This morning I decided to do a proper test of the possibilities of the Range, I have to say I am very impressed with it, and once I found the right workflow you can work extremely fast and efficiently.
I started with a scan of the engine bay, but it took me just a few minutes to scan and process, so I decided to keep pushing until I could, so I also scanned both wheelarches and the interior of the car. In total I have done 10 scans:
-General engine bay (3 scans for maximum detail, but could be only 1)
-Deep pocket area in the engine bay
-Left wheelarch
-Right wheelarch
-Interior from the left
-Interior from the right (to cover all shadows on the left)
-Interior roof area
-Dashboard detail
I have done all the scans using my phone, and I have sent the RAW data to the laptop to do the processing.
Unfortunately, the align feature in RevoScan is still not powerful enough when you need accurate alignment, so I had to use a external software tool for that.
The main problem I had when aligning the different scans is that RevoScan only allows you to align two point clouds, and when zooming in, it’s difficult to pick up one specific detail. If it was possible to align two meshes, this would be easier.
Also, in geomagic design you can just use your three dots, or you can go further and allow the software to do an additional local adjustment based on your picks, which usually results in an almost perfect match.
Buy yourself some sticker dots in a neon color , stick them in the places where you going to make partial scans , and scan using color , the sticker dots later would be guide for reference when merging using marker mode if your object don’t have as much features .
The feature where you can align 2 point cloud manually will return back in the future , it is right now available in Revo Studio as well .
Point cloud offers higher level of points , however the merging feature is not as sophisticated as in other programs , the registration is rather simple .
Remember,
You can always use CloudCompare (Free) if you require a more robust alignment/registration of meshes and pointclouds. Alignment and Registration - CloudCompareWiki.
Mike.
You have done a great job, can I ask what scan settings you used on your phone ?
I just cant get the same results on two pretty average white cars, scanned after sunset here in our Australian winter.
I am using feature mode on my android tablet, Marker Mode doesnt work at all.
These are both pretty perfect scanning conditions, can anyone offer any suggestions what I might be doing wrong ?
What would be the optimal scanner settings for this type of project ?
How would the Revopoint team handle this sort of project ?
Mike.
on standard accurracy/fast scanning you can choose very big object which gives you larger FOV. I would try that with feature mode and let those markers as features.
Hi Mike , the features/markets need to be separated between 4 to 8 cm from each other to works in Marker mode .
There was some issues with Range and marker mode but I think the last version fixed it .
As Ivan stated already , you can use Standard mode with Large object to get bigger FOV than body mode offers .
And please don’t put them in a row , you need to break the symmetry when placing features or markers , the same pattern will confuse the software .
There is one thing I’d recommend you to check… There are obviousely a lot of markers that are “double faced Janus” to the scanner. So, check it out: wouldn’t those make more problems to Range than solving them. I mean maybe a single markers will make it better. I don’t know this, it’s just an assumption.
Hi Mike
You know that I use the same since the beginning and they works the best for handheld scanning , early versions of Revo Scan 5 had issues with marker mode but it got fixed in the last release 5.0.7
The size of the markers really don’t matters , as long they are reflective enough .
Try to add some LED lighting to support the markers better , when is too dark markers not always works well .
I use only reflective markers .
If you have the markets apart 4 to max 8 cm in the FOV it should be ok , this get also for the white cubes .
If using white cubes with lots if flat surfaces , change the mode to Body Mode or Large object , you can’t try standard mode but I don’t think it is nesesery.
In my case I was using high accuracy and feature mode. I changed between dark object and general object depending on the specific area.
As you can see in my pictures, i didn’t scan any large body panels for the same reason as you: the feature mode didn’t work properly. However this was before the release of RS5.0.7 version, which theoretically improves this. I still have to check it.
Also, my car is inside an indoor garage illuminated with standard fluorescent lights, so this might improve the scanning results compared to scanning outdoors.
I also scanned another car very recently, which i haven’t showcased yet. This one was for a reverse engineering project for an actual customer. The conditions and the settings were very similar, however i used scan spray all over the place and took more time to ensure perfect results. Again, no body panels, only mechanical parts. And the results were extremely positive.
Generally I would recommend stick to what PUTV would say since she has a lot more experience =) I just point out that you can try this since I don’t know how does RevoScan inner logic behave in the situation where there are two close positioned markers and one of them can suddenly disappear when the cube will turn it’s side with the marker away from the scanner.
I ran some tests between the cubes with one sticker and the ones with 5 stickers.
The software certainly has an easier time with the single sticker cubes.(on my Tablet)
I ended up removing all but one sticker on each cube …~ 400
It makes sense, I spent over 10 years 3D scanning large buildings, bridges etc in the Surveying industry,
We would never put scan targets close to each other, or in a straight line as PUTV Mentioned.
