En:Antialiasing
Aus CINEMA 4D Wiki
Diesen Artikel auf Deutsch lesen
Inhaltsverzeichnis |
Overview
Antialiasing techniques are used to dissolve the grid of the ray tracer output. In order to achieve this goal, the ray tracer sends more search beams into the scene, than the output image has pixels. The return values within a pixel are averaged afterwards. This measure leads to an apparently increased resolution. Disturbing staggering at diagonal structures, object edge flickering and texture flickering in animations can be suppressed to a large extent. 2x2 antialiasing means that 2x2 = 4 search beams will be used within one pixel.
C4D offers an adaptive antialiasing. This means that the renderer can adapt the expenditure to the necessity within the picture. Compared to a non-adaptive antialiasing implementation this approach has the capability to save a lot of render time.
Two different principles of antialiasing are available with Cinema 4D:
Best mode
This is the Cinema 4D version of the “classical” antialiasing procedure, which was extended by an adaptive algorithm. According to the users parameter defaults the scene is scanned by means of search beams, whose horizontal and vertical distribution is denser, as the pictures resolution.
- Threshold
- While the antialiasing process runs within one pixel, a constant comparison of the interim result with the final results of the neighbour pixels takes place. If the colour value of the actual calculated pixel is similar enough to the colour values of its neighbours, the antialiasing process will be terminated. It is simply supposed that the pixels are in fact quite similar in colour. In most cases this is a good decision - sometimes not. If e.g. the user adjusts the threshold value to 5% the antialiasing process will be terminated if the colour difference between the actual calculated pixel and its neighbours is less than 5%. At least the number of sub pixels entered as the Min Level value will be computed. If no sufficient resemblance is achieved, the Max Level input value will be used as the final termination criterion. This method delivers a considerably faster computation of poor in detail areas.
- Min level
- Here we have the lower limit for the over sampling. 2x2 for example means that 4 sub pixel per pixel are computed at least. No other antialiasing parameter could overrule this value. The calculated colour value for the pixel is the average value of this 4 sensing.
- Max level
- This is the maximum value for the over sampling. 4x4 for example means that the algorithm will execute a maximum of 16 scans per pixel - even if the antialiasing result is not satisfactory.
Geometry mode
This type of antialiasing has been implemented for cleaning the contours of objects from disturbing staggering. The ray-tracing program scans the scene systematically from the left to the right and from above down. If it founds a change of objects between two search beams, a contour has been discovered obviously. This situation contains a high risk of disturbing effects. The algorithm examines the transition between both objects more exactly. It tries to investigate which portion of the pixel belongs to the first object and which portion already belongs to the second object. To achieve this aim it has to send out some search rays near the transition area between both objects. As soon as the position of the crossing within the pixel was determined exactly, the calculation of the proportionately weighted colour average for this pixel will be possible.
The geometry antialiasing is switched on compelling with any other antialiasing setting. Even if the switches are set to best with a Min Level and Max Level of 1 and a Threshold of 100% the geometry antialiasing becomes fully active, although this antialiasing setting is ineffective in every other respect.
Reflection and transparency
The geometry antialiasing principle is capable to recognize directly visible contouring. It is unable to find object transitions behind transparencies or through a mirror. The only way to antialiase these regions within the whole output, is the best mode process. If you want to apply the antialiasing to a limited area, e.g. a transparent object, you could do this by means of a compositing (render) tag. It is important to understand, that the render tag must be applied to the transparency or the reflective object - not to the object, which should be antialiased. This contouring are not treated any more by the geometry antialiasing algorithm described on top, but by the adaptive Best mode antialiasing process and will be handled like a texture on the mirror or on the transparency.
Render tag
Cinema-4D enables the user to adjust antialiasing settings object by object. Experienced users receive the possibility to adjust and optimize the antialiasing expenditure in a wide range and reduce the render time to a minimum. The accompanying settings are entered in the Render tag which must be associated to the suitable object. In addition the antialiasing mode must be switched to best. The Min- an Max level values are used to adjust the antialiasing quality of objects which do not wear a render tag by their own.
Attention: If an object can only be seen through a transparent or reflecting object, the antialiasing level (render tag) must be assigned to the transparent or reflecting object, not to the object which should be antialiased. You have to assign the anti-alias-level to the object which will be hit first by the search beam on its way from the camera through your scene.
Texture antialiasing
Textures can increase the degree of detail within a scene extremely. This can lead to a situation that the ray tracer has to process a complex pattern within the area of one pixel. It is the antialiasers task to form a sufficient exact average colour value from the pixels area - in this case the complex pattern. If this task cannot perform successful, the pixel will deviate from the expected colour structure in a disturbing way. for this kind of errors the coincidental situation of the pixel in the area and/or zoom shot and aspect angle of the camera are determining factors. The deviation would lead to speckled surfaces at still renders and flickering patterns within animations.
The so far delineated antialiasing principles are using a high number of search beams within one pixel to get an exact averaging of the area. These approaches take long computing time in the case of complex patterns. In some aspects this may lead to a bad relationship between the expenditure/time requirement and the antialiasers simple task, to dye only one pixel.
Mip Mapping & CO - Antialiasing in advance
The use of low dissolved textures for the antialiasing process offers one way out of this dilemma. Low dissolved variants of the existing textures have to be generated in advance. The ray tracer uses these low-dissolved variants only if the situation requires sending more than one search beam into one pixel - the antialiasing situation. In this case the ray tracer precepts a blurred variation of the original texture, which comes very close to the needed average value.
Among others, the well-known mip mapping is one of the variants of this principle, which had been implemented into Cinema 4D. The dedicated settings are put on in the Material Editor. In order to find it, please load or create a material, choose a material channel (e.g. colour) in that material, load or produce a texture and check the button sampling below.
Antialiasing and render time
Within Cinema 4D a multiplicity of iterative processes can be activated (representing material channels, shadows, global illumination et cetera). The antialiasing could be interpreted as the top layer of these processes. Aligned to certain optical phenomena, each process delivers its own portion to the pixels colour. Finally the antialiasing algorithm receives the outputs of the more deeply computing processes and decides - adaptive with respect to the interim result - if it is necessary to charge the deeper processes with calculating more sub pixels to get a sufficient colour result for the entire pixel.
The global illumination process (GI, scattered light computation) for example sends a certain number of search beams from the object into the scene. If the GI selector is adjusted to stochastic GI mode with 32 search beams, these 32 beams will be sent on every call of the process (optimisations discounted). If the antialiasing is switched off thereby (none), these beams are sent once per pixel. If it is active, e.g. at (best) 4x4 antialiasing, the GI-process will be called 16 times per pixel, which would result in 16 x 32 = 512 GI search beams per pixel. We can notice that the increased antialiasing attitude will not only reduce the lack of representation attributed to aliasing, but also improve the output quality of the global illumination task.
The goal of the parameter attitude is the economic employment of the render task, in order to obtain the expected expenditure quality as fast as possible. This includes some risk of negative effects that will affect to high time losses which could not be realized easily by less experienced users. If, for example, someone adjusts the GI parameters at an early stage of the project (antialiasing off) to the final quality this could produce a strongly exaggerated GI quality at the final rendering (at e.g. 8x8 or a 16x16 antialiasing) and lead to a substantial loss of time on complex renders or animations.
Tips
Motion blur
If you intend to apply motion blur to your scene, the antialiasing attitudes can often be reduced considerable. Of course the render time will downsize also.
Easy antialiasing setting for beginners
In practice Cinema 4D produces very good results at small AA-parameters (e.g. min value = 1x1 and max value = 2x2). Higher values would rarely lead to visibly better results, but to longer render time. However, in some cases extreme attitudes could be necessary. Experienced Cinema 4D users may be able to juggle some parameters in order to push the render time down or to defer the problem to the post work. Some beginners may be grateful for a method, which will supply the expected quality in an acceptable time, also if the approach may be not optimal with respect to the efficiency.
Switch the antialiasing to “best” mode. Min Level should be 1x1, Max Level 16x16. Both values are not changed thereafter. You are able now to use the full quality range of the cinema 4D antialiasing by simply adjusting the threshold parameter only. You will not have any quality problem with transparencies or mirrors.
Attitudes as examples:
- 50% for fast tests
- 5% delivers very good quality
- 0% delivers maximum quality under all conditions
