Rendering using VRay has its secrets, as any high end industry software has. Learn the basics on how to improve your final image by understanding light samples.
There is no other renderer that matches it´s quality when in the hands of a good professional, but its complexity on the almost infinite panels and switches makes big trouble to newcomers. I´m still trying to learn VRay, watched many videos, took many classes, but each project has its ow settings and behavior.
The Basics of Light Samples
Let´s take a look on the most basic settings to make rendering faster, sharper and with better quality by understanding how light samples influences the rendered image and CPU consumption. It´s a game where you need to tweak little bits of information to make huge changes.
This being said, let´s take a look at some test renders, with the time render stamped on the images.
1 Rectangular Light
For this example, I´m using only one Rectangular Light stretched on it´s V value (Y axis) and using a multiplier of 15. This multiplier is the intensity of the light, where 0 (zero) stands for off (no light emission) and 30 is the default value. As you stretch and make bigger your Rectangular Light, it will emit more light, therefore I had to decrease its multiplier.
Conclusions: As we can see, 8 samples is the default value and provides a poor quality light (and shadow) to our model. It´s used mostly in the early stages of the scene composition, and for real time rendering visualization (such as on RT Vray) on slower machines since it does not require much of CPU processing. The rendering itself took 14.2s. With almost no change to the rendering speed it´s perfectly possible to level up your light quality by increasing your Light Samples. Check the second image, I´m using 64 samples and the light and shadow quality are much higher without increasing significant CPU consumption, the render time was 14.6s. Almost the same time with two very different results!
Now, for the final render, to achieve better quality in terms of Light, I increased he Light Samples from 64 to 128 samples, and this time the light incidence and shadow required a lot of CPU, therefore increasing it´s render time from 14 to 36 seconds (2,5 x higher). The quality is better, sure, but really didn’t make a big change for this situation. You need to try out on your own scene.
2 Rectangular Lights
This time I added another Rectangular Light, positioned at the other side of the sphere and let´s check these results.
Conclusions: First of all, it´s incredible that by adding 2 lights in the scene at 8 samples each, instead of rendering slower, it´s faster! The first image shows us almost 2 seconds faster than using only 1 light. This is because it has less shadows to calculate at this low sample rate. The precision is not taken into consideration. On the other hand, this scenario begins to change drastically when increasing the Light Samples, where precision starts to take place.
The second image shows us a fairly well balanced rendered image, taking 21 seconds to render. We can see that the shadows are merging in a smooth way, and we have almost no noise present.
The third image shows us a very slow render (more than 1 minute) when using 128 samples on both lights. The shadow quality is perfect, there´s no apparent noise, just a very subtle noise pattern that would have been cleaned out if I were using a better image sampling rate (will be discussed on another Study Post, here on this Study book). But for this example, we could use only 64 samples, it looks good enough and the rendering time is ok.
Light Samples and GI (Global Illumination) Multipliers
This time I added GI (global illumination) for these render studies. I´m using just plain white background and white GI incidence. Because GI will washout the image I´m studying the multiplier effect. Let´s check the results.
Conclusions: Regarding render time, there wasn’t really big changes, all three images took almost the same time to render (1 minute and 13 seconds). What I want to make evident here is that the GI multiplier (for instance the Irradiance Map and Light Cache) affects the overall appearance of the final rendered image. The higher the multiplier (default is 1) the more the GI will influence the light incidence and shadows. According to your background and GI color (or maps) it will make brighter your final image.
Comparison with/without GI
Let´s put together, side by side, both images: with and without GI and compare the results and the rendering time.
On the left the image was rendered using 128 light samples (both lights) and GI at 0.4 multiplier.
On the right, the image was rendered using 64 light samples (both lights).
Conclusions: The result looks a little bit smoother on the left image, but at cost of significant render time. Both images looks similar, and if not put together, hardly anyone would notice difference, and most probably it´s very acceptable to speed up the renderings throwing away GI and balancing the Light Samples to a good relation between speed cost and quality benefits.