Quiz: which one is the Cycles rendering and which one is the Luxrender rendering?

Hi,

After doing a lot of Blender renderer feature research I ended up with two final contestants for my needs: Cycles and Luxrender.

I rendered the same scene with both renderers, using the best available rendering settings to my knowledge — Cycles: Full Global Illumination preset, and Luxrender: Bidirectional + Metropolis.

Materials and light strengths were also matched as close as possible.

My main goal was to find out if there’s a really noticeable difference in realism between Luxrender’s ‘physical’ unbiased rendering (MLT + spectral wavelength rendering) and the ‘non-physical’ unbiased rendering of Cycles (Quasi-MC + RGB color rendering).

Below are the results, only slightly post-processed (some minor adjustments such as the added vignette).

I’d like to ask you two questions:

1: Which image is rendered by which renderer (Cycles / Luxrender) ?

2: Which one is your preferred rendering?



The answer can be found in this thread.

Thanks,

Metin

P.S.: The model is available as a 3D print.

my tip:
A - cycles
B - lux

I like the B more, but the difference is really subtle

Thanks! I’m eagerly awaiting the next replies.

I also think it’s
a)cycles
b)lux

no preference really.

Thanks Photox!

A. Cycles
B. Luxrender
I’m liking A better.

  1. No idea
  2. I prefer A

A: Cycles
B: Lux
Prefer: B

B- Cycles.
A seems more pleasing.

How much time did you spend tweaking? can you get the color more the same, less red in A and more bright the hdr light spot reflection in A. Tweak A because I like B better. The red in A looks looks like it belongs to something different.

As @kzainger pointed out. There is a difference in lighting setup (lux hemi hdr is diff then cycles hdr).

A : cycles
B : Lux

Thanks for your replies guys!

I deliberately didn’t change the material colors, because I wanted to see how each renderer would treat the color shades, including gamma, tone mapping and spectral versus RGB. I did try to mimick the carpaint materials. I’ll tell some more about the process later.

I’ll wait for some more replies to get a better impression of your opinions, and then I’ll reveal which rendering is done with which renderer.

It is such a simple render it is hardly a test. My guess is they are both rendered in Blender Internal because BI could certainly achieve the posted look.

A better test would involve emission from the objects themselves and lighting other objects that surround the subject. This is where scanline fails and raytracers start to excel. It is also where Cycles would fall behind Lux because Lux supports more advanced algorithms. So it all comes back to the settings, right? Also what is the time? Is A faster than B?

The images demonstrate that you are fluent in Cycles and Lux based scene setup.

Thanks Atom.

I agree that an indirect lighting setup would be more representative for a comparison, but my goal was to see if an average scene setup of mine (usually a product presentation) would already look better / more realistic with a physical, spectral rendering approach. Next, I wanted to read your guesses and opinions.

My guess would also be
A Cycles
B Lux

I prefer B, mainly because of the DOF imho being too strong on A. Anyway, the difference between both images is subtle.

What bothers me more (especially if this ia a setup for product presentations) are the apparent shading artefacts along some of the edges… Did you use the standard OBJ importer or the modified one? (Sorry for getting off-topic…)

And I would also be interested in the render times.
The images are really head-to-head quality wise, so I would just use the solution that renders fastest.

Thanks for your replies guys, much appreciated and very interesting.

The results, also based on related discussions outside this thread, at the Luxrender forum, MoI 3D forum and on Facebook:

Most people think A is rendered with Cycles and B is rendered with LuxRender.

With regard to preference, the opinions are evenly divided. Some prefer the subtlety of rendering A, while others prefer the more pronounced reflections of rendering B.

Here’s the answer:

A is rendered using Luxrender,

B is rendered using Cycles.

For the Luxrender rendering I used the Carpaint material with a red / black diffuse shade and the three standard grey specular shades. I tried to mimick that as closely as possible with a Cycles shader containing three glossy layers with differing roughness, on top of a diffuse shader, and a Fresnel input for realistically converging reflectivity.

The other settings were also matched as closely as possible, including light strengths and DOF.

I used the exact same colors for the materials and the lights. The Luxrender rendering turned out quite saturated. I even slightly desaturated the reds of the Luxrender rendering in Photoshop. Cycles rendered the same colors with visibly less saturation, which helps establishing an impression of realism. I expected it to be the other way round: more saturation from a renderer using the limited RGB color range.

To get matching highlights I increased the Fresnel value of the Cycles shader, hence the slightly more pronounced reflections in that rendering.

As for my personal preference, I really can’t decide which one I like more, especially because I know that decreasing the Fresnel a little in the Cycles rendering would yield less pronounced reflections.

Please do not regard this little experiment as an instigator for another ‘my renderer is better than yours’ discussion. I just wanted to see if a non-complex rendering setup would already show differences in realism between a ‘physical’ spectral wavelength renderer (Luxrender) and a ‘non-physical’ RGB color renderer (Cycles).

All the best,

Metin

since both renders are “acceptable” the real question must be which is faster ?

been a while since I tried Lux (been told its a lot faster now)
but why wait an hour to find out I made a mistake ?