Big challenge here is managing the drift - this is test 1 and it looks like it’s possible.
I’ve done something similar with a $100,000 scanner and a boat, and still got drift over a few metres - so for the POP to manage this is very impressive and the drift is not only expected but you can manage it if you work clever.
Probably going to make a physically measured 3D model armature to mount the panels on and test the scan accuracy before accepting the data, piece by piece and then stitch it together - Gotta brush up on my Blender skills 1st and get rid of them holes where the dots are
Wow that’s impressive Jon. Can you elaborate a bit on how you use your tripod to get these kinds of scans? I’m picturing something on wheels that you slowly move around the chassis. Is that correct?
This was done by hand, the only time I used a tripod was when scanning the headlight pod well because I had to change the gain settings on the fly. Steady hands
Thanks, I’m upto 850 so far but no where near the end. The ones sent by Revopoint got seized at UK customs so they took a while to get here so I haven’t had time to let you guys know they arrived until now - many thanks.
@JonS - yes you can get scans that are un-usable, first few attempts were not ideal (see also the bumper scan I did on here) BUT I think I know how to get the best out of the device now. Basically I think it doesn’t like to be twisted or rotated out of plane (so hard to explain on a Monday morning) - so keep the scanner in the same orientation and scan like you’re airbrushing a model. Same technique for corners.
Its very tricky to do on your own actually, when I was using creaforms scanners, we often had two people working on the scan - one was watching the PC monitor to check the data, and the other was making sure the scanner was correct distance from the object etc. When someone got tired we swapped over. This is a similar situation where you need to keep your eye on two things.
The amount of drift I am getting is well acceptable for a device like this if used in a specific manner.
I’m lucky in that I have used top end scanners and CMMs so I know the little tricks you can use to get the best out of a device and realise why it’s not performing when a scan comes out poorly.
I’d love to do a youtube channel on this to explain better - just need to invent the 36 hour day first
I spent about 2 hours scanning the engine bay of the DeLorean last weekend and the results, while promising, are nowhere near what you’ve got here. Very impressive!
Cheers - its just a test right now to see if it can be done and where the accuracy is affected. As the body is stripped bare its hard to get the A / B pillars because they are so thin which throws the scanner off here and there (see the B pillar is a bit crazy on the screenshot) - but overall it did a great job.
I’m blessed the car is painted white though which allows the scanner to really dial back the gain and see the dots easier, I kept it in the ‘good’ range and was very liberal with the scanning motion. In the darker, black areas, I’m gonna have to spray it with white acrylic paint then wipe it off with alcohol when done - dots on a black/dark surface dont work too well.
I have had same problem with a bas relief into the dark area very bad…
I use a surface pro CP microsoft for making scan wich system you use…
wirh surface I keep onto a tripod so I have the two hand free… very usefull, my problem is lateral of the scan when I try to scan it loose the surface… like you say before if you change the orientation it loose very often the scan…
the part under the face that is black I have a hole into scan… for taking that I have made several scan… I don’t know better how to use the auto fnction of the scanner… and it’s difficolt to understand… and cannot find help on that… is difficolt also because the key to stop and start scan is very little and with my finger I must press two or three time to take the right pression… also looking the distnace and the gain and all by myself is difficolt…
Update… (object has been mirrored in post processing)
One thing I found after 3x solid days of scanning is if you’re doing a project like this you must be at TOO FAR distance. All the scans (probably did over 50 in total) that were in the Good/Excellent range heavily distorted. I’m guessing less is more…
I found scanning the car fast and fairly care free actually got better results - I’m guessing less points to stack tolerances upon, but faster seemed a lot better. The slower I went the more drift I got.
Too many dots actually caused problems as well, the computer doesn’t like it with 500x dots in the file and trying to line them back up after a miss scan can take a while.
Onto the details.
This is actually only 2x (two!) separate scans positioned in 3D software (not merged).
Is it accurate? Yes and no. Some dimensions are spot on, but some parts are distorted by 20mm or so which is not surprising really given that to do this properly you’d need to spend a LOT more money on hardware. The fact it sits on the CAD model chassis is actually quite mind blowing all things considered so there is definitely a workable solution here . There is just too much drift and the object is just way too complex. Still, the fact we even got here is astounding. With some editing you could do a better job by tacking multiple smaller scans onto an ‘armature’ with physical measurements but the time required is too much for me right now.
Where there are holes, a quick scan and merge can be done to patch things up, but again I have to move on with this project now - just thought I would share what can be achieved.
Possibly - I have no time to try it even if I had one now, the body is going off site soon. Just thought I’d share experience. The fact it got scans this good is really impressive so I am not complaining, as I say I know top end scanners that would struggle with this job.
