Example 1: What is antialiasing?

Example 2: Image samplers comparison

Example 3: Another image samplers comparison

Example 4: Yet another image samplers comparison

Example 5: Texture antialiasing

Example 6: G-Buffer based antialiasing

Example 7: Randomized antialiasing

Example 8: Antialiasing filters

Example 9: Antialiasing filters and moire effects

 

Example 1: What is antialiasing?

The following example shows the basic difference between an image with antialiasing, and one without:

No antialiasing (Fixed rate sampler, 1 subdiv)
Antialiasing on (Adaptive subdivision sampler, rate -1/2)

The left images are jagged around the edges of the sphere, while the right are smooth. Here are close-ups of the two images:

No antialiasing (Fixed rate sampler, 1 subdiv)
Antialiasing on (Adaptive subdivision sampler, rate -1/2)

Example 2: Image samplers comparison

Here are some examples demonstrating the quality vs speed of the image samplers. All the samplers were set to produce approximately the same image quality.

The first example is a normal smooth image (no blurry effects), click the images for a larger view:

Fixed rate (4 subdivs)
Adaptive DMC(subdivs 1/4)
Adaptive subdivision (rate -1/2)

Both the Adaptive DMC and the adaptive subdivision samplers are substantially faster than the fixed rate sampler.

Example3: Another image samplers comparison

Note: the Sponza Atrium model is created by Marko Dabrovic (http://www.rna.hr) and is one of the models for the CGTechniques Radiosity competition. The Athene model is a free model from the DeEspona Infografica model bank.

 

Here is a more complex example with some fine textures (lots of fine bump mapping) and an area light; this example uses a precalculated irradiance map - the render times are for the final rendering only. Click the images for a larger view.

Fixed rate (4 subdivs)
Adaptive DMC (subdivs 1/4)
Adaptive subdivision (rate 0/2, threshold 0.05)

 

In this case, the Adaptive DMC sampler performed best, and the adaptive subdivision - worst. Why is that? Here is the non-antialiased image (click it for full view), to give an idea of what the image samplers had to deal with.

 

Some parts of the image are quite "noisy" because of the fine bump map. Lots of image samples were required to smooth this out. Furthermore, each image sample was quite costly to compute - there is an irradiance map and an area light, which (especially the area light) need a lot of computations. With the fixed and Adaptive DMC samplers, V-Ray knows in advance how many image samples will be taken for a pixel; therefore it can optimize the computation of some values (the area light for example) so that the final image result is similar, while actually those values are computed with lower accuracy (i.e. tracing fewer shadow rays) for the individial image samples. This can not be done for the adaptive subdivision sampler - it does not know in advance how many samples will be computed for a pixel, therefore it needs to maintain a constant (high) accuracy. Constant accuracy is also required in order for the sampler to adapt correctly to the image. This is why, in this example, the adaptive sampler performed worse than the other two methods.

Example 4: Yet another image samplers comparison

Note: the dragon model is from one of the example scene files of 3ds Max 4

 

The third example is an image with direct GI and motion blur (click the images for full view):

Fixed rate (4 subdivs)
Adaptive DMC (subdivs 1/4)
Adaptive subdivision (rate 0/2, threshold 0.1)

 

In this case the fixed rate sampler was the fastest and the adaptive subdivision sampler - slowest of all (admittedly, the image computed with adaptive subivision is very smooth). This is because the cost of supersampling a pixel for the Adaptive DMC and the Adaptive Subdivision samplers becomes too great.

 

Here is the scene without motion blur, with irradiance map and with the Adaptive subdivision sampler (render time includes GI calculations):

Example 5: Texture antialiasing

This example deals with texture antialiasing and the effect of the Threshold for the Adaptive DMC and Adaptive subdivision samplers.

 

By default V-Ray antialiases everything in the image, including textures. This is especially useful for textures with small details or noisy bump maps as seen in Example 3. The Threshold parameter controls the extent to which texture antialiasing is performed. The effect of this parameter is most noticeable with the Adaptive subdivison sampler and low min rates. For the four images below, min/max rate of -3/2 was used:

Threshold 10.0
Threshold 5.0

 

Threshold 1.0
Threshold 0.1

 

Note that the edges of the object in those images is always sharp. This is because the Object outline option is turned on - see Example 6.

 

If you set the Threshold to a high value, you are effectively telling V-Ray not to antialias textures. You can use this fact to speed up the rendering of complex materials. Note however, that this will disable antialiasing of V-Ray shadows, reflections etc as well.

Example 6: G-Buffer based antialiasing

In Example 5, the edges of the object are always sharp, regardless of the value of the Threshold. This is because the Object outline option is turned on. Here are the first and the last of the images from the previous example, rendered with Object outline off:

 

Threshold 10.0, Object outline off
Threshold 0.1, Object outline off

 

Now the antialiasing of edges depends only on the Threshold. By default, the Object outline option is on, meaning that the outlines of objects are always antialiased. If there are many small objects in the scene, this may slow the rendering. In that case, it's better to turn this off and use only the Threshold to control image quality.

 

On the other hand, if there are lots of fine textures in the image, which you don't want supersampled, you can simply turn up the Threshold. In order to still keep object edges sharp, you'll need the Object outline option.

 

The Normals option allows you to always antialias internal object edges, in addition to the object outline, as shown in the example below (min/max rate -3/2, Threshold 10.0, Object outline on):

 

Normals off
Normals on

Example 7: Randomized antialiasing

Normally V-Ray places the image sample in a strict grid-like pattern. This may cause unwanted banding of edges that are nearly horizontal or nearly vertical. You can use the Jitter option to avoid that. Here is a comparison of an image rendered with and without the Jitter option:

Jitter off
Jitter on

Here are close-ups of the two images:

Jitter off
Jitter on

The Jitter option can be very useful for images with long thin lines as well.

Example 8: Antialiasing filters

Here is an example briefly demonstrating the effect of different antialiasing filters on the final result.

 

Note that rendering with a paricular filter is not the same as rendering without a filter and then blurring the image in a post-processing program like Adobe Photoshop. Filters are applied on a sub-pixel level, over the individual sub-pixel samples. Therefore, applying the filter at render time produces a much more accurate and subtle result than applying it as a post effect.

 

The Adaptive image sampler was used for the images below, with Min/Max rate of -1/3 and the Jitter option on.

 

Filter Image Zoomed-in image Comments
Filtering is off Applies an internal 1x1 pixel box filter
Area filter, size 1.5 (default setting) Slightly blurrs the image, visually more pleasing than the box filter.
Area filter, size 4.0 More blurring
Catmull-Rom Edge-enhancing filter, often used for architectural visualizations. Note that edge enhancing can produce "moire" effects on detailed geometry.
Mitchell-Netravali Allows control between edge-enhancement and blurring
MItchell-Netravali, ringing=1.5 Strong edge-enhancement.
Mitchell-Netravali, ringing=2.0 Even more edge enhancement; kind of cartoon-style effect

Example 9: Antialiasing filters and moire effects

This example demonstrates the effect antialiasing filters have on moire effects in your images. Sharpening filters (Mitchell-Netralavli, Catmull-Rom) may enhance moire effects, even if your image sampling rate is very high. Blurring filters (Area, Quadratic, Cubic) reduce moire effects.

 

Note that moire effects are not necessarily a result of poor image sampling. In general, moire effects appear simply because the image is discretized into square pixels. As such, they are inherent to digital images. The effect can be reduced through the usage of different antialiasing filters, but is not completely avoidable.

 

The scene is very simple: a sphere with a very fine checker map applied, texture filtering is off. The images were rendered with a very high sampling rate (15 subdivs, or 225 rays/pixel). This is enough to produce quite an accurate approximation to the pixel values. Note that the image looks quite different depending on the filter:

 

No filter

Area filter, size=1.5

Area filter, size=4.0

Catmull-Rom

Mitchell-Netravali, blur=0.333, ringing=0.333