Is it hard to use OpenGL in 2D plot?

[Leonid Rumatov]

Is it hard to use OpenGL in 2D plot?

We got some performance problem with 2d plotting, with big data. 

So, this question appears. 

Meybe, there is easy way - only to change some widget to QGLWidget and the usint of QPainter should kept.

[Luke Campagnola]

There are a few different answers to this question:

1) In 0.9.8, there are automatic downsampling features in PlotItem. Try right click -> Plot Options -> Downsample, then check 'Downsample' and 'Clip to view'. This will greatly expand the number of samples you can work with.

2) Have you tried using pg.setConfigOptions(useOpenGL=True) ? This will cause Qt to use OpenGL internally for rendering. On some platforms this leads to a performance increase, but I have generally been unimpressed by it.

3) Jochen Schroeder has done some work to bring 2D plotting to the existing OpenGL system; you can see his changes here: https://github.com/cycomanic/pyqtgraph/tree/glcamera

4) Ultimately, I hope to use VisPy to bring better GPU acceleration to pyqtgraph, but this is far from complete.

Luke

[Слава]

Hello, Luke.

Thank you, Luke, I'll try first and second variant.

--  You received this message because you are subscribed to a topic in the Google Groups "pyqtgraph" group. To unsubscribe from this topic, visit https://groups.google.com/d/topic/pyqtgraph/90vS5ZOSEhA/unsubscribe. To unsubscribe from this group and all its topics, send an email to pyqtgraph+...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/pyqtgraph/CACZXET9gvj1G5VgkPa1bN%3D8zZaNma6V1bEZ%2BFh0Gt6DqLub-DA%40mail.gmail.com. For more options, visit https://groups.google.com/groups/opt_out.

-- 
Best wishes!
 Слава                            mailto:dondu...@yandex.ru

[Samuel Garcia]

Hi Luke and list,
is there in pg (or do you plan?) a mechanism for plotting very big datasets that cannot be in memory.
For instance we can have a 10Gsample in a hdf5 files, and we want to load chunk by chunk when spannin/zoomin/draging. We never want to see the full window.
For that, we should need a hook or something like that could be call each time the zoom is change to load part of the dataset.
I known that Cyril Rossant already made some proof of concept with vispy with this kind of idea.
Do you plan to integrate such a mechanism in pg ? I would happy to help if you want.

Best,

Samuel

--
You received this message because you are subscribed to the Google Groups "pyqtgraph" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pyqtgraph+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/pyqtgraph/CACZXET9gvj1G5VgkPa1bN%3D8zZaNma6V1bEZ%2BFh0Gt6DqLub-DA%40mail.gmail.com.
For more options, visit https://groups.google.com/groups/opt_out.

-- 
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Samuel Garcia
Lyon Neuroscience
CNRS - UMR5292 -  INSERM U1028 -  Universite Claude Bernard LYON 1
Equipe R et D
50, avenue Tony Garnier
69366 LYON Cedex 07
FRANCE
Tél : 04 37 28 74 24
Fax : 04 37 28 76 01
http://olfac.univ-lyon1.fr/unite/equipe-07/
http://neuralensemble.org/trac/OpenElectrophy
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

[Luke Campagnola]

On Fri, Dec 6, 2013 at 4:40 AM, Samuel Garcia <sga...@olfac.univ-lyon1.fr> wrote:

Hi Luke and list,
is there in pg (or do you plan?) a mechanism for plotting very big datasets that cannot be in memory.
For instance we can have a 10Gsample in a hdf5 files, and we want to load chunk by chunk when spannin/zoomin/draging. We never want to see the full window.
For that, we should need a hook or something like that could be call each time the zoom is change to load part of the dataset.
I known that Cyril Rossant already made some proof of concept with vispy with this kind of idea.
Do you plan to integrate such a mechanism in pg ? I would happy to help if you want.

Hi Samuel!

There is nothing in pyqtgraph at present to handle this automatically. However, I have just pushed up an example demonstrating a simple PlotCurveItem subclass that will dynamically load and downsample data from HDF5 with minimal memory usage. See here:

https://github.com/pyqtgraph/pyqtgraph/blob/develop/examples/hdf5.py

The only thing it is missing is a mechanism for caching the downsampled data. Right now, if you wanted to display the entire data set at once, it would have to re-read all data for every repaint (although it would do so without any memory consumption issues). I suppose a cache would not be too difficult to implement..

Luke