Using fixed-rate sampling
Video: Using fixed-rate samplingAs V-Ray is a ray-trace rendering engine, and because image sampling we live the heart of the ray-trace rendering process, we can see perhaps why choosing which of the V-Ray's three image sampling engines to use in our scenes those become a fairly critical decision that needs to be made wisely. For this reason then, over the next three videos we're going to give you a quick overview of the controls for each engine type, as well as reviewing how using these controls will affect both the speed and quality of our rendered images.
- What's next?
Viewers: in countries Watching now:
Create highly realistic 3D architectural drawings with V-Ray, a popular third-party renderer for SketchUp. This course shows how to take a single scene with interior/exterior elements and add lights, move cameras, and enhance objects with translucent and reflective surfaces. Author Brian Bradley explains concepts like irradiance mapping, perspective correction, and fixed rate sampling, while showing how to leverage each of the V-Ray tools and its material and lighting types to achieve specific effects.
- Installing V-Ray
- Creating natural daylight with V-Ray Sun and Sky
- Bouncing light around with irradiance mapping and light caches
- Setting up a depth-of-field effect
- Creating diffuse and reflective surfaces
- Working with the Adaptive DMC engine
- Manipulating color mapping
- Adding caustic lighting and occlusion effects
Using fixed-rate sampling
As V-Ray is a ray-trace rendering engine, and because image sampling we live the heart of the ray-trace rendering process, we can see perhaps why choosing which of the V-Ray's three image sampling engines to use in our scenes those become a fairly critical decision that needs to be made wisely. For this reason then, over the next three videos we're going to give you a quick overview of the controls for each engine type, as well as reviewing how using these controls will affect both the speed and quality of our rendered images.
If you're unfamiliar with just what the image sampling process is all about, you can check out the Introduction to Image Sampling video in chapter 6 of my V-Ray 2.0 for 3ds Max Essential Training course, here on lynda.com. To get started, let's open up our Options dialog for ourselves. You'll notice in here we have an Image Sample rollout and if we use these dropdown, you can see the options available to us when it comes to choosing an Image Sampling Engine in V-Ray. As the first option is the fixed-rate engine, this is the one that we will be looking at in this particular video.
Now I do want to draw your attention to the fact that in order to focus on just what the image sampling engines are doing, we've turned off our anti-aliasing filters. Now this, generally speaking, is not something you will want to do as you're rendering. Filtering contributes an awful lot to the final look of our images. But because we don't want any extra contribution from our filters and because they do add to our render times, we have just disabled this option for now. Just to note also that as we examine our image sampling engines throughout this chapter, we're going to be making use off pre-rendered images to view the changes that our parameter changes would make in our rendered images.
This is because, firstly, as we're going to be increasing the quality settings inside of each of our image sampling engines, that means that our test renders will get longer and longer. So rather then wait around for those, we decided that working with some pre-rendered images would be beneficial there. It also means that we can quickly swap between our renders and just test; we can just check the changes that are occurring as we make changes to our Image Sampling controls. On to our fixed-rate sampling engine then. Well, the Fixed-Rate Sampler performs its task in a very straightforward manner.
It uses, as you would perhaps have guessed from the name, a fixed number of rays, or samples, per pixel to gather the information that it needs from our 3d environment. With this subdivs value set to 1, as it is here, a single sample is taken or cast from the center of each pixel required for our output resolution. Whatever objects in the scene that Ray encounters or hits, well, that will determine the Color value for that particular pixel. And if we just jump over into Adobe Photoshop, which we'll be using as our image viewer, we can see the render that we'd get from this setting.
Now as you can see, what we get is not incredibly impressive, in terms of finished quality, which is not surprising, seeing as we're using the very lowest value available inside of the Fixed Rate Sampler. But what we do see is that we can very quickly make an evaluation of many aspects of our scene at this moment in time. We can easily gauge the level and quality of illumination and global illumination that we getting from our lighting. We can easily gauge camera composition. We can tell whether or not that is working for ourselves. And to some extent, we can decide whether or not our materials are going in the right direction.
Of course, we do have a lot that is wrong with this image. We can see we have a huge noise problem that is occurring here, which to be honest, is actually our own fault. We have used the DMC GI solution for this particular interior render. We've chosen that deliberately because it does add an awful lot of noise into the scene, and so we want put our image sampling engines through their paces and just see how they handle cleaning up this noise, as well handling the materials and the lighting. Of course, the noise isn't the only problem.
If I just used the Ctrl+Plus keyboard shortcuts to just zoom in a little on our image, you can see that a lot of the materials are not really resolving too well either. Our lines sphere here really, we're not getting nice clean straight lines as they should be. We are getting a lot of breakup in there. You can see that our reflections too are suffering from noise problems; they are very bitty, very broken up. So clearly there is a lot of room for improvement in here, but the interesting thing--if I just use the spacebar to just pan down to the bottom left-hand corner, you can see we've a timestamp here--the interesting thing is that we're only at a minute and 23 seconds.
So from the point of view of just being able to gauge some of the aspects of our scene, just being able to figure out what is going on, the feedback that we're getting and the render times, very acceptable. Of course, we do want to improve the quality of our render, so again, let's jump back into SketchUp and see what happens as we increase our Subdivs Value. Let's double this up to a value of 2. This means that we're now getting four samples evenly distributed across each off the pixels in our render.
These four samples will be averaged together by the rendering engine to produce a final color value for that pixel. Now again, let's jump back into Photoshop, just to see what a difference that has made. So we go from a Subdiv of 1 to a Subdiv of 2 and as you can see, we do start to clean up some off the problems that we have in our original render. Particularly if we take a look at the lines on our sample sphere here, you can see that they start to resolve very nicely. We start to get something that looks much cleaner.
In fact, again, if we use Ctrl+Plus just to zoom in, you can see how they are starting to clean up and straighten out quite a bit. We are also cleaning up some of the noise on our walls. You can see very, very grainy there, and obviously, although not a finished quality again, we're seeing that things are cleaning up nicely. You'll also notice, if you take a look at our noisy sample materials--so we have a couple of materials that are very noisy, very bitty indeed-- if you watch those, you'll see that we actually start to find more detail inside each of those materials.
Now this is because with very, very low sampling settings, we're actually skipping over information in the scene. The render engine just doesn't have enough samples to be able to capture the fine detail. But now that we're increasing the sampling, we're starting to see things clean up very nicely. Obviously, we'll be interested in what has happened to our render times now. So if you remember, if we just go and check, you can see that we were at a value of 1 minute and 23 seconds for our single subdivision. And now what we're finding is that we've actually gone up to just over two minutes, which of course is not a tremendous increase at all.
What would happen then if we continue to increase our Subdivs value inside of the Fixed Rate Sampler? If we were to go from a value of 2 to a value of 8 inside of that control, we would go from this render to this, which, as you can see, does start to clean up some areas of our scene very, very nicely indeed, particularly if you keep an eye on the reflections in our sample sphere here. You can see that they go from very noisy, very grainy, to very nice and quite smooth indeed. In fact, again, the same is true for many other aspects of the scene.
We notice things cleaning up as we increase our Subdivs value. Again, the noise, the fine detail in our noisy materials, you can see, we start to capture even more and more detail. And of course, the noise on our wall does clean up a little more, although still a problem there. You can see we're increasing like improving that situation. Of course that does come at a cost. If we again just pan down to the bottom of our render, you see we're now up to 8 1/2 minutes, which in terms of percentage, is quite an increase.
But still, given the quality that we're getting, that probably is quite an acceptable render time, given the fact we're working at 1280x720 in terms of resolution here. Let's take one final jump from a value of 8 Subdivs up to 16, and we would go from this to this, which as you can see, starts to clean up even some of the real problem areas in the scene, such as the noise that we're getting on our wall. Of course, many of our materials don't improve a tremendous amount, except if you see this very, very finely detailed sphere down in the bottom right, if you keep an eye on that, you can see that those clean up very nicely indeed, in terms of really pulling out the detail that is in the material.
Of course again, the increase in quality comes at a high cost, because we're all the way up to 32 minutes and 19 seconds now. But again, given that we've deliberately set the scene up so that it has lots and lots of noise that would need to be cleaned up, that still is a fairly respectable render time. Now you may wonder why our render times increase so dramatically as we increase our Subdivs value. This highlights the weakness of the Fixed Rate Engine; you see, there is no adaptively whatsoever in this system.
V-Ray will always use this fixed value of samples or rays per pixel whether our scene, whether our image, requires them or not. This of course leads to long render times and still oftentimes can leave those with an unsatisfactory level of quality in our renders. You can see even in this image, we still have noise very obviously present on that back wall. For these reasons then, the Fixed Rate Engine is probably best suited to being used for quick and easy test renders. It is very, very simple to use.
All we have to do is choose the engine type enter a parameter into the Subdivs value, and away we go. There is nothing else that we need to worry about or tweak at all. So, having looked at the Fixed Rate Engine, let's move on in our next video to examining a system that does have adaptivity built into it, and this is the Adaptive DMC Engine.
There are currently no FAQs about SketchUp Rendering Using V-Ray.