V-Ray 3.3 for 3ds Max Review by Hammer Chen

[AndresAhumada]

Hi guys...

It's very interesting. It shows the good and bad of the new V-ray. But I see a lot of curiosity and surprise that there are big differences in performance, compared to the new method.
It really is that? ... Or there is an error in the review that showed these big differences in performance?

V-Ray 3.3 for 3ds Max Review by Hammer Chen


Cheers

[super gnu]

first thing i noticed, is he has got vray lens effects working with the vray sun?! doesnt work for me.

[FSGFX]

thats a good one...

VrayTriplanarTex has great potential for daily use. In fact, it is one of the major reason you should upgrade to SP3.

[Morbid Angel]

I don't think you can judge the performance 1 to 1, and I don't think he does either. He does state that he can get cleaner image overall with 3.3 Also I recall chaos saying that the new method is not faster, its probably a bit slower. But it gives more visually pleasing image. As computers get faster, I think there is less need to tweak each parameter to death. I for example have long abandoned the shader/light subdivs tweaking though I used to be crazy about that. But these days I let the computer do the thinking, just crank a few nobs and go home

[AndresAhumada]

I just wanted to raise the fact that in this review do enfassis this exponential loss in performance in some cases. What generates some mistrust on the engine.

[super gnu]

no automatic settings are perfect. in those cases you need to roll up sleeves and tweak like you did in 3.2 what is almost perfect is Chaos support.. send the scene in and they will help you fix it. 3.3 is a mindblowingly good release, and i dont understand where criticism can come from. in 90% of cases its faster for an equivalent looking result, and you still have the same options to optimise as before.

[Morbid Angel]

I think its a bit premature to say 40%-250% speed loss, by simply testing on one scene. Further more, he notes that though 3.3 renders take longer they are much cleaner. So, to produce an evenly clean image (especially in the darks) there is more render time needed.

[squintnic]

its only one scene, not really a proper review imo its just clickbait to get you to look at ads
i have been very impressed with the 3.3 speed improvements (working on detailed environments) i find it so much easier to dial in a reasonable render time and control noise levels - no more bloody subdiv fiddling!

[Joelaff]

We find we can raise our noise threshold with 3.3 and usually get renders faster than 3.2 that look better. You can't compare the same threshold numbers.

[Alan Iglesias]

We find we can raise our noise threshold with 3.3 and usually get renders faster than 3.2 that look better. You can't compare the same threshold numbers.

This is exactly what I am finding. Working with a complex interior with many lights and materials, 3.3 is rendering it very clean and with very reasonable times, cleaning up the noise significantly faster than 3.2 did. Also, it's very easy to adjust the Color Threshold for (much) faster iterative draft renderings.

It might be noted that this project was started fresh in 3.3 and besides some Vray VFB weirdness, it is working flawlessly.

-Alan

[^Lele^]

Hammer reads here and he's a forum frequenter, so i'm sure he'll explain himself (ie. he surely doesn't need me to help do so.), but i thought the review was quite interesting precisely because it shows scientific method: he has a thesis, he tries, the results say different, he adjusts the thesis, then tries again.
True to science, he's objective in noting all the minutiae BESIDES straight-up rendertime and noise threshold comparison.
Then again, who knows what google translate did with that blog post: we may all have read a different version, lol. (mine had rendering substituted for "nomogram". Which made me loopy for a few minutes.)

[hammerbchen]

Since the meaning of "Noise Threshold / Color Threshold" in V-Ray 3.3 is different with previous version. From scientific point of view, you can not compare the two cause you have two variables here. Unless you can define degree of noisy by number.