All posts related to V3
#390275
I don't normally look to FIRE as a way to preview displacement but I was curious about a test I recently did.

A simple cube. Each face 30x30.

A displacement map on each face.

Within 30 seconds on my machine the render was at SL 5 and looking pretty good.

It took 17 seconds in FIRE before it finished voxellizing and showed first image. And the image was quite rough. By 30 seconds it didn't look nearly as good as the full render.

Doesn't the full render also have to voxellize? Why does it produce an image SL 1 so fast, and SL 3 or 4 before FIRE shows first hint of an image?
#390280
The two engines simply require very different strategies (i.e. targeting realtime use vs. ultimate performance and quality). Say you had two clients, one who wanted to see progress as quickly as possible, and another who said, don't worry about showing me anything until you're done -- you'd approach jobs for each in very different ways.

As for FIRE being slower, in the long run, that is fully expected; its strengths are specifically a) that it renders direct lighting as quickly as possible, b) that it gives frequent updates (the production engine only updates at each SL), c) that it supports quick stop/restart, and d) that it supports dynamic changes to the scene -- all four of which are complete non-factors for the production engine, which is therefore free to use very different strategies for how it sets things up and renders the image.

Regarding your specific example, the point would precisely be that it is specific; given the requirements for FIRE, it can be that it will set things up slower or quicker than the production engine. Render to SL20 though, and compare the completeness of the result (caustics fully resolved, etc), and the time it took to get there, and you will invariably find that the production engine wins. There are effects that could take FIRE years to resolve, which are handled easily by the production engine, and that is fine, since FIRE only exists for the purpose of letting you know what's going on with your scene, while you are working.

As regards displacement in particular, you don't specify whether you are using pretesselated or on-the-fly displacement; the two entail work of a very different nature, but regardless, given what I point out above (especially item (d)), you can see how you might expect to see quite different behavior; in the end, it just depends on the scene.
Chocolate test with SSS

nice