Projectors... and Projector Mods.

Discuss about cameras, projectors, calibration panels, turntables etc.
mading
Posts: 307
Joined: 31 Jan 2017, 13:09

Re: Projectors... and Projector Mods.

Post by mading »

After restart it went again to 60fps at 1080.
But it was stuck at 15fps even with ICCapture.
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: Projectors... and Projector Mods.

Post by Micr0 »

Last night I made what I think are the most positionaly accurate 125mm panels I've ever made. I took 2 pieces of 1/4" (.635mm) glass which the CMM Measured at +-.06mm over 70mm for flatness. I covered them with flat black 3mil sign vinyl under a layer of flat white vinyl. I indicated them in the VMC and used a 1/4" end mill to remove the white vinyl and create the calibration pattern.

Youtube video here:

https://youtu.be/jZ25JoE7JgE

I set up the camera and projector and ran the calibration routine with dual camera optimization off, and initially got RMS error of ~.26. FWIW I milled registration marks into the vinyl so I could check the alignment of the panels and used the CNC'd aluminum holders I made so I know the panels themselves were positioned well. Doing the test where I scanned the panels, fine aligned them and checked the deviation (at .1 and .05mm) I was getting better results then the last. But still not great.
Cal Panel 2-10-2018.jpg
I did some experimenting with turning optimization on and off (off is better), and with focusing the cameras and or projector to different depth points in relation to the cal corner, and lastly moving the cameras and projector as s unit to different angles relative to the corner. Surprisingly everyone of these changes resulted in a different RMS number. I finally found a generally repeatable formula for set up that gave me the lowest RMS # of ~.2 for these panels. I also tried re scanning the Magneto body (it was still attached to my rotary) and got excellent results ( I forgot to take screen shots and don't have my license with me so I cant open the file). I can't say the results were that much better than any I have gotten before but they were definitely very good.

All this does make me question what exactly is the RMS error an aggregate of and more specifically what are the biggest influences on the number? Likewise how important that indicator?

The only variable I haven't been able to really change is my projector. I can see some of it's shortcomings (the lens I think is a big one). I do plan O trying a few more things. I'll post when I do.

Thoughts?
µ
bigbomber
Posts: 114
Joined: 14 Dec 2017, 18:08

Re: Projectors... and Projector Mods.

Post by bigbomber »

Micr0 wrote: 11 Feb 2018, 16:11 Last night I made what I think are the most positionaly accurate 125mm panels I've ever made. I took 2 pieces of 1/4" (.635mm) glass which the CMM Measured at +-.06mm over 70mm for flatness. I covered them with flat black 3mil sign vinyl under a layer of flat white vinyl. I indicated them in the VMC and used a 1/4" end mill to remove the white vinyl and create the calibration pattern.

Youtube video here:

https://youtu.be/jZ25JoE7JgE

I set up the camera and projector and ran the calibration routine with dual camera optimization off, and initially got RMS error of ~.26. FWIW I milled registration marks into the vinyl so I could check the alignment of the panels and used the CNC'd aluminum holders I made so I know the panels themselves were positioned well. Doing the test where I scanned the panels, fine aligned them and checked the deviation (at .1 and .05mm) I was getting better results then the last. But still not great.

Cal Panel 2-10-2018.jpg

I did some experimenting with turning optimization on and off (off is better), and with focusing the cameras and or projector to different depth points in relation to the cal corner, and lastly moving the cameras and projector as s unit to different angles relative to the corner. Surprisingly everyone of these changes resulted in a different RMS number. I finally found a generally repeatable formula for set up that gave me the lowest RMS # of ~.2 for these panels. I also tried re scanning the Magneto body (it was still attached to my rotary) and got excellent results ( I forgot to take screen shots and don't have my license with me so I cant open the file). I can't say the results were that much better than any I have gotten before but they were definitely very good.

All this does make me question what exactly is the RMS error an aggregate of and more specifically what are the biggest influences on the number? Likewise how important that indicator?

The only variable I haven't been able to really change is my projector. I can see some of it's shortcomings (the lens I think is a big one). I do plan O trying a few more things. I'll post when I do.

Thoughts?
Hi and thanks for your tries and tests.
How thick is white vinyl you drilled to black one? my scanner can scan a 0.1mm thickness of paper lied on a flat panel from 1Meter distance to my scanner. so I think if you use dents and deep marker points, your scanner can not see them correctly, specially you should think about shadows that projector light can cause.
About RMS, some times I get strange value too. can you discus more about repeatable formula you found?
About projector, I can help.
User avatar
Micr0
Posts: 586
Joined: 15 Nov 2016, 15:20
Location: New York City

Re: Projectors... and Projector Mods.

Post by Micr0 »

The white vinyl is about .06mm thick. I did think about the recess as being an issue. However last week I made a set of 225mm test panels out of an old melamine covered shelf board by drilling (in the same VMC BTW) the holes 20mm deep and painting the inside of the holes with black paint. The drill didn't leave very clear round holes as it chipped slightly the laminate. These panels gave me an RMS error number of .18. As for shadows, It would be the white vinyl casting a shadow on the black so I doubt shadows would be an issue. I also don't think the resolution of the cameras is going to see any noticeable (to David) distortion from a .04 - .06mm deviation in the surface.
µ
User avatar
Micr0
Posts: 586
Joined: 15 Nov 2016, 15:20
Location: New York City

Re: Projectors... and Projector Mods.

Post by Micr0 »

Tonight I made a set of 280mm panels out of melamine in the VMC. I put a couple of registration features to help align them during assembly into the corner. They worked great. RMS Error .12 and subsequent scans are absolutely excellent. My only worry with these is that the Melamine isn't really the best material for this and I'd like something more like glass or ground aluminum fixture plate. That said I did deviation measurement in David on a .1mm scale and the results are better but not great. The areas that are not green I am suspecting are cause by distortion induced by my projector. I'm pretty sure it's not coming from the cameras as I have very high quality lenses etc and a home projector usually don't.

I re scanned one of the engine parts I had lying around.
2018-02-13_21-01-28.jpg
2018-02-13_21-06-47.jpg

Setup.Custom.hp3dscansetup ...
Trying to connect forcing that format...
Connected.
Trying to connect forcing that format...
Connected.
Number of cameras: 2
Is system calibrated: yes
-> Setup is 'inner corner'
- RMS error is ok: 0.12224
<camera_model>CameraTsai</camera_model>
<cx>956.8791016937434506</cx>
<cy>582.70752417102187337</cy>
<f>3161.9658153128470985</f>
<sx>1.0011455156514770248</sx>
<kappa1>8.3600926419601409464e-09</kappa1>
µ
3dxcan
Posts: 74
Joined: 02 Jun 2017, 17:21

Re: Projectors... and Projector Mods.

Post by 3dxcan »

the problem of scan distortion is norrwed down on projector lens, we all have tried different possible causes, thats what i believe is the bottleneck here. I havent been able to find a projector with exchangable compact lenses which would be as reliable as the camera lens in terms of lower distortion.
User avatar
Micr0
Posts: 586
Joined: 15 Nov 2016, 15:20
Location: New York City

Re: Projectors... and Projector Mods.

Post by Micr0 »

David is supposed to calibrate around some of that but I have no idea how and or how successful it really is. Bomber says there are other factors in an unmodified projector at play too.

Of course, Ideally it is better to eliminate the source of error then to try a software workaround.
µ
NARMELLA

Re: Projectors... and Projector Mods.

Post by NARMELLA »

Micr0 wrote: 14 Feb 2018, 02:05 ...

I re scanned one of the engine parts I had lying around.

2018-02-13_21-01-28.jpg

2018-02-13_21-06-47.jpg


Setup.Custom.hp3dscansetup ...
Trying to connect forcing that format...
Connected.
Trying to connect forcing that format...
Connected.
Number of cameras: 2
Is system calibrated: yes
-> Setup is 'inner corner'
- RMS error is ok: 0.12224
<camera_model>CameraTsai</camera_model>
<cx>956.8791016937434506</cx>
<cy>582.70752417102187337</cy>
<f>3161.9658153128470985</f>
<sx>1.0011455156514770248</sx>
<kappa1>8.3600926419601409464e-09</kappa1>
Amazing .
bigbomber
Posts: 114
Joined: 14 Dec 2017, 18:08

Re: Projectors... and Projector Mods.

Post by bigbomber »

Hi friends. This is my own self-made projector specially designed to work with SLS systems. It directly drive one single crystal chip to display photos. Light is emited by a 20W LED. I plan to change the LED with laser to get focusless projector.
Screenshot_2018-02-26-13-46-18-1.png
mading
Posts: 307
Joined: 31 Jan 2017, 13:09

Re: Projectors... and Projector Mods.

Post by mading »

Well, it seems quite different from a standard dlp!
What is the main difference? No color wheel?
I'm happy to see something really in topic! :)
LG PF50, LG PF1500, RangeVision DIY: 2x DahengMer630, 2X12 and 2X16 mm 5Mp ZK lenses, RV turntable
Post Reply