Scanning issues
Phil Kneen
I put a posting on the new Fabscan Pi web forum but as yet no one has replied so I am trying here:
Hi, I am having a few issue with the Fabscan, since the new software upgrade the scanner wont operate properly, I have tried a fresh SD card with the software on, done a calibration with the official card and holder, this seems to work but I have noticed when it goes through the process there is no mention that the calibration is succesfull, I can only assume it has as there is no warning that flash up, once it looks like calibration is over I can check the new setting area, this confirms the camera, led's, laser and turntable are all functional, once confirmed I adjust settings, the colour box is unticked, resolution slider 25%, saturation, contrast, and brightness 50%, when the scanning starts its usually 15 mins foa scan but at the end when I check there is nothing showing, when scanning is in progress nothing comes on the screen, can anyone help?
Noel Leitzman
There is a new image file released
(fabscanpi_basic_stretch_v.2019-02-21-00-00.zip) it can be found on the
Fabscanpi home page. This is based on the lastest Raspian Stretch lite
image. you do need to use raspi-config to set up the country, time and
keyboard. I suggest that you also change the password. Do not change
the selections for ssh, etc. I would use an attached hdmi monitor/tv
screen and a mouse and keyboard for this setup. With ssh operation and
you reboot, the only indication you have that it worked is the ssh
aplication disconnects. It can be an appreciable amount of time before
the shutdown and reboot process works so be patient.
The raspistill command will not work and you need to be in the web based
screen and click the scan button. After that a screen comes up on which
you can click the calibration button to run the calibration as you are
used to. At the beginning of the scan button screen you can also turn
the laser on/off and the turntable on/off. This allows you to focus the
camera and/or set the laser vertical and centered on the turntable.
The scan settings you are using will work-don't ask which are best. I
don't know and I don't know how to use the colored light
adjustments-leave them off at present. I like this release and it covers
the wifi and perhaps the keyboard issues.
My thoughts below,may be incorrect so use carefully:
You may have noticed that the resolution slide only settles at 5 spots
-the min. the max and 3 intermediate spots which I think are for 16, 8,
4, 2, or 1 microsteps between data readings. This means that for a 400
step 0.9 degree stepper motor the data is taken at 360 degrees divided
by 400, 800, 1600, 3200 or 6400 (wonder who needs 6400 sets of points).
I don't believe that the meshing routine works well with the maximum
resolution number of points as there is much data. The camera is turned
sideways to give the maximum field of view in the vertical direction and
depending upon the object height takes like 350-500 or so points for
each scan position. I have noticed that the bottom of my scans are
missing so I use a clear acrylic spacer to raise the object to get the
complete scan top to bottom. You can download a Meshlab application for
your operating system to view the meshed files and use its
capabillities. I need to learn how to use it. I think it may also be
able to load the .ply files too. I like the way the .ply and mesh files
are labeled but it also behooves you to keep a table of descriptions for
each date/time if you have many.
Noel
Phil Kneen
The calibration starts but unsure if it has finished as nothing come up to say it has/has not completed.
The earlier version was a bit hit and miss but at least it gave images of a shape, not a great image but it was an image.
Noel Leitzman
Sorry to take a while. I'm attaching my work in progress set of
instructions for you to look at and see if it helps. Keep in mind that
83 years old i have made one or two or more?? errors so they will likely
be corrections.
Noel
Phil Kneen
Phil Kneen
Noel Leitzman
Phil:
It looks like you ran the calibration on the object you wanted to scan as the green reference cross does not center on the turntable center. First of all run the calibration using the chart of squares as OI show in my write-up. you only need to run it once unless you change the laser position or something else physically. The laser looks like it is vertical and centered on the turntable as it should be . Work with the pill bottle only until you know the process is working. With the bottle in place and your are ready to start the scan the laser trace will show following the profile of the bottle but the green cross vertical should align with the turntable center. Then click start scan.
Noel
--
You received this message because you are subscribed to the Google Groups "fabscan" group.
To unsubscribe from this group and stop receiving emails from it, send an email to fabscan+u...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Noel Leitzman
Phil:
In thinking overnight I decided to send you something to aid in your scanner calibration process as an attachment. Your PDF viewer should be able to increase the screen size to where you can read the printing in the pictures. For your information, if you are using Windows10 you can take a screen shot by pushing the windows key and prt sc at the smae time and it will be saved in Pictures/Screenshots.
Noel
--
Robert Kränzlein
Noel Leitzman
No they don't have to. I think the vertical line should be in
the center of the turntable and may well align there after a
calibration is run at least mine does. I believe the lines are
for centering the camera view and obviously center the object
vertically in the view.
Noel
--
Phil Kneen
Phil Kneen
Noel Leitzman
Phil:
Things look pretty good, I'm encouraged. Upon thinking about
the green line let's change to the green lines show a reference of
where the center of your camera view is aimed and of course is
ideal when the vertical line is centered on the turntable. Take a
look at your calibration sheet position as it seems to be the
reverse of what they show. Also the
/etc/fabscanpi/default.config.json file needs to be looked at and
possibly changed as the vertical distance of turntable to the top
of the bottom square may need to be corrected. The latest changes
and as far as that goes some of the other changes have not been
updated but keep in mind that this all seems to be volunteer work
which does provide money for groceries,etc. I like this latest
change.
The IP of the FabscanPi can can change dynamically although it
usually doesn't. If you have a monitor attached to the PI as you
do it shows toward the end of the bootup screen. Or you can use
http://routerlogin.net (if your router supports it) to find the
location of attached devices. I think the latest release has a
utility to also find the IP but I haven't used it.
The FAQ's show a "My scan is cut-off at top/bottom" with their
answer being to make sure the dimension of turntable to top of
first square is corrected in the default calibration file. I
really haven't had much luck with that. When I looked at the
point cloud files (.ply) with the Notepad text editor my scans do
not go down to the top of the turntable at the bottom but stop
before then. I placed a clear acrylic spacer under the object and
edited the mesh file (.stl) in my drawing program to remove the
bottom garbage and make the bottom flat. The Meshlab program
(you can download free for your operating system) allows you to
load the mesh into it for viewing and you can use it to reduce the
number of points in the mesh to the point where it can be handled
in a 3D drafting program. If you can load meshed .stl files in
your drafting program, trim the bottom and export back to .stl,
which is what I did for the shaker girl that I 3D printed You
have a picture of it in the material which I sent to you. The
point cloud .ply files show 6 values for each point (X,Y,Z-the
vertical, and 3 values for the color/texture. I believe that 3D
printer people aren't really very interested in the last 3
values. A large number of vertical points are recorded for each
turntable rotation position. I suspect that Meshlab can do many
good things to the mesh files but I'm just learning how to use it
a little bit at a time.
You are definitely making progress.
Noel
--
Noel Leitzman
Phil:
I've also noticed that you had trouble with the wood laser mount which came with the case kit. There is a thingiverse file for a previous version of the fabscan cube which mounts in the laser location and fits a nema 17 stepper mounting pattern on the case. It is "Laser-Cradle for FabscanPi Cube". I used the parts except for the cradle which has to be higher than the thingiverse design. It places the laser about 75 mm above the case floor (perhaps 3 mm higher than the wooden mount. This will allow the laser to be centered on the intended location. The file I'm attaching is Laser-Cradle-12mm-Fabscan. The notch in the base of the original design has no use for this application. I think the important part is that the laser trace be vertical goes over the turntable center and extends from below the edge of the turntable to as high as it goes. You may have to wedge your laser slightly tilt it up or down to achieve this.
Noel
--