You are not logged in. Please login or register.
SoliForum - 3D Printing Community → ATLAS 3D & FreeLSS → New version of FreeLSS committed to Github
Those values look much more reasonable. Have you done a test scan yet? Hmm, looking at your setup, it's possible that the laser is being picked up on the back-drop and throwing off the calibration as well.
I just ran a scan of a phone holder...
There looks to be distortion, some feature doubling, and separation of front/back surfaces.
I'll try to find a new backdrop. In the meantime, any other suggestions?
Thanks for the feedback tonyno. I'm looking into this now.
The lasers look like they should be fine. Are you using the 5MP camera or the 8MP? Also, which Pi version are you using? Lastly, place the calibration item on the table liner so that it doesn't move during the calibration.
Are we able to use the 8MP camera now?
@knowack You can still make a single scan from both lasers. The main difference is that the merging would take place in another application such as MeshLab instead of directly on the ATLAS 3D.
I guess I need a Meshlab tutorial before I upgrade.
FreeLSS version 1.24 is now available via the ATLAS 3D auto-update system.
This releases fixes the reported issues from 1.23 and also adds a few new features.
Faster Calibration Speed
The calibration processes is up to 10 times faster on the Pi 1 and 2 to 3 times faster on Pi 3.
OBJ File Support
OBJ files can now be generated in addition to STL, PLY, and XYZ. The option must be enabled on the SETTINGS page.
Fixed Object Size Bug
FreeLSS 1.23 contained a bug where the max object size wasn't being handled properly. This resulted in laser results being used on backdrops and other objects close to the scanner. This bug would have resulted in bad results for calibrations and scans.
8MP Camera Verified
The calibration process has been verified to work with the 8MP camera. If you are using an 8MP camera, make sure that you have one of the newer Pi firmwares that support it properly. On the Setup page, Camera Sensor should show "IMX219". If it shows OV5647 then you need to update the Pi firmware.
Visual Calibration Information
Atfer calibration is performed, a file is now generated that visually shows the results. In general, the user will not need to use this information but it will help diagnose any problems or just to track changes over time.
Below is what the PLY contains (as viewed from MeshLab) along with a description.
Thanks,
Uriah Liggett
Ok, so I reprinted the calibration pattern at the correct size and installed the 1.25 update. Now I'm getting funky shaped scans and the scanner does not complete a full 360 even though it is set to do so. The front and back are curved and no longer parallel. I have also been getting this issue where the bottom side are clipped off in a curve. The curved clipping at the bottom has been going on since before any updates and is the reason for the little platforn under the scanned object. The front and back used to be straight though. Misaligned shells for left and right but straight.
This is a rectangular xacto knife box that I have been using for tests. This is a low frames scan (200 per rev) at 5MP. What causes such as this?
overview:
tilted front view:
top down:
corner view showing curved clipping:
calibration info:
calibration ply:
http://isurf2.com/1/1501522915.ply
Hi Brink,
For the table...
Please double check that the Steps per Revolution is set to 3200 on the Setup page.
On the CAMERA page, if you enter 360 and click Rotate, does it complete a full rotation?
After you click Enable Motor is the table hard to turn to where the motor kinda jerks when you try or does it turn fairly easily?
Are you certain that the box isn't sliding slightly as the table turns?
For the calibration...
Can you perform the calibration 2 - 3 times moving the checker board around slightly and then post the results. I would like to see if they are all very close.
In the meantime, I will try to find a similar object here and scan it for reference.
Thanks,
Uriah
Brink,
I found a box of bolts that's the approximate shape of what you were scanning. I then tilted the camera at an extreme angle to make it more challenging for the calibration process. I performed a 200 frames per rev scan and posted pictures of the setup and the resulting files below.
You can see how the box is going through the table due to how far the camera was tilted. The thresholding algorithm is currently rejecting all of the points below the table, this should be changed next release.
http://www.murobo.com/misc/1501629477_Right.ply
http://www.murobo.com/misc/1501629477_Left.ply
- Uriah
Getting Beautiful scans now. I couldn't tell you what fixed it. Was something on the settings page. When I clicked the "calibration" preset it changed something that I had set wrong further down on the page. Anyway... works great now. BTW the shells are aligned and everything. Is it a possibility we could get a both option back for downloading the left and right pi aligned clouds? This was already requested by knowack in post 66 of this thread. 
That's fantastic Brink. I'm happy to hear that you are producing great scans again. The merging option is likely to return but there are several other features and improvements that have higher priority right now.
- Uriah
The merging option is likely to return but there are several other features and improvements that have higher priority right now.
- Uriah
Understood. Since I have good results, and no issues with aligning my lasers, I'm just going to wait to upgrade.
I just applied the update and it can't find the chessboard. Yes, I remembered to use a background this time. 
Here's the preview.
Update: Added some blank paper to the background and it's seeing it now.
Hmm. Here's the scan of that phone holder again. Seems like it's not seeing much.
Stand by, I may have just spotted user-error...
The detection algorithm hasn't changed. Can you double check your noise removal and threshold settings? Also, you can see what the laser will pick up by clicking TEST on the camera page.
- Uriah
Thanks Uriah. Got it working a couple posts above. New issue is shown in the phone-holder post above.
No thoughts on this?
Is that a black or translucent item?
tonyno, my previous post was in response to the cell phone holder issue. Please double check your noise reduction and threshold settings. Use the "Test" button to ensure that the object will be detected when scanning.
- Uriah
It's black.
Thanks Uriah. Will check that.
Update: Immediately noticed that it was set to Calibration, not Default. o.O Pretty sure that was it. Testing now.
Not it.
Please double check your noise reduction and threshold settings.
A little help with this? Just try different settings?
Turn off noise reduction and set threshold to its lowest value. Set exposure to auto and reduce the amount of light shining on the object as well. Please post a picture of the image generated when you click the Test button.
It's black.
Thanks Uriah. Will check that.
Update: Immediately noticed that it was set to Calibration, not Default. o.O Pretty sure that was it. Testing now.
If it's black, you may need to dust it with talcum powder; the camera has a hard time seeing the laser line against black objects, since black absorbs energy rather than reflecting it. ...a characteristic of all 3D scanners.
SoliForum - 3D Printing Community → ATLAS 3D & FreeLSS → New version of FreeLSS committed to Github
Powered by PunBB, supported by Informer Technologies, Inc.
