A quick post to mention a 3D visualization tool for UML diagrams developed using Google SketchUp . The tool has been created as part of a case study to discover whether using a 3D perspective has a substantial impact on model comprehension.
Looking at the current status of the tool I’m NOT convinced about this claim (I’d be more interested in having a good zoom in/out operator for models, able to show/hide details of the models when zooming in/out but this is another story) though I guess I should give them a chance. You can get a feeling of the tool by looking at this youtube video
FNR Pearl Chair. Head of the Software Engineering RDI Unit at LIST. Affiliate Professor at University of Luxembourg. More about me.
Your post says, “The tool has been created as part of a case study to prove that using a 3D perspective has a substantial impact on model comprehension.” I presume you mean “…a case study to _discover whether_ using a 3D perspective has a[n]… impact.” 🙂
You are right, my wording implied that there was an impact. My mistake
There are currently two students taking the previous work on 3d models for increase model comprehensibility further. Their thesis are underway, one thing they have done is to implement a functional demo (with zoom and panning options). I’ll try TO see if they ARE willing TO contribute WITH a blog post ON their discoveries ON http://blog.ituniv.org
Best regards,
Marcus (archie)
I’m not sure that a 3D view adds much to compositive like structures, although from an implementation perspective an easier task to accomplish.
I’ve often thought that Colaboration Diagrams and Sequence Diagrams lend themselves to usefully blending.
Would be interested in seeing some further development of the concept, in a more professional look & feel, some more robust and serious approach for using in practice, provided it was shown effective.
Found some other shots as well, not sure whether the same team/tool: https://www.youtube.com/watch?v=gcgQajTXVrA
Still, seems like a an interesting idea. I’m very curious about it’s promise, what value it can add, and where (or if) something like this could take us.
I would be amongst the very early adopters if it offers the same features as current tools, but at least could greatly facilitate model manipulation and visualization, so an easier development if the tool does what any should: assist.
Was there any progress related to this concept since this post was published in 2009? Has it been proved useful/effective?
Let’s see where it goes, I hope somebody seriously pushes it.
Hi Alex, I haven´t really seen anything new on this topic (except for this other post I wrote: X3D-UML – Implementation of UML 3D with X3D which links to the video you pointed out as well.
Let’s hope someboyd gives it another shoot. I’d also like to see some more work on this area before deciding whether it has some value or not.
Alex,
To be honest, I keep buying my hardware with oversized 3D capabilities in the hope I can justify myself ( namely, a time or materials budget) the development of such presentation/interaction tooling.
In meetings with Integrated Teams, where many of us where discussing UseCases(legs)/User Stories (realizations), with Team Members all the way from Business, Req.Analysis, Design&Impl, DBAs, Testers, as a Process Engineer/Discipline Lead I observed that the models we where discussing and walking through on the whole, had at least 2 defects:
they were whether partial slices or presentation views to focus on the particulars relevant for each team/layer,
or they were huge renderings we had to produce by stitching (tape, actually) up to 4 A3 sheets.
Therefore, I envision a multi-screen meeting environment, where a facilitator/ conductor may proceed introducing the main course of a scenario walk-through in a “central?” screen, and when appropriate, specific specialist Team Members, could focus on the details more relevant to them in separate screens.
This would allow to squeeze the maximum feedback from all involved, while consuming the least of Teams Members’ valuable time.
Indeed, current 3D/alpha rendering capabilities of graphical chipsets, display subsystems and OpenGL/WebGL/CSS3 would help a lot in the quality and amenability of the presentation experience.
Let’s hope we find somebody to pull some moneys for this – or we just hit even a smallish lotery price – and we’ll be able to enjoy such facilities.
… sorry:
the “other” defect I found, was that some or others Teams Members were bound to get bored to death, when exposing them to details that were whether less relevant to their work, or they actually knew by heart all ready.
I think that 3D will bring better understanding in different aspects of software design, implementations, workflow, stack traces – you name it. The problem as I see currently is that the tool just tries to depict flat diagrams in 3D. They were developed to be flat. Something new has to be invented to be used in 3D. To give you an analogy – present software languages known to me don’t support multi-core processors out of box. New algorithms have to be developed to make it happen. For example, mutexes work well but for one processor. In multi-core environment we have to remember, that all processors not just one could be affected by context switch due to a mutex-based sync. So, it’s not just one context, but many. Thus non-blocking algorithms come in result.
Thanks Jordi. Yes, 3D &| WebGL could better serve modelling, by providing dynamic zooming, or even better allowing different audiences to drill-down in the perespective more relevant to them – specially for multi-tiered diagrams where more than one architectural layer is modelled.
This is specially relevant for the example, which uses ROOM-like capsules of UML2, and inherently bring us to the very useful paradigm of mechanics’ and electricians’ pry-open/fold-closed of devices AND cables/connectors, bluring the distinctions of black/gray/white boxes in the usual flat UML representation diagrams.
Cavan, yes Sequence Ds can time-track behavior happening in a composite structure, but I observe more often an isomorphism between composite structures and UML Activities, where call-behavior actions delegate on yet another Activity. Also, Activity in-out parameters and action input/output pins may quite isomorphically match CompositeStructure Ports on Capsules.
In this sense, Activity Diagrams bring the “actual-behavior-happening” over various scenarios (over time) which may be accomplished by executions in a composite structure.
This was actually what some (of us) intended in the duality between the Process Components and Business Process views (with their common Component Collaboration Architecture) of UML Profile for EDOC and also the related UML Profile for Enterprise Application Integration – precursors and drivers of the UML2 standard.
Jordi,
May I ask how may I change the avatar appearing by my comments ?
Mind you – not that I don’t like it (as native of Valencia I like anyting with flapping bat-like wings ! — note for non-valencians: the totemic animal of our city is a bat —
Indeed, I am just joking – but would surely change it to a likeness closer to mine !
Cheers,
ACV
Creating an account in gravatar.com with the same email that you used to leave the comment should make the trick (even for past comments). Let me know if it doesn’t
Thanks Jordi. Yep, it works.
Forget multi-screen. Let’s try HoloLens http://techxplore.com/news/2015-12-hololens.html
https://www.microsoft.com/microsoft-hololens/en-us
Bet you even freakiest hackers start modeling like there’s no tomorrow.
Thanks for your contribution Jordi. I am a big State Machines fan and I extensively use Hyerarchical State Machines in my day to day work. Using hierarchy simplifies A LOT models and reduces modeling time… definetly it makes sense to have 3D modeling tools.
Just think in biologists when they visualize proteins, architects when they design a building with CAD tools… the degree of model complexity used in SW is becomming so high 3D will become a must.