Conference follow up

One of the offerings from the pre-Marketgraph PLA4 seminar, this may seem a little “cartooney” but I found this presentation most interesting. As I tend to do most of my design work in SketchUp (but it could be any 3D modeling software, really) one of the biggest problems I have is that palimpsest of design history, preserving the iterations that came before, the design legacy that lead you to where you are. Recognizing this in 3D character design , this paper focuses on the creation of a software that preserves the construction lines similar to those used by comic and graphic book artists. A wonderful addition to some older sketch based modeling software, this is a very interesting foray into equipping the computer with the tools to innovative aid in the design process, rather than just be another form of pencil. You can read the technical paper here, which is a lot better than the video and focuses on the design history.

Another offering PLA4 Conference had that I found rather pertinent [I promise I’ll get to the meat of Marketgraph soon] is the following research. That’s a rather fancy way of saying that the researchers were looking to translate the edges of 3D models into line structures that more realistically reflects hand drawn renderings in their tone, emphasis and variation, none of which a computer can do well right now.

You can view a PDF of the entire presentation here or a copy of the technical paper here.

If you’ve ever worked in Sketchup’s new style features, at first you’re all excited because now you never have to sketch again, you can just design aways to you little heart’s content and when the client says “let me see your conceptual design” you can fire off a few scenes in the pen style and, viola, hand renderings. Right?


Now, don’t get me wrong, SU’s new styles are a great start and they have helped me a lot, but the fact that the sketchy lines are basically the software mapping bitmapped images of lines over the vector data is a problem. There is no thinning of lines to show distance, no thickening to emphasis prominent morphologies, and when you have a long facade disappearing off into the distance all the damn horizontal features of the windows disappear, the verticals stay, but the horizontals are all gone (the fine people at Google tell me this is because the software looks to eliminate small lines, hence those horizontal ones at the perspective narrows in the distance). And don’t get me started on the flat, solid, hard edged shadows out of a Meier rendering.

But the problem isn’t just with SU, and that is why this paper is so interesting. I’ve found that designing in 3D is much better that most 2D process sketches, and we really need a way to communicate the warm and fuzzy of hand drawings, but with traditional architectural drafting conventions. This does that, at least theoretically. It takes traditional modeling and uses algorithms to interpret the geometry in a more traditional sketchy technique.

Now, I had a few choice works with the nice people at Google today during the exhibition. According to them, their emphasis right now has very little to do with NPR but with “Model size and performance”, like we need to start building entire detailed cities in SU – it’s a design tool, not a full 3D uber-render platform. He actually dared ask me if I used the Layouts feature and I explained through gritted teeth that until they have a rendering solution that I don’t have to tweak in Photoshop, I can’t use their layouts because the model is inappropriate to present. I told him that I’m personally at a crossroads – I need an NPR solution, and either I A) Hope Google fixes this in the immediate future, or B) Start to integrate a software like Maya and it’s toon shaders into my pipeline in order to achieve the results I want. He merely smiled. Maybe if I had thrown something, that would have helped. All well, I got them back, I too two of the free swag pens instead of just the one they told me I could have. Vengeance is mine.