I’m a Swiss animation student doing my final graduation-thesis. The topic is about render engines, comparing realtime-renderer such as Unreal with more common engines in the field if animation such as Arnold or Redshift.
I prepared some questions and wanted to give this board a shot getting my thesis started!
No right or wrong, just let me know what you think:
1 - Raytracing-engines look amazing but take some time to render, whereas unreal engine has no rendertime but is not calculated as accurately. Or is it?
2 - Why is there no non-game-engine-based-renderer available yet for animators (e.g. for Maya or Cinema4D)?
3 - What is the reason, raytracing-renderer take so much longer to render compared to the same setup in Unreal?
I’d be super grateful if some of you guys would take a minute to answer some questions!
I think there are many, maybe even plenty… So much computer animation series are produced nowadays, that I think there are many
Raytracing differs based on used system of rules. The complexity of raytracing depends on this rules. How many times to reflect light, how to make usage of diffraction, refraction, what to do with smoothing, how to eval shadows… How many traces to throw in unit volume and in how many directions… As you can see much metrics to think of. Realtime render systems are cheating in some moments and we get rather good results…
there is a big difference in sampling too, as offline renderers always do lots more samples per pixels than realtimes renderers. (it implie lots denoising and filtering algorithms )
hey there!
here are a few humble answers for you :
when you are referening to raytracing engine , you maight want to call them offline renderers (like Corona, Vray, Brazil , mental ray, as even if they use raytracing, they have lot of other features/optimisations.
for accuracy, offline renderers are “scientificaly” more accurate dans realtime renderers. lots of things like caustics, refractions, diffractions are simulated more than computed in realtime engines.
it s true and false, as for exemple, you have renderers like octane and hypershot, wich are not realtime, but tend to get a more efficient way of computing pictures than “regular” offline renderers. and lots of people (including me) attend tu use Unreal engine as an offline renderer .
one of the main concern is that in a physicaly accurate world, each object impact each other during rendering, (not only transparent objects). realtime renderers tend to approximate everything, (base on perception more thant accuracy) to limit computation.