Question about render quality and the time it takes.

Usernamenottaken

There are times where I am using HDRI's to render and I can get a great image out in 5 minutes using a 2080ti. However, I also have scenes that are only lit using actual objects are emitters. While these MUCH longer, they often times seem to possess a different feel entirely. One worth waiting for because the objects and elements are actually there and not just an imagine.

So my question is - If I am rendering for 3 hours at around 3,000 iterations, how much better is it going to look at 6,000 iterations? At what point can you look at an imagine and say "Okay, that's enough" or will you get unrivaled quality if I let this go for 8 hours? Aside from being speckled with unfinished pixels, will the final product change or do final products just offer clarity?

I am only asking this to see if it is worth waiting this extremely long amounts of time or if it's just better to cancel the render at 3,000 iterations instead of waiting for 7,000 if the difference will be minimal.

Gordig

There are too many variables at play to give a meaningful answer to that question.

felis

To my knowledge, you can't set a rule.

A key to getting fast renders is having enough light. The less light there is the longer it will take to converge.

nicstt

Don't use iterations as a judge of how done a render is.

Look at it... and ask: "Am I happy with how it looks?"

If you're happy with it, it doesn't matter how many itterations it's had. If you're not happy then the same applies and you wait for more to pass.

If you get your image to perfectly clear, it will look fake, particularly if there is no depth of field; photos aren't like that and neither is real life.

 

outrider42

The iteration count doesn't mean a whole lot, and really convergence doesn't really mean that much either. Some scenes you create might run thousands of iterations and yet accomplish very little, while another scene might finish in a few hundred iterations. One factor that increases how many iterations you need is how many times the light needs to bounce off things, basically a more complex scene that requires more calculations is going to take more iterations.

Paintbox

Think of it as this, the more light sources (and HDRI counts as a bunch of light sources) and the more objects of which light rays can bounce of, the more time a render will take before it will be done.

margrave

Paintbox said:

Think of it as this, the more light sources (and HDRI counts as a bunch of light sources) and the more objects of which light rays can bounce of, the more time a render will take before it will be done.

If you use a spotlight and change it from "Point" to "Disk" or "Rectangle", does that still count as a single light source?

fastbike1

@margrave "If you use a spotlight and change it from "Point" to "Disk" or "Rectangle", does that still count as a single light source?"

Yes. The advantage of geometry is that (if you give it dimensions) it will act like a real world photography softbox. Rule of thumb for photographic lighting is bigger/closer equals softer shadows. In general, using geometry for a spotlight is never a bad thing.

margrave

fastbike1 said:

@margrave "If you use a spotlight and change it from "Point" to "Disk" or "Rectangle", does that still count as a single light source?"

Yes. The advantage of geometry is that (if you give it dimensions) it will act like a real world photography softbox. Rule of thumb for photographic lighting is bigger/closer equals softer shadows. In general, using geometry for a spotlight is never a bad thing.

I always use geometric spotlights, for precisely that reason. Just wanted to make sure the same photographic principles apply in Iray world.