From mconway@microsoft.com Tue May 26 15:22:30 1998 Received: from burdell.cc.gatech.edu (root@burdell.cc.gatech.edu [130.207.3.207]) by lennon.cc.gatech.edu (8.8.4/8.6.9) with ESMTP id PAA15622 for ; Tue, 26 May 1998 15:20:22 -0400 (EDT) Received: from wheaten.hitl.washington.edu (EcK4WI5eHIFPkf+zcR0vBppzJ+4aOtPe@[128.95.73.60]) by burdell.cc.gatech.edu (8.8.4/8.6.9) with ESMTP id PAA14722 for ; Tue, 26 May 1998 15:20:11 -0400 (EDT) Received: from mail4-b.microsoft.com (mail4-b.microsoft.com [131.107.3.122]) by wheaten.hitl.washington.edu (8.8.8/8.6.12) with ESMTP id MAA29708 for <3d-ui@hitl.washington.edu>; Tue, 26 May 1998 12:20:01 -0700 (PDT) Received: by mail4-b.microsoft.com with Internet Mail Service (5.5.2328.0) id ; Tue, 26 May 1998 12:20:50 -0700 Message-ID: <4FD6422BE942D111908D00805F3158DF05B26743@red-msg-52.dns.microsoft.com> From: Matt Conway To: "'Ernst Kruijff'" , 3d-ui@hitl.washington.edu Subject: RE: extremely hungry for polygons and texels Date: Tue, 26 May 1998 12:20:48 -0700 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2328.0) Content-Type: text/plain; charset="windows-1250" Status: RO Making peace time: I think there's more consensus than contention here... > -----Original Message----- > From: Ernst Kruijff [mailto:ernst@kwetal.ms.mff.cuni.cz] > Sent: Tuesday, May 26, 1998 9:28 AM > To: 3d-ui@hitl.washington.edu > Subject: extremely hungry for polygons and texels > > > > > >> On a semi-technical point, my experience with artists > working in VR is > >> that they are extremely hungry for polygons and texels. > > > >In my experience, the best artists understand the > constraints of the medium > >that they're working in, whether they work in watercolor, > oil, polygons, or > >pixels. It is certainly an difficult skill to acquire but > not an impossible > >one. > > > I don't agree with either of you, I am afraid. I experience, > and so do many > other artists and architects, that simple primitives (low > polygon count) and > NO texels work very well in some design processes. I think we are more in agreement than in disgreement. I didn't say that artists START with high-res, or that high-res is always the end product, or that textures are always appropriate. Sorry if what I wrote left you with that impression. What I AM asserting is that artists who create textures and models should (and can) understand the limitations of the media they work with, and that the good ones do. One of the hallmarks of a skilled artist is that they can deliver a good model on a low polygon budget and can cover it with a texture with a low-pixel budget. > If you consider conceptual design tools (CDS, JDCAD and so > on), users do not > start with using highly detailed primitives and texels. If > you consider the > approximate modeling time with HMD based modeling tools - > approximately 30 > to 45 minutes - you don't have time for advanced editing. > There is not even > a high need to allow textures and advanced deformation of forms during > conceptual modeling - non-realism is often needed. > That is also why tools like Sketch use non-realistic > rendering techniques. A very powerful argument for the non-realistic rendering is that it is useful when your work needs to communicate teh meta-point that "this is a work in progress." Conceptual renditions in 2D or 3D should have the look of a tentative design, not a finished product. Presentations that look "too polished" do not encourage people to suggest changes or to give the kinds of agressive feedback that "sketchier" presentations can give you, which is often important in the design phase. Generating high-quality textures doesn't need to be back-breaking work. NOTE: I am not implying that the often times the best textures are not high resolution, just high-quality, often hand-painted (not a stock wood-grain). > Modeling for presentation purposes indeed requires a high > amount of polygons > and texels - in that case, you are right. To clarify my point: I believe that SOME but not all presentations need hi pgon count. More often one needs high texture count, and sometimes, rarely, you need both. Rather than focussing on polygons, I think we have to see polygons for what they are: a means to an end. The end we seek is the perception of high-detail ('high spatial frequency') and it is often possible to get that fidelity from a skillfully rendered texture map, even if it is wrapped around a very small number of polygons. I am in complete agreement that there are places where you can't play authoring tricks like this. Furthermore, there are incremental technologies on our horizons for both texures and polygons that don't make us choose these sorts of tradeoffs: authors create content at the highest level of fidelity possible and the system degrades things as needed (progressive meshes, wavelets, etc..) Matt