Timelapse pictures & videos with Duet and webcam on layer change
-
@aniron said in Timelapse pictures & videos with Duet and webcam on layer change:
Is there anything in principle preventing this from working on the older Duets, like the 0.6?
Could you restate what "this" is? Lots of things discussed above...
-
@Danal sorry, I was referring to the OP. Didn't realise how old it was.
"Requirements
DuetWifi or Duet Ethernet or Duet 2 Maestro controlled printer
RepRapFirmware v1.21 or v2.0 or higher" -
Ah, got it. OP will need to answer.
I've been intentionally not "stepping on toes" because both @resam and I have tools that do lapse-to-video. Each has its strengths and weaknesses.
But I will mention mine... The one linked below "duetLapse" should be able to interact with any Firmware V2 or V3. It was developed to run on a Pi, but could potentially run anywhere that Python runs. It only needs to be able to open the camera, and to reach the printer via network.
-
I don't see any specific reason why it should not work on a Duet 0.6 - as long as it runs RRF 1.21 or later and has Telnet active.
-
This post is deleted! -
Hi all,
Finally got a Raspberry Pi 4B, have it setup sand installed and working with Duet RFF Timeplase. I've got the Pi taking pics of the print on layer change.
My question is regarding the layer change script. How should I add a line to the script to tell the print head to move to a specified place before each pic, without messing up the core programming?
Thanks in advance!!
-
@Damien I have modified my set up to sync with layer change- it requires no wiring or hacks other than adding a little custom g-code on layer change. I use a bluetooth button to trigger the cell phone camera.
The custom g-code is in the printer profile tab of the slicer, so I made a layer sync specific printer profile. When I want to make layer synced video, I use that profile when I slice, otherwise I use a normal profile.
-
Hi everybody,
is that solution able to do both? Stream a video to the DUET3D web front end to check if the print is still working as expected AND creating time lapse (time based or on layer change) at the same print in parallel?
Or is it only for making time lapse?
My plan ist to get a raspberry pi (3, 4 or zero) in combination with a logitech HD920 cam.
-
@Hugo-Hiasl This mainly depends on the software you use to provide the stream. Some of these will allow you to have a continuous video stream that will just be paused for a second to take a snapshot image. I think both
mjpg-streamer
anduStreamer
support this. But I have never tried it. -
@wilriker Keepalive post.
In answer to your question, yes
mjpeg-streamer
can be used this way and it is how theoctolapse
plugin works inoctoprint
. I have a Vivedino Troodon (Voron-clone) and it's clone Duet2 Wifi card is so good it has the same issues with Octoprint that the OEM board suffers. -
@wilriker
I cant speak forustreamer
butmjpeg-streamer
provides a http stream onhttp://<ip:port>/?action=stream
and can accept multiple connections. I regularly have it streaming on my desktop and media-pc simultaneously without issues or any noticeable load on the Pi.OctoPrint's inbuilt timelapse system, (or octolapse , if you love needless complexity) requests snapshots via
http://<ip:port>/?action=snapshot
, there is no detectable pause on the streams when this is invoked.Conventionally: To get a timelapse you use a script to capture frames at set intervals/events, timestamp them if desired and save them. Once the capture is complete you stitch them together using ffmpeg.
However.. when I want a timelapse on my CNC I start a screen session (nohup shell) on it's attendant Pi and run this:
pi@laserweb:~ $ ffmpeg -r 600 -i http://localhost:8080/?action=stream -vf "drawtext=fontcolor=white:fontsize=16:box=1:boxcolor=black@0.3:x=(w-text_w-10):y=(h-text_h-5):expansion=strftime::text='CNC \:\ %H\:%M\:%S'" -r 20 lapse.`date +%Y%m%d%H%M%S`.avi
Which tricks
ffmpeg
into thinking that my 30fps stream is really a 600fps stream, and the asks it to timestamp that and convert to a 20fps framerate.
The end result is a 20x speedup video with timestamp, but the quality is noticeably lower than the capture/assemble method. And the Pi3 is devoting 32->40% of it's CPU to ffmpeg while it is running. Not recommended as a permanent solution, it's a classic quick'n'dirty one liner.