You know unbiased engines take a looooooong time to render, right? Such is not practical for animations unless you have a render farm to back up the job.
Maxwell Renderer is, theoretically, the most "accurate" unbiased engine out there. It is more or less designed to simulate the physics behind an IRL camera.
V-Ray has a PPT (progressive path tracing) feature that somewhat simulates an unbiased renderer, but not quite, and there's limitations on what kind of shading level you can achieve with it (so there's a chance you won't be able to render out a truly noise free image).
Thea Render has both biased and unbiased rendering engines, which all share the same material system. In my experience, Thea is maybe ~80% the renderer that Maxwell is. The fact that you can switch engines really easily and not have to retexture everything is pretty handy though.
There are others (Octane, Indigo, etc) but I've never tried them. I only deal with V-Ray and Maxwell on a daily basis (and I only use Maxwell for things that need to look 100% physically correct).
I've tried a few unbiased renderers; quality wise, they are all pretty similar (Indigo, Arion, Octane, Maxwell, all of them can do photorealistic renders without much effort). But the best I used yet is Octane, because if you have a good Nvidia card, it's lightning fast compared to others; I have a GTX 780ti and played with the demo version, and was really amazed by its speed.
The only drawback is the missing displacement support, but I read it will be implemented in the version 2.0
https://corona-renderer.com/ By far the best IMO. Its free, development is very active and it has a unique blend of unbiased/biased rendering system. For Max only at the moment though.
physically (unbiased) rendering engines generally are a bit easier to operate ? since its rendering is based off of calculated light pathing - though it gives the user less control compared to the biased engines... and in general, takes longer to render a product. while the biased engines allows users more control, renders faster, but are less accurate unless under a skillful user? (so a biased software can have the best of both world, but only if the renderer is very capable)
and because of the rendering time, biased engines are usually better for animation, unless you have a lot of resource power to hasten the rendering speed of these unbiased (physically based) renders...
so if i was to work on animation, generally speaking, a biased engine is usually more suitable... while an unbiased (physically based) engine might be better (or perhaps easier) in achieving photorealism if you're only working on a small project (perhaps a single high quality image) where you dont care much about the rendering time, but only the quality of your photorealistic image?
is that why some people say unbiased are better for characters? because these projects are generally less complex than a big movie scene where there's tons of architecture designs and countless objects in the scene... since they are usually just 1 or a handful of characters in a small scene.. so instead of controlling every fine details of the rendering process and lighting to maximize time and perfection, a unbiased (physically based) engine might be more suitable on these small character portraits/scenes ?
is that why some people say unbiased are better for characters? because these projects are generally less complex than a big movie scene where there's tons of architecture designs and countless objects in the scene...
Take a look at gravity, its completely rendered with arnold (unbiased renderer). So you see that unbiased renderers are suitable for heavy scenes too
We use arnold too for the rendering of tv spots. Its really fast and easy to set up and you don´t have artifacts which happen when you use final gathering or lightmaps like in vray or mental ray which cause animation flickering.
The only visible "artifact" in arnold is noise, which can be compensated by more sampling steps. Ahh and something about complex scenes. A current shot for a tv commercial we do has around 3 billion triangles at rendertime (~500 characters and environment) and its still rendering.
We rendered this Playmobil spot entirely with arnold, most scenes only have one skylight and a direct sun/moonlight.
Replies
by that i mean, i dont mean only portrait shots though, but character + scene (interior of a room, etc) however, the focus is on the character.
for still images as well as animation
Maxwell Renderer is, theoretically, the most "accurate" unbiased engine out there. It is more or less designed to simulate the physics behind an IRL camera.
V-Ray has a PPT (progressive path tracing) feature that somewhat simulates an unbiased renderer, but not quite, and there's limitations on what kind of shading level you can achieve with it (so there's a chance you won't be able to render out a truly noise free image).
Thea Render has both biased and unbiased rendering engines, which all share the same material system. In my experience, Thea is maybe ~80% the renderer that Maxwell is. The fact that you can switch engines really easily and not have to retexture everything is pretty handy though.
There are others (Octane, Indigo, etc) but I've never tried them. I only deal with V-Ray and Maxwell on a daily basis (and I only use Maxwell for things that need to look 100% physically correct).
-CMPX
The only drawback is the missing displacement support, but I read it will be implemented in the version 2.0
physically (unbiased) rendering engines generally are a bit easier to operate ? since its rendering is based off of calculated light pathing - though it gives the user less control compared to the biased engines... and in general, takes longer to render a product. while the biased engines allows users more control, renders faster, but are less accurate unless under a skillful user? (so a biased software can have the best of both world, but only if the renderer is very capable)
and because of the rendering time, biased engines are usually better for animation, unless you have a lot of resource power to hasten the rendering speed of these unbiased (physically based) renders...
so if i was to work on animation, generally speaking, a biased engine is usually more suitable... while an unbiased (physically based) engine might be better (or perhaps easier) in achieving photorealism if you're only working on a small project (perhaps a single high quality image) where you dont care much about the rendering time, but only the quality of your photorealistic image?
is that why some people say unbiased are better for characters? because these projects are generally less complex than a big movie scene where there's tons of architecture designs and countless objects in the scene... since they are usually just 1 or a handful of characters in a small scene.. so instead of controlling every fine details of the rendering process and lighting to maximize time and perfection, a unbiased (physically based) engine might be more suitable on these small character portraits/scenes ?
We use arnold too for the rendering of tv spots. Its really fast and easy to set up and you don´t have artifacts which happen when you use final gathering or lightmaps like in vray or mental ray which cause animation flickering.
The only visible "artifact" in arnold is noise, which can be compensated by more sampling steps. Ahh and something about complex scenes. A current shot for a tv commercial we do has around 3 billion triangles at rendertime (~500 characters and environment) and its still rendering.
We rendered this Playmobil spot entirely with arnold, most scenes only have one skylight and a direct sun/moonlight.