It should not be a hard update, I believe all the functionality if the pylon API is not needed, as these cameras have their user profile to be loaded at startup. Image processing into pix_video is already a good solution. But let's plan it for USB3, the speed is incomparable to GigE cameras.

Popesz

On Thu, Aug 9, 2018 at 7:35 AM Jean-Marie Adrien jm.adrien.mnt@gmail.com wrote:

by the way i use extensively vintage PAL cameras with GEM *on mac* since 15 years, and i would love some update allowing for faster cameras and higher definition cameras because i fear things will simply not work anymore one of these days… means it won’t be possible to find standard definition cameras for instance, cameras which long ago disappeared from macs.

i react to the *donation* issue : i would be ready to give some personal money for quicktime compatibility, maybe some other pd users would also do so … because this is in my case personal money and not institutional money, maybe there could be a way to set up something like a kickstarter project with many contributors (backers) all of this resulting in a decent amount … ? just thinking ... JM

Le 8 août 2018 à 09:31, Csaba Láng langcsaba@gmail.com a écrit :

Dear list! as we know Gem is already quite outdated, the videoPYLON plugin was written in 2011. Since that time USB3 has arrived and the GigE cameras are not as efficient as the new technology (2.5x faster connection is the USB3) Pylon has already version 5.x (opensource) and provides driver to both USB3 and GigE cameras.

Anyway, I would like to ask if someone has the time to update the code for the fresh USB3 cameras, *donation is possible!*

here are the link to Pylon:

https://www.baslerweb.com/en/sales-support/downloads/software-downloads/pylo... and the videoPYLON code on git: https://github.com/umlaeute/Gem/tree/master/plugins/videoPYLON

Please let me know if anybody interested!

Best,

Popesz _______________________________________________ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list

Antoine, thanks for your declaration, however I have a huge project which should be finished in December. I already read the Pylon5 documentation, compared the GigE commands to the USB3, made some changes (possibly mess) in the CameraProperties.cpp - deleted the unused commands like IP address, replaced the old commands with the new or similar ones etc, but cannot compile on ubuntu 18.04. As I wrote earlier, most of the functionalities are not urgent, image processing is needed as the rest can be loaded from the user settings at start up. Maybe we should talk to Iohannes, as he wrote the original plugin to GigE cameras, but he is on holiday until the weekend. Best, Popesz

On Thu, Aug 9, 2018 at 10:51 AM Antoine Villeret antoine.villeret@gmail.com wrote:

Hi,

I could have a look into Pylon pluging in the next few months, but I do not garantee any deadline since my todo list is almost full until december If no one else did the trick, ping me in one month or so to remind me what I'm writing today :-)

a

Le jeu. 9 août 2018 à 09:36, Csaba Láng langcsaba@gmail.com a écrit :

...

It should not be a hard update, I believe all the functionality if the pylon API is not needed, as these cameras have their user profile to be loaded at startup. Image processing into pix_video is already a good solution. But let's plan it for USB3, the speed is incomparable to GigE cameras.

Popesz

On Thu, Aug 9, 2018 at 7:35 AM Jean-Marie Adrien jm.adrien.mnt@gmail.com wrote:

...

by the way i use extensively vintage PAL cameras with GEM *on mac* since 15 years, and i would love some update allowing for faster cameras and higher definition cameras because i fear things will simply not work anymore one of these days… means it won’t be possible to find standard definition cameras for instance, cameras which long ago disappeared from macs.

i react to the *donation* issue : i would be ready to give some personal money for quicktime compatibility, maybe some other pd users would also do so … because this is in my case personal money and not institutional money, maybe there could be a way to set up something like a kickstarter project with many contributors (backers) all of this resulting in a decent amount … ? just thinking ... JM

Le 8 août 2018 à 09:31, Csaba Láng langcsaba@gmail.com a écrit :

Dear list! as we know Gem is already quite outdated, the videoPYLON plugin was written in 2011. Since that time USB3 has arrived and the GigE cameras are not as efficient as the new technology (2.5x faster connection is the USB3) Pylon has already version 5.x (opensource) and provides driver to both USB3 and GigE cameras.

Anyway, I would like to ask if someone has the time to update the code for the fresh USB3 cameras, *donation is possible!*

here are the link to Pylon:

https://www.baslerweb.com/en/sales-support/downloads/software-downloads/pylo... and the videoPYLON code on git: https://github.com/umlaeute/Gem/tree/master/plugins/videoPYLON

Please let me know if anybody interested!

Best,

Popesz _______________________________________________ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list

---

Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list

Iohannes,

the only link i got from Basler is a python wrapper:

https://github.com/basler/pypylon

the rest is binary with compiled folders and files headers on the original Basler website.

Does it help you?

On Mon, Aug 13, 2018 at 2:15 PM IOhannes m zmoelnig zmoelnig@iem.at wrote:

On 2018-08-09 10:51, Antoine Villeret wrote:

...

Pylon has already version 5.x (opensource) and provides driver to both

where's an opensourced version of Pylon? all i can find are proprietary, EULA encumbered SDKs for free download.

fgamsdr IOhannes

---

Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list

On 2018-08-17 12:52, Csaba Láng wrote:

Iohannes,

do you think porting to pylon5 is feasible?

yes.

I have 2 pieces of USB3 Basler cameras until 1st September, so I can check if they work properly, at least the image processing.

cool.

Please, let me know if you have time to port it.

i think so.

fgamsdr IOhannes

Iohannes,

after compiling freshly Gem from git with videoPYLON support, gem_videoPYLON.la and .so is in the Gem folder, however after creating pix_video, can not see videoPYLON among the plugins: [pix_video]: backend #0='v4l2' : v4l2 analog [pix_video]: backend #1='v4l' : v4l analog [pix_video]: backend #2='dv4l' : dv4l dv [pix_video]: backend #3='dc1394' : dc1394 iidc [pix_video]: backend #4='unicap' : unicap analog [pix_video]: backend #5='vlc' <--- DISABLED

I have checked on which bus the USB3 Basler cvameras are connected - e.g. Bus 002 Device 002: ID 2676:ba02 Basler AG ace Therefor added instead of dev/video0 /dev/bus/usb/002/002 as a message.

This is what pd prints out: Failed to query video capabilities: Inappropriate ioctl for device libv4l2: error getting capabilities: Inappropriate ioctl for device [GEM:videoV4L] VIDIOCGPICT: Inappropriate ioctl for device

Let me know if you need more info or any log.

Best, Popesz

On Wed, Aug 22, 2018 at 5:51 PM IOhannes m zmoelnig zmoelnig@iem.at wrote:

On 2018-08-20 15:46, IOhannes m zmoelnig wrote:

...

...

Please, let me know if you have time to port it.

i think so.

i've updated the videoPYLON backend in the git-repository of Gem, so it now works with Pylon-5.0.12 please test it and report back. (you should compile the entire Gem package)

fgamsdr IOhannes

---

Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list

GEM: Graphics Environment for Multimedia GEM: ver: 0.93.git 3550ede3 GEM: compiled on Aug 29 2018

$ ls -l Gem.pd_linux gem_videoPYLON.so -rwxr-xr-x 1 root root 85108760 sie 29 12:24 Gem.pd_linux -rwxr-xr-x 1 root root 903600 sie 29 12:24 gem_videoPYLON.so

$ ldd gem_videoPYLON.so linux-vdso.so.1 (0x00007fffe9d89000) libpylonbase-5.0.12.so => /opt/pylon5/lib64/libpylonbase-5.0.12.so (0x00007f692a424000) libpylonutility-5.0.12.so => /opt/pylon5/lib64/libpylonutility-5.0.12.so (0x00007f692a0a9000) libstdc++.so.6 => /usr/lib/x86_64-linux-gnu/libstdc++.so.6 (0x00007f6929d1b000) libc.so.6 => /lib/x86_64-linux-gnu/libc.so.6 (0x00007f692992a000) libgcc_s.so.1 => /lib/x86_64-linux-gnu/libgcc_s.so.1 (0x00007f6929712000) librt.so.1 => /lib/x86_64-linux-gnu/librt.so.1 (0x00007f692950a000) libGenApi_gcc_v3_0_Basler_pylon_v5_0.so => /opt/pylon5/lib64/libGenApi_gcc_v3_0_Basler_pylon_v5_0.so (0x00007f6928e6e000) libGCBase_gcc_v3_0_Basler_pylon_v5_0.so => /opt/pylon5/lib64/libGCBase_gcc_v3_0_Basler_pylon_v5_0.so (0x00007f6928c55000) libLog_gcc_v3_0_Basler_pylon_v5_0.so => /opt/pylon5/lib64/libLog_gcc_v3_0_Basler_pylon_v5_0.so (0x00007f6928a4c000) libpthread.so.0 => /lib/x86_64-linux-gnu/libpthread.so.0 (0x00007f692882d000) libdl.so.2 => /lib/x86_64-linux-gnu/libdl.so.2 (0x00007f6928629000) libm.so.6 => /lib/x86_64-linux-gnu/libm.so.6 (0x00007f692828b000) /lib64/ld-linux-x86-64.so.2 (0x00007f692a934000) libMathParser_gcc_v3_0_Basler_pylon_v5_0.so => /opt/pylon5/lib64/libMathParser_gcc_v3_0_Basler_pylon_v5_0.so (0x00007f6928081000) libXmlParser_gcc_v3_0_Basler_pylon_v5_0.so => /opt/pylon5/lib64/libXmlParser_gcc_v3_0_Basler_pylon_v5_0.so (0x00007f6927d22000) libNodeMapData_gcc_v3_0_Basler_pylon_v5_0.so => /opt/pylon5/lib64/libNodeMapData_gcc_v3_0_Basler_pylon_v5_0.so (0x00007f6927b08000)

On Wed, Aug 29, 2018 at 1:16 PM IOhannes m zmoelnig zmoelnig@iem.at wrote:

On 2018-08-29 12:47, Csaba Láng wrote:

...

Iohannes,

after compiling freshly Gem from git with videoPYLON support, gem_videoPYLON.la and .so is in the Gem folder, however after creating pix_video, can not see videoPYLON among the plugins: [pix_video]: backend #0='v4l2' : v4l2 analog [pix_video]: backend #1='v4l' : v4l analog [pix_video]: backend #2='dv4l' : dv4l dv [pix_video]: backend #3='dc1394' : dc1394 iidc [pix_video]: backend #4='unicap' : unicap analog [pix_video]: backend #5='vlc' <--- DISABLED

for testing, please remove all the other (video) backends: $ mkdir disabled $ mv gem_*.so disabled/

then please provide the exact output of the following two commands: $ ls -l Gem.pd_linux gem_videoPYLON.so $ ldd gem_videoPYLON.so

and the exact version string Gem prints.

...

I have checked on which bus the USB3 Basler cvameras are connected - e.g. Bus 002 Device 002: ID 2676:ba02 Basler AG ace Therefor added instead of dev/video0 /dev/bus/usb/002/002 as a message.

i have no idea whether this is the appropriate device string for a Basler cvcamera. most likely it is not (but we need to get the videoPYLON backend working before we can do anything else).

ghamdf IOhannes

---

Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list

Looks almost good, there is image from the camera

Let's see what is wrong:

• Terminal output: $ puredata

priority 8 scheduling failed. open: /etc/pd/gem.conf: No such file or directory open: /home/is/.config/pure-data/gem.conf: No such file or directory open: ./gem.conf: No such file or directory load plugins 'film' in './' pattern : ./gem_film*.so load plugins 'image' in './' pattern : ./gem_image*.so load plugins 'image' in './' pattern : ./gem_image*.so load plugins 'image' in './' pattern : ./gem_image*.so load plugins 'model' in './' pattern : ./gem_model*.so load plugins 'record' in './' pattern : ./gem_record*.so load plugins 'video' in './' pattern : ./gem_video*.so dylib loading file './gem_videoPYLON.so'! not reloading 'video' plugins (already 1 loaded)

• gemwin cannot be created if I add it to the patch, but from help it

works

• device0 gives image after creating gemwin from help: PD console:

[pix_video]: backend #0='pylon' : pylon USB usb GigE gige GLEW version 1.10.0 [gemglxwindow]: Direct Rendering enabled!

with some invalid command name (Tcl) INVALID COMMAND NAME: invalid command name ".x55e350de23c0.c" while executing ".x55e350de23c0.c delete 55e350de2dd0BASE0" ("uplevel" body line 139) invoked from within "uplevel #0 $docmds" Not sure if important

• trying to change the input from device0 to device1 - which only at the

very first time managed to do, no image from the 2nd camera, and coming back to device0:watchdog: signaling pd... watchdog: signaling pd... and forever like this.

On Wed, Aug 29, 2018 at 2:57 PM IOhannes m zmoelnig zmoelnig@iem.at wrote:

On 2018-08-29 13:30, Csaba Láng wrote:

...

GEM: Graphics Environment for Multimedia GEM: ver: 0.93.git 3550ede3 GEM: compiled on Aug 29 2018

$ ls -l Gem.pd_linux gem_videoPYLON.so -rwxr-xr-x 1 root root 85108760 sie 29 12:24 Gem.pd_linux -rwxr-xr-x 1 root root 903600 sie 29 12:24 gem_videoPYLON.so

ok. looks good.

so what's the full output on the console (not the Pd-console, but your terminal!).

fgsdmrt IOhannes

---

Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list

Looks like for me that pix_video does not disconnect the cameras, that is why I cannot get image back after changing device. After turning off puredata, if not frozen and killall it, the pylon viewer cannot get image from the cameras, 0 frames and an error message.

Probably I will not need it to jump from device to device, but makes life easier if does not have to restart the computer after each new test.

On Wed, Aug 29, 2018 at 5:46 PM Csaba Láng langcsaba@gmail.com wrote:

gemwin behaves strange:

if I open pd from terminal with puredata command, gemwin cannot be created, but in help menu works if doubleclick on .pd file, gemwin is created

On Wed, Aug 29, 2018 at 5:37 PM Csaba Láng langcsaba@gmail.com wrote:

...

Looks almost good, there is image from the camera

Let's see what is wrong:

• Terminal output: $ puredata

priority 8 scheduling failed. open: /etc/pd/gem.conf: No such file or directory open: /home/is/.config/pure-data/gem.conf: No such file or directory open: ./gem.conf: No such file or directory load plugins 'film' in './' pattern : ./gem_film*.so load plugins 'image' in './' pattern : ./gem_image*.so load plugins 'image' in './' pattern : ./gem_image*.so load plugins 'image' in './' pattern : ./gem_image*.so load plugins 'model' in './' pattern : ./gem_model*.so load plugins 'record' in './' pattern : ./gem_record*.so load plugins 'video' in './' pattern : ./gem_video*.so dylib loading file './gem_videoPYLON.so'! not reloading 'video' plugins (already 1 loaded)

• gemwin cannot be created if I add it to the patch, but from help it

works

• device0 gives image after creating gemwin from help: PD console:

[pix_video]: backend #0='pylon' : pylon USB usb GigE gige GLEW version 1.10.0 [gemglxwindow]: Direct Rendering enabled!

with some invalid command name (Tcl) INVALID COMMAND NAME: invalid command name ".x55e350de23c0.c" while executing ".x55e350de23c0.c delete 55e350de2dd0BASE0" ("uplevel" body line 139) invoked from within "uplevel #0 $docmds" Not sure if important

• trying to change the input from device0 to device1 - which only at

the very first time managed to do, no image from the 2nd camera, and coming back to device0:watchdog: signaling pd... watchdog: signaling pd... and forever like this.

On Wed, Aug 29, 2018 at 2:57 PM IOhannes m zmoelnig zmoelnig@iem.at wrote:

...

On 2018-08-29 13:30, Csaba Láng wrote:

...

GEM: Graphics Environment for Multimedia GEM: ver: 0.93.git 3550ede3 GEM: compiled on Aug 29 2018

$ ls -l Gem.pd_linux gem_videoPYLON.so -rwxr-xr-x 1 root root 85108760 sie 29 12:24 Gem.pd_linux -rwxr-xr-x 1 root root 903600 sie 29 12:24 gem_videoPYLON.so

ok. looks good.

so what's the full output on the console (not the Pd-console, but your terminal!).

fgsdmrt IOhannes

---

Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list

On 8/29/18 7:19 PM, Csaba Láng wrote:

Iohannes, how can I have an impact on the frame rate of the camera?

i guess via the "AcquisitionFrameRateAbs" property (see [enumProps( for a full list of available properties). you should also be able to set the frame-rate via the PylonViewerApp, before opening the camera with Gem.

gfamrds IOhannes

On 8/29/18 6:25 PM, Csaba Láng wrote:

Looks like for me that pix_video does not disconnect the cameras, that is why I cannot get image back after changing device. After turning off puredata, if not frozen and killall it, the pylon viewer cannot get image from the cameras, 0 frames and an error message.

Probably I will not need it to jump from device to device, but makes life easier if does not have to restart the computer after each new test.

sure. however, this is hard to debug without a device at hand. it might be related to https://github.com/umlaeute/Gem/issues/192 (again check the terminal for error messages)

also: did you try to phyisically unplug the camera?

gasdmr IOhannes

On 8/29/18 9:55 PM, Csaba Láng wrote:

Gemwin has been solved after a restart of the computer.

Sure I tried to unplug the cameras, but then pd crashes and switches off. The disconnect solution would be a big help, as creating a complex patch with frequent computer restart is a pain in the a***s.

the idea was, that unplugging the camera (and crashing Pd) would be less obtrusive than having to power-cycle your computer. i'm not sure what you mean by "switches off" though...

anyhow, for starters you could try to confirm whether you are seeing something similar as https://github.com/umlaeute/Gem/issues/192

gfmadsr IOhannes

On 8/29/18 10:30 PM, Csaba Láng wrote:

I meant under crash that first freezes and than closes all opened pd windows

ok. this *is* better than having to reboot, no? (assuming that plugging the camera back in makes it re-appear).

Tried the what works and not kind of messages from issue 192:

[0( and [1( works fine, 0 stops image rendering while the last frame stays on the screen, 1 will tirn on the rendering

ok. i was more thinking about whether you get a similar error message when Pd-crashes, namely:

microsleep select: Interrupted system call

or simile.

you might also want to try getting a backtrace using gdb.

[device 0185-0000( is unknown info for me however the image has disappeared and only white screen it shows. As I do not know what numbers should be used for my cameras, can not get back the image

the "0815-0000" is the camera-emulator device provided by the Pylon SDK. you obviously have to fill in meaningful values, e.g. the ones obtained by [enumerate(. but anyhow, that wasn't the part is was interested in.

gaksrd IOhannes

On 8/29/18 11:35 PM, Csaba Láng wrote:

The other problem is that can not identify within pd the cameras. Anything I send to pix_video, fails. I called the cams test1 and test2, however [device test1( does not work, therefor enumerate as well or enumProp.

???

It says that a valid device is needed, but f I have an image from the camera at startup the patch, why I cannot enumerate it.

you need a valid device for [enumProps(, but not for [enumerate(. you need to hook something like a [print] to the 2nd outlet of [pix_video] in order to get anything meaningful from these messages.

unfortunately did not understand your gdb or microsleep select advices. please specify

Pd prints an error like "microsleep select: Interrupted system call" on the console before it hangs here. does something similar happen?

about gdb, something like this should do the trick (i'm sure i've posted this before):

$ gdb --args pd -nrt -stderr -lib /path/to/Gen /path/to/pix_video-help.pd
[...]
(gdb) thread apply all bt
(gdb) run
[...make it crash...]
(gdb) bt
[...paste the following outputet]
(gdb) quit

Iohannes,

managed to make a crash, check the gdb log in the attachment. Basically there is a Segmentation Fault. What does it mean and how can I fix it?

On Thu, Aug 30, 2018 at 12:52 AM Csaba Láng langcsaba@gmail.com wrote:

could not manage to make it crash :) jumps between devices smoothly. Will test it the next days, but anyway I big thanks to you! It looks amazing. Now I test with 227fps.

On Thu, Aug 30, 2018 at 12:28 AM IOhannes m zmölnig zmoelnig@iem.at wrote:

...

On 8/29/18 11:35 PM, Csaba Láng wrote:

...

The other problem is that can not identify within pd the cameras. Anything I send to pix_video, fails. I called the cams test1 and test2, however [device test1( does not work, therefor enumerate as well or enumProp.

???

...

It says that a valid device is needed, but f I have an image from the camera at startup the patch, why

I

...

cannot enumerate it.

you need a valid device for [enumProps(, but not for [enumerate(. you need to hook something like a [print] to the 2nd outlet of [pix_video] in order to get anything meaningful from these messages.

...

unfortunately did not understand your gdb or microsleep select advices. please specify

Pd prints an error like "microsleep select: Interrupted system call" on the console before it hangs here. does something similar happen?

about gdb, something like this should do the trick (i'm sure i've posted this before):

$ gdb --args pd -nrt -stderr -lib /path/to/Gen /path/to/pix_video-help.pd
[...]
(gdb) thread apply all bt
(gdb) run
[...make it crash...]
(gdb) bt
[...paste the following outputet]
(gdb) quit

---

Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list

indeed the bt at the end just missed full log in the attachment Thanks for pointing it out,

Popesz

On Thu, Aug 30, 2018 at 4:11 PM Christof Ressi christof.ressi@gmx.at wrote:

I think you forgot the last part:

(gdb) bt [...paste the following outputet] (gdb) quit

IOhannes wants to see a full backtrace. it should look like this: https://ftp.gnu.org/old-gnu/Manuals/gdb/html_node/gdb_42.html

Christof

Gesendet: Donnerstag, 30. August 2018 um 15:36 Uhr Von: "Csaba Láng" langcsaba@gmail.com An: Pd-List pd-list@lists.iem.at Betreff: Re: [PD] videoPYLON C++ update needed

Iohannes,

managed to make a crash, check the gdb log in the attachment. Basically there is a Segmentation Fault. What does it mean and how can I fix it?

On Thu, Aug 30, 2018 at 12:52 AM Csaba Láng <langcsaba@gmail.com[mailto: langcsaba@gmail.com]> wrote:

could not manage to make it crash :) jumps between devices smoothly. Will test it the next days, but anyway I big thanks to you! It looks amazing. Now I test with 227fps.

On Thu, Aug 30, 2018 at 12:28 AM IOhannes m zmölnig <zmoelnig@iem.at [mailto:zmoelnig@iem.at]> wrote:On 8/29/18 11:35 PM, Csaba Láng wrote:

...

The other problem is that can not identify within pd the cameras. Anything I send to pix_video, fails. I called the cams test1 and test2, however [device test1( does not work, therefor enumerate as well or enumProp.

???

...

It says that a valid device is needed, but f I have an image from the camera at startup the patch, why I cannot enumerate it.

you need a valid device for [enumProps(, but not for [enumerate(. you need to hook something like a [print] to the 2nd outlet of [pix_video] in order to get anything meaningful from these messages.

...

unfortunately did not understand your gdb or microsleep select advices. please specify

Pd prints an error like "microsleep select: Interrupted system call" on the console before it hangs here. does something similar happen?

about gdb, something like this should do the trick (i'm sure i've posted this before):

$ gdb --args pd -nrt -stderr -lib /path/to/Gen /path/to/pix_video-help.pd
[...]
(gdb) thread apply all bt
(gdb) run
[...make it crash...]
(gdb) bt
[...paste the following outputet]
(gdb) quit

---

Pd-list@lists.iem.at[mailto:Pd-list@lists.iem.at] mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list%5Bhttps://lists.puredata.info/l... Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list%5Bhttps://lists.puredata.info/l...]

here you are :)

On Thu, Aug 30, 2018 at 4:44 PM IOhannes m zmoelnig zmoelnig@iem.at wrote:

On 2018-08-30 16:32, Csaba Láng wrote:

...

indeed the bt at the end just missed full log in the attachment Thanks for pointing it out,

thanks. that's better. however, i think there's still some parts missing of the backtrace. could you hit <return> rather than <q>+<return> to get another page of the backtrace (and repeat until no more pages appear).

fgmasdr IOhannes

---

Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list

it is much more stable with a simple patch, however can be crashed only clicking on a [device 1( message. The crash in the log what I sent you happenes when I want to use it with pix_blob and changing the trashold in the pix_background object.

On Thu, Aug 30, 2018 at 5:12 PM IOhannes m zmoelnig zmoelnig@iem.at wrote:

On 2018-08-30 16:48, Csaba Láng wrote:

...

here you are :)

cool.

does it also crash if you only use a very simplistic patch without any image processing? (basically just [gemwin], [gemhead], [pix_video]; if you want add [pix_texture] + [square])

gmasdr IOhannes

Just in case I send you a new Segmentation Fault log. Here I have already no method for Gemstate. I send you the patches to have a better idea what I am trying to do. trackerping.pd is the main patch, containing filecontrol. after gemwin is rendered, open languagechoice.pd there are 2 cameras placed in the main window sending to the pd ping1 and pd ping2 to languagechoice.

Basically the 2 cameras are creating one touch surface by placing them one above the wall facing down, the other one facing to the wall. When camera1 gets the wall hit tells the event to camera 2 which gets the position x y from pix_blob. Now when the 2 cameras send to each other info about the presence of an object, I get the segmentation fault.

On Thu, Aug 30, 2018 at 6:11 PM Csaba Láng langcsaba@gmail.com wrote:

I will try to rewrite my patches from scratch as the patch which was simple and contained 2 pix_video with both of my usb3 cameras using 227fps, but I added pix_gain, pix_background and pix_blob, looks good and smoothly works. check the attachment My original idea was to send gemlist from one patch to an other via the send object, in this case closing the patch and opening a new one does not have to switch of pix_video, and does not have to restart the device as it does not work. I hope i explained it clearly :)

On Thu, Aug 30, 2018 at 5:31 PM Csaba Láng langcsaba@gmail.com wrote:

...

it is much more stable with a simple patch, however can be crashed only clicking on a [device 1( message. The crash in the log what I sent you happenes when I want to use it with pix_blob and changing the trashold in the pix_background object.

On Thu, Aug 30, 2018 at 5:12 PM IOhannes m zmoelnig zmoelnig@iem.at wrote:

...

On 2018-08-30 16:48, Csaba Láng wrote:

...

here you are :)

cool.

does it also crash if you only use a very simplistic patch without any image processing? (basically just [gemwin], [gemhead], [pix_video]; if you want add [pix_texture] + [square])

gmasdr IOhannes

On 8/31/18 2:42 PM, Csaba Láng wrote:

Here I have already no method for Gemstate.

that's pretty simple: languagechoice.pd -> [pd ping1] -> [pd map]

the [inlet] (1st outlet of [pix_blob] is a GemList, which you hooked up to a [* 2] object. (btw, when printing to the Pd-console (that is: without the "-stderr" flag), you can Ctrl-click most (red) error-messages, and Pd will highlight the object that has emitted them.)

fgmadsr IOhannes

PS: it doesn't explain the crash though.

On 9/1/18 12:30 AM, Csaba Láng wrote:

Thanks for the hint. The crash I eliminated with a delay 1 between send and receive objects. Very strange that it helped but that was the only idea I had.

the beauty of memory corruption is, that you don't really know when it's going to happen. adding a single line of code might change the memory layout of the program, does moving the corruption somewhere else where you won't immediately trigger it. however, the memory corruption is still there, waiting to bite when you are absent-minded.

fgmdsa IOhannes

Iohannes, I am not sure if understood you well, did you mean that even I have a good graphic card with my patch I am unable to exceed 60 fps? Or you meant a [metro 5] to [gemhead] to achieve 200fps and analyzing the center of gravity with pix_blob could do the job. Popesz

On Sat, 1 Sep 2018 at 01:02, IOhannes m zmölnig zmoelnig@iem.at wrote:

On 8/31/18 2:42 PM, Csaba Láng wrote:

...

Just in case I send you a new Segmentation Fault log. Here I have already no method for Gemstate. I send you the patches to have a better idea what I am trying to do. trackerping.pd is the main patch, containing filecontrol. after gemwin is rendered, open languagechoice.pd there are 2 cameras placed in the main window sending to the pd ping1 and pd ping2 to languagechoice.

Basically the 2 cameras are creating one touch surface by placing them

one

...

above the wall facing down, the other one facing to the wall. When

camera1

...

gets the wall hit tells the event to camera 2 which gets the position x y

to avoid confusion, it's probably better to reserve the term "camera" for the physical device.

...

from pix_blob. Now when the 2 cameras send to each other info about the presence of an object, I get the segmentation fault.

so the crash appears when you you send something to [r ball] (which in turn triggers a [s english] resp [s polski]; which in turn triggers the patch-open and patch-close).

i'm pretty sure that dynamically opening and closing patches is a bad idea. what's wrong with just using [spigot] to let the gemlist through or not?

apart from that, it's crashing because of an illegal memory access in a [bng] object (that is: a graphical bang). this is either an initialization problem or some memory corruption.

anyhow, parts of the patch are missing, (the dynamically loaded patches), and i don't know where [once] is coming from.

i have a slight suspicion that you are running yet another [pix_blob] in your dynamically loaded patches, which you probably don't need (you already have a [pix_blob] on all your video sources), no?

oh, and your cameras might already support setting gain and contrast and cropping the image. if so, you might want to do these things on the camera rather than within Gem.

as for the framerate: even if you run your cams at 1000fps, Gem will run at it's own fixed framerate (which is 20fps by default, and i don't see where you've changed that). Gem's framerate is usually bound by the gfx-card (so often you cannot really go beyond 60fps). however, if you don't do any openGL rendering (most Gem-objects will do openGL, with the big exception of the pix_* family - apart from pix_texture, which does openGL), then you can have a single [gemhead] run at a higher rate by simply banging it to a [metro] running at that rate.

gmads IOhannes

---

Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list

On 9/3/18 9:50 PM, Csaba Láng wrote:

Iohannes,

did you have a chance to check where my memory corruption is coming from?

no.

i'm not totally sure that the problem is in the videoPYLON code; that's why i wrote:

anyhow, parts of the patch are missing, (the dynamically loaded patches), and i don't know where [once] is coming from.

so i'd need your full patch (send it to me privately)

gamrd IOhannes

On 8/29/18 5:46 PM, Csaba Láng wrote:

if I open pd from terminal with puredata command, gemwin cannot be created,

did you do a proper "make install"?

gfamdsr IOhannes