Published on the first version of Metinseven.com in January 2005
Next Limit’s Maxwell renderer offers a very interesting approach to the setup and rendering of 3D CG scenes. To approximate reality as close as possible every available Maxwell element carefully imitates its real-life equivalent. In other words: a camera offers shutterspeed and diaphragm settings, light calculations are performed in spectral colorspace in stead of limited RGB space, physically correct outdoor lighting conditions are determined by specifying a desired date, time and location, and so on. The virtual world has never before been this close to the real world.
Maxwell’s true-to-life approach requires a certain level of knowledge about real-life photography, lighting and material properties. This article is an initiative to clarify as much as possible about the world of Maxwell, ranging from basic information about photography and light types to tips and tricks for increasing rendering speed. The knowledge I share in this article is based upon the 3ds Max plug-in implementation of Maxwell, as well as general Maxwell functionality.
Please do not e-mail me with Maxwell questions, thank you. The official Maxwell forum provides support for Maxwell.
Please note that unauthorized publication of this text is not permitted. If you would like to publish this text, a part of it and/or a translation of it elsewhere, then you can contact me to ask for permission, thank you.
Now fasten your seatbelts, here we go …
Theory
The Maxwell camera simulates the most important properties of a real-life physical camera.
Practice
Always use a Maxwell camera for your Maxwell scenes. You can choose between a target-oriented camera or a free camera.
Theory
An fStop (often notated as “f/stop”) is the focal length of a camera lens (F) divided by the diameter of the lens opening (D). The lens opening is also referred to as “aperture” and the focal length is the distance between the camera lens and the camera sensor (you can find both of these elements in Illustration 1).
Illustration 1 - At the left is scenery, in the middle you see the camera lens and the camera sensor with the inversely captured scenery is at the right. The focal distance (or focal length) is the space between the lens and the sensor.
The lens opening diameter is determined by a blades-based mechanism called the diaphragm (discussed later on in this article).
The fStop system was conceived to establish a consistent exposure to light, regardless of the focal length of the lens being used. With this system an fStop of 16 on a 1000 mm lens lens will expose the same amount of light as an fStop of 16 on a 28 mm lens.
Example: if you’ve got a lens with a Focal length of 30 and a lens Diameter of 60, then F/D = 30/60 = an fStop of 0.5.
Practice
Higher fStop values allow less light to enter through the lens. Lower values allow more light to pass through the lens. An fStop that is too low will cause the resulting image to be over-exposed to light, resulting in areas becoming flattened because of too much brightness (see Illustration 2). An fStop that is too high will cause the opposite effect.
Illustration 2 - two quick �n� dirty test renders - the pot on the right has an fStop value that is too low, washing out subtle details due to a light surplus.
Theory
The shutterspeed determines how long the camera sensor is exposed to light. In real-life cameras this is achieved by a mechanical shutter between the lens and the sensor which opens and closes for the amount of time indicated by the shutterspeed. The shutter should not be confused with the diaphragm (see below), as these are separate mechanisms, each with their own function.
A shutter’s speed usually indicates a fraction of a second. Example: a shutterspeed value of 1/100 will expose light to the camera sensor for 1/100th of a second.
Practice
Longer shutterspeeds will expose a scene to the camera sensor for a longer time. This results in blurriness of moving objects and/or blurriness of the whole scene caused by movement of the camera itself. But longer shutterspeeds also make the camera sensor detect more light from a scene and thus you will get better images of dimly lit scenes.
For sharper imagery increase the shutterspeed. A picture taken at a shutterspeed of 1/1000th of a second will enable you to freeze relatively fast motion, while the same motion shot at 1/10th of a second will probably cause a blurry result. You will need more light in a scene that’s shot at a higher shutter speed though, with daylight on a bright sunny day being ideal.
If you follow a moving object with your camera and keep the shutterspeed relatively low, the object you’re following will remain sharp and the background will show motion blur.
Theory
A diaphragm is a dilating mechanism in a camera that exists of a number of blades arranged in a circular fashion. By shifting the blades a diaphragm can widen or narrow a circular opening (aperture) between the camera lens and the camera sensor, resulting in more or less light to reach the sensor. The diaphragm should not be confused with the shutter (see above), as these are separate mechanisms, each with their own function.
A camera’s focal length divided by the diaphragm’s aperture results in an fStop value (see above).
Practice
The blades of a diaphragm can be of a polygonal or circular structure. The difference becomes apparent when there is Depth Of Field (DOF, see further on) in your imagery. When you focus on a centric element in your image, areas that are closer and further away from the focus area get a characteristic out of focus blur. The best-known characteristic of these out-of-focus areas is the shape of the blurred light, also known as bokeh. The bokeh effect is particularly visible in highlight areas.
A diaphragm with six blades results in typical hexagonally shaped bokeh light aberrations. A higher number of blades and a circular diaphragm type will result in a more circular bokeh effect, but will require more rendertime in Maxwell.
Rotating the angle of the blades will result in the polygonal bokeh shapes changing their general orientation. Adjust it as desired in the Maxwell camera.
Theory
When you focus on a centric element in your image, areas that are out of focus typically become unsharp. This phenomenon is called Depth Of Field (DOF).
Practice
The DOF amount depends on the fStop value and the lens diameter, so the focal length dictates the amount of DOF. Decrease the fStop value and/or the lens diameter for a more exaggerated contrast between areas that are in focus (sharp) and areas that are out of focus (blurred).
Use the Maxwell camera target or target distance value to enshrine the DOF’s focal point. As you work in real life dimensions when using Maxwell you can easily determine by the target’s distance from the camera whether you’re photographing at a macro level and adjust the fStop value and/or lens diameter accordingly, in order to control the amount of DOF.
To avoid unwanted DOF effects make sure your scene’s dimensions are accurate by adjusting the Scene Scale value in the Maxwell renderer dialog or by scaling the scene to correct dimensions.
If DOF-affected objects are positioned in front of a bright light source, the DOF blur surrounding those objects is likely to become absorbed by the bright light source, resulting in edge sharpness in that area. Avoid this by lowering the light’s strength and compensating for the light loss by lowering the fStop value or the shutterspeed.
Check out this site for help while determining the amount of DOF in your scene: www.dofmaster.com.
Theory
The relationship of fStop versus shutter speed is as follows: halve the shutter speed to get 1 stop increase (for example 1/100 to 1/50), and divide the fStop by 1.4 to get 1 stop increase (for example f5.6 to f4.0). Vice versa, if you multiply the fStop by 1.4 you get 1 stop decrease. So if you want to keep the light strength the same, make sure you compensate for an fStop or shutter speed increase/decrease.
In order to provide some generalized support what fStop usually corresponds to what shutterspeed under common lighting conditions you can find a table below. Please keep in mind that this is only a guideline. Each individual lighting and motion condition requires customized settings.
Practice
|
OUTDOOR LIGHT |
INDOOR LIGHT |
||
| FSTOP | SHUTTERSPEED | FSTOP | SHUTTERSPEED |
| 1.0 | 1/8000 | 1.0 | 1/1000 |
| 1.4 | 1/4000 | 1.4 | 1/500 |
| 2.0 | 1/2000 | 2.0 | 1/250 |
| 2.8 | 1/1000 | 2.8 | 1/125 |
| 4.0 | 1/500 | 4.0 | 1/60 |
| 5.6 | 1/250 | 5.6 | 1/30 |
| 8.0 | 1/125 | 8.0 | 1/15 |
| 11.0 | 1/60 | 11.0 | 1/8 |
| 16.0 | 1/30 | 16.0 | 1/4 |
| 22.0 | 1/15 | 22.0 | 1/2 |
| 32.0 | 1/8 | 32.0 | 1 |
Each incremental fStop value in the fStop-shutterspeed table allows half as much light to enter the camera and each lower shutterspeed value allows double as much light to enter the camera, in order to compensate each other and achieve a stable light brightness.
Theory
Light as we perceive it is part of a broad electromagnetic energy spectrum, expressed in wavelengths. For example blue colors have relatively shorter wavelengths than red colors. That’s why we can distinguish more values of the red color than values of the blue color.
When we perceive a color, we’re under the impression that the surface we’re looking at has that color. In fact, the opposite is true: the surface absorbs all color components of the light spectrum except the color we perceive, which bounces from the surface into our eyes.
Maxwell’s render calculations are performed in spectral color space and subsequently converted to RGB values, for ultimate realism.
The spectral data representing light source properties is referred to as an illuminant, determined by the internationally accepted CIE standard. The Maxwell emitter material allows the user to turn any object into a light source and choose the type of illuminant in the emitter material.
Illuminants should not be confused with the strength of a light source. An illuminant is better described as the indication of a light’s color temperature being emitted from a physical light source.
Practice
Light intensity is expressed in Watt (W). The amount of light energy from a light emitter depends on both the Watt value and the volume of the light source. Studio lighting equipment usually ranges from 1000 Watt to 2000 Watt. With indoor photography you will have to use a lower fStop value and/or lower shutterspeed than with outdoor photography to get the right amount of light into the camera.
The color temperature determines the spectral properties of an illuminant, expressed in degrees Kelvin. The higher the temperature, the more a light tint shifts along the color spectrum, starting with reddish tints at low Kelvin values and gradually moving towards bluish tints as the color temperature increases. See Illustration 3 for a visual survey of Kelvin color temperatures.
As a general rule of thumb you can keep in mind that the numerical value next to an illuminant’s letter often indicates the color temperature of the illuminant in an abbreviated fashion. For example: D50 equals illuminant type D at a color temperature of 5000 degrees Kelvin.
A Maxwell emitter material can turn any scene object into a light source (see the paragraph about the Maxwell Emitter material). When applying a Maxwell light emitter material to a scene object, bear in mind that more polygons in a light emitting object cause the render time to increase. In other words: using a simple box as a light emitter is more advantageous than a sphere with many faces.
Illustration 3 - A table of color temperatures expressed in degrees Kelvin.
Below you can find a brief elucidation of a selection from the available illuminant standards.
Theory
Next to the illuminant types that will be discussed, the “Custom” type from the illuminants rollout acts as a neutral light type next to the predetermined color temperatures of the available illuminant types.
Practice
The Custom light category is the type to choose if you want to see the effect of the adjustable custom color. In other words you can only customize the color (temperature) of your light emitter using the Custom type from the Illuminants rollout. Selecting a preset such as Cool White will always result in the emission of a yellowish / brownish color tint (despite the confusing name “Cool White”), regardless of what color you determine for the color swatch.
Theory
A-type illuminants indicate so-called incandescent light at a relatively low (reddish) temperature range: about 2856 degrees Kelvin.
Incandescent light arises from heated atoms that release some of their thermal vibration as electromagnetic radiation. Incandescent light sources may range from a light bulb to the sun.
Practice
Use the A-type of illuminants for simulating artificial light, like emitted from a light bulb or a candlelight.
Theory
Illuminant B equals direct sunlight at about 4874 degrees Kelvin.
Practice
Use the B-type of illuminant for simulating a bright direct sunlight tint without much interference of atmospheric skylight.
Theory
CIE standard illuminant C was introduced to represent average daylight with a color temperature of 6774 degrees Kelvin.
Practice
Use the C-type of illuminant for simulating very bright light, such as direct sunlight together with a slightly overcast daylight sky.
Theory
The “D” in D-type illuminants can be memorized as Daylight. The value behind the D indicates an abbreviation of its color temperature.
Practice
Use the D-type of illuminant for simulating very bright light, such as direct sunlight along with atmospheric skylight. The influence of the skylight depends on the color temperature of the D variant (see Illustration 3 for color temperature reference).
D50 is the equivalent of reddish tinted sunlight at sunrise or sunset at a color temperature of 5000 degrees Kelvin.
D65 is equal to average noon daylight at a color temperature of 6500 degrees Kelvin. So this light has a slightly warmer tint than the C illuminant.
D75 equals an overcast sky at 7500 degrees Kelvin.
Theory
F-type illuminants indicate Fluorescent light. Fluorescence is conceived by gas and phosphors and has a soft, glowing quality.
Practice
Illuminant F2 equals Maxwell’s “Cool White” preset, at about 4200 degrees Kelvin. In contrast with what “Cool White” suggests it is applicable for a soft, warmly tinted indoor lighting setup. For a more neutral lighting tint choose illuminant type C, D65 or D75.
Theory
Maxwell materials accurately simulate the most important real-life surface properties.
Practice
Use the U Roughness and V Roughness values that are present in several Maxwell material types to simulate a glossy reflection / refraction appearance (blurry reflections / refractions, see Illustration 4).
Illustration 4 - Establish glossy reflections and refractions by adjusting the U and V roughness values.
Use differing U Roughness and V Roughness values to achieve an anisotropic effect (unequal reflection / refraction blurring along different axes). Useful for simulating surfaces with miniscule directional grooves, such as brushed metal.
If present in a Maxwell material, the Scattering section properties determine the way light is absorbed by a surface and reflected from a surface, using the so-called Bi-directional Scattering Distribution Function (BSDF) approach. In practice this is usually referred to as surface shaders, such as Phong, Blinn and Ward.
Theory
A dielectric material is essentially a substance that is a poor conductor of electricity, but an efficient supporter of electrostatic fields.
Practice
Most dielectric materials are solid and a second common characteristic is transparency. Examples include glass and plastics. Some liquids and gases can be good dielectric materials. Dry air is an excellent dielectric, distilled water is a fair dielectric and a vacuum is an exceptionally efficient dielectric.
A higher Absorbance value causes more light to be absorbed by the dielectric volume, resulting in the material becoming darker. Please note that the Absorbance parameter depends on the scale of the scene. For example: if you have an absorbance value of 0.1 for a window that’s 0.01 meters thick, and use the same absorbance value for a window with a thickness of 10 meters (something unrealistic because of a wrong scene scale) then in the last example you will get a very dark window.
A dielectric material’s “Abbe” value indicates its amount of dispersion (also known as diffraction: the prismatic separation of a light’s spectral color components). A material with a high Abbe number means that the different light wavelengths will have nearly the same index of refraction, resulting in less separation between the light’s spectral colors.
The dielectric material’s Nd number indicates the index of refraction (IOR).
Set the Abbe value in a Maxwell dielectric material to a value higher than 150 to decrease dispersion, but also to decrease noise and increase render speed.
For plastic dielectric types a general rule of thumb is that polyurethane is mostly used for more solid, durable plastic applications and polycarbonate is generally used for common household plastic applications such as plastic milk bottles.
Theory
Maxwell’s diffuse material is a general purpose material simulating non-shiny surfaces that catch light in a diffuse manner and optionally absorb a given amount of light. In Maxwell Studio’s material editor this is indicated as Lambertian shading.
Practice
The higher a diffuse color value and saturation, the more light will reflect from the surface, resulting in more color bleeding, but it will also require more render time. In general, try to avoid very high surface color values and saturation. Another advantage of this advice is that less saturated colors increase realism.
Theory
A Maxwell emitter material can turn any scene object into a light source.
There are different types of indicating light energy distribution. W equals power indicated in Watts, W/m^2 indicates watts of power per square meter, intensity is expressed in W/sr, and and W/(m^2*sr) stands for radiance.
See the “Light” section in this article for more theoretical light knowledge.
Practice
When applying a Maxwell light emitter material to a scene object, bear in mind that more polygons in a light emitting object cause the render time to increase. In other words: using a simple box as a light emitter is more advantageous than a sphere with many faces.
Light intensity is usually expressed in Watt (W). When using W the size of your light emitting object does not influence the amount of emitted light energy.
if you work with W/m^2, then the emitter object emits one unit of Watt power per square meter, so if you resize your emitting object, the amount of light it emits into the scene will change.
The W/m^2sr units variation is very suitable for simulating a large distant light emitter, such as the sun.
Studio lighting equipment usually ranges from 1000 Watt to 2000 Watt.
With indoor photography you will have to use a lower fStop value and/or lower shutterspeed than with outdoor photography to get the right amount of light into the camera.
See the “Light” section in this article for more practical light knowledge.
Theory
A Maxwell Plastic material accurately simulates a shiny plastic surface.
Practice
The Plastic material is a general purpose material for creating shiny surfaces. Next to plastic you can also use it to give a varnished look to a wooden floor and many other shiny applications.
Use the Specular color to control the intensity of reflections on your plastic surface.
Theory
Maxwell rendering is both time-based as well as sample-based. You enter a target time and Maxwell will try to achieve a rendering result as good as possible within that time frame. The more time you reserve for a Maxwell render, the less noise the resulting imagery will contain and the better the image quality will be.
Practice
Speedier processors yield better results within a given amount of time than slower processors (CPU speed x Time = Quality).
The time value does not affect the end result at the end of a given time. In other words: setting the render time to 240 minutes yields the same result as setting the render time to 2400 minutes and stopping the render after 240 minutes. Rendering with a time limit is mainly useful when you are rendering an image sequence for animation, otherwise I suggest to increase the render time limit to 1440, which equals 24 hours. You can then always end the render when you’re satisfied with the result.
An effective way of determining the final light exposure in your rendered scene is to check out the render preview that can be found in the Maxwell renderer interface.
Theory
This value indicates a target quality level for the render. Maxwell will try to reach the entered quality level, but whether the renderer achieves that within the given time frame depends on several circumstances in your 3D scene. When a higher quality level is reached this will result in a more complete lighting solution and less noise in the rendered image.
Practice
Speedier processors reach higher sampling values (better quality) within a given amount of time than slower processors.
Lower the sampling value for earlier render termination when this value is reached, but also less quality in the result.
The image file that is being rendered is regularly updated during rendering, showing the result of new subsamples between whole sampling levels. You can terminate a Maxwell render at any desired point in time and can then use the resulting image at the time of render termination.
For single image renders it’s generally better to keep the sampling level limit high (the default of 25 is fine). You can then always stop the render if you’re already satisfied with a lower sampling level.
Theory
The Maxwell renderer’s Environment section is dedicated to accurately simulating environmental conditions, such as physical skylight with or without sunlight, based upon real-world coordinates, date and time.
The absorption function in the Fog Options section refers to the fraction of light that is absorbed when penetrating the fog. The scattering function is related to the intensity of light being reflected by fog molecules.
Practice
For exteriors and for interiors with a lot of outside light penetration use Maxwell’s physical sky. For interiors with less exterior light penetration use the skydome. The skydome generally renders faster, resulting in less noise when the render is finished.
Use absorption and scattering values above 0 to activate fog.
Increasing the absorption value makes the fog look darker.
Theory
The number of threads refers to the number of processing pipelines to be utilized by Maxwell. Basically one processor equals one thread, but machines that support hyperthreading can have two threads per processor.
Practice
Make sure the scene scale value corresponds with your scene’s proportions as accurately as possible, to achieve proper realism and to avoid unwanted consequences, such as an incorrect DOF effect.
Set the number of threads to the maximum amount of available threads on your system in order to maximize render capacity for Maxwell. For example if you’ve got a hyperthreading dual processor machine, set the number of threads to 4.
Theory
The Film ISO value originates from traditional photography. It is an indication of the film speed, where higher values refer to faster film types with the ability to capture more light. However, the resulting images of higher ISO film types also display more graininess.
The Maxwell renderer’s tone mapping section offers some optional tweaks for the appearance of the rendering.
Practice
Adjust the Film ISO value to increase or decrease the brightness in your render. A value of 200 means that twice as much light is captured from the scene by the Maxwell camera compared to using an ISO value of 100. A Film ISO value of 50 means that half the amount of light is captured from the scene by the Maxwell camera compared to using an ISO value of 100.
Unlike traditional film ISO types increasing the Film ISO value in Maxwell doesn’t induce graininess.
Burn is a tone-mapping parameter that controls the RGB conversion from internal high-dynamic range calculations.
Increase the gamma value in order to brighten dark areas in your image.
Phenomenon
� Scene objects suddenly aren’t visible anymore when rendering.
Possible solution(s)
� It might be a normals problem. Check if the normal vectors of your polygonal object all point towards the right direction. If appropriate correct this until all normals point in the right direction.
Phenomenon
� Glass, plastic or an other dielectric element causes light strength to decrease too much when going through the element, resulting in the object(s) becoming too dark.
Possible solution(s)
� The type of glass / plastic / etc. you chose might not be suitable for the application in the scene. Try some different types.
� The scene object that contains the dielectric material may be out of proportion compared to a real-world equivalent. For example, a glass window may be too thick. Check your object dimensions compared to the scene’s scale.
� Be sure all caustic options are switched on in the render dialog.
� Decrease the Absorbance value in order to lower the amount of light that will be absorbed by the dielectric volume.
� Try raising the color value of the dielectric material.
� If your dielectric object type does not need crystal-like colorful dispersion / diffraction effects, increase the Abbe value for more clearness (Abbe values of 175 and above yield clear results).
� Make sure the scene contains enough surfaces that are not too dark, in order to sufficiently reflect light throughout the scene.
� Simply render for a longer time. As Maxwell reaches a higher sampling level, the more complete the light solution will be.
Phenomenon
� Noise is still apparent when rendering has completed.
Possible solution(s)
� Render for a longer time to reach a higher sampling level and eventually get rid of all noise.
� If your target sampling level is reached and the result still shows noise, rerender with a higher number of samples.
� Decrease the number of bounces in your scene, as fewer bounces require less calculation and thus the renderer gets more opportunity to get rid of noise during the given rendertime.
� Make sure your scene is lit well enough. Scenes with a coarse light distribution require more effort for the renderer to get free of noise. This is comparable with the phenomenon of more noise in digital photos in areas of the photo that are insufficiently lit.
� For exteriors and for interiors with a lot of outside light penetration use Maxwell’s physical sky. For interiors with less exterior light penetration use the skydome. The skydome generally renders faster, resulting in less noise when the render is finished.
� Use less faces / triangles in your light emitting objects and (if possible) use less light emitting objects. More light emitters as well as more polygons in your light emitters require more calculation.
� A reasonably effective approach to reduce noise in a final result is to render at a higher resolution than your desired size and scaling down the result to the desired size. The higher the render resolution, the more you will be able to downsize the result and the more the noise will be averaged into a clean picture by the scaling, without losing sharpness.
� Try to avoid very high surface color values and saturation.
� Try to use as few as possible noise-inducing elements in your scene, such as dielectrics (transparent elements such as glass) and glossy reflections (using U and V roughness). If you do use dielectrics in your scene, increase the Abbe value to 150 or higher to suppress noise.
� Try to decrease polygonal overhead in your scene wherever possible, in order to simplify the scene and the necessary calculations.
� Use a so-called “denoiser” to eliminate noise from your image after it’s rendered, such as ABSoft’s effective application and Photoshop plug-in Neat Image.
Jan 01
This entry was posted on Saturday, January 1st, 2005 at 9:38 pmand is filed under Computer, Articles. You can follow any responses to this entry through the RSS 2.0 feed. Responses are currently closed, but you can trackback from your own site.
Comments are closed.