Few questions about ScanCenter

Write here about construction, ideas, equipment, tips n tricks etc. related to structured light scanning
User avatar
OBNRacerMan
Posts: 372
Joined: 12 Aug 2019, 09:46
Location: Obninsk, Russia

Re: Few questions about ScanCenter

Post by OBNRacerMan »

mdesign wrote: 25 Apr 2022, 17:33 How much I can move the calibration panel between vertical and tilted modes. I`ve thought that If I`m able to move the base of the calibration panel then it`s equal to moving a tripod with a projector/camera.

Maybe it`s needed to stay the base in the same place and only move calibration panels from vertical to tiled position with remain in the base primary position. What do you think about that?
I made a small video on which I am calibrating for the archival "4th" zone of size XS (the smallest one that RV had in the old days (then it was abandoned, as well as the largest zone 500 * 500mm)).
Here is a link to this video on Google Drive
Soft: RangeVision DIY, Calibration panel/Rotary tables - RangeVision, Prj: LG PF1500G, Cam: Daheng Mercury MER2-630-60U3M-L (USB3.0, monochrome) x2, Obj: ZLKC FM12036MP5 (F2.8/5Mp/12-36mm) x2. And a handheld scanner Creality Ferret
mdesign
Posts: 37
Joined: 03 Jan 2022, 14:51

Re: Few questions about ScanCenter

Post by mdesign »

Thanks for the movie. Now it`s clear to me. So I can move calibration panels to hit the centre of the camera view.
mdesign
Posts: 37
Joined: 03 Jan 2022, 14:51

Re: Few questions about ScanCenter

Post by mdesign »

I have two additional questions:

1) I`m scanning with a middle size calibration panel.
I`ve scanned measure tape 1.5m with markers and I have a 3.5mm mismatch per 1.5m scans.
So I`ve measured a single patch scan to see how accurate a single patch scan is.
You may see a single patch scan below. It has a 0.76mm mismatch on every single scan. I know it`s not a perfect measure of error but I know something is not ok if I lose 3.5mm per 1.5m. I`ve noticed that I have a camera angle of 10.62 instead of 10.65 as you write above (it`s my error). I think that should not influence bad measures so much.

Question: Have you checked the measures of your scanner before? Maybe that 0.5mm per single patch scan is normal I will not reach anything better with that (after calibration with middle size calibration panel).
measure.jpg
2) Maybe it`s involved with the last point.
My 12mm cameras with that LG projector have a quite big waste of projector screen size. Maybe it decreases my accuracy.
waste.jpg
Is it allowed to push back cameras into the middle of a projector (it won't work with the smallest calibration panel)? I would like to make the camera view frame more equal to the projector picture frame. Maybe it will increase the accuracy of each patch. What do you think about it?
User avatar
OBNRacerMan
Posts: 372
Joined: 12 Aug 2019, 09:46
Location: Obninsk, Russia

Re: Few questions about ScanCenter

Post by OBNRacerMan »

I only once (and then, especially for the test) did the calibration on the middle zone "M". The main "working" zones for me are "S" and "XS". Therefore, it is difficult for me to say offhand what kind of run-up comes out on large calibration zones ... Perhaps I will find time to repeat the test more carefully, at least on the "M" zone.

The accuracy itself depends on the calibration zone (the larger the zone, the lower the resulting accuracy). Also (to a lesser extent) the quality of the calibration plays a role (the value that the program shows in the main window (where the current zone and accuracy are indicated)).
According to the size of the projection and the field of view of the camera - first you need to make sure that the zoom in the projector is set to the minimum.
You also need to understand that a lens with a focal length of 12mm, originally designed for a 2/3" matrix, will have a real value of about 14mm on a 1/1.8" matrix. However, due to such "cutting", possible edge distortions of the optics are almost completely removed. The "extra" edges of the projection are really not used, but only when projecting onto a plane that is perpendicular to the camera axis. When a real object is being scanned, this size can change in the frame. Also, at the edges of the projection, the sharpness of the patterns drops a little (that's why the "cropping" of the frame helps here too).

The accuracy of the scan itself, first of all, depends on how accurately the calibration table is measured at the factory (this value is written on it). But only the detail (density of points) that the scanner can produce depends on the ratio of the projection size and the field of view of the cameras - the number of scanned points itself is (conditionally) unchanged for any size and is determined (ideally) by the number of pixels of the camera matrix (and, to a much lesser extent - on the resolution of the projector (more precisely, the density of the patterns that it creates on the object)). But the minimum distance between the points of the resulting "cloud" is directly related to the size of the object.

P.S. As a result - if you really want to "adjust" the field of view of the cameras to the size of the projection, then you can reduce the distance between the cameras and the projector, provided that they move (if you look at the attached picture) along the "red" line, and not along the "blue" ".
Changing the distance from the cameras to the projector
Changing the distance from the cameras to the projector
P.P.S. I am currently testing 12-36mm varifocal lenses in an attempt to save on buying two extra pairs of 12mm and 25mm (in addition to the 16mm I have). Due to their large dimensions, they even had to make new adjustable mounts to increase the angle in the "XS" zone, avoiding the complete blocking of the light flux from the projector from too close lenses :)
Size comparison between 16mm and 12-36
Size comparison between 16mm and 12-36
Here the scanner is set to the XS zone
Here the scanner is set to the XS zone
Soft: RangeVision DIY, Calibration panel/Rotary tables - RangeVision, Prj: LG PF1500G, Cam: Daheng Mercury MER2-630-60U3M-L (USB3.0, monochrome) x2, Obj: ZLKC FM12036MP5 (F2.8/5Mp/12-36mm) x2. And a handheld scanner Creality Ferret
mdesign
Posts: 37
Joined: 03 Jan 2022, 14:51

Re: Few questions about ScanCenter

Post by mdesign »

Thanks for your advice. I like your varifocal ZLKC lenses but I`m afraid of light loss with that type of lens.
I have to improve my scanner build because it has some bad sides. I`m working to have fewer vibrations and to improve overall quality.

I had an error also because I moved the projector zoom handle to the wrong side. It was zoomed on max. My bad. After that, all changed in the right direction.
mdesign
Posts: 37
Joined: 03 Jan 2022, 14:51

Re: Few questions about ScanCenter

Post by mdesign »

OBNRacerMan wrote: 03 Apr 2022, 17:30 The vertical angle of deviation of the optical projection from the lens axis can be calculated experimentally - by placing the projector on a flat surface / table at a distance, for example, 1 meter, from the wall. Then measure the height from the table to the center of the lens and from the level of the table to the center of the projection - then calculate the value of the angle. For LG PF1500/HF60 projectors this angle is 10.65° - this is exactly the angle I set when designing the "paws" on which a linear rail with carriages for cameras is installed.
Have You measured that 10.65° angle when the projector stood on the table or was screwed by 3 screws in the bottom?

I`m asking because I`ve noticed that when the projector stands on its legs then the front of the projector is a bit lower than the rear.
projector.jpg
User avatar
OBNRacerMan
Posts: 372
Joined: 12 Aug 2019, 09:46
Location: Obninsk, Russia

Re: Few questions about ScanCenter

Post by OBNRacerMan »

This angle was not calculated by me personally and, as far as I know, the projector at that time was simply installed on a plane.
It may be necessary to use shims when setting up the geometry of the setup with mechanical fastening.
Soft: RangeVision DIY, Calibration panel/Rotary tables - RangeVision, Prj: LG PF1500G, Cam: Daheng Mercury MER2-630-60U3M-L (USB3.0, monochrome) x2, Obj: ZLKC FM12036MP5 (F2.8/5Mp/12-36mm) x2. And a handheld scanner Creality Ferret
mdesign
Posts: 37
Joined: 03 Jan 2022, 14:51

Re: Few questions about ScanCenter

Post by mdesign »

So if the angle was measured on legs then cameras should be installed on the projector base also when it stands on legs. My bad was that I bought a 15cm arca swiss quick release plate which was shorter than the legs so the whole projector was more flat. My lesson is that If I will use your angle I should place it also on the legs (not on the three mounting screws) because then the camera angle will be different than 10.65.

That`s how I did it:
flat.jpg
That`s how it should be done if I will use your 10.65 degrees angle:
on_legs.jpg
Am I right?

For me, it would be better to mount it to three screws (which are used to mount it to the ceiling) because there I will be sure about the perpendicular angle between the projector and the rail. For that, I should measure that angle and I`m a little afraid if I will do it right.

For now, I think about designing the solution to use all four screws and to stand on legs.

Using three or all four has also one risk: vibrations. Legs have got rubber pads to decrease vibrations. When you use screws for mounting the projector there would be needed to give there some silicone pads to remove vibrations. I`m working on it.

I hope to remove the vibrations alert in ScanCenter. I hope it will increase quality and accuracy a bit too.
User avatar
OBNRacerMan
Posts: 372
Joined: 12 Aug 2019, 09:46
Location: Obninsk, Russia

Re: Few questions about ScanCenter

Post by OBNRacerMan »

I just now noticed that your base plate strongly protrudes back and goes under the rear "soft" support of the projector, while my plate is shorter and ends at the mounting holes.
photo_2021-11-29_22-02-41 (3).jpg
photo_2021-11-29_22-02-41 (3).jpg (12.67 KiB) Viewed 5750 times
Soft: RangeVision DIY, Calibration panel/Rotary tables - RangeVision, Prj: LG PF1500G, Cam: Daheng Mercury MER2-630-60U3M-L (USB3.0, monochrome) x2, Obj: ZLKC FM12036MP5 (F2.8/5Mp/12-36mm) x2. And a handheld scanner Creality Ferret
mdesign
Posts: 37
Joined: 03 Jan 2022, 14:51

Re: Few questions about ScanCenter

Post by mdesign »

I have both versions because I mount it on a quick release plate and I have short and long versions. So I can choose as I want but I don`t know which version is right to have that 10.65degrees angle valid.

I don`t see that in your photo. Have you raised your version to the height of the projector legs?
Is your projector tilted forward (the front leg looks to be shorter than the back one)?

Another question is how it was measured. If it was laying flat on the table during measurements as you said before, then it should lay in the same way on the scanner base.
Post Reply