First Scans With Flex.

What have you scanned recently?
User avatar
Micr0
Posts: 586
Joined: 15 Nov 2016, 15:20
Location: New York City

First Scans With Flex.

Post by Micr0 »

Got my first chance to spend some time with FlexScan. I did a careful calibration with 15mm cal panel and scanned the gear case that I had posted here a month ago. I tried to be a through as possible so I could compare apples to apples. The settings in flex are all at the default including smoothing and clean up. I have a bit of a sync issue between my projector and IS cameras and it is leading to a bit of surface roughness. This required applying a bit more smoothing than I would like to use normally. David seems to be a bit less sensitive to this but I haven't tweaked anything in flex so I'm sure I can improve things here. Here are screen grabs of the final model (red-Flex, Blue/Green- David). The final mesh in both is a fusion of about 65 scans.
FS Gear Case 1.jpg
FS Gear Case 3.jpg
GearCase2-1.jpg
µ
Alonso
Posts: 109
Joined: 05 Dec 2016, 05:55
Location: USA

Re: First Scans With Flex.

Post by Alonso »

Yea the david/HP scan looks sharper for the reasons you mentioned.

I'm curious as to which software was faster when aligning the scans? That's the part of HP3DScan that always irked me the most

Also roughly what size is the part? I'm just trying to get a better idea of how big of a thing you can scan in relation to the size if the calibration board you used.
contactpolyga
Posts: 73
Joined: 26 Nov 2020, 23:26

Re: First Scans With Flex.

Post by contactpolyga »

Looks like a great first scan! Scan looks nice for your first test.

As you've mentioned the sync issue might be causing some loss of sharpness, the only other issue I can think of is if smooth or precise merge was used.

Lastly FS3D is sensitive regarding the calibration panel, so for the last touch of sharpness, we generally recommend using one of our cal boards or a precise checkerboard, as you'll lose that last bit of of crispness due to a bit of calibration drift.
mading
Posts: 307
Joined: 31 Jan 2017, 13:09

Re: First Scans With Flex.

Post by mading »

Nice job Micr0!
contactpolyga wrote: 28 Jan 2021, 18:26 Lastly FS3D is sensitive regarding the calibration panel, so for the last touch of sharpness, we generally recommend using one of our cal boards or a precise checkerboard, as you'll lose that last bit of of crispness due to a bit of calibration drift.
I was suprised by the low number of projections/pattern required by FS3D.
Is there a way to tune the number or dimension of stripes to increase the sharpness?

I understand the importance of patterns. Mine were printed with an inket, the worst scenario.
But I started in David with the same printed pattern's quality, and I got good sharpness (and bad alignments) very soon.

Differences in the two scans of Micro are evident. We have to say that he comes from a huge experience in 3D scanning with David HP.
Do you think this sharpness can be easily matched tuning FS3D?
LG PF50, LG PF1500, RangeVision DIY: 2x DahengMer630, 2X12 and 2X16 mm 5Mp ZK lenses, RV turntable
User avatar
Micr0
Posts: 586
Joined: 15 Nov 2016, 15:20
Location: New York City

Re: First Scans With Flex.

Post by Micr0 »

contactpolyga wrote: 28 Jan 2021, 18:26 Looks like a great first scan! Scan looks nice for your first test.

As you've mentioned the sync issue might be causing some loss of sharpness, the only other issue I can think of is if smooth or precise merge was used.

Lastly FS3D is sensitive regarding the calibration panel, so for the last touch of sharpness, we generally recommend using one of our cal boards or a precise checkerboard, as you'll lose that last bit of of crispness due to a bit of calibration drift.
Re: Calibration,

Would that be reflected in the values reported for the reprojection and error deviation?
What would ideal Reprojection Error and Error Deviation numbers be?

For this scan/calibration I have:

Reprojection Error = 32.69µ
Deviation Error + 19.7µ

BTW I get Deviation error , but What is "Reprojection Error"?
µ
contactpolyga
Posts: 73
Joined: 26 Nov 2020, 23:26

Re: First Scans With Flex.

Post by contactpolyga »


I was suprised by the low number of projections/pattern required by FS3D.
Is there a way to tune the number or dimension of stripes to increase the sharpness?

I understand the importance of patterns. Mine were printed with an inket, the worst scenario.
But I started in David with the same printed pattern's quality, and I got good sharpness (and bad alignments) very soon.

Differences in the two scans of Micro are evident. We have to say that he comes from a huge experience in 3D scanning with David HP.
Do you think this sharpness can be easily matched tuning FS3D?
Yes, there's advanced settings with a bunch of settings including number of patterns - increasing them will improve detail slightly.

On sharpness in FS3D, yes definitely given that FOV, i would expect it should be producing sharpness closer to the David output, although I don't know the exact issue on why it's much worse but Projector, camera and calibration variations or camera settings as well that the IS camera driver is pretty new.

For existing examples of semi pro FS3D DIY configuration, I think this is one of the coolest examples. It's not quite DIY it's the same general idea...
https://www.mechscan.co.uk/macro-3d-scanner/

If there's time in the next couple weeks, i can ask one of the guys here to put together a reference system with reference scans -
contactpolyga
Posts: 73
Joined: 26 Nov 2020, 23:26

Re: First Scans With Flex.

Post by contactpolyga »

Micr0 wrote: 28 Jan 2021, 22:12
Re: Calibration,

Would that be reflected in the values reported for the reprojection and error deviation?
What would ideal Reprojection Error and Error Deviation numbers be?

For this scan/calibration I have:

Reprojection Error = 32.69µ
Deviation Error + 19.7µ

BTW I get Deviation error , but What is "Reprojection Error"?
Those numbers look pretty good given a 200mm or so FOV. TBH it's been many years so i don't remember the exact math on these numbers but as far as I remember.

Reprojection is that it took given a known, position pixel location, it ran the triangulation and that's how far it was off from your calibration board. ie. Your square size was suppose to be 10mm given the calibration, it took those pixels and ran it through the calibration and that's how far it was off from the ideal size it was suppose to be.

Deviation if i remember correctly is the step size from a single pixel, x to x or xy to xy..
User avatar
Micr0
Posts: 586
Joined: 15 Nov 2016, 15:20
Location: New York City

Re: First Scans With Flex.

Post by Micr0 »

contactpolyga wrote: 29 Jan 2021, 00:16
Micr0 wrote: 28 Jan 2021, 22:12
Re: Calibration,

Would that be reflected in the values reported for the reprojection and error deviation?
What would ideal Reprojection Error and Error Deviation numbers be?

For this scan/calibration I have:

Reprojection Error = 32.69µ
Deviation Error + 19.7µ

BTW I get Deviation error , but What is "Reprojection Error"?
Those numbers look pretty good given a 200mm or so FOV. TBH it's been many years so i don't remember the exact math on these numbers but as far as I remember.

Reprojection is that it took given a known, position pixel location, it ran the triangulation and that's how far it was off from your calibration board. ie. Your square size was suppose to be 10mm given the calibration, it took those pixels and ran it through the calibration and that's how far it was off from the ideal size it was suppose to be.

Deviation if i remember correctly is the step size from a single pixel, x to x or xy to xy..
Thanks

I did a bit of manual editing and cleaning and tried Precise Fusion. It's a bit better.

I'll mess around with it and re scan it when I get a chance.
FS Gear Case 4.jpg
µ
User avatar
Micr0
Posts: 586
Joined: 15 Nov 2016, 15:20
Location: New York City

Re: First Scans With Flex.

Post by Micr0 »

contactpolyga wrote: 29 Jan 2021, 00:05
Yes, there's advanced settings with a bunch of settings including number of patterns - increasing them will improve detail slightly.

On sharpness in FS3D, yes definitely given that FOV, i would expect it should be producing sharpness closer to the David output, although I don't know the exact issue on why it's much worse but Projector, camera and calibration variations or camera settings as well that the IS camera driver is pretty new.

For existing examples of semi pro FS3D DIY configuration, I think this is one of the coolest examples. It's not quite DIY it's the same general idea...
https://www.mechscan.co.uk/macro-3d-scanner/

If there's time in the next couple weeks, i can ask one of the guys here to put together a reference system with reference scans -
I'd be very interested to know more about how he adapted the lens to that Acer projector. :D
µ
contactpolyga
Posts: 73
Joined: 26 Nov 2020, 23:26

Re: First Scans With Flex.

Post by contactpolyga »

Well you remove the old one and put in a new one.. :lol:
Post Reply