thanks for comments.

I wasn't after a detailed scientific result, but just looking for rough indications. But I will make a second round given your responses. Basically, I used defaults / out of the box settings, click and run, at a "no visible artefacts" quality level.

For Vue, the Broadcast level is considered fine for web gallery stills and professional animation, and (both being biased renderers) it should compete with Firefly's high-end settings as mentioned: bounces 12, irr cache 90, idl quality 90, samples 3. Firefly's result were not a surprise, but Vue's results were as I recently spend some serious rendertime on images which took say 10 hours on average (and smaller in size and lower settings). And of course Firefly results can be halved or doubled by altering values, same for Vue, but that's not going to turn about 7 hours into less than 7 minutes on about comparable quality output. Note that Firefly ran on 12 threads while Vue could only use 8.

Same for Lux, I mentioned the No Acceleration, Hybrid and Pure settings and each seems to trigger another tracing routine. As the Poser scene was set up using Paolo's gear (lights, studio) I only had to ensure that the mirrors were reflective indeed. I'm sure everything can be tweaked here too, and the 4800 samples/pixel is completely arbitrary although earlier tests revealed that 1200 - 2400 is required for final web-gallery quality. But even at the lowest limit Lux takes twice the time of Firefly.

Octane was the one driving it all, as I was about to replace my version 1.53 by the new 2.0. Same settings etc and it took slightly longer indeed. You can compare quality, the image above is the Octane 2 one at 4800 s/p. There is far more shine on the girls skirt (and a lot of other places) which hampered a meaningful detailed Photoshop image subtraction for comparison. There is slightly more noise in the mirror above the girl's shoulder. Both compared to the Octane 1.2 render in the gallery. Again, another image could have done with a 1200 s/p quality limit turning the Octane 2 number into 20 minutes.

But in the end, comparing figures for renderers on CPU-only with one at GPU-only can only be indicative. Everyone has a different kit. The differences by the CPU-only tools (Vue on 8 threads, Firefly and Lux-NoAccell both on 12) are clear.

I'll look into Octane 2 alternative settings - just for my own good, isn't it?

- - - - - 

Usually I'm wrong. But to be effective and efficient, I don't need to be correct or accurate.

visit www.aRtBeeWeb.nl (works) or Missing Manuals (tutorials & reviews) - both need an update though