Hi guy,
what do you think about this comparison between Brazil 2.0 and V-Ray 1.5?
Brazil 2.0 is so fast than VRay 1.5, or what?
Bye
cecofuli
Well, its an amateur comparison…
Terrible secene to do the testing on. And so none of those results can be conclusive.
Im sure that brazil 2 is fast.
Really tired of the which is faster comparisons nowadays.
opss…sorry 
. Anyway, i quote Morbid Angel: very stupid and inconclusive “pseudo-review”…
Wasn’t this comparison done by the guys from Uniform ?
They cant have spent much time on it if so.
I dont really care who did the comparison.
When you compare renders there are certain things you must look to, in order to get correct speed/visual result.
In most cases if the comparison is correct, visual result should match quite closly.
In this result visual difference is significant. So is the render time. Due to the points outlined in the link, there are some major errors done in the comparison.
Secondly the scene in question is a bad choice; bad lighting, bad texturing.
What you must do to compare renders is render separate components one by one, i.e. reflections, shadows, gi, refraction, displacement.
Evaluate each one on its own. Then combine each ofthe results to total, and this will be your correct result.
If for example you place a sphere on a plane and shine an area light at it, and compare pure glossy reflections from direct and indirect then you can correctly judge render time differences.
You must also account for things like sampling algorythms, filtering alrogrythms and so on.
If you compare a vray render that uses 1/24 samples of adaptive dmc vs mental ray 1/3 adaptive subdivision then the actual number of samples taken will be significantly different. That alone will alter the render quality and also time required.
The list may go on…
haha…basically what I said. But in a brief defenition 
Vray v’s brazil its a no contest for me so I’m not going to waste time on that, however uninformed articles like such only reiterate what I have thought for the past few years i.e. 3D World is nothing more than a rag, I see better tuts, product reviews, industry news floating around on the web (all for free). I recall posting a question on their forum about how they can actually charge people €10 approx for such low quality material, do we really need a another f&%^&ing tutorial on how to model a hand???
I hope I’ll be able to finish tweaking the scene for QUALITY, as I honestly can’t be bothered to get into the “mine is bigger than yours” debate.
Just so I can show VRay doesn’t “cheat” normals, or reflections, or GI, as stated somewhere else.
I was under the impression that the author of the blog works for Brazil…
As always… Mine is bigger than yours
It’s very hard to compare 2 renderers…
You can compare feature by feature, not an entiere project… too hard…
Best regards.
sheesh. not sure what the point is comparing things.
guess its the oldest one in the book (just don’t mention the war, must be the other one)
I love vray, and now i too have sinned. Moving some types of projects over the renderer called Brazil, and Im happy to use both. Right tools for the job.
back on track. 3dworld mag has been good over the years, but the need to include everyone has meant its a bit short on quality stuff, very short. CGI magazine, god rest its soul, was much better. a pro mag would go down well, but like mentioned most good stuff can be found on the net.
cgtalk has gone the same way. loads of poser people, loads of cracked software people, and loads of noobs asking, how do you do water, how do you extrude a poly, how do i make my render look like ILM’s etc etc.
It could be interresting to test all renderers feature by feature :
Only the mtion blur, only the displacement, only heavy geometry, only heavy geometry instances…
Best regards.
Ok, here’s what I’ve come up with.
The original image, rendered as extracted from the provided archive:
and a fixed/optimised one:
Notes on the new scene:
Every surface has glossy reflections. There is no “specular” highlight in the strictest sense.Walls, bricks, floors, props, even the cloth has speculars generated as a subset of reflections. Noise on the table is a mistake (kept the sampling too low, my bad) that on the couches is due to the very fine bump map, and lack of AA filtering.
The GI is calculated somewhat roughly, but it’s helped in precision by the Detail Enhancement at rendertime.
The image is rendered as Clamped to 1, as no post was to be done, and i could see some comments coming from some user about the “wrong” AA at the windows.
The optimised scene rendered in 3:03 minutes (on 32 ( ! ) 2.66ghz Intel CPUs), the original one rendered in 3:24 minutes on the same hardware setup.
A net gain of roughly 10% speed, for an objectively superior image quality (not BEAUTY, simple quality).
The only colormapping performed is a slight gamma @ 1.6 in linear space to brighten the scene (very dark walls, very slanted light angle), which could have been done in Post anyway.
Did you have to adjust the system units? In my tests I couldn’t get any of the nterior lights to show up until I set the units to mm. IMO the setup of the scene is a little FU’d.
yes, I forgot that.
I set the scale to proper units.
It was entirely off the bonk to start with.
been following the thread for a couple of days, but I hadn’t realized how ugly that render looked so far. it really makes you cringe. I mean on what basis you write a review, a comparison or whatever, when your test scene looks like that?
anyway, good work there lele, it made me feel much better after having a good look at the untouched scene.
As a note: it seems that some texture went amiss in the passage to the renderfarm (which Vergefio kindly provided). Mainly the diffuse on the couches, that I can see.
Here’s how they looked rendered locally on my very slow rig, in a WiP shot (so don’t consider anything else in it. it’s just to show it wasn’t done “on purpose” to try and cheat):
EDIT: tracked down the issue with the couches to a missing colorcorrect map on Vergefio’s side (hence the black rendering)