I just recently learned that it's AD's shortcomings of properly exposing MR features.
So the blog you linked to features two Mill employees. Both who I feel we're very lucky to have and have been instrumental in a re-education of the MR and AD relationship along with what can be done with something that will be packaged for the foreseeable future. If it's going to be bolted on, might as well make it useful right? Trust me, your frustrations are shared by every artist here. At the same time, I feel like we have an extremely hostile environment for renderers considering the variety of styles, scope, volume of work, along with the usual technical hurdles that comes with every job. Arnold and Mantra gets it's fair share criticisms here too. We have people here that are intimately involved with MR who understand real production problems and to be fair, the blame has been wrongly placed in some respects on MR when it should have lied with the AD translators. But that is hopefully changing. Whether or not it's done in time I don't know. I'm a cg artist, not a crystal ball. The intention of our development here in our hostile environment is to have it eventually be integrated at the top with AD, and that link to AD is in place as well. For now, all I can say is that I do finally feel we can make progress on a renderer despite its unpopularity. At the end of the day, they all shoot ray bundles and return pixels. The speed, quality, and balance between control vs ease of use determines if it's good or not. Working on making those attributes better results a better tool for production. Here, developers are entrenched with artists so they know exactly what we're up against. David Hackett sits right next to me and hears me bitch about every single artifact, flicker, additional second of render time, and has expressed positive moves on MR part to address all those issues. We make him use Arnold too and light shots because we are mean. The renders out now are all pretty damn amazing, but it really does require accessibility to the features. I just read not 2 days ago on this very list SI can't render curves in Arnold when the alpha version of MtoA does via a checkbox. Wat? So it's not just specific to MR. The problem exists for every render engine and every platform (Vray still sounds like magic tho). Here's hoping Autodesk hears us out. -Lu On Wed, Nov 7, 2012 at 10:26 AM, Ed Manning <[email protected]> wrote: > Hi all -- > > Not to start another flame war, but after months struggling with what > should be simple things, I have to ask: > > Why is Mental Ray integration so haphazard in AD products in general and > Softimage in particular? > > For example, MR now supports much-improved IBL, ptex, iRay, and per-object > sampling settings, as well as a set of new BSDF-based surface shaders. > > NONE of these are exposed in Softimage. Third-party means of exposing > some of these do work, but not very well. IBL, for example, seems not to > support transparent shadows at all. In Maya, they work. Having only global > settings for unified sampling is a crapshoot -- for some shaders, it's like > super-speed, while others actually get noisier and slower. Ray-depth-based > optimizations, which should be simple, have to be manually set, per > parameter, IF you can even get your hands on a third-party shader that > provides accurate counts of raydepth and type. Framebuffers only work > properly with third-party shaders (they slow down renders ridiculously when > used with the native "x" shaders) and don't properly account for > reflections and refractions that are more than one ray-hit deep. The list > goes on. > > The few features newly-exposed in Softimage, such as Unified Sampling and > MetaSL, are poorly documented if at all. The only help for working with > these tools, which we pay Autodesk for, comes from third parties, NVidia's > forums, and Maya users. I have to spend time translating tutorials and blog > posts from Maya-speak to glean the most basic information > > The failure on Autodesk's part seems to be universal, if worst in > Softimage-land -- even though more things seem to work in Maya (or even > MAX), there's little in the way of documentation or tutorials from AD. For > example, because Maya's render settings are so lame and poorly-oriented for > Mental Ray, there is a 3rd-party plug-in (Mental Core) simply to make it > possible for users not working at fully-pipelined facilities to set up MR > renders and get useful framebuffers and passes out. There is also this: > > http://elementalray.wordpress.com/2012/08/10/new-maya-rendering-ui-testing/ > > Basically, if I understand this, NVidia, not Autodesk, has written a new > MR render UI for Maya, which has to be installed as a plug-in, and which > bears a striking resemblance in its organization to the venerable Softimage > Render Options. > > So AD's devs can't even port a UI that they developed from one 3D package > to another? NVidia has to do it for them? > > Am I the only one frustrated & disappointed by this? > > > etm > >
