Window close immediately

[alessandro-zomparelli]

Hi @maweigert , thank you for your great work, this visualizer is really amazing. Unfortunately when I try to execute it, the window immediately close without any kind of messages in the console...

If I try to run spimagine, it work perfectly, but I'm not really able to run your examples... Do you have clues?

Thanks!

[maweigert]

Hi, I'm confused as to when this error appears, so i would need more information to pin that down:

• Could you post a minimal example that crashes for you?
• Did you run it from ipython, from the command line, or as standalone app?
• Which OS are you on?

[alessandro-zomparelli]

Sorry, I wasn't really specific: Ubuntu 16.04, Python 3, Nvidia GTX960M. This is my test code:

from spimagine import volshow import numpy as np

data = np.linspace(0,1,100*3).reshape((100,)3)

w = volshow(data)

w.transform.setRotation(.1,1,0,1) w.set_colormap("hot")

w.saveFrame("/media/DATI/PythonCode/scene.png")

The generated image is that:

The problem occurs on both Jupyter and simple command line. Trying a simpler code, same problem:

import numpy as np from spimagine import volshow

data = np.einsum("ij,k",np.ones((100,)*2), np.linspace(0,1,100)) w = volshow(data)

If I type dmesg | tail I get this:

[ 33.233310] nvidia-modeset: Allocated GPU:0 (GPU-daa1fc2f-e793-5d4f-fd9a-08262cc7c7e7) @ PCI:0000:02:00.0 [ 33.233501] nvidia-modeset: Freed GPU:0 (GPU-daa1fc2f-e793-5d4f-fd9a-08262cc7c7e7) @ PCI:0000:02:00.0 [ 34.407216] nvidia-modeset: Allocated GPU:0 (GPU-daa1fc2f-e793-5d4f-fd9a-08262cc7c7e7) @ PCI:0000:02:00.0 [ 34.407353] nvidia-modeset: Freed GPU:0 (GPU-daa1fc2f-e793-5d4f-fd9a-08262cc7c7e7) @ PCI:0000:02:00.0 [ 38.743852] nvidia-modeset: Allocated GPU:0 (GPU-daa1fc2f-e793-5d4f-fd9a-08262cc7c7e7) @ PCI:0000:02:00.0 [ 38.744090] nvidia-modeset: Freed GPU:0 (GPU-daa1fc2f-e793-5d4f-fd9a-08262cc7c7e7) @ PCI:0000:02:00.0 [ 39.215370] nvidia-modeset: Allocated GPU:0 (GPU-daa1fc2f-e793-5d4f-fd9a-08262cc7c7e7) @ PCI:0000:02:00.0 [ 39.215514] nvidia-modeset: Freed GPU:0 (GPU-daa1fc2f-e793-5d4f-fd9a-08262cc7c7e7) @ PCI:0000:02:00.0 [ 77.630615] nvidia-modeset: Allocated GPU:0 (GPU-daa1fc2f-e793-5d4f-fd9a-08262cc7c7e7) @ PCI:0000:02:00.0 [ 77.630976] nvidia-modeset: Freed GPU:0 (GPU-daa1fc2f-e793-5d4f-fd9a-08262cc7c7e7) @ PCI:0000:02:00.0

Can be something related to my system, maybe with the incoming version 18.04 LTS everything will work fine...

[maweigert]

Hi,

The output is the same as for me and this is expected behaviour (its the cube rendered with the maximal intensity side front). Likewise your png seemes to be saved just fine.

So does it crash inbetween? Can you rotate the cube when you're grabbing it with the mouse?

Did you start ipython with ipython3 --gui qt5

or in jupyter included %gui qt5 at the beginning?

[alessandro-zomparelli]

Hi @maweigert , many thanks, now in Jupyter it works. But i'm still not able to run it with from the command line. Ipython is the only way to execute it from the command line?

[alessandro-zomparelli]

Solved! I was missing the blocking=True argument for volshow: w = volshow(data, blocking=True)

Thanks again! Alessandro

[maweigert]

Cool :)

Strange still, as ipython3 --gui qt5 should give you the default non-blocking behaviour (which is what one normally wants). Was there an actual error message, when you run it from within ipython (without blocking=True)?

[alessandro-zomparelli]

Well, I know nothing about ipython, but because I'm running it under python 2.7 (maybe this is the reason?), I typed: ipython --gui qt5 /media/DATI/PythonCode/spimagine_test_02.py

And without using blocking=True, it doesn't work. Why should I use ipython instead of just python?

[maweigert]

Ahhh, now I see :)

Yes, there are several ways to use it:

1) interactively in ipython, where one first starts the ipython shell with ipython --gui qt5 and then just types in the commands line by line as in the example. This is the way I mostly use it, whenever I have some arrays that is alread loaded or processed in ipython, i just do

import spimagine
volshow(my_array)

This is my preferred way to interact with 3d data (works in Jupyter the same way).

2) as skript to output the rendered scenes: This is like you did it, when you run a script with ipython my_script.py the intepreter will directly return after the window is created. This is normally what I want, as it enables to run lots of scripts that produce some output, without user interruption (e.g. to close the window).

3) What I typically do:

• start ipython --gui qt5
• run the skript from within ipython: run spimagine_test_02.py that should work with blocking=False

Finally, I would always use ipython over the standard python shell. It has tab completion, syntax highlighting, and lots of cool stuff (e.g. you can run normal shell commands from within).

[alessandro-zomparelli]

Ah ok, now it's more clear. In my case I'm interested at the moment in running real-time simulations and see the outputs with volshow. It works perfectly!

But I will probably used it for generating catalogs of images, and for that I will probably use your way :)

Many thanks, and great job again!