Thanks Andrew, Plenty to go at. I will start experimenting tomorrow.
Chris ----- Original Message ----- From: "Andrew Berge" <[EMAIL PROTECTED]> To: <[email protected]> Sent: Friday, March 23, 2007 2:22 PM Subject: Re: Displacement mapping render times > Chris, > > Not specifically dealing with displacement , but as a general speedup > tip i have had some success doing this : > > In the render settings try setting your X x Y supersampling to 10x10 > and then slide your threshold down to around 5 or less and experiment > till the effect becomes disturbing .. this can really speed up a reander > and in some cases it can improve the look and remove some noise at the > same time, what you are basically doing is not rendering adjacent pixels > which are very close in colour. > > Another one is that AA can be VERY intesive computationally and > sometimes if your scene requires a high AA of 3 or 4 you may find that > turning it all the way down to 1 or even off and then instead using the > post effects scaling instead can be faster. > In the case of a seven hour render, I think i would be looking at any > possibilites of breaking the render into layers and then compositing > them together after, even if this dosnt speed the render up much, > (ofcourse it may dramatically speed the render) It will mean that god > forbid if you have to make any alterations you wont have to re-render > the whole thing to change one small ellement. > > If you are network rendering... dont.. :) Unfortuantely realsoft > performs better and faster if you assign a group of frames to each > machine and then save out seperate projects for eg frame > 1-100,101-200,201-300 etc, varrying the number of frames according to > the relative speed of the machines. > > Couple more things worth looking at that dont usually mean a big > performance increase but worth looking at: > > remove any constructors.. rotate,sweep, etc and extract the gemetry from > the constructor and delete them setting the constructor to none.in the > objects spec props. > > dont leave anything in edit mode > > avoid making complex and deep hierachies > > optimize any sds objects to as few vertices as possible, if its a flat > object with sharp corners, or in paces it dosnt show then set them to > poly mode. > > use smaller texture maps if possible > > Well thats about all I can think of at moment, I can really relate to > your situation, its no fun facing a deadline with render times like > that, worse comes to the worse render to a smaller rez and scale up, > youd be surprised how often they dont notice ;) .. good luck with it and > if I can think of anything else I'll let you know > > happy rendering, > > Andrew > > > Chris Sellars wrote: > > Hi Vesa, > > to be in the region of 7 times faster to make them commercially viable. > > Obviously we will be increasing the size of our render farm as we can afford > > to do so, but in the meantime we will just have to limit or exclude the use > > of displacement mapping. The difference with displacement disabled compared > > to the same scene with it enabled, is in the region of 25 times faster, > > which I think you will agree is a huge difference. > > > > Thanks, > > > > Chris > > Morsel Animation > > > > ----- Original Message ----- > > From: "Vesa Meskanen" <[EMAIL PROTECTED]> > > To: <[email protected]> > > Sent: Friday, March 23, 2007 11:57 AM > > Subject: Re: Displacement mapping render times > > > > > > > >> Hi Chris > >> > >> > >>> Unless anyone can come up with a super speed up method for Displacement > >>> mapping, or Realsoft can vastly improve its efficiency, I will have to > >>> avoid > >>> its use and just model displacements in my geometry where possible in > >>> future > >>> projects. > >>> > >> I expect that V6 will render complex displacement mapped scenes typically > >> > > 2 > > > >> times faster. Hopefully this helps. > >> > >> > >> Kind regards, > >> > >> Vesa > >> > >> > >> > >> -- > >> No virus found in this incoming message. > >> Checked by AVG Free Edition. > >> Version: 7.5.446 / Virus Database: 268.18.17/730 - Release Date: > >> > > 22/03/2007 07:44 > > > >> > > > > > > > > > > > -- > No virus found in this incoming message. > Checked by AVG Free Edition. > Version: 7.5.446 / Virus Database: 268.18.17/730 - Release Date: 22/03/2007 07:44 > >
