glad it tracks better. I will try scanning finer parts with more details. Is denoiser 3 always a good setting? I feel it might be a bit too strong smoothing? But with 1 or 2 there are a lot of random blobs floating around…

I will continue to test. Have already contacted support, lets see what they say about the exposure difference between the cameras. Thanks for you help!

Yes that is not normal behavior , impossible , so I guess the calibration fixed some things here .
Also please avoid yellow lighting on your scan , you can have cover it and scan in dark , but no yellow colored light as I see under your RGB camera , yellow absorb blue messing up the pattern .

What I am testing my MINI with is an old bad 3D print that has 0.2mm layers , if I can see the layers after scanning I am sure the scanner resolution is working fine regarding the quality .

that is how I found bugs when testing out MINI prototype for Revopoint

Admittedly in the old video my Dual Axis orientation was wrong, it was tilting left/right not forward/backward, and it was too far away (only good) but still it should not have lost tracking so bad.

it lost tracking the moment you started , that is bad , I never lost tracking with the bust , or anything like in your video .
I usually use regular turn at 0 and then tilt at 29 away from MINI , that’s enough .

1 Like

Thanks, that makes sense. I always felt the tilting towards the MINI was a bit redundant too…

Will try 0 and 29 degrees only.

The idea with a 3D printed part is smart, 0.2mm layers should be visible.

1 Like

Yes, even 0.1 layer should be visible , since the accuracy and resolution is 0.02mm , the smallest details I got was between 0.03mm to 0.075 mm , so 0.2mm should be no problem

Those lines are reminiscent of times I’ve tried to record video of my computer screen using my cellphone, almost like there is some interaction or interference between the energy of my cellphone’s RF connections and the camera.

I may be totally off base here, but an interesting experiment would be to ensure that your cellphone is placed WELL away from the scanner while scanning. ISTM that you should be able to notice a change right away if there is some correlation.

2 Likes

It is strange but never happened to me , there are normally slightly lines always ( beside my last edition of POP2 without any lines ) everything else have it to some degrees. One of my unit have half screen brighter than the other half .

However what @hugonl38 shows in the preview, I never saw before , to regular however it seems not affecting the scanned surface at all .

I have very similar problems to @hugonl38 ‘loses tracking’ (not the red warning but you can see it’s not tracking) and the horizontal banding on the depth camera:

The lines do not affect the scanning results so that should be not a problem , if something was wrong you would see the lines in the preview slready , but after @hugonl38 made video , it is all fine .

You can only see that is not tracking if you see multiple objects appearance in one scan and usually red warning .

So make a screen shot or video of that isdue , and before try to recalibrate your device and follow the test as @hugonl38 did .
Try first 3 tests of the calibration , to find the lowest values, anything below 0.08000 Is just fine .

Then do scan

Object at 10cm distance from MiNi
One rotation at 0 degrees 360 frames, fuse it at 0.02mm and mesh at 6 with denoiser at 3 .

OK, so I calibrated once with a result of 0.06… (reference 0.5 - not sure what that means)

Performed another scan:

Fused point cloud:

Ok :ok_hand: you fused point cloud, and where is the mesh ?
The point cloud is exactly as it is in the software at this moment , but you need to mesh it after to clean out the artifacts .

And have actual meshed model .

The point cloud in Revo Scan is not exactly what you would expect at this moment, but that is the new algorithms issue and not your device .