diff --git a/src/integrators/photonmapper/photonmapper.cpp b/src/integrators/photonmapper/photonmapper.cpp index cbc847fa..caae6ef9 100644 --- a/src/integrators/photonmapper/photonmapper.cpp +++ b/src/integrators/photonmapper/photonmapper.cpp @@ -75,7 +75,7 @@ MTS_NAMESPACE_BEGIN * } * * When the scene contains participating media, the Beam Radiance Estimate \cite{Jarosz2008Beam} - * is used to estimate the illumination due to volumetric scattering. + * by Jarosz et al. is used to estimate the illumination due to volumetric scattering. * * \remarks{ * \item Currently, only homogeneous participating media are supported by this implementation diff --git a/src/mtsgui/resources/docs.xml b/src/mtsgui/resources/docs.xml index f713104b..1bc04460 100644 --- a/src/mtsgui/resources/docs.xml +++ b/src/mtsgui/resources/docs.xml @@ -1,19 +1,5 @@ - - Scene root element - Tweaking the kd-tree construction parameters is generally not required. - kd-tree construction: Enable primitive clipping? Generally leads to a significant improvement of the resulting tree. - kd-tree construction: Relative cost of a triangle intersection operation in the surface area heuristic. - kd-tree construction: Relative cost of a kd-tree traversal operation in the surface area heuristic. - kd-tree construction: Bonus factor for cutting away regions of empty space - kd-tree construction: A kd-tree node containing this many or fewer primitives will not be split - Requires an integrator - Requires a camera - Requires one or more luminaires - Requires one or more shapes - - Base class of all sampling-based integrators Base class of all sampling-based integrators @@ -299,17 +285,33 @@ - Rasterization-based global illuminated technique using hardware - accelerated renderings of the scene under point source illumination. Based on - "Instant Radiosity" by Alexander Keller in Computer Graphics Proceedings, - Annual Conference Series, SIGGRAPH 97, pp. 49-56. +

This integrator implements a hardware-accelerated global illumination + rendering technique based on the Instant Radiosity method by Keller. + This is the same approach that is also used in + Mitsuba's real-time preview; the reason for providing it as a separate + integrator plugin is to enable automated (e.g. scripted) usage.

+ +

The method roughly works as follows: during a pre-process pass, any present direct + and indirect illumination is converted into a set of virtual point light + sources (VPLs). The scene is then separately rendered many times, each time using + a different VPL as a source of illumination. All of the renderings created in this + manner are accumulated to create the final output image.

+ +

Because the individual rendering steps can be exectuted on a + graphics card, it is possible to render many (i.e. 100-1000) VPLs + per second. The method is not without problems, however. In particular, + it performs poorly when rendering glossy materials, and it produces + artifacts in corners and creases . Mitsuba automatically limits + the "glossyness" of materials to reduce the effects of the former + problem. A clamping parameter is provided to control the latter + (see the documentation for details). The number of samples per pixel specified to + the sampler is interpreted as the number of VPLs that should be rendered.

 - Total number of virtual point lights that should be rendered - Shadow map resolution + Resolution of the shadow maps that are used to compute the point-to-point visibility - Longest visualized path length (-1=infinite). When a positive value is - specified, it must be greater or equal to 2, which corresponds to single-bounce - (direct-only) illumination. + Specifies the longest path depth in the generated output image (where -1 + corresponds to ∞). A value of 1 will only render directly visible light sources. + 2 will lead to single-bounce (direct-only) illumination, and so on. Relative clamping factor (0=no clamping, 1=full clamping) @@ -326,6 +328,9 @@ unless an extremely large number of photons is used. A simple remedy is to combine the photon mapper with an irradiance cache, which performs final gathering to remove these artifacts. Due to its caching nature, the rendering process will be faster as well.

+ +

When the scene contains homogeneous participating media, the Beam Radiance Estimate by Jarosz et al. + is used to estimate the illumination due to volumetric scattering.

Number of luminaire samples for direct illumination Number of glossy samples for direct illumination @@ -350,17 +355,31 @@ - Progressive photon mapping implementation. Only handles surface - interactions. Parallelization is limited to the local cores. +

This plugin implements the progressive photon mapping algorithm by Hachisuka et al. + Progressive photon mapping is a variant of photon + mapping that alternates between photon shooting and gathering passes that involve + a relatively small (e.g. 250K) numbers of photons that are subsequently discarded.

+ +

This is done in a way such that the variance and bias of the resulting output + vanish as the number of passes tends to infinity. The progressive nature of this + method enables renderings with an effectively arbitrary number of photons + without exhausting the available system memory.

+ +

The desired sample count specified in the sample generator configuration + determines how many photon query points are created per pixel. It should not be + set too high, since the rendering time is approximately proportional to + this number. For good results, use between 2-4 samples along with the + low-discrepancy sampler. Once started, the rendering process continues indefinitely + until it is manually stopped.

 - Longest visualized path length (-1=infinite). When a positive value is - specified, it must be greater or equal to 2, which corresponds to single-bounce - (direct-only) illumination. + Specifies the longest path depth in the generated output image (where -1 + corresponds to ∞). A value of 1 will only render directly visible light sources. + 2 will lead to single-bounce (direct-only) illumination, and so on. - Initial photon query radius (0 = infer based on scene size and camera resolution) + Initial photon query radius (0 = infer based on scene size and camera resolution) Number of photons to shoot in each iteration - Granularity of photon tracing work units (in shot particles, 0 => decide automatically) + Granularity of photon tracing work units (in numbers of traced particles, 0 = decide automatically) Specifies the minimum path depth, after which the implementation will start to use the "russian roulette" path termination criterion when tracing photons (set to -1 to disable). @@ -371,17 +390,23 @@ - Stochastic progressive photon mapping implementation. Only handles surface - interactions. Parallelization is limited to the local cores. +

This integrator implements stochastic progressive photon mapping (PPM) by Hachisuka et al. + This algorithm is an extension of progressive photon mapping that improves convergence + when rendering scenes involving depth-of-field, motion blur, and glossy reflections.

+ +

Note that this integrator ignores the sampler + configuration---hence, the usual steps of choosing a sample generator and a desired + number of samples per pixel are not necessary. As with PPM, once started, + the rendering process continues indefinitely until it is manually stopped.

 - Longest visualized path length (-1=infinite). When a positive value is - specified, it must be greater or equal to 2, which corresponds to single-bounce - (direct-only) illumination. + Specifies the longest path depth in the generated output image (where -1 + corresponds to ∞). A value of 1 will only render directly visible light sources. + 2 will lead to single-bounce (direct-only) illumination, and so on. - Initial photon query radius (0 = infer based on scene size and camera resolution) + Initial photon query radius (0 = infer based on scene size and camera resolution) Number of photons to shoot in each iteration - Granularity of photon tracing work units (in shot particles, 0 => decide automatically) + Granularity of photon tracing work units (in numbers of traced particles, 0 = decide automatically) Specifies the minimum path depth, after which the implementation will start to use the "russian roulette" path termination criterion (set to -1 to disable). @@ -462,52 +487,95 @@ - - Kelemen-style Metropolis Light Transport implementation. This - integrator can operate on top of either a simple path tracer or a - fully-fledged bidirectional path tracer with multiple importance - sampling. Various optimizations are implemented, namely stratified - mutations, separate direct illumination, two-stage MLT, and the - improved weights proposed by Kelemen et al. For details, see - the respective parameter descriptions. + +

Primary Sample Space Metropolis Light Transport (PSSMLT) is a rendering + technique developed by Kelemen et al. which is + based on Markov Chain Monte Carlo (MCMC) integration.

+ +

In contrast to simple methods like path tracing that render + images by performing a naive and memoryless random search for light paths, + PSSMLT actively searches for relevant light paths (as is the case + for other MCMC methods). Once such a path is found, the algorithm tries to + explore neighboring paths to amortize the cost of the search. This can + significantly improve the convergence rate of difficult input. + Scenes that were already relatively easy to render usually don't benefit + much from PSSMLT, since the MCMC data management causes additional + computational overheads.

+ +

The PSSMLT implementation in Mitsuba can operate on top of either a simple + unidirectional volumetric path tracer or a fully-fledged bidirectional path + tracer with multiple importance sampling, and this choice is controlled by the + bidirectional flag. The unidirectional path tracer is generally + much faster, but it produces lower-quality samples. Depending on the input, either may be preferable.

+ +

Caveats: + There are a few general caveats about MLT-type algorithms that are good + to know. The first one is that they only render "relative" output images, + meaning that there is a missing scale factor that must be applied to + obtain proper scene radiance values. The implementation in Mitsuba relies + on an additional Monte Carlo estimator to recover this scale factor. By + default, it uses 100K samples (controlled by the luminanceSamples + parameter), which should be adequate for most applications.

+ +

The second caveat is that the amount of computational expense + associated with a pixel in the output image is roughly proportional to + its intensity. This means that when a bright object (e.g. the sun) is + visible in a rendering, most resources are committed to rendering the + sun disk at the cost of increased variance everywhere else. Since this is + usually not desired, the twoStage parameter can be used to + enable Two-stage MLT in this case.

+ +

In this mode of operation, the renderer first creates a low-resolution + version of the output image to determine the approximate distribution of + luminance values. The second stage then performs the actual rendering, while + using the previously collected information to ensure that + the amount of time spent rendering each pixel is uniform.

+ +

The third caveat is that, while PSMLT can work with scenes that are extremely + difficult for other methods to handle, it is not particularly efficient + when rendering simple things such as direct illumination (which is more easily + handled by a brute-force type algorithm). By default, the + implementation in Mitsuba therefore delegates this to such a method + (with the desired quality being controlled by the directSamples parameter). + In very rare cases when direct illumination paths are very difficult to find, + it is preferable to disable this separation so that PSSMLT is responsible + for everything. This can be accomplished by setting + directSamples=-1.

 + + If set to true, the MLT algorithm runs on top of a + bidirectional path tracer with multiple importance sampling. + Otherwise, the implementation reverts to a unidirectional path tracer. + Specifies the longest path depth in the generated output image (where -1 corresponds to ∞). A value of 1 will only render directly visible light sources. 2 will lead to single-bounce (direct-only) illumination, and so on. - When separateDirect is set to true, this parameter can - be used to specify the samples per pixel used to render the - direct component. Should be a power of two (otherwise, it will be - rounded to the next one). When set to zero or less, the - direct illumination component will be hidden, which is useful - for analyzing the component rendered by MLT. + By default, this plugin renders the direct illumination component + separately using an optimized direct illumination sampling strategy + that uses low-discrepancy number sequences for superior performance + (in other words, it is not rendered by PSSMLT). This + parameter specifies the number of samples allocated to that method. To + force PSSMLT to be responsible for the direct illumination + component as well, set this parameter to -1. - - If set to true, the MLT algorithm runs on top of a - bidirectional path tracer with multiple importance sampling. - Otherwise, the implementation reverts to a basic path tracer. - Generally, the bidirectional path tracer should be noticably - better, so it's best to this setting at its default. - - - This setting can be very useful to reduce noise in dark regions - of the image: it activates two-stage MLT, where a nested MLT renderer - first creates a tiny version of the output image. In a second pass, - the full version is then rendered, while making use of information - about the image-space luminance distribution found in the first - pass. Two-stage MLT is very useful in making the noise characteristics - more uniform over time image -- specifically, since MLT tends to get - stuck in very bright regions at the cost of the remainder of the image. + + Use two-stage MLT? Please see the documentation for details. - Number of samples used to estimate the total luminance - received by the camera's sensor. + MLT-type algorithms create output images that are only + relative The algorithm can e.g. determine that a certain pixel + is approximately twice as bright as another one, but the absolute + scale is unknown. To recover it, this plugin computes + the average luminance arriving at the sensor by generating a + number of samples. - Probability of creating large mutations in the [Kelemen et. al] - MLT variant. The default is 0.3. + Rate at which the implementation tries to replace the current path + with a completely new one. Usually, there is little need to change + this. Specifies the minimum path depth, after which the implementation will start to use the @@ -645,249 +713,6 @@ - - Homogeneous participating medium - - Homogeneous participating medium. An arbitrary (manifold) shape - must be specified as a child object. - - Specifies the shape of the medium - - - - Heterogeneous participating medium - - Heterogeneous medium class using trilinear interpolation, - Simpson quadrature and one of several possible sampling - strategies. Data files have to be provided in an ASCII - format as follows:
-
    -
  • The first three numbers determine the X,Y and Z resolution, - each of which has to be larger than 2.
    • -
  • The next six numbers determine the minimum X, Y and Z - as well as the maximum X, Y and Z values of the enclosing - axis-aligned bounding box.
    • -
  • Afterwards, (xres*yres*zres) density samples follow in - XYZ order, (e.g. the second entry has coordinate x=2)
    • -
- - - File containing sampled volume densities - - - Specifies the used strategy - the four choices are: -
    -
  • standard: generate a 'desired' accumulated density by sampling an - exponentially distributed random variable. Afterwards, try to find the - distance, at which this much density has been accumulated. Here, the - composite Simpson's rule is used to integrate density along the ray.
    • -
  • coleman: Sampling technique by [Coleman et al., 1967]. Only for media with - a wavelength-independent extinction coefficient.
    • -
  • uniform: Naive variant for verification purposes: uniformly - sample a distance along the ray segment, which intersects the volume
    • -
  • double: Double integral approach - stupid and slow, but it also works..
    • -
- - - - While integrating density along a ray, approximately one sample - per voxel is taken - that number can be changed here - - - - Can be used to override the extinction coefficient used to sample distances - in the in-scatter line integral. By default, the smallest spectral sample of - sigmaA+sigmaT is used. - - - - - - Isotropic phase function - - Basic isotropic phase function - - - - - Henyey-Greenstein phase function - - Phase function by Henyey and Greenstein (1941). Parameterizable - from backward- through isotropic- to forward scattering. - - - Asymmetry parameter of the Henyey-Greenstein phase function. Must - lie in [-1, 1] where >0 is forward scattering and <0 is backward - scattering. - - - - - Abstract luminaire - Abstract implementation of a luminaire. Supports emission and shadow ray sampling and computes related probabilities. - Affine luminaire space to world space transformation - - - - Lambertian area luminaire - - Lambertian area light source - can be attached to an arbitrary shape - contained inside the scene. Shadow rays are generally sampled - uniformly with respect to surface area, which may lead to high - variance (e.g. many of the generated samples are facing away - from the point to be shaded). - When the shape in question is a sphere, rays are sampled uniformly - wrt. solid angle, which significantly reduces the variance. - Thus, spheres are recommended whenever there is some flexibility - in choosing the luminaire shape. - - Intensity of the luminaire - - - - Collimated beam luminaire - Collimated beam with a configurable thickness. Points from (0,0,0)->(0,0,1) by default - - use the toWorld parameter to change this. - - World-space beam radius - Intensity of the luminaire. - Specify only one of intensity and power. - Power of the luminaire. - Specify only one of intensity and power. - - - - Isotropic point source - Positioned at (0,0,0) by default - use the - toWorld parameter to change this. - Intensity of the luminaire - - - - VRML SpotLight-equivalent luminaire - - In its local coordinate system, the spot light is positioned at the origin and points into the positive Z - direction. Its intensity linearly ramps up between cutoffAngle - and beamWidth, after which it remains at the maximum value. - - Intensity of the luminaire - Cutoff angle in degrees - Beam width in degrees - - - - Constant background luminaire - Simulates a diffuse, infinitely far-away emitter - Intensity of the luminaire - - - - Environment map luminaire - - Basic environment map implementation without importance sampling. - Uses the scene's bounding sphere to simulate an infinitely far-away - light source. Expects an EXR image in latitude-longitude (equirectangular) format. - - Intensity of the luminaire - - - - Camera base class - - A camera turns a sample on the image plane into a 3D ray. - For this, it requires two supporting objects: a Sampler and a Film instance. - - Affine camera space to world space transformation - - - - Pinhole camera base class - - Provides solid angle computation - routines useful for importance-based integrators. - - Field of view of the camera (in degrees) - Specifies which side of the image plane should cover the field of view specified in the fov parameter - - - - Perspective camera model - - Depth of field can optionally be activated by specifying both focal - distance and lens radius (uses a thin lens approximation) - - Distance to the focal plane - World-space lens radius - Near clipping plane distance - Far clipping plane distance - - - - Orthographic camera model - - Simple orthographic camera model - - Near clipping plane distance - Far clipping plane distance - Specifies which side of the image plane - maps to normalized device coordinates in [0,1] - - - - - Abstract Film base class - Used to store samples generated by an Integrator. - Horizontal sensor resolution in pixels - Vertical sensor resolution in pixels - Horizontal pixel position of the crop window relative to the upper left corner - Vertical pixel position of the crop window relative to the upper left corner - Width of the crop window in pixels - Height of the crop window in pixels - - If set to true, regions slightly outside of the film plane will also be sampled, which - improves the image quality at the edges especially with large reconstruction filters. - - - - - High dynamic-range film - - Simple film implementation, which stores the captured image - as an RGBA-based high dynamic-range EXR file. - No gamma correction is applied and spectral radiance values - are converted to linear RGB using the CIE 1931 XYZ color matching - functions and ITU-R Rec. BT.709 - - Should an alpha channel be added to the output image? - Should a program logo be added to the output image? - - - - PNG low dynamic-range film - - Simple film implementation, which stores the captured image - as an RGBA-based low dynamic-range PNG file with gamma correction. - Spectral radiance values are converted to linear RGB using - the CIE 1931 XYZ color matching functions and ITU-R Rec. BT.709 - - Should an alpha channel be added to the output image? - Bits per pixel including alpha (must be 8, 16, 24 or 32) - Gamma value for the correction. Negative values switch to sRGB - - - - MATLAB film - - Debugging film, which dumps raw luminance values as a 2D array in - MATLAB M-file format. When test case mode is active, the film writes - a triple (luminance, variance, sample count) for every pixel. - - Should all spectral samples be exported? In this case, - each pixel will generate several numbers (matching the configured discretization of the color spectrum). - By default, Mitsuba exports the luminances found using the CIE XYZ color matching curves. - - - Box filter: the fastest, but also about the worst possible reconstruction filter, since it is extremely prone to aliasing.