In this video I gonna show you the main differences between Vray, Corona and specially Octane Render

which is the renderer that I actually using more in my pipeline

I´ll also talk about cpu x gpu rendering

Biased and Unbiased renderers

To maybe answer that common questions of...

Which one is the best? Which one is the faster?

And obviously make some advertise about my newnest Octane Render Course

So that´s it, I´ll try to keep it simple and I hope you like it

First of all I´ll talk about the difference between Cpu x Gpu render engines

In therms of processing, on cpu renderers the Processor is the responsible for the rendering process

On Gpu renders we use one or more Video Cards to process the rendering

The first difference between them is the amount of simultaneous threads of each method

For example the Processor has tens or hundreds of cores/threads to make a lot of processes

And Video cards has thousands of cores/threads to make a single process

On Nvidia video cards, these Cores are called CUDA cores

which can be used by the Gpu render engines to make the processing faster

So it means that every Gpu renderer is faster than Cpu renderers?

No! And I´ll talk about it soon.

The second difference between them is the memory usage

For me this is the worst thing about Gpu render engines

In the CPU render engines we use the Ram memory

which in standard personal computers can be expanded to 64gb or more than that if you use high performance workstations

on the other hand, GPU renderers like Octane, uses the Video Card(s) memory / VRAM

So at this moment if you look for the best video card, you can find a TitanX, with 12gb of VRAM

By a huge amount of money (at least here in Brazil)

Is good to know that Octane Render doesn´t take advantage of SLI

It means that if you has TWO video cards, the first one with 3GBs of VRAM and the second one with 2GBs of VRAM

Octane won´t add the both memories, It will just add the Speed of them (Cuda Cores)

always leveling by the Video Card with less VRAM

in other words, even having one 3GBs VRAM video card, you could use just 2GBs of VRAM

If both video cards were activated to render at the same time

This memory issue can limit us a lot

Cause more memory, means you can render more polys in your scene, more high quality textures, bigger the render size

And wider the amount of rendering solutions as fur, displacement, motion blur, etc

For this reason I´ll show in this video soon, how the Octane Render memory usage actually works

And how to optimize your scene in the best way

Now talking about compatibility

The CPU renderers has advantages again

Cause all the kind of Processors in the market are compatible with CPU renderers

On the other side, a lot of GPU renderers just use the Cuda Technology, exclusive to Nvidia Video Cards

So, AMD/ATI cards can´t be used on Octane Render

If you were planning to buy Octane Render and you just have AMD/ATI Video Cards

Is better to wait a little more to eventually Octane get this kind of support, which actually the developers has in their plans

So, balancing these informations I think we can talk about Cost X Benefit of each method.

All the equipments in Pink are the ones which needs to have really high performance, and probably a little bit expensive

In yellow are the moderate in therms of price and quality

And finally in blue, the ones that you don´t need to spend so much money on it.

So in this case for CPU renderers we have the Processor, Motherboard and RAM with high performance

but the Video Card don´t need to be awesome as it is responsable most part of the time just for the viewport processing

On Gpu Renderers, in this case, Octane Render

The processor is mostly used just for load the scene and geometry, but has almost no usage at the time of rendering process

So you can choose if you wanna invest or not in a high performance processor

The Motherboard doesn´t need to be the best one, but as advice try to find some with a lot PCIexpress slots

For you upgrade your system with new video cards in future

About RAM memory you can invest in less amount of GBs at least using Octane Render

cause there is other GPU renderings on the market that uses Ram memory too

But for sure it will be implemented on next Octane Render versions

And last but most important, the Video Card. The most expensive item cause you will really need awesome video card(s)

Cause better the video cards, means faster your renders and more complex your scenes can be, depending of your VRAM amount

So, as a conclusion, for both kind of Rendering methods, CPU or GPU

we can save some money in some itens in relation to others

But balancing all these informations I can say that invest in GPU rendering, normally is a bit more expensive, at least in a first moment

Cause there is something really important that we can´t forget which are the future upgrades on our workstation

that is, we have to think in long-term investments

Using CPU, we could be forced in some cases to buy a new computer, new processor, motherboard, ram, etc.

On the other hand, for GPU renderers we just need to buy a new Video Card and thats it. Our speed and processing has been increased.

It means, in Long-term, probably the investments on GPU rendering turns out cheaper.

The Next thing I wanna show you in a simplified way is the difference between Biased and Unbiased renderers

Starting by the biased ones, basically to be faster they calculate just a few samples/pixels on our scene,

representend by the yellow points on the first image

after that the Renderer use this points as references to estimate the intermediate samples, represented here by the blue points

this approximation calculation is called Interpolation,

that way the renderer saves a lot of time on calculating every sample/pixel in maximum quality

The good points: Interpolation can save a lot of render time being highly customizable

The bad points: The Interpolation method can cause some splotches, flickering on animation or errors on light calculation

as I represented on the third image

The Unbiased renderers, by default, renders all the samples/pixel in a brute force way, it means, without interpolation

The good points: the brute force calculation normally avoid the splotches, light leaks, errors on lighting and flickering on animations.

The bad points: as the renderer calculates every sample/pixel in a brute force method, you can have longer render times in comparison with biased ones.

So, as we can see, biased x unbiased has no relation with cpu x gpu rendering methods

but, the way they calculate the samples of our scenes

Is good to remember that Vray, even being a biased renderer has a lot of ways to work in an unbiased way.

For example using the GI with Brute Force on primary and secondary bounces or in some cases Light cache x Light Cache

and using the progressive method of Anti Aliasing you will have vray simulating an Unbiased Render Engine Calculation

similarly, Corona and Octane render, even being unbiased has a lot of techniques to optimize them to cheat a little

the longer render times common on unbiased engines

So, enough of theory, let´s see some action

I gonna show you a comparison between the lighting process on the 3 main renderers

And the idea is not to just focus on render time, but specially the whole process until the final render

with a good lighting without noise, splotches, etc

For these tests I used the newnest versions of each one: Vray 3.3, Corona 1.3, Octane 2.24.2-2,16 (beta) for 3D Studio Max

But you can find the Octane Render plugin for a huge list of softwares, like: Maya, Cinema 4D, Revit, Modo and a lot more.

Just for your information, the next videos are 3x faster than real speed.

As the 3 renderers has planar lights I lit the scenes equally, using just a planar light on the right window

Leaving the lights and cameras with default values and after that just playing with the camera, lights and render settings

until reach the final render, so this way, every test on each render engine will start at the same point to make it fair

Vray and Corona Renderers were made with my I7 4960X and Octane Render image, with just One GTX 780ti

Let´s see the tests

Starting with Vray, we can see in the first test the light appears overexposed using the default values

So I changed the light intensity and after that I adjusted the Exposure and Contrast on VFB, which sincerely I do not like so much

I setup the render with settings that I know that gonna result in a fast and clean render in the end

It´s important to say that I made all my efforts to optimize the maximum I could each one of the renderers using all my knowledge

But maybe using your own techniques you could have different results in therms of time and quality

Jumping to Corona I just gonna change to UHD Cache mode which is a little bit faster than default hdcache

gonna make 2 more renders just to adjust the light intensity

and finally just playing with the contrast we are good to go

So in this moment is clear how easier is to work with a unbiased render like Corona, without worry about light issues and other artifacts

And finally on Octane, we just need to adjust the realtime exposure, the enviroment color to control the indirect light and that´s it.

Comparing all of them at same time, we have a higher workflow time when using vray

Which is the time needed to configure the whole lighting perfecly to just Render the final image

In other words, we waste more time setting up vray

Corona was the second and Octane the fastest one.

Comparing render time Corona was the slowest

but for sure we could speed up the rendering using for example bucket mode and playing a little more with the settings

and again, Octane render was the fastest.

Then, we can already conclude that Octane is the best one?

In fact, NO!

Octane render had a best behaviour in this specific simple scene, using this specific one light source and specific video card

maybe in more complex situations for example interior scenes with a lot of lights

or more complex landscapes we could have very different results

In these presented situations the time were no more than 2 minutes

but if you expand it to more complex scenes and the time compared in hours

you could have almos 2x time setting up Vray in comparison to Corona, but vray with faster render time

at least in this presented situation

So, the question is: Which one is the best? or Wich one is the best for you?!

There is some people who can´t take longer render times, (this is my case).

And others who don´t wanna waste time learning how to configure a lot render settings

This way we can see that the best one will depends of your workflow, the time you have for each step of your project

And of course, your experience and personal taste

Now making another comparison

In this case I redid all the renders on each Render Engine, trying to achieve the almost exactly same result of light and realism

Here we can see how is possible to achieve similar light results on Octane in the first image, Vray in the second and Corona the last one

So when people ask me, Which one can achieve the most phorealistic results?

The answer is whathever! Cause you can reach realism in any of them, and a lot of other renderers

however each renderer has its particularities and its own ways to reach the photorealism

Then, again. You need to know what render engine fits better on your pipeline and workflow

At this moment probably a lot of you are already looking at these 3 images looking for errors,

expecting to show to your friends how your favorite render engine is the best one and crushed the others

So that´s why, in this case I LIED!

In fact this is the correct renderers order

The first image was rendered in Corona, the second one in Octane and the last one Vray

I lied intentionally to you detach from this kind of little stuff, which in the end are almost unnoticeable

specialy if we aren´t seeying the images side by side to the other.

Then I´lll keep the mistery.

Right now, Did I told the truth about the order?

Well, I´ll let you sleep with this question which is so much funnier

I´ll start to talk about Octane Render into details

About the positive points and specially negative points

Cause I´ll not here to talk about good stuff and sell courses

First of all I gonna show you how the Video Card memory (VRAM) usage works.

For this purpose I downloaded the GPU-Z software free software, which shows every detail about your video card(s)

On sensors tab we can see the actual VRAM usage, which in my case its a GTX 780ti with 3GB of VRAM

The most important here is to see what processes our computer is doing right now

At this moment I just have the 3D Studio Max opened with this scene and obviously the windows

which is being calculated as part of the VRAM memory (displays too)

It means that as my Video Card is being used for this processes I´ll not be able to use all the 3GB of VRAM in my render

I´ll have a bit more than 2GBs to use in my scene.

Remember that you need to have enough space of VRAM to calculate, scene geometry, your textures

which in case of Octane it already have this Option, Use Cpu Memory, to calculate your textures with your RAM memory

but the other functions as Displacement, Motion blur stil need to be loaded on your VRAM

I´ll render this scene even without any textures, and this aircraft which is very detailed with screws, etc

uses 185mb of VRAM and this amount of VRAM used will variates a lot depending of the Render Output Size.

Cause bigger the render size, more memory is needed to render out your image

So in this example, with this option activated(lock resolution), I can make Octane adapts the render size to the window size

and also we can see how the memory usage variates according to the window size

then we can think. well, in this scene I could put a just few airplanes in this scene or we could extrapolate the VRAM limit,

Well, so... Octane just work with simple scenes with just a few poligons? Well it depends on your scene

Octane has an option which can help us a lot in some cases

selecting the object and right clicking on it you´ll find the octane properties and inside it we have the Movable Proxy option

when activated it works as a proxy, using the first model as a base and any instance of it will be treated as a reference