Bad scanning results

Good afternoon,

I’ve been trying to achieve a ‘baseline’ first scan to see that my new POP2 functions.

Opening RevoScan shows an impressive and consistent depth image and colour image, clearly the hardware works to some degree! Following kondzio29’s fantastic advice in this thread I’ve begun using a powered USB bank, and this takes the scanning frame rate from ~3 to ~20.

Unfortunately I am having persistent problems with tracking. I have been experimenting a lot, for example with lighting intensity and RevoScan settings, but the two headline failure cases are as follows:

Scanning the sample bust included in the package on the included marker covered turntable in feature tracking mode cannot hold tracking to the object, averaging perhaps three serious (>10 degree) dislocations in one revolution (about a minute).

Scanning the sample bust in marker tracking mode is not possible, as of the 40 marker points on the included turntable at most six are identified by the scanner, typically between 1-3 (5%?). This achieved only by configuring the turntable, scanner, and lighting explicitly to try and track as many markers as possible.

I have now recalibrated the scanner, and whilst the preview data remains reasonable, and perhaps I sometimes see a seventh marker point for a frame now, but feature tracking and marker tracking remain impossible.

It’s entirely possible I’m doing something wrong! Or perhaps that my scanner has some defect. With that said, I’m not sure what else I should try. Without working tracking or the ability to use individual frame data the scanner cannot do much for me.

Many thanks,
David

The scanner is capable to produce no more that 10 FPS , so you can’t get more than that as that is the scanner hardware limitation .

The markets are recognized by the angle , think here about a mirror reflecting the light at angle .
Usually setting the scanner at 45 degrees to the turntable will pick up more markers.
Personally I use my own more reflective markers , the one white paper markets are not so good on turntable and they need good angle to reflect the IR light back to the sensor.

I have no idea why you can’t scan the bust as that is the most easy object to scan.
Just set the POP2 at 45 degrees slightly above the bust , make sure the distance is between 15-20 cm , scan one 360 degrees rotation and you should have it done .
No rocket science here about .

Portable scanners are not suitable for per frame scanning , the projector do not produce steady pattern, it is DLP motion projector , it produce 10 frames per second and the frames are not aligned before they enter the software , so using the frames from the hardware will do nothing , the frames that are aligned by the software algorithms are the one to be used in fusing process after scanning .

However if you can’t get this to work, your frames are not usable at all , if they were you would have a perfect scan .

So please try again setting the way I told you , get away from windows light as it include Infrared light, indoor light do not affect POP2 while scanning .
Also remember about the gain settings , the red square need to be set only on your scanned surface and not on the background , after it get adjusted , switch to Manual and start scanning .

I would like to see the screenshot video of this process .
Then I can determine if your device need replacement or something else.

Waiting for your reply .

Hi there Mr Volume, thanks for coming across quickly to help me out.

Interesting to note about the bust, markers et cetera. I had assumed the bust and the provided turntable would be the ideal test scenario. As you said, I’d found I could see the most markers at a high angle (45 degrees like you say, or I was using near perpendicular [70?] to pick out six or seven). It seems the printed white markers on the turntable are not actually ideal. In the last hour I used some of the silvered reflective marker stickers to replace all of the printer markers, and now marker identification is Massively more reliable, typically I see 10-20, losing them at more shallow angles or at distance. This does let me do a reasonable bust scan! As I cannot move it relative to the markers I have to combine a few scans in other software to get a scan including all angles.

Tracking still not great without markers. I appreciate your angle and distance suggestions, but I fear I’ve tried all angles and distances at this point. I get the best (non marker) results at about 130mm to surface and 35 degrees below horizontal.

I would really really love the frames. I’m fairly impressed by the hardware, especially given the price point, but this is not how I would expect to leverage the point data. I don’t particularly understand your comments about framerate and alignment, those rapid unaligned frames are exactly what I’d like. I work with vaguely similar data in a professional capacity and I feel the raw device output would let me produce better composites. I see no reason that fusing and baking and meshing should be some obligatory part of the pipeline, or why I shouldn’t be doing them in my own way.

Cheers.
David

P.S. I guess my time estimation is pretty bad? Maybe I had 1fps and got up to 10. I’d have been safer describing that as a 5-10 fold improvement I guess.

You welcome and it is Mrs PUTV :wink:

They are not and I use reflective markers myself .
You don’t need 20 markers to be visible , at least 3 per frame (view)

That should be no issue at all as that is more as you need .

I guess you need to recalibrate your Scanner again if it don’t works .
You did not provided me with any informations I asked for so it is hard here to guessing anything in this case .

“Just don’t work” means not much , the bust provided is the most easy to scan object , my 8 years old scan more difficult objects than that so definitely something is not right with the device .

Again I suggest you to record the process of failure , send email to customer@revopoint3d.com with the link to this thread and your forum ID .

If another calibration don’t make the scanning better, there is nothing else you can do as you did already everything.

Once you make the video , the customer service will determine if this is device issue and if it is you will get replacement, so don’t waste your time on hoping for another solutions .

It is not gonna happen tomorrow, I explained to you already , however the team said they will provide stuff for developers once the Revo Scan software reach it’s pick in development. When it going to happen I have no idea .

However I don’t have any issues with tracking or markers with any of my devices , not the one that are released not the future builds .
3D Structured Scanners relay heavily on Tracking points for scanning , without it it can’t process anything , the bust have more than enough tracking points on its own and don’t need to be scanned with markers .

Marker mode is for scanning objects that are more or less features, like for example coffee cup or a box . The Markers represent the tracking points , you can’t move the object while scanning using marker mode unless the markets are on the object and nowhere else , it need to be steady in one position to the markers .

Another thing , if you scanning using turntable, you can’t have anything else be visible in your preview other than the object , a small fragment of a table or anything that do not move in preview will prevent the tracking .

It is a learning curve and there are not shortcuts .
However I believe that your device is not working properly in this case if neither feature or marker mode don’t works with the bust .

So please create the video as requested and follow to guide so your issue going to be fixed .

I’m not sure anything about frame data timelines has been explained, in your first or second post. I don’t see that there’s any great crest breaking in terms of the Revo Scan development. Give us the tools to leverage the cool hardware. Peak or no peak.