With Arnold, you have to embrace the concept of simplicity in settings for tweaking wholeheartedly and accept it´s meant to be unbiased, lean&mean brute force physical correct.
There are ways to optimize on a per shader level to influence what an individual rays "sees" or "hits" as well as optimizations here and there that help get cleaner results with less than an infinite number of rays. What´s impressing is the memory footprint Arnold can household with and the impression it gives in terms of chewing through geomery in general. That´s the quick part... You may end up not caring at all about the size of the scene but just the amount of rays you can justify for rendering and how best to distribte those rays. E.g. once you have your rendertime ballpark figures for a total amount of rays/pixel it may seem as if it´ll always render similar amounts of time with similar rays/pixel settings. I can´t comment on fur/feathers. For mapped transparencies, VRay may sometimes be as tricky as Arnold. I like VRay, the render quality, the chaos group and the licensing model and would recommend to look at it as a qualified alternative to Arnold depending on the kind of shots you have to do and artists available. In terms of 3D lighting in general, it´s worth giving both renderers a hard test in what their lighting tools can actually provide you with and if baking to textures makes for a big part of your final image. This also applies to how you plan to UV texture map things. Make sure you check for limitations or workflows you may need come up with Cheers, tim On 15.03.2013 21:41, Sebastien Sterling wrote:
Good evening/day everyone ! The people i am working for are currently debating whether or not to take on Arnold, as their official renderer on their next feature film. One of my character fx friends has been Beta testing it to great effect, and now has to make a case for the switch from 3DLIGHT to Arnold. If there are any Arnoldites out their, i was wondering if you had any test renders demonstrating Arnolds efficiencies\ deficiencies VS other renderers; It doesn't matter if its softimage maya or modo.. or mental ray, PR man, vray, maxwell... The extra data would be very much appreciated especially with identical scenes, written observations are equally welcome. So if you've anything at all to share, many thanks ;)
