Open Beta: New toolkit for high-quality plots

[heliosphere]

I'd like to let everyone know that's we're running an open beta for a brand-new LabVIEW toolkit, aimed at making publication-quality plots directly from the block diagram. In addition to regular line, scatter, and bar plots, this toolkit also supports advanced features like 2D contours, vector fields, streamlines, alpha transparency, and inline math using LaTeX. You can save plots directly to disk in half a dozen formats, including PDF and PNG.

We call it the "Advanced Plotting Toolkit", and it's available at http://www.advancedplotting.com along with example VIs and documentation.  The beta is free and distributed as a standard .vip package with no dependencies.  If you have the time, we'd love to hear your feedback.  That doesn't just mean bug reports... this is an amazing community and we'd love feature suggestions as well.  Comments are welcome here or by email (contact@advancedplotting.com).

I've included some screenshots of our example VIs below to give you an idea of what the output looks like. Btw, there is also a stand-alone demo app (built using Application Builder) that you can use to test-drive the toolkit without installing anything in LabVIEW.

[altenbach]

Looks nice. I'll try to play with it.

[Henrik_Volkers]

Looks nice 🙂

For line plots: Can I fill the area between two plots?  (Create a [50%,95%,..]  zone) 

As far as I understood the manual the error bar is tricky for non symmetric values..... How to create  X and Y errors with both none symmetric ranges? Use it twice?

For line plots:  Add symbols not at each single point ...  option to define a x, y or (x²+y²) (or???)  symbol distance ....  (nice if you want to show multiple plots in B&W and want to avoid a handmade decimation) .. ends in more different linestyles ...

[heliosphere]

Hi Henrik,

Filled areas... not at the moment but it's on our roadmap!  You're right... the current errorbar model assumes that the errors are symmetric.  You could try calling it twice and overplotting.  I'll add that feature to our tracker... we could add a polymorphic version where the X/Y error inputs are 2D.

To make a line plot with only a subset of the markers, simply call the VI twice: once with the line visible and markers hidden, and again with just the markers you want and the line style set to invisible.  This is a general feature of the Toolkit: rather than making complex VIs, we encourage layered plots to get the result you want.

[Intaris]

Looks very nice.

I'm downloading the demo right now, we might have a use for this in the near future.  All depends on performance, will report back as soon as I have numbers.

[altenbach]

---

@Intaris wrote:

All depends on performance, will report back as soon as I have numbers.

---

Performance does not seem very high but I also don't think that is the main purpose. You don't use this to update a graph millions of times per second. Instead, it is mostly used to show a finalized result for puplication quality output or for a report.

The toolkit looks well put together, and contains extensive help. The installation does not add examples to the example finder, so I recommend to immediately download the examples from their web site.

Some features need a more detailed explanation, e.g. the definition of "symlog" for the axis mapping.

My understanding (wild guess!) is that this is a LabVIEW interface to an existing open source graphing package (matplotlib?) and as such is pretty polished. Nothing wrong wit this. The generation of publication quality advanced plots is a weak point in LabVIEW and it is a pain to make your own from scratch (example).

Overall, I like it. Still need to play a little more. 😄

[heliosphere]

Thanks for the note on Example Finder... btw you can still get at the examples by browsing through "Directory Structure", although there's some kind of glitch that makes them not show up in the "Task" view.

You're right about matplotlib!  Although this isn't an official matplotlib project, we're in touch with the developers and the Toolkit contains an embedded copy of the library.  We credit them on the About page of the site (under "What Can I Plot") as well as in the acknowledgements section of the manual.  In fact, the project that became the Toolkit began when I got tired of having to export data to Python every time I wanted a publishable plot. 🙂

[AaronTheAE]

Heliosphere! This is some great content! Tried out the demo and like it so much myself that I downloaded it from VIPM. We always love a new good toolkit to add on! Thanks and keep it up!! 

[thoult]

It's very rare that I come across toolkits that seem genuinely useful to me, but this one looks fantastic.

Will have a look at the beta with anticipation.

[thoult]

Some thoughts:

- Setting the legend string to an empty string means it doesn't display on the legend, as I would imagine is expected behaviour (certainly is in gnuplot). Maybe you could document this in the help?

- Descriptions for the typedefs wouild be useful for those of us who use palettes with icons only, as Context Help helps you figure out which one you're after when you're unfamiliar with the palette.

- Have you got any plans for future matplotlib features to include? I'm thinking pie charts (possibly currently doable via polar charts?), mplot3d, boxplots...

It looks great though!