General Information & Tips for Newcomers and those interested in scanning

Here are some general tips and information about scanners that I’ve compiled over 3 years as a scanning technician for aerospace investment castings, browsing this forum, and having my Pop2 for about a year now.

  • Make sure you’re in a proper lighting environment – sunlight, incandescent bulbs, some LED’s, etc. will emit infrared and interfere with POP/Range scanners. No Structured Light scanner (all of Revopoints’ Products) will do great outdoors -EXCEPT on overcast days or evenings. When I scan, I will turn off all lights that are unnecessary or overhead of object.
  • Scanner uses two sensors to make measurements using trigonometric algorithms. No scanner can sense a feature if both sensors cannot see the feature (think of your eyes and putting a piece of cardboard between them). To get past this limitation – rotate the scanner vertical (180 degrees) so get scans of tight areas.
  • Moving the work piece, in my opinion, is better than moving the scanner as moving the scanner by hand is shaky and operators can drift in/out of the scanning volume (in and out of focus). However, depending on the subject – you may have to move the scanner.
  • Light used (Infared vs. Bluelight) has no impact on accuracy. However, there is a major impact on noise rejection – which can greatly improve accuracy. Blue light is the least prone to light interference from the environment and thus the best wavelength to use for scanner.
  • Speaking of accuracy – accuracy has to do with optics (camera/sensor lenses) and nothing else. Unfortunately, in our price point we have cheap optics – but issues of optics are corrected with a proper calibration. To increase the accuracy of scans, you have to increase the pixel density per volume (this is reported as point distance on Revopoints Specifications). On Revopoint systems our lenses, and thus, “Resolution” or accuracy is fixed.
    • Range has a point distance of 0.3mm
    • Pop2 has a point distance of .15mm
    • Mini has a point distance of .05mm!

This means the smallest feature (in theory) each scanner can infer is .3mm, .15mm, and .05mm) though it will probably be more likely double or triple this number in reality. With everything else in the world there are tradeoffs. Higher resolution means you need a bigger camera (more Pixels) or have a much smaller scan volume. On the flip side, having the ability to scan a large area (Range) comes at a cost of less resolution/accuracy.

  • To get the highest resolution/accuracy as possible:
    • Reduce light pollution
    • Keep lenses/scanner clean, dirty or scratched lenses will impact scan.
    • Ensure you’re in scan volume (“excellent distance” in Revoscan software).
    • Adjust exposure so that some areas are just overexposed – but not entirely red.
    • Scan in High Accuracy mode.
    • Marker scanning (using points) may increase accuracy if Revoscan uses photogrammetry in conjunction to structured light – but I do not believe they do. However, ATOS by GOM/Zeiss does.
    • Avoid scanning layers more than once – but overlap scans to improve tracking. Unfortunately, when meshing in Revoscan – the software cannot do a gaussian best fit. Instead of averaging the surface data – double scans will confuse and ruin scans.
    • Attempt to collect as much data as possible – filling holes injects “Fake” data into the scan – useful for aesthetics – but not for engineering work.
    • In mesh/point cloud editing – do not use smoothing algorithms or point reduction as this alters data/deletes data.
    • Make sure your scanner is calibrated to the advertised point distance or better. Poor calibration will introduce artifacts and may make merging scans/meshing difficult or impossible. You should check calibration regularly or if the camera is transported, dropped, shaken, etc.
  • To scan Symmetric or very flat components/items:
    • Use marker mode only
  • To scan large Items:
    • Use body mode – it reduces accuracy but increases scan volume.
  • To scan shiny/metal items, you must paint/coat item in matte white or grey. To do this:
    • Use scanning spray (most expensive)
    • Mix talc & isopropyl alcohol and use spray bottle (most practical)
    • Mix ~25% finger paint and 75% isopropyl alcohol to use with air spray gun (most practical for large pieces or for those who scan a lot/often).
    • *Note over spraying item will add material to the item – and by doing so you may reduce accuracy of scan if scanning for quality/reverse engineering purposes.
    • *Talc or fingerpaint should be removable with either water, alcohol, or mineral spirits if necessary (depending on material of item).
  • Random tips/tricks:
    • Markers/point can be used even in feature scanning mode to help merge scans by marker.
    • Scanner supplied stands/handles use standard ¼-20 hardware which can be used with any standard camera tripods or accessories (steady cams, etc.).
    • When not in use – store the camera in its box!
    • Experiment with new software:
      • GOM Inspect is freemium – used in Aerospace/Automotive.
      • CloudCompare is popular & open source.
      • Meshlab
      • Others (search the forum).
  • Troubleshooting:
    • Spiraling scans
      • User error caused by scanning symmetric or flat surfaces – use markers and marker mode.
    • Bumpy meshes
      • Light pollution/noise – check surrounding lights.
      • Check exposure.
      • Check calibration.
    • Distorted point cloud/trouble merging scans:
      • Check calibration.
      • Reduce “scan size” – longer scans tend to drift (ie curl) due to calibration/tracking error that gets stacked up.
    • Small features not meshing
      • Meshing algorithm cannot infer some small features vs. area of less-dense data.
      • Try a different piece of software.
    • Having a difficult time scanning and getting data:
      • Check light pollution
      • Check calibration
      • Check exposure
      • Check that both sensors can “See” the feature.
8 Likes

This would be true for 3D modeling , but not exactly true in this case here .
Resolution has nothing to do with scanner accuracy , you can have accuracy of 0.01mm with resolution of 0.1mm

The accuracy depends of the resolution of the projected pattern on the object and distance between the scanner and the objects . ( And of course the depth camera sensors that capture the pattern)

However the accuracy is not steady However the volume precision is .

The closest scanning distance that the device offers provide the best accuracy , Mini at 10cm, POP2 at 15cm, Range at 30 cm

How far the distance between the object and scanner , how lower the accuracy .
Why? Because of the resolution of the projected pattern lines that will get bigger with the greater distance , the sensors registering less finer destruction details in the pattern producing less details on the surface .

However, it is possible to archive more details while meshing the fused point cloud outside the limitation of the software .
Since the software don’t use every data of the point cloud , as that would result in a monster mesh with huge size and hard to edit after .

Especially true when scanning very big objects around 2 meters, merged or not .
Normal size objects works just fine .

And one note : the accuracy you mentioned are specification accuracy what is above the actual true numbers .

According the measurements Range reaching accuracy from around 0.1mm to 0.3mm depends of the scanning distance . It would never capture the details if the true accuracy would be just 0.3mm .

Range has a point distance of 0.1mm
Pop2 has a point distance of .05mm
Mini has a point distance of .02mm

According to my point distance measurements while beta testing the prototypes .

1 Like

Yes, you’re correct - And I was making large generalizations about accuracy/resolution here to make it more user friendly to beginners who are not familiar with the principles of scan hardware and precision vs. repeatability, etc.- I realize there are other factors that determine resolution/accuracy.

My points on accuracy is specifically talking about lenses and depth of focus (I am not an engineer of optics, photographer, or an expert on this - I just have practical experience of using these systems).

Assuming sensor (pixels) are fixed - if you have wide lenses - you capture more data but your points or number of pixels are spread thinner than if you have narrow lenses… The way I think of this is pixel-density (Pixels per square inch - or cubic inch). Devices like Range have less “resolution” in my explanation because the pixel density of its lenses are lesser than that of the mini… However the projection resolution and pattern also have an important role in it as well…

My simplified statements on accuracy was to explain to those not familiar with any structured light technology to understand that blue light vs. infrared light is a moot argument (but does reduce errors from light pollution and noise) and that the software has more impact on quality of scan than the physical hardware as hardware limitations - except the scan volume limitation and point distances (both are proportional to the hardware optics) can be overcome.

I also agree the point distances of the Pop2 are much better than what is advertised - however - I used advertised specifications as they are in the literature and did not want to confuse anyone (or pull numbers out of my rear end).

1 Like

That’s ok I like what you prepared here , very good job @jheins3 and very informative for new users .

The Range lenses are actually new and improved depth sensors and better than MINI .
But because of the wide position it can’t capture higher density and is limited to 30 cm with the FOV .
The camera resolutions really don’t matters here much but the quality of the glass lenses do , the distance and resolution of the pattern plays significant role of the capture of frame cells .
Revo scanners are structured light scanners that do not capture images of the surface . Only the changes of the infrared laser pattern on the surface itself and nothing else , so how cleaner the infrared reflective pattern is captured by the sensor how better the accuracy of the cell frames .
Imagine a night vision , and only object that reflects infrared light are visible by the sensors ( in this case the distorted pattern )
For that reason reflective surfaces can create so much noises reflecting scattered light back into the sensors .

I know many people think here just about depth camera function , but that is not exactly what is happening while capture , as only the infrared pattern is captured reflecting back the IR light to the sensors and the rest don’t matters .

If Revopoint was allowed to use much stronger infrared laster than only 1 class IR laser , you would get even better results regarding accuracy . For that reason Blue laser was used in MINI to reduce the noise and improve the resolution at the distance of min 10 cm …
How finer and sharper the pattern , how higher the accuracy will be .
The DLP projector shutting vertical lesser lines at 2 separate angles back and forth , one for the left and one for the right sensor that recreate the surface in real time from the reflections in the main window (OpenGL)

The left top preview window is just video feedback of the left depth Sensor and do not represents the actual view , just for references.

That’s very good … I do agree , neither Revopoint want to do it , they downgrade the specifications on purpose , it is better to get a better results as surprise than buying a lies … or create controversy online .
But that is also good to have it explained for a new users that would expect a 0.05mm accuracy at any range , but that’s not how it really works , the accuracy is between 2 points , not between 2 features or measurements via a software , you need a professional tools for that , that are very expensive and used mostly in land for certificates .

I saw many people mislead others with the findings not educating themself regarding the hardware but only using their common sense , and as we all know, common sense is not common at all :wink:

Anyway thanks for the tread , really nice of you to create it and take the time to write it down …

2 Likes

Hey Hi,
Thank you for all of the great tips,

it is called CloudCompare, ( CloudCompare - Open Source project )
It is actually very powerful and can handle huge pointclouds in many formats.
I use it nearly ever day for cleaning / subsampling / segmenting pointclouds, QA of asbuilt 3D models, Meshing etc.
Dont forget to donate to Daniel if you use it for commercial work :slight_smile:
Mike.

1 Like

Thank you! I’ve updated my post to reflect the correct name. I’ve not used it yet but plan on giving it a try!

One of the best and most important contribution in the forum. Thank you very much. It will greatly improve my experience with all scanning scenarios. I appreciate.

1 Like

You mentioned OpenGL used for determining the surface from the fringes, if it were ported to Metal would it be possible to improve utilization of mac graphics hardware for improved performance?

You also noted improved lenses on range compared to mini - how does the lens quality in the Pop 3 compare to mini/pop2?

Hi @mikefish the OpenGL do not have any influence over the performance , it is simply a display guide , the processing already happening inside the processor (CPU ) of the scanner and what you see is actually just data feedback as a reference ( visualization) The magic of Revopoint scanners is to not relay on your computer power or performance while scanning , what makes it truly portable .
The only thing you can improve is fast connection between the scanner and your mobile or computer device as snow, or low data connection will interrupt the performance.

I don’t have exact specification on the Depth Camera lenses , but the RGB camera have 30% larger aperture letting more light reach the sensor. Together with the 2 dual white LEDs that illuminate and remove shadows from the scanned object, POP 3 can capture more vibrant colors.
I know the Range Depth camera lenses was updated becouse Range needed to capture objects from larger distance , and the Depth camera sensors are very powerful as well as I tested distance to 3 meters , so very powerful .

I did not go very deep with POP3 to push its limits yet since I was busy with all stuff before the release , but I retired my POP 2 and donated it already , as POP 3 do exactly what I always want it to do , the white LEDs was on my Wishlist for so long so happy the designers decided to finally add it to the scanner as it improves so much scanning medium size assets in color as never before .
Now don’t forget about the POP 3’s 9-axis IMU (Inertial Measurement Unit) that continuously measure the scanner’s movement eliminating at the same time faulty frames caused by sudden movement or shaking.
I could scan an apple without a marker mode for first time ever what is a huge progress .

2 Likes

A very helpful description of a lot of the limits I’ve encountered trying to scan items outdoors with my Range 2.

I am traveling to a UK aircraft museum and will attempt to scan the intake shapes of a large aircraft for a 3D printing/modeling project. You can usually count on outdoors to be overcast, but you never know. And because the top of the wing is above my head, I have an extendable pole mount to assist, but unfortunately it won’t let me view my iPhone to determine if I’m in the sweet spot for useful scanning information.

Bringing markers just in case.

1 Like

Finally,
I’ve found the EXACT information I have been too lazy to look for properly, but needed badly in order to properly bond with my scanner…

I’ll delete the Gumtree listing (like ebay) of it now, remedy the 18 things I was doing wrong and ready myself to try and scan again