  | | | connecting outputGeometry[0] - > inMesh? | connecting outputGeometry[0] - > inMesh? 2003-11-21 - By Joe Spadaro
Back Szabolcs and Gary, Lets not under estimate Gary's thought pattern here. Note he suggested using a surfacing tool this would be better than creating a low res model from scratch. In fact I have based entire crowd pipelines on this workflow. I would not keep the high res in the file though nor would I constrain parts to increase performance, a good LR bind on one light poly mesh will far out perform constrains to the point of real-time.
The workflow I would recommend would be to:
* Create a Low res and a high res model that has a trans and a no trans hierarchy. * The trans hierarchy should be identical in both case containing your skeleton and other motion controllers, this is what you animate. * The noTrans hierarchy should be everything that is not animation dependent like bind skins history driven surfaces and anything else that changes from LR to HR. * It is perfectly fine to group these two hierarchies under a main DAG for organizational reasons but don't animate the notrans hierarchy or it could break your referencing. * Then reference your LR while animating and only animate attributes under the trans hierarchy. Switch your reference file to the HR at render time.
If you decide to avoid referencing this WF will also provide for easy animation (hierarchy below) import or paste.
Joe Spadaro http://www.shinyredguitar.com
> > >-- ---- ---- ---- ---- ---- -- > >Date: Thu, 20 Nov 2003 15:07:22 +0100 >From: =?us-ascii?Q?=22Horv=E1tth_Szabolcs=22?= <szabolcs@(protected)> >Subject: Re: connecting outputGeometry[0] -> inMesh? > >The thing you say is perfectly valid, but won't get you too far. Even an `optimized lowpoly` >version of a scanned model is too high for most animation cases. You can hide the binded mesh >and that basicaly turns of any calculation of the bind (except if you constrain something visible to the animated >surface), but it might still be too slow for realtime animation. You have to really clean things >up to the basics, or build a lowpoly version of the model to get realtime feedback with >more characters interacting on the screen. > >Personaly I'd suggest building a simple version of the model from scratch, just to give you >the visual feedback for dimensions. You can cut the scanned model to pieces and link to bones, >to get more precise body contact, but most of the animation can be done using only the lowpoly version. >And use the final version for major animation tests only, and to test intersections. > > >Cheers, >Szabolcs > >ps. Take every word as IMHO. The scale of the project migh seriously affect the way you set things up. > > > >>>Hi, I was thinking along the lines of the topology of the different >>>meshes being different, they will basically conform to the same shape, >>>but the [highres] version will not be the result of running a >>>smoothing operation on the [lowres] mesh, let me explain ... We plan is >>>to start off with a [highres] cyberscan of a sculpted maquette from >>>this we are looking the derive a [lowres] mesh using something like >>>Cyslice. Basically I was trying to find a way that I could rig the >>>[lowres] mesh (for animation) and have this drive a [highres] mesh for >>>rendering, basically as a way to get round having to drive a whole load >>>of points whilst animating. >>> >>>Now maybe I am getting Mayas model of operation slightly wrong, as you >>>seem to be pointing out in your (PS2 see below), maybe my best bet is >>>to just rig the [highres] mesh and then hide this, maybe then cutting >>>the [lowres] mesh into sections that can be parented directly to bones >>>for doing / previewing animation. One question with this would be "Can >>>I turn off the [highres] bind, i.e. stop it calculating to help speed >>>things up"? Maybe this just happens with hiding or maybe its something >>>that I can explicitly set ........ >>> >>>thanks again to everyone that answered, I am just trying to get a >>>handle on the kind of workflow I am going to need to deploy. >>> >>>cheers gary. >> > > > > >
-- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- --- List-help: <mailto:listar@(protected)?Subject=help> List-unsubscribe: <http://www.highend3d.com/maya/listserver/> List-subscribe: <http://www.highend3d.com/maya/listserver/> List-archive: <http://www.highend3d.com/maya/archive/>
|
|
 |