Sorry to hear about the calibration issues , it really sucks , the second part is not easy to pass with the weird angles , But you know what , I find scanning on one angle better than another , the worse at 90 degree angle . It always need to be below that for best results . Gyroscope is used for measuring the 3D object in space , and Depth camera is used to create Depth maps only , but with the Revopoint technology I was wrong not once already . It always seems something else that we thought we knows about . What is not bad thing . Same with the upcoming Revopoint MINI blue light high accuracy scanner 0.02mm , where the blue light technology is not exactly what everybody else is using on the market.
My POP2 restored itself again . I am getting the best performance from using WiFi and external power source . I connected Mobile Cable and plugged the USB A to my PC using 6ft USB extension cable and connected it for data via WiFi 5Ghz and works so much better than via USB only . I guess there is some issue with the power feed if you use just one USB for both data and power.
Also worked for me on my PC via USB C for a while and now stopped working for no reason . On My phone no issue via wifi or USB C again with external power source only and Mobile cable connector .
Thanks PopUp for the extra info! Yes it has been difficult to determine what role the accelerometers/gyros play in the entire process, mine will at times confuse itself as to what “level” is, or at others get stuck “out of level” where no change in orientation (relative to the subject or not) will affect the level indicator. The USB thing makes sense, what you describe is a common issue with analog synthesizers; power over USB will cause grounding feedback if/when the USB connected is on the same power supply. The solution to these scenarios is to power those devices from a battery pack or a different power source with a different ground. This problem mostly, however, is due to the fact that you are dealing with an analog signal output; same device when its audio output is routed via that same USB cable digitally will also resolve the issue. I would think this device would fall into the latter scenario. Nevertheless if it can potentially clean things up enough for me to have a chance at a better calibration, and ultimately usable model, then I will have to give it a go. To be clear, are you saying that you are connecting to the device via USB, and connecting to it via your wifi in addition to that on the PC (you mentioned running the A to your PC, which I assume would also create a communication channel)? Or are you connecting the USB to only a power source and no data connection on that side, leaving WIFI the sole data connection? I have a feeling it will not do much for me, however, as when connected to my iPhone 13Pro Max it is essentially creating that scenario. In my case, however, the Mobile tracking issues are quite a bit worse. The fact that mine is getting steadily worse and seeing no gains, though, has anything but a hardware issue of some kind becoming almost 0 in probability.
I was considering backing the new BL version so that I could review/assess the + and - of each form factor and conditions based scanning performance, however until I am able to get at least this one working I simply cannot justify that step. Given that I am waiting on communication I will likely still give this a go, if nothing but to add another data point. I’ll report back if I find anything of interest!
@PUTV, sorry for the double reply, but I think I just had an epiphany. I have been thinking about how the calibration process runs, and how many of us may be anchored in similar yet different methodologies for calibration thus making assumptions about the objectives. We do a LOT of camera calibration with our DJI drone fleet, and other devices that we use, and when following the instructions in the tutorial I found myself conflicted with what I thought I should do, vs. what is being asked of me. Now, when asked to align the dots, the evolutionary start of my approach was to get all 5 dots to line up, no matter the orientation etc. in hopes that the various sensor inputs on the device would learn the deviation from its perceived norm and correct for it. With that approach I DID have success in calibrations, but dwindling success. On closer inspection I realized that each iteration it appeared that it may have been attempting to start at a lack of perfect alignment, and work its way toward full overlap iteratively. I decided that I would change process, and shift attention to the physical orientation/level of the device and aim for “as close as is optically possible” when lining up the dots. In many cases this meant starting out with only 3 of the circles actually over their targets, the others off. Phase 1 always went through all aligned every time. Phase 2 would start off passing green on the “as in place as possible”, and in each iteration would start coming in closer. At times Phase 2 would end with all of the circles in place but the top, in others it would manage to get them all. Phase 3, like Phase 1 has always gone quickly and seemed more validation. That one with the change in approach seemingly concludes much faster.
This still has not netted me better than a .17 shift, but it did bring me back from spiraling into the 2+. This was enough of a revelation that the users mindset and objectives in a test like this can have a huge impact on the end result. This is why I want to ask you; what alignment approach have you been taking? Does your Phase 2 always start basically impossible to align visually without contorting the devices orientation? I believe that there may be a little bit of lateral vs. literal thinking going on here which may be obscuring the intent and objective of the test, and it comes down to my first simple question, whether the expectation is to get as close as possible while maintaining the orientational structure of the device during the test? Or to “align the dots” - which if left open to interpretation, at least in my case, and with my experience on other types of calibrations, was to change perspective until they did line up. My thoughts watching the person in the demo, is that #1 is right and my assumption was wrong since you never see him changing orientation at all in any phase. However, at the same time on his camera view his dots never looked like mine have from the start in phase 2, and that is completely NOT optically possible to get in the frame without serious contortions. Knowing your experience in this context might go a long way to help isolate what’s going on here, and if not, it might at least help some other people better define their process for calibration as intended.
Again, this didn’t bring my alignments to where I think they should be, and it did not fix the tracking issues I have, but I think it is and will be a contributing factor to getting things as dialed in as is possible. And perhaps as is true in in most things, the reality of what should be done may be somewhere in the middle.
@EventHorizon
Richard let me start with with the Calibration , the step # 17 is most difficult , it seems impossible to pass , but I did and twice , holding the POP2 completely horizontally is the key as I believe the calibration here is to adjust the position of the depth cameras , if the 2 cameras are not aligned precisely to focus at the same angle it will result in bad scans , but since there are no motors to move the depth cameras what is really adjusted , is it possible that the Infrared light source can adjust itself ? because if it is just code based I don’t see how . I did also cheat test moving just the calibration plate and the results was around 0.9000 . I do have couple of 3D laser scanners put the process of calibration is simple .
Coming back to POP2 in the video reference the step # 17 is wrong the plate is in the wrong position , it should be on the right side and it shows at the left side what make the dots impossible to even align.
I tried to use an arm perfect leveling in horizontal position but once I hit the #17 the trouble started .
I have zero tips here for you because everything I tried leading me to the same issue .
One tip only , keep the table with the plate leveled , make sure the white leg of the plate is inserted all way in .
USB:
Yes that make very much sense , thanks for the info regarding the power issue via the USB.
Regarding the connections via USB/WiFi here is my diagram I made for give people idea for connecting the POP2 to their devices , it works the same way with smartphones as it do with computer , you may try any of it and see how it works for you , I tested them all on my smart phone and my PC .
There is 3 more way not included in this diagram using extension cables with the Mobile connector
where the USB A goes always for the power source and USB C for the Data
And another using the single Micro B to USB A male cable , power and data with one cable but it never worked for me on my PC
Or using Micro B to USB C for power and data on one cable , it worked for me on my PC but sometimes it failing , and I have no control over it.
