Closing the Testing Round
ma...@makr.zone
Pipelines can now expose essential stage properties as parameters to be controlled directly from vision settings, without having to go to the pipeline editor. These parameters can be controlled using customizable sliders. During adjustment, the camera view shows a preview of the affected stage and/or the pipeline end result. The solutions makes tuning pipelines much easier. See the Wiki:
https://github.com/openpnp/openpnp/wiki/Exposed-Pipeline-Parameters
2022-02-19 Background CalibrationAutomatic calibration of the bottom camera background can be enabled to drive the color- keyed removal of the background in a bottom vision pipelines. The calibration is done with no part is on the nozzle tip, during the nozzle tip calibration (for run-out compensation).
The MaskHSV stage for knocking out the green key-color of Juki nozzle tips (and similar) is fully controlled. A trouble-shooting function detects bad background conditions such as missing shading, nozzle tip reflections etc. See the Wiki:
https://github.com/openpnp/openpnp/wiki/Nozzle-Tip-Background-Calibration
2022-01-23 Advanced Camera CalibrationCamera setup and calibration has been improved to eliminate the need for manual setup of camera flips, rotation, position, units-per-pixel scaling, and lens distortion compensation. In addition, camera tilt is also now corrected. See the Wiki:
- https://github.com/openpnp/openpnp/wiki/Advanced-Camera-Calibration
- https://github.com/openpnp/openpnp/wiki/Advance-Camera-Calibration---Camera-Mounting-Errors
Alignment (Bottom Vision) and Fiducial Locator vision settings (including the pipelines) are now stored in separate Vision Settings entities that can be assigned to Parts, Packages and to the default Machine Setup Vision presets.
If a setting is not assigned on any of these levels, it will be inherited from the more general level. The system allows for more efficient, more centralized management of vision settings. The vision settings are now also available on the GUI for all these levels. Quick specialization and generalization functions are provided. The OpenPnP stock settings and pipelines are always present in the central list and can be assigned and copied.
Existing configurations with old part settings are migrated automatically. All combinations of settings and pipelines that happen to be equal will be unified into one new Vision Setting. Inheritance will be established, where Vision Settings are common among Parts or Packages.
A new table linking feature (in the menu) can be used to link a selected Part to its Package, Feeder and Vision Settings. Selecting one, automatically selects the other across tabs.
Shai
mark maker
It is all Issues & Solutions guided. Just give it a try!
To answer your question: due to some chicken and egg considerations, it is still a two-step process:
- A quick preliminary calibration that is used to initially
coordinate machine motion and camera orientation/position etc.
- Tony's advanced calibration, that is much more precise and compensates lens distortion and camera tilt.
All the manual calibration steps are still available in OpenPnP,
but they are disabled (grayed out) once you let Advanced Camera
Calibration do its magic. The reason it is still available is that
some machines might not support all technical aspects that are
needed to perform the automatic calibration.
Plus we also don't want to break existing configurations, i.e.
the old stored settings are still fully supported.
_Mark
--
You received this message because you are subscribed to a topic in the Google Groups "OpenPnP" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/openpnp/2GHhEK8UpMQ/unsubscribe.
To unsubscribe from this group and all its topics, send an email to openpnp+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/4b4df078-5b64-43dc-8141-63c12c1cff04n%40googlegroups.com.
ThePoll100
I have a well-functioning machine running on the latest regular (not test) version. I would like to change to the trial version but this is the second time I approach it and I have a problem every time.
I generally do this:
I install a new trial version and run with the old configuration working. There are three points in solving problems
- advanced calibration of the upper camera
- advanced calibration of the lower camera
- request to add additional commands
I go through all the steps and there is a problem with the cameras, namely the camera image becomes narrowed, shifted and sometimes warped.
Another error that I noticed is the AdvancedLoosePartFeeder error, i.e. when it does not detect an element on the feeder, the machine goes crazy, it is impossible to move the head because it screams about crossing the safe zone.
What could be causing such problems?
Best regards and thank you in advance for your help.
PaulM
mark maker
The symptoms you describe, might indicate that your camera is
quite tilted, i.e not looking down perfectly perpendicular to the
machine X/Y plane. I have the same problem on my machine, and had
to disable Advanced Camera Calibration, until I can physically
adjust the camera.
There is a checkbox, on that tab, that can easily be switched
off.
The Advanced Camera Calibration is doubtlessly a very valuable
addition, but the testing version runs fine without it. Camera
tilt compensation is only needed for feeders that are much
higher/lower in Z than the PCB surface Z, and that use
vision.
With disabled Advanced Camera Calibration, confirm that the
machine still runs at least as fine as before.
Anything that needed to be migrated, like the new Vision
Settings, should still work as before.
_Mark
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/a45ae954-5542-427e-ad03-0e7d4a39ab44n%40googlegroups.com.
Jan
May I ask you to check an issue I've seen since quite some time: if
OpenPnP is just started and the camera view disabled (Cameras -> Show
None) one thread is running all the time consuming 100% CPU load (on one
core). I don't see a reason for that so it's likely a synchronization or
signalling issue. Thank You!
Jan
On 15.05.2022 12:04, ma...@makr.zone wrote:
> Hi all,
>
> *- IMPORTANT -*
> I will be merging the testing version into the regular OpenPnP Version
> soon.
>
> Those who would like to make sure no bugs remain, should take the
> opportunity to test with their configuration within the next week. If
> nothing new/dramatic crops up, I will merge it after that.
>
> I've just updated the CHANGES file. The most important new features are
> as follows:
>
> Pipelines can now expose essential stage properties as parameters to be
> controlled directly from vision settings, without having to go to the
> pipeline editor. These parameters can be controlled using customizable
> sliders. During adjustment, the camera view shows a preview of the
> affected stage and/or the pipeline end result. The solutions makes
> tuning pipelines much easier. See the Wiki:
>
> https://github.com/openpnp/openpnp/wiki/Exposed-Pipeline-Parameters
>
> *2022-02-19 Background Calibration*
> Automatic calibration of the bottom camera background can be enabled to
> drive the color- keyed removal of the background in a bottom vision
> pipelines. The calibration is done with no part is on the nozzle tip,
> during the nozzle tip calibration (for run-out compensation).
>
> The MaskHSV stage for knocking out the green key-color of Juki nozzle
> tips (and similar) is fully controlled. A trouble-shooting function
> detects bad background conditions such as missing shading, nozzle tip
> reflections etc. See the Wiki:
>
>
> *2022-01-23 Advanced Camera Calibration*
> Camera setup and calibration has been improved to eliminate the need for
> manual setup of camera flips, rotation, position, units-per-pixel
> scaling, and lens distortion compensation. In addition, camera tilt is
> also now corrected. See the Wiki:
>
> <https://github.com/openpnp/openpnp/wiki/Advanced-Camera-Calibration>
> * https://github.com/openpnp/openpnp/wiki/Advance-Camera-Calibration---Camera-Mounting-Errors
> <https://github.com/openpnp/openpnp/wiki/Advance-Camera-Calibration---Camera-Mounting-Errors>
>
> *2021-12-30 Vision Settings*
> Alignment (Bottom Vision) and Fiducial Locator vision settings
> (including the pipelines) are now stored in separate Vision Settings
> entities that can be assigned to Parts, Packages and to the default
> Machine Setup Vision presets.
>
> If a setting is not assigned on any of these levels, it will be
> inherited from the more general level. The system allows for more
> efficient, more centralized management of vision settings. The vision
> settings are now also available on the GUI for all these levels. Quick
> specialization and generalization functions are provided. The OpenPnP
> stock settings and pipelines are always present in the central list and
> can be assigned and copied.
>
> Existing configurations with old part settings are migrated
> automatically. All combinations of settings and pipelines that happen to
> be equal will be unified into one new Vision Setting. Inheritance will
> be established, where Vision Settings are common among Parts or Packages.
>
> A new table linking feature (in the menu) can be used to link a selected
> Part to its Package, Feeder and Vision Settings. Selecting one,
> automatically selects the other across tabs.
>
> See also:
> https://github.com/openpnp/openpnp/blob/test/CHANGES.md
> <https://github.com/openpnp/openpnp/blob/test/CHANGES.md>
>
> Needless to say there are many, many small improvements on top of these
> large changes:
>
> https://github.com/openpnp/openpnp/pulls?q=is%3Apr+is%3Amerged+base%3Atest+
>
> _Mark
>
> --
> Groups "OpenPnP" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to openpnp+u...@googlegroups.com
> <https://groups.google.com/d/msgid/openpnp/12b3f423-9900-4c79-8759-920c7a9b08e7n%40googlegroups.com?utm_medium=email&utm_source=footer>.
mark maker
Cannot reproduce it. On what OS is that?
Can you provide the machine.xml?
Jan
Jan
> <https://groups.google.com/d/msgid/openpnp/1c1d5e67-724b-a906-fac2-dcf6729af259%40makr.zone?utm_medium=email&utm_source=footer>.
mark maker
I don't have a clue, all your related settings look good. As I do not have your cameras, I obviously can't reproduce.
Do you know Process Explorer? It could at least give some hints
towards thread CPU usage:
https://docs.microsoft.com/en-us/sysinternals/downloads/process-explorer
Select "Properties..." on the javaw.exe process, then go to the
"Threads" tab. Send a screenshot.
Its a bit helpless...
_Mark
Jan
Please find the requested screenshot attached. I hope you can see
something...
If you need anything else, please let me know.
Jan
> <https://groups.google.com/d/msgid/openpnp/7e6d1612-8371-d28e-0a49-1ffa15e3a3ab%40makr.zone?utm_medium=email&utm_source=footer>.
Dave McGuire
> I will be merging the testing version into the regular OpenPnP Version
> soon.
>
> Those who would like to make sure no bugs remain, should take the
> opportunity to test with their configuration within the next week. If
> nothing new/dramatic crops up, I will merge it after that.
Back on April 30th, I mentioned a possible bug in the strip feeder
code, in which the job stops when a strip hits its feed limit. I
believe it should search for another feeder with the same component and
continue, and only stop the job if there are no others.
This constitutes the only real manual job interventions that I have
here on my system when doing builds. Would it be possible for you to
have a look at that before the final merge?
Thanks,
-Dave
--
Dave McGuire
McGuire Scientific Services, LLC
New Kensington, PA
ThePoll100
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/4dd7a95f-f754-3518-889e-d7771c46c8d7%40makr.zone.
Jan
I just remembered, that a GUI for panel fiducials would be nice. Tony
checked the code and adding them manually to the job already works well
(at least for single sided panels).
Jan
On 15.05.2022 12:04, ma...@makr.zone wrote:
> Hi all,
>
> *- IMPORTANT -*
>
> soon.
>
> Those who would like to make sure no bugs remain, should take the
> opportunity to test with their configuration within the next week. If
> nothing new/dramatic crops up, I will merge it after that.
>
> I've just updated the CHANGES file. The most important new features are
> as follows:
>
> Pipelines can now expose essential stage properties as parameters to be
> controlled directly from vision settings, without having to go to the
> pipeline editor. These parameters can be controlled using customizable
> sliders. During adjustment, the camera view shows a preview of the
> affected stage and/or the pipeline end result. The solutions makes
> tuning pipelines much easier. See the Wiki:
>
> https://github.com/openpnp/openpnp/wiki/Exposed-Pipeline-Parameters
>
> *2022-02-19 Background Calibration*
>
> drive the color- keyed removal of the background in a bottom vision
> pipelines. The calibration is done with no part is on the nozzle tip,
> during the nozzle tip calibration (for run-out compensation).
>
> The MaskHSV stage for knocking out the green key-color of Juki nozzle
> tips (and similar) is fully controlled. A trouble-shooting function
> detects bad background conditions such as missing shading, nozzle tip
> reflections etc. See the Wiki:
>
>
> *2022-01-23 Advanced Camera Calibration*
>
> manual setup of camera flips, rotation, position, units-per-pixel
> scaling, and lens distortion compensation. In addition, camera tilt is
> also now corrected. See the Wiki:
>
> <https://github.com/openpnp/openpnp/wiki/Advanced-Camera-Calibration>
> * https://github.com/openpnp/openpnp/wiki/Advance-Camera-Calibration---Camera-Mounting-Errors
> <https://github.com/openpnp/openpnp/wiki/Advance-Camera-Calibration---Camera-Mounting-Errors>
>
> *2021-12-30 Vision Settings*
>
> (including the pipelines) are now stored in separate Vision Settings
> entities that can be assigned to Parts, Packages and to the default
> Machine Setup Vision presets.
>
> If a setting is not assigned on any of these levels, it will be
> inherited from the more general level. The system allows for more
> efficient, more centralized management of vision settings. The vision
> settings are now also available on the GUI for all these levels. Quick
> specialization and generalization functions are provided. The OpenPnP
> stock settings and pipelines are always present in the central list and
> can be assigned and copied.
>
> Existing configurations with old part settings are migrated
> automatically. All combinations of settings and pipelines that happen to
> be equal will be unified into one new Vision Setting. Inheritance will
> be established, where Vision Settings are common among Parts or Packages.
>
> A new table linking feature (in the menu) can be used to link a selected
> Part to its Package, Feeder and Vision Settings. Selecting one,
> automatically selects the other across tabs.
>
> See also:
> https://github.com/openpnp/openpnp/blob/test/CHANGES.md
> <https://github.com/openpnp/openpnp/blob/test/CHANGES.md>
>
> Needless to say there are many, many small improvements on top of these
> large changes:
>
> https://github.com/openpnp/openpnp/pulls?q=is%3Apr+is%3Amerged+base%3Atest+
>
> _Mark
>
> --
> Groups "OpenPnP" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to openpnp+u...@googlegroups.com
> <https://groups.google.com/d/msgid/openpnp/12b3f423-9900-4c79-8759-920c7a9b08e7n%40googlegroups.com?utm_medium=email&utm_source=footer>.
tonyl...@gmail.com
mark maker
Hi Jan,
The result is what I expected. You see most of the CPU load is
used in the qcap.dll which is a part of Windows Direct X. It is a
known issue, that it uses a lot of CPU load, even if OpenPnP does
nothing with the frames.
Maybe you improved lighting recently?
Are you sure, the load actually drops when you show the camera views (not "None")? That would be very unexpected!
When I examined the issues, I confirmed that OpenPnP or Java is
not to blame. Other Non-Java programs have the exact same CPU load
when using two cameras (tested with OBS studio).
There is a newer Windows API for multimedia that moves the load
to the GPU. But that does not seem to really change the result on
my notebook: same amount of fan cooling required, same noise!
Given the load is not actually limiting on today's multi-cores,
this is zero-sum. And those with cheapo/weak CPUs won't have
dedicated GPUs either.
If you think about it: in my case with two Full HD ELP cams,
there are two good quality/rather high bandwidth Full HD
1080p/30fps or two HD 720p/60fps streams being constantly decoded.
That is not nothing!
When I started editing video in DV format, at tiny 576i resolution, I needed a dedicated very expensive hardware just to copy footage from the DV tapes to the HD. No real-time viewing or editing, of course. Real-time viewing was only possible from the camera, playing the tape, using an analog video overlay window on a special graphics card. 😂
For background (you might also follow the links at the end):
https://makr.zone/camera-fps-cpu-load-and-lighting-exposure/519/
_Mark
mark maker
Hi Dave,
As far as I know, this was not introduced by the testing version, i.e. it is the same in the develop branch. Or do you have information to the contrary?
I'd like to close this round now, without opening new
constructions sites. But that does not mean that development will
stop after that. The testing version is then open for new stuff
again.
_Mark
Jan
Many thanks! Lets do it like that.
I'm happy to test it. (I just received a bunch of double sided panels
with four fiducials on each side. Some components are still missing and
I'm not in a hurry.)
Jan
> <https://groups.google.com/d/msgid/openpnp/12b3f423-9900-4c79-8759-920c7a9b08e7n%40googlegroups.com?utm_medium=email&utm_source=footer>>.
>
> --
> You received this message because you are subscribed to the Google
> Groups "OpenPnP" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to openpnp+u...@googlegroups.com
> <mailto:openpnp+u...@googlegroups.com>.
> To view this discussion on the web visit
> <https://groups.google.com/d/msgid/openpnp/5e926c18-9551-439e-8a9b-6ef8c102f46en%40googlegroups.com?utm_medium=email&utm_source=footer>.
mark maker
With "troubleshooting" you mean Issues & Solutions?
You can do it both ways:
- Either just disable Advanced Camera Calibration and check if
it solves the issues.
- Or start from the old version machine.xml and test if
everything is behaving as before.
> Is there any way to accurately position the camera? It
is known that in the camera the matrix itself may not be in the
plane and it is hard to tell if it is even. Maybe there are
some markers in the camera view to have a reference point?
Not easy. The best I can tell you...
- Perform preliminary calibration to get Units per Pixel.
https://github.com/openpnp/openpnp/wiki/Vision-Solutions#calibration-rig
- Use a millimeter paper that you glue down on a flat table surface that you know to be good.
- Enable the grid reticle in the camera view (context menu).
- Then try to adjust the camera tilt until those two match.
You can also perform Advanced Camera Calibration and then read
the tilt angle on the tab. But it is hard to know which way it
goes. Maybe @Tony can help.
_Mark
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/CA%2BSVFECf2ByseTevkO_%3D%2BWenMTWPrBL5A_ZSH9uQusy4RE9VoA%40mail.gmail.com.
mark maker
Like I said:
I'd like to close this round now, without opening new
constructions sites. But that does not mean that development
will stop after that. The testing version is then open for new
stuff again.
😁
Jan
Many thanks for the explanation. This CPU load is nasty, but I need
both cameras, so I guess, I've to live with it...
If I enable the camera view, I can see some addition load on the GPU.
The CPU load is the same.
I've seen your blog about FPS and exposure. Its again this type of
nasty game that numbers on paper are made to look good but in reality
they can be quite different. There has to be a reason why consumer
cameras are about 50,- while scientific cameras can be about 5000,- and
more...
I think a dedicated camera trigger from the controller might be a good
idea (at least for bottom vision and global shutter cameras): if the
move to the camera is finished, a single picture is taken for the
alignment. Then there is no overhead for the camera anymore.
Jan
On 18.05.2022 09:01, mark maker wrote:
> Hi Jan,
>
> The result is what I expected. You see most of the CPU load is used in
> the qcap.dll which is a part of Windows Direct X. It is a known issue,
> that it uses a lot of CPU load, even if OpenPnP does nothing with the
> frames.
>
> exposure device setting, /the higher /your raw FPS will be, and/the more
> /CPU load you have in qcap.dll. Which makes sense: it decodes more
>
> Maybe you improved lighting recently?
>
> Are you sure, the load actually drops when you show the camera views
> When I examined the issues, I confirmed that OpenPnP or Java is not to
> blame. Other Non-Java programs have the exact same CPU load when using
> two cameras (tested with OBS studio).
>
> There is a newer Windows API for multimedia that moves the load to the
> GPU. But that does not seem to really change the result on my notebook:
> same amount of fan cooling required, same noise! Given the load is not
> actually limiting on today's multi-cores, this is zero-sum. And those
> with cheapo/weak CPUs won't have dedicated GPUs either.
>
> If you think about it: in my case with two Full HD ELP cams, there are
> 720p/60fps streams being constantly decoded.*That is not nothing! *
> When I started editing video in DV format, at tiny 576i resolution, I
> Real-time viewing was only possible from the camera, playing the tape,
> <https://groups.google.com/d/msgid/openpnp/ccdee52c-e581-6e3d-50c2-b731a2959388%40makr.zone?utm_medium=email&utm_source=footer>.
mark maker
Yes, machine triggered global shutter still frames would lighten
the load significantly.
The only down-side I see would be adaptive Camera Settling, that obviously only works with a video stream. But that's only necessary on mechanically "challenged" machines, like mine, that shake and vibrate after moves 😅
Dave McGuire
Hi Mark, ok, I understand the process now. I don't know if that
feature was introduced in the testing version or not; it just "appeared"
at some point over the past couple of years. I have no specific
knowledge of when/how it appeared.
-Dave
On 5/18/22 03:05, mark maker wrote:
> Hi Dave,
>
> As far as I know, this was not introduced by the testing version, i.e.
> contrary?*
> I'd like to close this round now, without opening new constructions
> sites. But that does not mean that development will stop after that. The
> testing version is then open for new stuff again.
>
> _Mark
>
> On 17.05.22 21:32, Dave McGuire wrote:
>> On 5/15/22 06:04, ma...@makr.zone wrote:
>>> I will be merging the testing version into the regular OpenPnP
>>> Version soon.
>>>
>>> Those who would like to make sure no bugs remain, should take the
>>> opportunity to test with their configuration within the next week. If
>>> nothing new/dramatic crops up, I will merge it after that.
>>
>> Thank you for all of your hard work on this!
>>
>> Back on April 30th, I mentioned a possible bug in the strip feeder
>> code, in which the job stops when a strip hits its feed limit. I
>> believe it should search for another feeder with the same component
>> and continue, and only stop the job if there are no others.
>>
>> This constitutes the only real manual job interventions that I have
>> here on my system when doing builds. Would it be possible for you to
>> have a look at that before the final merge?
>>
>> Thanks,
>> -Dave
>>
> You received this message because you are subscribed to the Google
> Groups "OpenPnP" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to openpnp+u...@googlegroups.com
> <mailto:openpnp+u...@googlegroups.com>.
> To view this discussion on the web visit
> <https://groups.google.com/d/msgid/openpnp/122bbf0b-1be5-47f7-f0c0-7da51523b46b%40makr.zone?utm_medium=email&utm_source=footer>.
ma...@makr.zone
Jim Young
mark maker
Sorry, I don't know. I would assume the latter.
--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/28bb3333-4819-4c51-ac5d-142dfb4da740n%40googlegroups.com.