Follow

"we have so much data these days, why aren't we using 3D visualizations?" - discuss

@dominikus I'd say: it's not about the dimensionality of the data space, but the dimensionality of our processing space, i.e. our brains that are better at interpreting 2D projections than 3D volumes.

@dutchdatadesign @dominikus much? ;)

I would add that *designing* for 3D is harder — in part due to the perceptual factors, but also just in terms of UI design. It's more actual work, so fewer people explore it. Yet that makes it a great, wide open opportunity space.

@scott @dominikus I'm serious, I'm not sure 'where do we leave all that data' is the question to answer when it comes to visualizations for human consumption.

We could represent data in 16D, but our brains don't know how to make sense of it. Of course, we do better at 3D, but my experience from data analysis is that I've always gained my insight from 2D projections, not higher-dimensional representations.

Maybe the question is: How do we find the right (sequence of) 2D projections?

@dutchdatadesign @dominikus yes agreed! Though: Globes have value in a world of mostly 2D maps. I remain open to the possibility that there may be limited, specific uses of 3D that can enhance our understanding of specific kinds of data relationships.

Two examples (happen to both have defined narrative structures, so maybe that's a factor in successful 3D):

@scott @dominikus @gka I didn't think of globes, kinda is a "world" in its own :)

The yield curve is a good example (but note that understanding is greatly enhanced by stepping thru multiple more-or-less 2D projections). I think the usefulness of 3D depends more on the relationships in the dataset than on the quantity of data.

In the Bolt piece I find the 3D effect not very helpful for analysis, mostly eye candy that makes the presentation more accessible.

@dutchdatadesign @dominikus @gka Nice pun! :)

And yes, definitely, "bigger" data doesn't necessitate more dimensions. :)

I don't think either of those pieces are intended for analysis; they are stories. If someone (not me) were to survey good and bad 3D vis out there, it would be helpful to differentiate between which tasks each excel at (if any!), e.g. storytelling or exploratory understanding or structured analysis…

@scott @dominikus @gka So now we have come full circle, back to answering Dominikus question again: "we have so much data these days, why aren't we using 3D visualizations?"

@dutchdatadesign :) Right you are! And @dominikus has been notably silent on the issue…

@scott @dutchdatadesign @dominikus My take is that 3d simply isn’t working in more than 95% of the cases. and spotting the few cases where it actually might work (if done properly that is) is hard enough, and then criticism is huge if you made the wrong call. lots to lose so we often rather “play safe”. also 3d excel bar charts ruined this dimension entirely for many

@scott @dutchdatadesign @dominikus also i’ve used the “avoid 3d charts” advice in workshops myself because that’s just easier than explaining when 3d makes sense. even now i struggle to come up with a clear and simple answer. maybe this one: “use 3d only if the extra burden of reading and interpreting the perspective dimension is justified by a significant gain in insights that can’t be seen in a 2d alternative form.” but this answer really just opens more questions...

@scott @dutchdatadesign @dominikus err , replace “perspective dimension” with “perspective distortion” ^

@gka @scott @dutchdatadesign thank you everyone for the super-insightful discussion and sorry about launching this whole thing and then disappearing 🕳😊

the quote was verbatim from a conference talk by someone who's doing sci-fi interfaces for movies, so I was really curious what you would think about it.
one major thing I found problematic about his talk was the UX/interaction aspect or: that no one had ever actually USED one of their interfaces 😀

@gka @scott @dutchdatadesign but to get back to the topic: "never use 3D in viz" seems to me like one of those design rules that you should only violate if you're absolutely sure what you're doing...
but i'm convinced that 3D could be interesting, also in combination with AR or some other future stuff (see hackernoon.com/can-augmented-r)

@gka @scott i took more offense with the quote since simply adding one more (visual) dimension doesn't change the overall scalability of a visualization (what @dutchdatadesign already said before & much better).

if you really want to troll me, say something like "we need 3D visualization to solve big data" 😜

@dutchdatadesign @dominikus @gka @scott (bit late on this and probably obsolete) I once started collecting 3D vizzes on my Pinterest board. (nl.pinterest.com/studioterp/in) Think I've stopped collecting since there are not many great examples. I think communicating the axis is where most projects fail and the fact that our eyes/mind just can't handle that much at once and just need overview.
And 16D, wow, what dimensions are that?! 🤢 🤣

@dominikus @gka @scott @dutchdatadesign I recently learned there’s a term of art for such things: “FUI” for ”Future User Interface” or “Fantasy User Interface.” Their main use cases are setting tone and contributing to overall fictional world building, as well as informing and moving the fictional narrative forward. Lots of pretty things on this ReddIt sub reddit.com/r/FUI/

Sign in to participate in the conversation
vis.social

vis.social is an open social platform for creative people, especially anyone in SciArt, SciComm, data, visualization, creative coding, and related arts and research. English is the common language of the instance.