Recording images of scenes

[psomers]

Hi Sofa team,

I saw that the python3 bindings for implementing the pygame example mentioned in this thread were removed shortly after the conversation here took place (glewinit() and draw()). Is there a plan to re-implement them or replace them? Or maybe I’m missing the replacement. Since Sofa can be driven entirely from a python environment, it is pretty handy to be able to implement a simplified viewer without handing over the main thread. I’ve added back in the bindings in my copy of sofa and it’s pretty easy to have a nice gui using pyqt and python debugging capabilities aren’t lost.

I guess I’m asking more out of curiosity than need at the moment.

Keep up the good work!

Cheers,

Peter

[jnbrunet]

Hey Peter,

I just answer a similar question here: https://www.sofa-framework.org/community/forum/topic/sofapython3-pygame-rendering-example-not-working/#post-18513

Let me know there if you have any further questions.

J-N

[etagliabue]

Hello everyone,

I am also looking for a way to add a virtual RGB-D sensor in my simulation (thus, both 2D image and depth, giving the whole visible 3D point cloud) and I have come into this thread. Since last post is of some months ago, I am wondering if any of you has made any progress meanwhile and/or has any suggestion? @jieying @balazs @jangirrishabh @pierreshg

Thanks!
Eleonora

[psomers]

Hi Eleonora,

I think this is what you could be looking for. I use it to do exactly what you‘re describing.

Cheers,
Peter

[etagliabue]

I don’t know what happened but my post was supposed to go here (I see the title of the post is the same).

Thanks for your reply @psomers! I’ll try that out.

Eleonora

[Arthur]

A solution for recording a scene without using SofaPython3. In fact it is a different process entirely that fakes inputs to the gui

If you are running sofa on a linux desktop you can record a running sofa scene with the following python script (N.B. you might have add the –start flag to the runSofa command).

import os
import pyautogui 
import time

##This is needed if you want to record on a desktop with more than one window to make the sofa
# gui the active window
os.system("wmctrl -l > /tmp/window_list.txt")
sofa_window = None
with open("/tmp/window_list.txt", "r") as ping:
        line = ping.readline()
        while line:
            if "Sofa -" in line:
                print(line)
                i = line.index("Sofa -")
                sofa_window = line[i:-1]
            line = ping.readline()

cmd_string = 'wmctrl -a "'+sofa_window+ '" '
print(cmd_string)
os.system(cmd_string)

# this will begin recording wait ten wall clock seconds (not simulator secdons and end the recording
pyautogui.write('v', 0.5)
time.sleep(10)
pyautogui.write('v', 0.5)

Also if you are running a version of sofa that is not compatible with the recording functionality in SofaPython3 or for whatever reason you can’t get that recoding to work there is a work around I have found. If you have a headless server with sofa installed on it you can record videos with the following (tested with python3 code).

import time
import subprocess

class SofaRecorder:
    def __init__(self, running_on_headless_server=False):
        self.x_server = None
        self.pyautogui = None
        self.headless = running_on_headless_server
    def start_x_server(self):
        if self.x_server == None and self.headless:
            self.x_server = subprocess.Popen(["sudo","X","-config", "/dummy.conf"])
            time.sleep(3) #wait for process to get online

        # pyautogui cannot be imported until there is a display to connect to process launches
        import pyautogui
        self.pyautogui = pyautogui

    def start_recording(self):
        self.start_x_server()
        time.sleep(1)
        self.pyautogui.write('v')

    def stop_recording(self):
        self.pyautogui.write('v')
        time.sleep(1)

    def stop_x_server(self):
        subprocess.Popen(["sudo", "kill", str(self.x_server.pid)])
        time.sleep(1)
        self.x_server = None

# example of usage
if __name__ == "__main__":
    recorder = SofaRecorder(running_on_headless_server=True)
    # if the x_server is not started before the sofa gui launches
    # the sofa process will seg fault
    recorder.start_x_server()
    # without the -g batch option this launches a sofa gui
    sofa_process = subprocess.Popen(["runSofa", "--start"])
    recorder.start_recording()
    print('recording started')
    time.sleep(30) #note this records for 10 wall clock seconds not seconds in the simulator
    recorder.stop_recording()
    print('recording stoppped')
    time.sleep(3)
    sofa_process.terminate()
    print("sofa terminated")
    time.sleep(3)
    recorder.stop_x_server()
    print('x server terminated')

These are the dependencies I had to install on my docker container which already had sofa set up in it:

sudo PIP_TARGET=/usr/lib/python3.7/dist-packages python3.7 -m pip install \
    pynput pyautogui

sudo DEBIAN_FRONTEND=noninteractive  apt-get install -y python3-tk python3-dev
sudo apt-get update
sudo apt-get install -y xserver-xorg-video-dummy x11-utilss
export DISPLAY=:0

and dummy.conf is

Section "Monitor"
  Identifier "Monitor0"
  HorizSync 28.0-80.0
  VertRefresh 48.0-75.0
  # https://arachnoid.com/modelines/
  # 1920x1080 @ 60.00 Hz (GTF) hsync: 67.08 kHz; pclk: 172.80 MHz
  Modeline "1920x1080_60.00" 172.80 1920 2040 2248 2576 1080 1081 1084 1118 -HSync +Vsync
EndSection
Section "Device"
  Identifier "Card0"
  Driver "dummy"
  VideoRam 256000
EndSection
Section "Screen"
  DefaultDepth 24
  Identifier "Screen0"
  Device "Card0"
  Monitor "Monitor0"
  SubSection "Display"
    Depth 24
    Modes "1920x1080_60.00"
  EndSubSection
EndSection

[Author]

Posts