Join Brian Bradley for an in-depth discussion in this video Using the fixed-rate engine, part of V-Ray 3.0 for 3ds Max Essential Training.
- Having already taken a somewhat brief look at image sampling theory in chapter one of our course, what we want to do here is move on to an examination of V-Ray's image sampling engines themselves, taking a look at the controls and parameters that each of them make available to us. The scene that we will be working with consists of a fairly innocuous looking setup, that does neverthless present a number of fairly typical rendering challenges to our image sampling engines. We have plenty of straight lines in the scene, both in terms of materials and on the geometry itself.
These of course can be the bane of pixel-based graphic applications. We have both glossy and blurred reflections on the materials as well as noisy surface patterns and flat, solid colors. We're making use of the V-Ray dome light in the scene using the default shadow "Subdiv" setting of eight, which can of course throw a lot of shadow noise into an environment, with the potential for noise having been heightened by the fact that we are also using a high dynamic range image in the dome light. Indeed, if we go ahead and load the first of our saved images from the history list, we can see that using the current settings, we are getting a pretty rough-looking image from V-Ray.
Middle mouse zooming to 100% shows that we have lots of missing texture information, we have jagged edges on the geometry, and at 200%, we also see lots of small noise in shadowed areas, particularly so on solid or flat color surfaces. Now you may recall, also back in chapter one, that we learned how ray trace renderers such as V-Ray, use rays and samples as a means of gathering information from our 3D environments. The end result of that gathering process being the computing of final pixel color values.
The really cool thing about V-Ray is that it actually gives us a choice of four different sampling engines that can be used in order to handle this information gathering part of the rendering process. Well, strictly speaking, we actually get two distinct engine types because the fixed, adaptive, and progressive options available, are all simply variations of the same basic technology. Because they do all have independent controls however, and for the sake of clarity in this chapter, we will refer to them as separate and distinct sampling engines.
Indeed, if I just open up the "Render Setup" dialogue by means of the F10 key and jump into the "V-Ray" tab, we can see inside the "Image sampler" roll out, the four choices available in the "Type" drop down. In this particular video, we're going to focus on making use of the "Fixed" or fixed rate sampling engine. Now the fixed rate sampler does its job in a very straightforward way, using, as you would, perhaps have guessed, a user-specified or fixed number of rays per pixel in order to gather information.
With a "Subdiv" value of 1 set, as we have it here, V-Ray will shoot a single ray from the center of each pixel, specified in our rendering resolution. This means that in order to complete the 1200 by 1500 image that we have here, V-Ray's had to shoot and trace from the rendering camera, 1.8 million rays. Now whilst this may sound like a lot, in terms of ray trace rendering, it actually isn't, as testified to by the poor image quality that we are seeing.
Whilst a fixed "Subdiv" value of 1 may be fine then if we are rendering a quick test image, it clearly isn't enough to give us a final, high-quality render. To get more out of this particular sampling engine, we clearly need to increase the quality settings. Let's set the subdivision value of 8, so quite a significant jump in terms of the number of samples being used, adding up to a total of 150,200,000 all told. If we go ahead and load our second render, though, we can see just how big a difference those extra samples start to make, with things very obviously starting to clean up nicely.
At 100% zoom, we are now picking up the fine detail and our diffuse textures, our shadow noise is cleaning up nicely. Even the bump in the wood material has cleaned up quite a bit. The problem, of course, is that we have made a really significant jump in terms of render times. So whereas our initial render came in at just over one minute and 30 seconds, we have now jumped way up to just over 60 minutes, and we still, of course, have a way to go before we could say that our image has cleaned up enough to be considered acceptable as a final deliverable.
In order to do that, of course, the only thing we can really do here is continue to brute force our way through the cleanup process, and once again increase the "Subdiv" value. This time, jumping all the way up to 16 subdivisions, meaning we are now shooting a total of 460,800,000 rays into the scene. Which will, as you can imagine, take quite some time to calculate on just a single machine. What we do get from the render though, if I just load our third and final image into the frame buffer, is in most respects, a pretty much a noise-free image.
No noise to speak of on our solid color materials, shadowed areas and geometry edges look nice, even the detailed material such as the counter top are looking good. In fact, the only real downside to what we have here, is the fact that our render times have gone through the roof, the image now taking over an hour and two minutes to complete. The problem you see with the fixed rate engine is that is has no adaptivity whatsoever built into the system. V-Ray will use the fixed number of samples per pixel that we specify whether all areas of the final image require them or not.
This of course, can lead to some very long render times, as you have seen. For this reason then, the fixed rate engine, typically speaking, is probably best used for taking quick and easy to settle test renders. Indeed we saw right at the start of this exercise, how we got a reasonable amount of visual feedback from a low setting of just one subdiv. At the same time, getting pretty fast render returned to us of course. Well, let's move on to our next exercise and take a look at a system that does have adaptivity built into it, this being the very aptly named, adaptive sampling engine.
- Using the new UI elements, Quick Settings, and revamped Frame Buffer
- Understanding color mapping modes
- Adding V-Ray light types
- Working with the V-Ray Sun and Sky systems and dome light
- Using irradiance mapping and light cache
- Working with diffuse color maps
- Making reflective materials
- Creating a translucency effect
- Using the new SSS and skin shaders
- Ensuring quality with image sampling
- Working with the adaptive subdivision engine
- Controlling the physical camera
- Working with FX tools such as VRayFur and VRayMetaball
- Stereoscopic 3D rendering
- Using Render Mask