I wouldn't base render quality on render times of a simple scene as renderers organize and optimize differently. What is one renderer's strength could be another's weakness. Case in point:
Back in 2002 I ran side-by-side tests to demonstrate to my students how the Softimage|3D renderer and mental ray differed. For those that can remember, the Softimage|3D renderer was blazing fast out of the box and everybody loved it except for it lacked customizable shaders, stability, and just about anything else you wanted in a production renderer. The students loved it because it was so fast and used it even though it couldn't produce the looks they wanted. I had to twist their arms to even look at mental ray. The test scene consisted of a single texture mapped sphere at the world origin with a single raytraced shadow casting point light, and the heaviest anti-aliasing filter (Bartlett 4). It took the Softimage|3D renderer about 2 seconds to render from press of the button to completion message. My students were impressed. When I ran the same test using mental ray, mental ray thought about it for about 40 seconds, then quickly rendered out the tiles to finish just under 45 seconds. Students groaned. Then I made a simple adjustment - I moved the camera up to the surface of the sphere so only a couple of polygons filled the entire frame. When I re-ran the test the Softimage renderer now took more than 5 minutes! Mental ray took the same 45 seconds. The reason for the difference was the Softimage|3D renderer was view dependent and as affected by what the camera could see. Mental ray operated in world space and didn't care where the camera was placed unless a shader was applied providing additional instructions. To evaluate a renderer, you have to push it through many tests with all kinds of variables. In the end you'll discover any renderer will do some things very well and other things not so well. You should use the renderer that caters to your type of work. Matt From: softimage-boun...@listproc.autodesk.com [mailto:softimage-boun...@listproc.autodesk.com] On Behalf Of Christopher Sent: Monday, April 01, 2013 6:47 PM To: softimage@listproc.autodesk.com Subject: Re: Vray vs Arnold Displacement Arnold wins, can we compare Arnold to 3Delight ? Arnold is at 13 seconds, and VRay is at 18 seconds, maybe it is true nothing can beat the speed of Arnold, no wonder it gets the attention. Christopher [cid:image001.jpg@01CE2F0B.DBC883A0] Raffaele Fragapane<mailto:raffsxsil...@googlemail.com> Monday, April 01, 2013 9:30 PM http://mashable.com/2010/08/22/how-to-undo-send-in-gmail/ [cid:image001.jpg@01CE2F0B.DBC883A0] Gustavo Eggert Boehs<mailto:gustav...@gmail.com> Monday, April 01, 2013 9:20 PM where is that undo email function it is 2013 already! -- Gustavo E Boehs http://www.gustavoeb.com.br/blog [cid:image001.jpg@01CE2F0B.DBC883A0] Gustavo Eggert Boehs<mailto:gustav...@gmail.com> Monday, April 01, 2013 9:19 PM i'll hide now -- Gustavo E Boehs http://www.gustavoeb.com.br/blog [cid:image002.jpg@01CE2F0B.DBC883A0] Steven Caron<mailto:car...@gmail.com> Monday, April 01, 2013 9:08 PM really? i am pretty sure the unaccounted time (4.82) is actually what makes his render take 18+ secs. and i also believe the subdivision, displacement, accel build etc is indented under the bucket rendering because its part of the bucket rendering and not in addition to it. [cid:image001.jpg@01CE2F0B.DBC883A0] Gustavo Eggert Boehs<mailto:gustav...@gmail.com> Monday, April 01, 2013 8:54 PM Thats actually 18 seconds, 13 just for bucket rendering +3 for subd, +1 for displacement +others... Still thats 18 seconds with brute force GI and Arnold scores :) -- Gustavo E Boehs http://www.gustavoeb.com.br/blog
<<inline: image001.jpg>>
<<inline: image002.jpg>>
