# 3
# location change via translate.x
* STTransform(translate=(self.image.size[0] * (-loc0 - 0.25), 0.0))
Please note, that this only works for
subdivide method. impostor needs a bit more thought.
4 replies
https://vispy.org/getting_started/plot.html
^ "... coming soon"
This is kinda core doc, no? How to code a "Hello World" plot with vispy...
^ "... coming soon"
This is kinda core doc, no? How to code a "Hello World" plot with vispy...
@p-i- You'll have to check the examples in the vispy repository. We are a small team with limited time and haven't had time to include all of this information. If you are attending the SciPy 2021 conference this year (going on now) we are doing a sprint on Saturday and Sunday as part of the conference. This could be a task to be worked on.
I've moved from hybriddesktop/mobile/fullstack to 90% numpy/python/visualizations lately and I'm fascinated to find this project.
Previously I used https://github.com/unconed/mathbox but the cognitive load / learning curve (and lack of community) fried my brain.
I just can't guarantee to deliver on time with that tech as it rabbitholes everywhere.
I just can't guarantee to deliver on time with that tech as it rabbitholes everywhere.
I'm working on 2 projects that I might be able to migrate to vispy.
First one is plotting points. Each point is a trade. Intensity is volume.
A few thousand dots per frame.
I get 20-30fps rendering into an XxYxRGB numpy array and blitting it to screen thru opencv. And I can catch mouseclicks.
But it gets sluggish as I increase the dots.
And I have to do numpy voodoo for multiple trades occupying the same pixel, that I'd love to gut out from my code.
And the other, same deal just plotting an array of coloured rectangles.
It's for visualizing a stock graph, the rectangles represent trades or orderbook info.
And I click on a trade to zoom into it & observe market activity in close temporal proximity to that trade. I want to do this as an overlay.
I figured I can pass a numpy vertex array into a shader.
@djhoese it might be an idea to package all of the examples as a handful of .ipynb-s. That way,
(1) it's really easy for people to simply run the notebook, and see everything on offer.
(2) it's possible to publish the notebook with output, just render it to HTML, so it could double as doc.
I know that with small teams it's a pressure to update doc (& so it can get out of sync), as limited resources should be focused on the trunk.
(1) it's really easy for people to simply run the notebook, and see everything on offer.
(2) it's possible to publish the notebook with output, just render it to HTML, so it could double as doc.
I know that with small teams it's a pressure to update doc (& so it can get out of sync), as limited resources should be focused on the trunk.
gotta start using a calendar.
@p-i- Thanks for the suggestions. There is an issue on github already about updating the gallery for the vispy website with more of the example scripts (something on the todo list for the scipy sprint). The problem or confusion that notebooks cause is that vispy's jupyter widget is experimental and does not currently function very well. I would not want people to think that they should expect everything in vispy to work in a jupyter widget. If the widget backend wasn't used then a separate window is popped up which can just be awkward. Lastly, some of the cool stuff in vispy is done by embedding it into a larger Qt (or other GUI framework) application
And how to plot a large array of points, so that if multiple points land on the same pixel the intensities add?
Also two_qt_widgets.py segfaults:
RuntimeError: OpenGL got errors (Check before draw): GL_INVALID_VALUE
Everything else seems to work from /examples
For the oscilloscope you might need pyopengl installed.
Yes, it should be no separate Visuals but a collection. Each visual is treated as separate GL program and this affects performance if there are hundreds or thousands of them. A collection circumvent this by using only one program to draw all items. You might skim through the vispy issues. There have been discussions recently (this and last year I think).
pyopengl is a library and some parts of vispy use it. Other parts, most parts, use a custom linking to your systems GL libraries
@p-i- ^
the plotting interface as a whole is very limited and unfinished. It was put together by the original vispy developers as a proof of concept but needs a lot of work
I reverted to doing it with matplotlib for the time being (as I have a deadline):
https://stackoverflow.com/questions/68404385/dynamically-redraw-second-axis-subfigure-upon-click-event-in-first-in-matplotlib
I would be interested to know how much work it involves to achieve this with vispy.
If anyone's interested to take it on as a contract, do message me.
https://stackoverflow.com/questions/68404385/dynamically-redraw-second-axis-subfigure-upon-click-event-in-first-in-matplotlib
I would be interested to know how much work it involves to achieve this with vispy.
If anyone's interested to take it on as a contract, do message me.
@djhoese I like the pattern of 'doc-thru-code'.
I kinda wish all doc was centered around this.
It cuts out using English as an intermediate language.
I kinda wish all doc was centered around this.
It cuts out using English as an intermediate language.
@BaiSource Keep reading the documentation below that section. There is a Windows section and Linux section that describe switchable graphics for those platforms. The easiest solution is to configure your sysyem to use the nvidia gpu for everything. This isn't a vispy thing, it is up to your system to tell vispy what it should use.
@p-i- I'm not really sure what you expected, but coming to a volunteer open source project on a deadline is rarely going to go well. And I'm not sure if you're criticizing our documentation or saying you like the examples. I've already said we have limited time and that we're working on the documentation so repeating that our docs aren't up to your standards isn't accomplishing anything except making me feel bad. If you have improvements in mind and they aren't already in github issues please make some. Even better, submit a pull request to improve what you can.