DuetLapse available for Alpha testing

Danal

Python honors symlinks, that is exactly how I'm keeping one central copy of DuetWebAPI and not copying it around to each DuetLapse, TAMV, etc, etc. as I develop this stuff. If you want it not in the current directory, and/or don't want to use symlinks, then properly put it in Python's path please.

By the way, git clone will NOT update things, 'wget' has all kinds of edge cases, I haven't found a really bullet proof way to distribute scripts from github (besides set up a package server. Which I am reluctant to do, and even that requires entries in /etc directories on user endpoints).

Anyway, symlinks work fine, I'm curious why yours don't.

ln -s ../DuetWebAPI/DuetWebAPI.py DuetWebAPI.py right?

Screenshot to get colors:

arhi

@Danal said in DuetLapse available for Alpha testing:

Python honors symlinks, that is exactly how I'm keeping one central copy of DuetWebAPI and not copying it around to each DuetLapse, TAMV, etc, etc. as I develop this stuff. If you want it not in the current directory,

kill me if I know why but it didn't work .. I'll try again later but this work as is now so ...

and/or don't want to use symlinks, then properly put it in Python's path please.

Need to try that again, when I did with PYTHONPATH=/root/DuetLapse/DuetWebAPI python3 DuetLapse.py it was finding api ok but failed something else, but maybe now that I fixed the other thing that will work, will try

By the way, git clone will NOT update things

of course, git pull does ( git checkout or stash first if files are modified like they are now)

'wget' has all kinds of edge cases, I haven't found a really bullet proof way to distribute scripts from github (besides set up a package server. Which I am reluctant to do, and even that requires entries in /etc directories on user endpoints).

git clone
git pull
git checkout/stash

works awesome, the only issue I need to handle is this path for libs so it knows how to find api, will retest this but even as is it's not a biggy

hahaha retested the symthing now, it works, the %^$#*&^#

PYTHONPATH=/root/DuetLapse/DuetWebAPI python3 DuetLapse.py -h works too, I'll use this way .. the problem was obviously the numpy or whatever is that thing called...

anyhow, these are irrelevant parts of that diff .. what about that change in the onePhoto() ? makes sense?

arhi

Danal

@arhi said in DuetLapse available for Alpha testing:

hahaha retested the symthing now, it works, the %^$#*&^#

Don't you just love things like that!

@arhi said in DuetLapse available for Alpha testing:

anyhow, these are irrelevant parts of that diff .. what about that change in the onePhoto() ? makes sense?

Makes perfect sense, I just need time to look at it... probably next day or two.

arhi

@Danal said in DuetLapse available for Alpha testing:

Don't you just love things like that!

I'm not 40 any more, doing things between 3 and 5am obviously need to be doublechecked

Danal

@Adrian52 said in DuetLapse available for Alpha testing:

Great program, but I still have not got it quite right. I am currently using. /DuetLapse - duet 192.168.1.16 -camera pi -detect pause, with M25 and M400 added to each layer in gcode (easy to do in kisslicer). The problem is that two frames are taken, one when the head is moved away by pause.g, the second a bit later, when the print has resumed. I also get an error message on the console 'm24: cannot print because no file is selected'. It looks as though the m24 is being sent twice.
I have edited my pause.g so that there is no z shift.

Just pushed changes to the detect pause logic that should eliminate this. It has been tested to ensure no syntax errors, but has not been run through a print.

Adrian52

@Danal here is my first go

https://vimeo.com/415969620
Worked fine with M226 in the gcode file. Excellent. Will try detecting layer changes later. Great work.

phdrumlineka

I was attempting to run this on the same PI as my OctoPrint setup. I kept getting errors saying numpy wasn't installed. After attempting the suggested command to get numpy installed, it claimed that pip wasn't installed for python3. After a lot of googling I got pip and numpy installed, but I was still getting the same error saying numpy was required.

I finally added some additional output to DuetLapse.py file and found that I was getting the following error:
Original error was: libf77blas.so.3: cannot open shared object file: No such file or directory
It looks like the answer to that error was to run
sudo apt-get install libatlas-base-dev

I just thought I would post this here in case anyone else runs into this issue, or if anyone has a better way to solve it. I am novice with Linux and Python, so this may not be the best way to handle things.

I am using the following versions of OctoPrint/OctoPi:
OctoPrint version: 1.4.0
OctoPi version: 0.17.0

Danal

Interesting.

I developed the script on the "Duet" provided image for a Pi, and I'm sure that has things installed that were transparent to me. I've tried to have the script produce meaningful errors when something is missing, and I will need to do that for Numpy. Also, I will look into the libatals-base-dev.

pip is not needed by the script at runtime; it is needed to install things the script needs. As you say, python3 does not install pip by default.

A lot of instructions show how to "bootstrap" pip install via python itself. The actual proper way to install pip on a Python3 (or mixed) system is sudo apt install python-pip. This would (should?) get the libatlas dependence automatically.

I've updated the script to show the required commands, if numpy is not found.

Thank you.

phdrumlineka

@Danal
I am pretty sure that is not how I did it . Good to know. Thanks for looking into that. Hopefully it will help others in the future.

I would also like to mention that it is working now, happily taking pictures of my current print. Good work. I appreciate it.

Danal

@phdrumlineka said in DuetLapse available for Alpha testing:

I am pretty sure that is not how I did it

No worries. Most of the instructions and tips that turn up in a search are for older versions.

arhi

just figured out fswebcam will fetch image from rpi camera too .. and its "title" works better than "annotate" with raspistill for adding txt on the image

arhi

I managed to make the "dual channel video"
sample

now don't know how to properly view it. In potplayer you can switch between the channel using ALT+V but I'd like some picture in picture or something similar more .. will experiment more with the result.. I forked the original repo so I can keep track of changes and so it's safe to play with it as I'm not a python dev so I'm probbly doing things wrong

Danal

Python is pretty easy in some ways.

A pic-n-pic video "editor" would make a very cool result.

arhi

@Danal I managed to get used to "no ;" somehow with go but this whitespace structuring is really annoying ... irrelevant ... enough ppl bitching about it online on programming forums no need to spill it here, I just mentioned 'cause probably things I did can/should be done differently (properly - in the spirit of language).

A pic-n-pic video "editor" would make a very cool result.

I actually expected that I'll easily find a video player that will do just that, allow 2 tracks to be played pip style or one next to another or something like that, but most players don't even support multi track video and best I could find is "easy switching" with ALT+V on the potplayer. I generate first both videos as mk4 and then use ffmpeg to copy them into single mkv with two tracks. Will investigate further, maybe there is better way. So far I like that I don't have to use raspistill and that the fs.. can do both rpi camera and usb camera, and I want to be able to capture both and that works somehow ok. Now I need to see how to make a best output from the two sets of images

Danal

@arhi said in DuetLapse available for Alpha testing:

@Danal I managed to get used to "no ;" somehow with go but this whitespace structuring is really annoying ... irrelevant ... enough ppl bitching about it online on programming forums no need to spill it here, I just mentioned 'cause probably things I did can/should be done differently (properly - in the spirit of language).

I am doing a BUNCH of Python, regular old C, C++, and C# right now (different projects). Subtleties of syntax difference are driving me crazy. (As well as C string/pointer stuff, which I'd forgotten how bad that really is/was).

It is really funny how your mind works. I am really good about no ";" in Python, and yes ";" in the various C flavors... except... on printf statements. The syntax checking editor gets me every time. I just can't make my mind do it.

And, I really, really, really, like the indentation based structure of Python. It is 1/2 the reason I use it instead of PERL. To each his own...

Danal

@arhi said in DuetLapse available for Alpha testing:

I actually expected that I'll easily find a video player that will do just that, allow 2 tracks to be played pip style or one next to another or something like that, but most players don't even support multi track video and best I could find is "easy switching" with ALT+V on the potplayer. I generate first both videos as mk4 and then use ffmpeg to copy them into single mkv with two tracks. Will investigate further, maybe there is better way. So far I like that I don't have to use raspistill and that the fs.. can do both rpi camera and usb camera, and I want to be able to capture both and that works somehow ok. Now I need to see how to make a best output from the two sets of images

I was thinking more of an editor, so there is nothing at "my end". Just a regular vid, with side-by-side or p-n-p.

Speaking of which, I don't have any players installed on this machine, and I haven't been able to look at that mkv quite yet.

arhi

@Danal yes, everyone has their preferences. I do system service programming 10 hours a day in c/c++ and since not many of us in my group know how to debug a core file without symbols that was dumped by a 64bit sparc cpu I kinda do some of those too so asm/c/c++ .. and while I can carry syntax of "whatever you throw at me" including python, I rarely write it "in the spirit of ..." but rather "hack it somehow so it produces required results" (as you can see if you diff your tree to my fork)

arhi

@Danal said in DuetLapse available for Alpha testing:

and I haven't been able to look at that mkv quite yet

not much to see really, usb camera looking at the print, rpi camera looking at ceeling, you play the video and if your player support only one track you will see either part being printed or ceeling of my workshop, if player support multitrack you can switch between one or second video... so far did not find a way to see both tracks in the same time, exploring if that's possible, if not I'll try to make different videos then. What I do is make IMGA*.jpg and IMGB*.jpg, I make two mp4 videos (one from IMGA* other from IMGB*) and then I make a third video .mkv that combines first and second video

Danal

@arhi said in DuetLapse available for Alpha testing:

I do system service programming 10 hours a day in c/c++

Then I have a really off the wall question for you:

Do you know how to FORCE "Visual Studio Code" (Note: not Visual Studio) with PlatformIO installed running the Espressif32/ESPIDF platform, force it to compile a mixed source module project with C compiler for .c and .h files and C++ compiler for .cpp and .hpp files? I can handle the 'extern' and such for the cross-calling, so the linker is happy, no problem. I just need it to use the right #$% compiler.

I thought the whole "use the right compiler by file extension" thing was automatic, and it seemed like it has worked before in VSC. However, at the moment, I have a project that throws tons of errors in the .cpp files, because it is treating them as .c.

Like I said, off the wall, because it is a VERY narrow question: VSC (not VS), PlatformIO and Espressif32/ESPIDF...