Raytracing Manual 4gi5e

with a value between 0.05 and 0.01 for in the commandline field of the expert mode dialog. This however considerably longer calculation times. © Relux Informatik AG, Fit for Raytracing, 27.07.2011

may result in

INDIRECT LIGHT CALCULATION

18/52

5 INDIRECT LIGHT CALCULATION 5.1

Indirect light calculation in the raytracing context

Fig. 9 Analysing indirect illumination with sample rays The direct and indirect lighting are common for lighting analysis and need no further explanation. Unfortunately, the situation is not so trivial anymore when it comes to simulating illumination scenarios. Here, the indirect part of the calculation is by far more challenging than the direct part. Within the raytracing context, also a different algorithm is used for the indirect part. The most important characteristics of this approach are explained in the following, because this basic knowledge is necessary to be able to sucessfully configure and use the tool for the wide variety of cases occuring in todays poject work. In contrast to the direct light calculation, one cannot rely on localised light sources at known positions anymore when it comes to determine the amount of indirect light reaching a point. It is necessary to add up all contributions from the surrounding, which can be described theoretically in form of an integral equation:

L denotes the 'seen' luminance dependent on direction for a point in a scene. By the way, the special case of direct illumination of chapt. 4 can be deducted also from this general formula.

© Relux Informatik AG, Fit for Raytracing, 27.07.2011


19/52

For the indirect contribution, the integration is performed over the whole hemisphere 'above' a surface element. Within the alogrithm, this is done by sending out a certain number of sample rays into the surounding scene and adding up their returned contributions (Fig. 9). The number of rays sent out has a high influence on the accuracy of the process. Also, the adequate amount of rays needed for an accurate sampling of the surrounding does not only depend on the geometry, but also on the lighting scenario (e.g.uniform or highly varying illumination).

Fig. 10 Details of the indirect algorithm. Left: calculation at distant points and interpolation inbetween, right: iterative simulation with several ray bounces Clearly, this approach is very time consuming, and therefore cannot be done for each pixel separately. To circumvent this difficulty, the indirect calculation is done for points at certain distances, and the illumination inbetween is determined by interpolation routines (Fig. 10, left). In addition, stochastic methods come into play, too, to guarantee a higher reliability of simulation results in the average but still keep calculation times within reasonable bounds. It is worthwile noting that two assumptions are silently made in this indirect interpolation approach: -

the indirect illumination varies by far less strong than the direct illumination the majority of the indirect illumination stems from bigger and more or less uniformely illuminated areas in the surrrounding.

The latter assumption may not always be applicable. In these cases the indirect simulation process becomes extremely difficult. Of course, the stepwidth of interpolation distance is a very important criterium. The module contains self adjusting routines, which adapt this distance locally dependent on the detected circumstances. Nevertheless it can be necessary to set a lower threshhold value for this spatial resolution because in practice, the demand for a higher or lower accuracy can vary strongly for different scenes. (This spatial resolution value is comparable to the calculation grid width setting in the ReluxPro standard radiosity module). A last point needs to be considered: the indirect illumination happens in reality by multiple reflection of light between all surfaces in the scene, which theoretically proceeds infinitely often. But the amount of light distributed in this process decreases © Relux Informatik AG, Fit for Raytracing, 27.07.2011


20/52

quickly with increasing number of interreflection steps (Fig. 10, right). For the simulation, one makes use of this fact by stopping the calculation after a certain number of ray bounces. The rest portion of illumination which is negelected by this approximation can either be ignored completely if it falls below a certain tolerance threshhold, or it can be coarsely reperesented by an ambient background illumination value. The necessary minimum number of reflection steps needed for a reliable simulation can generally be deducted by simple geometric considerations. Take for an example a standard office scene with uplighters: with 1 interreflection step the light path luminaire->ceiling->workplane is taken into , with 2 steps additionally, a. o., the paths luminaire->ceiling->walls->workplane get considered, and so on. In most cases, the minimum number of interreflection steps is 2 or 3, lower settings only make sense for quick test previews.

Note: The necessary amount of interreflections for an adequate simulation does not

only depend on the scene geometry, but also on the illumination scenario and the materials. If semitransparent surfaces modeled with the milky glass material (RADIANCE 'trans' material) are present in the scene, it is important to augment the interreflection setting by 1 or 2, because the diffuse part of the transmission in this case also demands a separate interreflection step.

5.2

Using the method in practice – hints for parameter settings

In addition to this background information recommendations for parameter settings are given now for some typical types of scenes. Interreflections For common architectural scenarios with usual building materials (i.e. no exotic setups like a room full of mirrors or all surfaces having unnaturally high diffuse reflection coefficients of 0.9 or higher), the indirect calculation algorithm converges at 7-9 interreflections. This setting is of course only needed for cases, in which a demand for a high calculation accuracy exists. In general, for illuminance value calculations, a setting between 4 and 7 can be used in order to get results within a tolerance range of 10% around the convergence value.

Fig. 11 Convergence of the calculation results with higher number of traced diffuse interreflections within a scene simulated solely by the indirect light algorithm. For visualisations, often 2-3 interreflections plus an ambient background illumination are sufficient. Only in purely indirectly illuminated scenes (e.g. daylight scenes © Relux Informatik AG, Fit for Raytracing, 27.07.2011


21/52

without window precalculation) a higher setting of 4 is recommended. The Fig. 11 depicts the above mentioned convergence of the calculation. The graph on the right shows the illuminance over the number of interreflections for a small test area in a complex lighting situation (daylight simulation without precalculation in a badly lit part of the scene). The scene and the location of the test area is shown on the left. The graph also shows the slight fluctuation in the values, which stems from the use of stochastic methods in the algorithm. Initial number of rays Recommendations for the appropriate number of initial indirect sampling rays are listed in the table below: simple scenes, with well and uniform lighting (artificial and/or daylight with window precalculation, cf. chapt.6.4)

800 - 1200

complex scenes (high level of detail, lighting scenario with high variability (e.g. with narrow beam spots or sunlight)

1500 - 3000

daylight without window precalculation (cf. chapt 6.4)

3000 - 6000

Spatial resolution The spatial resolution value is a critical parameter with a strong influence on calculation time. for an understanding it is important, that it is a kind of guidance parameter for the calculation, whose accuracy relaxes for distances below this value. The default setting of 0.4 m is generally sufficient, because it guarantees accurate results on the bigger surfaces of the scene, whereas edges, corners, small fixture geometry and other small scene detail receive a lower attention, which can be tolerated as for this kind of detail a high simulation accuracy is not needed anyway. For cases, in which the default setting does not suffice, the value may be reduced to 0.3 or 0.2m. Values below 0.1 m should only be used in extreme circumstances. In contrary, one can try to augment the value to 0.6 - 1m for very big scenes. Of course, a commom problem is that big scenes often also contain fine grain detail locally, so the opportunities to use coarse resolution values are limited in practice. Background ambient illumination For visualisations, one can use the approach of simulating the indirect illumination with a limited number of interreflection steps and approximating the rest with a constant background illumination value. This value should be handled and understood as a manual setting based on the usual 'try and error' principle, because a priori, one cannot know how much of the illuminance gets ignored by stopping the calculation already at a low number of interreflection steps. In interior scenes with artificial light, values of 0.01-0.1 are reasonable, for daylight scenes 0.01-1, for exterior scenes with daylight and sunlight recommended settings lie between 1-10, for exterior scenes at night with artificial illumination, values of 0.001 or lower should be used. The ambient background illumination can be set in 3 different colour channels. © Relux Informatik AG, Fit for Raytracing, 27.07.2011


22/52

Note: This ambient background illumination only makes sense for more or less uniformly illuminated scenes. In special cases, e.g. scenes with large scarcely illuminated areas or for value calculations with a high number of interreflection steps and a high demand on accuracy, it is recommended to use a zero ambient background value to avoid adding too much light without justification.

All parameter recommendations should be seen as first proposals, and mostly they represent the minimal requirements, too. Higher values may be necessary in special circumstances. In contrary, for quick test previews, lower settings also suffice. It definitely pays off to experiment with different settings to gain some experience in handling the tool and chosing the right settings for each different scenario. Additional option: soft interpolation Within the raytracing program, the focus is clearly pointing at a physically correct simulation of the light propagation. For the sake of efficiency, the algorithm relies on stochastic methods for the calculation. This is justified, because for value generation (e.g. in a grid of points), a single value has a reduced significance. A slight fluctuation can be tolerated, as always the average over a certain area is the important criterium. In image generation, this stochastic approach may of course lead to visible artifacts, recognisable e.g. by more or less disturbing splotches in the image. To optimize the raytracing module also for image generation, an enhanced interpolation option is in the stage of development, which already can lead to much cleaner images in cases where the original version runs into problems.This option can be activated in different levels (low, medium, high) with the soft interpolation parameter. The default setting is off.

Note: This option can only work, if the other settings are at least coarsely adequate.

The development is still in progress, so optimal results under all circumstances cannot be guaranteed. The option should only be used for visualisations, because in the medium and high setting, it can have locally varying influences on the accuracy of the indirect calculation.

5.3

Problematic aspects and limitations of the method

The indirect calculation is a very challenging task, so lot of optimizations have to be used to be able to tackle the problem at all. Because the demands vary widely in practice, some further adjustments additional to the already mentioned ones may be necessary in some cases. Also, like every technical tool, the current indirect algorithm has its limitations which make it unsuitable for certain simulation scenarios. Scenes with huge external obstructions Normally, the indirect calculation is optimized for cases, where the area which is of interest for illuminance values or picture generation more or less comprises the whole scene. In some scenarios this may not be the case. Take for example a daylight simulation, where a room of, say, 10-20m dimensions is of interest, but it is modeled within a surrounding set of external obstrcutions, making up a whole scene of 200300m dimensions. For these kind of scenarios, a special command line parameter should be set to disable some optimization features and enforce a higher accuracy. The parameter is -am 0 and it has to be entered in the commandline text field at the bottom of the expert mode dialog. © Relux Informatik AG, Fit for Raytracing, 27.07.2011


23/52

Scenes with extreme variability in illumination The indirect calculation algorithm runs into difficulties when it comes to simulate certain extreme illumination scenarios, especially those, where the second one of the assumptions mentioned in the introduction of this chapter is no longer justified. If the indirect illumination no longer stems from big, well lit areas in the surrounding, the limitations of the sample ray approach become apparent. Look for example at the scene depicted in the Fig. 12. The staircase is almost not illuminated at all, except some objects on the table which are lighted by very narrow beam spots. Sampling the reflected light from this small illuminated area (small in comparison to the whole scene) from every other point in the staircase is a very challenging task. To achieve satisfying results even in such cases, a high number of initial indirect rays (5000 – 10000) is necessary, and, dependent on the circumstances, additionally the above mentioned commandline parameter –am 0 should be set, too.

Fig. 12 Example scene with extremely non-uniform illumination.


DAYLIGHT SIMULATIONS

24/52

6 DAYLIGHT SIMULATIONS Daylight simulations are generally more challenging in of modeling accuracy and parameter settings than artifical light calculations. The most important aspects are described below:

6.1

Illumination type and results

Daylight simulations can be done when the appropriate illumination type is selected in the main raytracing dialog (either daylight or artificial and daylight). Also important is the correct date and time setting and the sky model type. If direct sunlight is wanted, this has to activated separately. For setting the position data of the location (longitude and latitude) see the main ReluxPro documentation. The generated results are mostly similar to the case of artificial light simulation. Only when the CIE overcast sky is chosen, daylight factors for each measurement plane are generated additionally to the standard value output.

6.2

Daylight openings

For ReluxPro interior scenes one can chose from windows or skylights for daylight openings. In a corresponding property dialog, dimensions and materials for the window pane, etc., can be entered. Often, daylight openings are characterised by the pure opening width. The effect of window frames and the pollution of the window panes in the final building is then considered with help of reduction factors.

Note: If the option generate additional geometry is activated, the raytracing module

generates window frames automatically (as long as a frame geometry is selected for the actual window). In this case, of course the frame reduction factor is not considered in the raytracing calculation. Rectangular frames of different dimensions can be generated, the dialog for entering the frame dimension and material parameters can be accessed from the window property dialog. A list of frame geometries can be set up, and then for each window an individual geometry can be selected from the list. For exterior scenes, the standard ReluxPro window and skylight types cannot be used. Here daylight openings including frames and window panes have to be entirely modelled by the (cf. also chapt. 8.1 ).

6.3

External floor reflection - the mirrored sky concept

For daylight simulations, the reflection of sky and sunlight by external surfaces may play a certain role for the scene illumination. In the default case, the raytracing module only considers the ground plane reflection with the concept of the mirrored sky (Fig. 13): Instead of modelling a huge exterior plane, the reflection of the light from the sky hemisphere is approximated by mirroring the sky luminance in a direction perpendicular to the ground plane and multiplying it with the reflection coefficient of the exterior area. This reflection coefficient can be set in the property dialog of the outer floor plane, it is the same which is also used for the radiosity calculation.



25/52

Fig. 13 left: mirrored sky concept to simulate ground plane reflection without need to model the plane itself. For comparison, on the right the case with an extra ground plane is shown. This mirrored sky concept (RADIANCE s might be more familiar with the term 'ground glow') is the most common, basic form of consideration of external ground plane reflection. An interesting further benefit is, that the concept is independent from other surfaces and objects added to the scene and does not conflict with them, i.e. for special scenarios separate planes etc. in the vicinity of the simulated building may of course be added by the ..

Note: The mirrored sky is always included, but it is of course only capable of

simulating the diffuse reflection of the light of the sky hemisphere, which is sufficient for the most cases of daylight analysis. For special cases (e.g. simulating reflected sunlight, or ground planes with specularly reflecting material), a separate surface has to modeled explicitely by the (Fig. 13, right). The standard ground plane used by the ReluxPro radiosity calculation is not used by the raytracing module. If an external plane is modeled, attention should be payed to avoid overlapping of outer and inner floor planes. The best way is to really draw the outer plane(s) only in the exterior area of the ReluxPro scene. By this approach, also the light leak problem, i.e. resolution problems within the indirect interpolation, resulting in erroneously considering lighting values from the outer area within the interior scene, can be avoided.



6.4

26/52

Window precalculation

One reason why daylight calculations are more challenging than artificial light simulations is the fact that the sky hemisphere cannot be treated simply as a localised light source comparable to a common luminaire. Within the raytracing module, the lighting by the sky hemisphere is evaluated with the means of the indirect calculation (see chapt. 5). The sun in contrary cannot be modeled as a localised source either (because of its great distance), but it can be treated easily within the direct calculation by means of an angular source with approx. 0.5° solid angle. To better capture sky hemisphere light entering a closed scene with relatively small openings, it is advisable to perform the simulation in two steps. At first, a precalculation is performed for each daylight opening, which samples the light entering through it for the current situation (sky, date, time, external obstructions, etc.). This information is then connected to the opening in form of a luminous intensity distribution curve (LDC). In the second step of the simulation, the daylight openings can be evaulated like localized sources for the interior part of the scene. Of course these generated LDCs are only valid temporarily, they have to be regenerated after each change of the input configuration (date, window pane material, sky model, etc.)

Fig. 14 Schematics of the window precalculation: Information about entering daylight is connected to the openings in form of a data set (LDC) for later evaluation during the lighting simulation of the interior scene.

Note: The window precalculation feature currently is only available for ReluxPro interior scenes by using the standard daylight elements window and skylight. Window surfaces are usually bigger than common light emitting areas of artificial luminaires, so for reliable simulation results, it often is necessary to subdivide the window pane for the calculation, as the raytracing method has difficulties in coping with large area sources. This subdivision is performed automatically by the program (see below). Of course there are also cases in which the window panes are huge in relation to the scene (e.g. a wall made up completely by big windows), then the precalulation is not recommended anymore and should be deactivated. Also, the virtual light emitting faces at the position of the window panes can generate disturbing highlights in conjunction with specular surfaces, in such circumstances it might also be advisable to perform the rendering without the precalculation step. © Relux Informatik AG, Fit for Raytracing, 27.07.2011


27/52

The window precalulation is a flexible and powerful tool, and is able to provide reliable simulation results if used properly. The default parameters are useful for generating quick results in the most basic cases of simple windows with usual sizes and without external geometry like lightshelfs or venetian blinds. For more complex scenarios the calculation parameters have to be adapted accordingly. The possible settings are explained in the following sections. External interreflections This value specifies, how many bounces (see Fig. 15) of diffuse interreflections of daylight are considered in the exterior area for finally determining the amount of light reaching the daylight opening. The default value of 0 only considers light from the sky vault and directly illuminated objects (e.g. sunlit areas). With higher values, more diffuse interreflection steps between surfaces in the exterior (ground plane, buildings, building elements directly in front of the windows, etc.) are taken into , according to the number given.

Fig. 15 External interreflections for the window precalculation When both direct sunlight and light from the sky should be evaluated, sorting out the contributions is not trivial at all. Therefore, Fig. 16 explains the details once more: the default value of 0 means that the direct illumination in the exterior area is evaluated, i.e. light from the sky, surfaces lit by the sun plus the ground plane reflection via the mirrored sky concept (see chapt. 6.3). Further contributions stemming from diffuse interreflections of light between modeled objects in the exterior area need settings of 1 or more external interreflections (e.g. to sample the amount of sky hemisphere light reflected diffusely by an exterior building).

 Note: The setting only has an effect on the diffuse interreflections, specular

reflection (e.g from metal slabs with significant specular reflection coefficient) get evaluated independently. But here, only the light from the sky vault is considered. If the mirror-like reflection of direct sunlight into the scene should be analysed, the mirror material needs to be used for the relevant objects. Further attention has to be paid on the material of the window panes. The above explanation is valid for standard glass windows. If a milky glass material (RADIANCE © Relux Informatik AG, Fit for Raytracing, 27.07.2011


28/52

'trans' material) is used, the number of external interreflections has to be at least 1, to sample the diffuse contribution of the window pane transmission.

Fig. 16 Contributions gathered with the window precalculation: in the default case of 0 interreflections, only the direct illumination is considered (left), with a higher value further diffuse indirect contributions are taken into , too (right).

Number of sampling rays With this parameter, the amount of rays are set which are sent out to sample the surrounding luminance in the exterior area (Fig. 17, left). This also defines the angular resolution of the generated window LDC . The value can be seen in analogy to the number of rays in the normal indirect calculation. The default value of 100 is somewhat coarse, if external geometry is present, the value should be augmented. The following table depicts the resulting angular resolution of the generated window LDC, vertically and radially around the normal direction to the window pane. LDC angular resolution (°) vertical

radial

60

16.4

30.0

100

13.8

22.5

300

7.8

13.3

600

5.8

9.2

1000

4.4

7.2

3000

2.5

4.1

rays



29/52

Fig. 17 Details of the window precalculation: number of sample rays (left), number of sample points (middle) and subdivision (right).

Number of sample points The sky luminance distribution does not change over the width of the window pane, but the influence of external geometry or thick walls does. So, dependent on the circumstances, performing the precalculation with several stochastically distributed sample points is reasonable (Fig. 17, middle). Assuming segment dimensions (see below) of 0.5m and 1m, the default value of 4 points is OK for thin walls without external geometry. When lightshelfs are present, some more points (4-16) are recommended, in case of venetian blinds or otherwise complex surroundings even more (e.g. 20-40) should be used. The maximum is 64. For quick tests or visualizations, also the minimum setting of one point may be used, by accepting the price of higher inaccuracies of calculated values in the vicinity of the window. Subdivision threshhold As mentioned in the introduction, window panes get automatically subdivided by the precalculation algorithm, to avoid too big surface areas for the resulting virtual light emitters and thus to guarantee reliable simulation results (Fig. 17, right). This subdivision is controlled with a threshhold value. If a dimension of the window pane exceeds this length, subdivision occurs in that direction. The resulting segment sizes of course can differ somewhat from the threshhold value. For usual interior scene segment sizes bigger than 1 m side are rarely appropriate. In special scenarios, e.g. an atrium building where the top opening is modeled with a huge skylight, segment sizes of more than 1 m side still can be reasonable.

Positioning of external geometry The standard ReluxPro windows and skylights are placed automatically in the middle of the wall resp. the ceiling. Wall and ceiling thickness is set to 0.2 m by default, it can be changed in the basic settings in the 'Options' dialog. The virtual light emitting face generated by the precalculation is located directly on the interior side of the pane (approx. 1cm in front of it), see Fig. 18. © Relux Informatik AG, Fit for Raytracing, 27.07.2011


30/52

When further external geometry is modeled, care should be taken that no intersection occurs with the window and light emitter pane geometry.

Fig. 18 Positions of window pane (A) and virtual light emitting surface (B)generated by the precalculation. 1) denotes the exterior side, 2) the interior side of the window wall. Examples and recommendations for parameter settings

Fig. 19 Typical cases for the window precalculation Fig. 19 depicts common cases for the window precalculation, the listing below gives corresponding recommendations for parameter values. As always, subdivision into segments of 0.5 – 1m sidelengths is assumed. simple, unobstructed windows (left): default settings, 4 sample points, 60-120 rays lightshelfs and similar geometry(middle): 4-16 sample points, 100 – 300 rays venetian blinds (right): 20-40 sample points, 400-1000 rays By comparison with this example cases, appropriate settings for individual scenarios can be found easily. Fig. 20 finally shows a case where whole walls are made up entirely out of big windows. Here the precalculation would produce a very high number of virtual window light sources, resulting in extreme calculation times. So it is reasonable to deactivate the precalculation in these cases. This is justified, because the big window openings guarantee that a lot of the indirect sample rays can leave the scene and thus the incoming daylight can be captured well also by the standard indirect calculation algorithm. Of course, the indirect parameters have to be set appropriately (see. chapt 5.2)



31/52

Fig. 20 Example case where a simulation without window precalculation is recommended.

6.5

Daylight redirection

Undoubtedly, the simulation of daylight redirection systems is one of the most challenging tasks in lighting analyis. In the current state of technology, mainly 2 methods are common, whose applicability depends on the type of redirection object. 1) Light pipes, which mostly are used to redirect sunlight over greater distances into deeper section of a building, in general need Forward Raytracers or Photon Map programs for an efficient and straightforward simulation. 2) Facade installations (CFS, ‘Complex Fenestration System’), i.e. special window pane materials, films or certain types of highly reflection blinds, which either replace the window pane or are installed additionally, usually get measured in a goniophotometer, in order to produce a description of the redirection characteristic in form of a BSDF data set. With such a material description, the amount of daylight entering through a CFS then can be determined by the means of the window precalculation (cf. chapt. 6.4) like it is done for normal glass panes.



32/52

From version 2011.3.0 onward, ReluxPro is capable of performing simulations according to the method 2) described above. Currently, 4 CFS materials are included in the material database. Their BSDF datasets have been measured at the LESO/EPFL in Lausanne (CH). Assinging these materials happens just like with the normal glass pane materials. The calculation algorithm however is rather complex, and also is still developed further at the moment. So the following hints are important:

Note: 1) The CFS materials currently are provided for use on vertical Relux windows only, they should not be applied to skylights.

2) The visualisation is not yet completely realistic. On the surfaces, the angular dependent luminance is displayed, which results out of the calculated temporary LDC for the specific viewing direction. 3) The redirection effect is of course active for both sky and sunlight. However, a significant contribution to the illumination only results out of the redirected sunlight component. In fact, the ability to use direct sunlight for indoor illumination, which normally is blocked completely e.g. by blinds, is the main benefit of these new materials.

Fig. 21 Often, direct sunlight is blocked by e.g. standard venetian blinds. These however reflect only a little amount of light towards the ceiling (image on the left). By using redirection materials in the upper part of the window, a far greater portion of sunlight can be used for indoor illumination via diffuse reflection at the ceiling (right image).


MATERIALS

33/52

7 MATERIALS Correct material modelling plays a vital role in a simulation, but in a general sense, finding the appropriate abstractions and mathematical models to describe light interaction is a rather difficult task.. Many common materials show a complex behaviour, being partly diffuse, partly specularly reflecting. A material may additionally exhibit transparency, both in a diffuse and/or specular way. On ing through transparent media, light refraction, absorption and scattering occur. Generally, all mentioned effects are dependent on the wavelength, in the visible part of the spectrum this of course is the source of the different colors seen in our everyday surrroundings. Most generally, these effects can be described with a bi-directional reflection and transmission distribution function or data-set (BRDF, BRTF, etc.), giving reflection and transmission dependent on both incidence and outgoing light direction. Really employing such detailed methods however is only needed for special tasks lilke e.g. simulation light-redirection elements in daylight design. For the most common cases of practical simulations, a set of straightforward algorithms exist, to treat each of the above mentioned aspects separately. Material description then can be done with a few parameters for e.g. diffuse reflection, specularity, transparency, etc. After a short introduction into the material models offered by the raytracing module, this chapter contains some important detail on a few selected material types which need closer attention in order to produce reliable simulation results.

7.1

Raytracing materials overview

The provided material models come with different levels of abstraction, keeping this in mind is important for a general understanding of the matter. In their simplest form, they are described with a few parameters. Objects modeled with these materials then show the properties uniformely and isotropically over their whole surface. The basic opaque materials (given by their original RADIANCE names) are 'plastic', 'metal' and 'mirror', basic transparent or partly transparent materials are 'glass', 'trans' and 'dielectric'. A second set is made up out of so-called procedural textures. These are complex material description at a higher level of abstraction, where common building materials like e.g. a brick wall or a stone floor with different tile and grout material, surface normal effects (grouts, beveled edges) and color patterns (e.g. marble or woodgrain) are simulated. These materials appear as Raytracer Materials in the ReluxPro material editor, with the opportunity to adjust their properties. Editing is facilitated by a small preview, showing directly the result of the new settings. Furthermore, one can use pictures (either from the editor or provided by the in JPG, GIF or BMP format) for mapping them onto object surfaces. For working with the material editor and asg materials, see also the ReluxPro documentation. © Relux Informatik AG, Fit for Raytracing, 27.07.2011

MATERIALS

34/52

Note: With a few exceptions (mirror, visible light emitting face in luminaire models),

materials are 'two-sided', i.e. they show the same appearance on either side of a surface. The procedural textures (raytracer materials) currently are provided mainly for the use on walls and ReluxPro primitives (cube, workplane, etc.). They are not yet suited well for use on 3D objects with curved shapes. All raytracing materials can in principle also be used within the standard ReluxPro radiosity calculation module, however a lot of their important effects (especially transparency and specularity) are not considered then.

7.2

Opaque materials

Opaque materials are the most simple ones. In their default state, only the diffuse reflection coefficient is set, in three different color channels. You additionally can set a value for specularity (or 'glossy' reflection) and a surface roughness, which simulates a microscopic surface structure and produces more realistic distributed reflections instead of perfect mirror-like appearances which are seldom present in normal scenarios. The surface roughness only has an effect on the glossy reflections. The basic types 'plastic' and 'metal' differ in the color of their specular highlights: for the 'plastic' material, specular reflection is always neutral (i.e. in the color of the incident light), in the 'metal' type, specular reflection is modifed by the material color.

Note: The RGB values are diffuse reflection coefficients, not colors like in graphics programs. E.g. commonly available white wall paint has RGB coefficients around 0.8 and not 1.0. Also, setting RGB completely to zero to model black surfaces is equally unrealistic and should thus be avoided, too.

Although a lot of materials like porcelain or polished marble appear very glossy in reality, their specularity in of the physical coefficient is rather low, in the range of 0 – 0.2. Higher values are only realistic for metals. The surface roughness parameter should be used with care, values above 0.2 are not realistic either, they may of course be used to create some special effects.

7.3

Transparent / semitransparent materials

Here unfortunately the most difficulties and misunderstandings arise in the normal simulation process, so it is important to understand correctly the assumptions made in the different transparent material types. Glass With respect to parameter setting, the 'glass' material is the simplest one of the transparent types, here only the RGB values for the transmissivity can be set. Internally, however, the material uses a higher degree of abstraction. It is intended to model typically thin glass panes used in e.g. in windows, and additionally to the © Relux Informatik AG, Fit for Raytracing, 27.07.2011

MATERIALS

35/52

transparency, a specular reflection dependent on the incidence angle is considered, too (Fresnel-equations for n=1.52). Also, the mutliple reflection occuring between the two sides of a pane is principally included in the formulas of the used alogrithm. So the real transmittance (i.e. the fraction of light which really es through the pane) is always smaller than the above transmissivity, which only denotes the fraction not absorbed in the interior of the pane. This resulting transmittance value is shown in the editor dialog as percentage value. Refraction of light rays is ingored, which is justified for the case of thin glass panes. Thus, to provide correct results, the material is intended to be used on single surfaces only (ReluxPro workplane primitve), thereby modelling the effect of a complete thin glass pane. See also chapt. 8.1 . In the accuracy range of common lighting simulations, the use of this material type is also justified for todays windows made out of multiple glass panes. Trans This material allows to model semi-transparent behaviour, shown e.g. by the so-called 'milky glass' or thin sheets of cloth. An arbitrary combination of glossy and diffuse reflection and direct and diffuse transmission can be set, limited only by the physical constraint that the sum of all coefficients must not exceed 1.0. (The editor interface and the internal algorithm guarantees that). In contrast to the glass type above, specularity here is not dependent on the incidence anlge anymore. Because of the great variety of possible settings, some experimenting with demo scenes is strongly recommended to get aquainted with the material type and the effect of the settings.

Note: If the diffuse fraction of transmission is nonzero, you need the indirect calculation (interreflections parameter greater than zero, cf. chapt. 5) to correctly evaluate all properties. Also, if the roughness value is nonzero, all transmission (diffuse and direct) is only evaluated by the indirect calculation. Dielectric A 'dielectric' material is suitable for modeling transparent objects like thick glass panes or water volumes, because here the refraction upon incidence and leaving the medium is considered. The index of refraction has to be greater than 1 for this effect to show up. The RGB parameters describe the transmission per 1 m path length through the material.

Note: This material generally is intended to be used for visualization purposes only. The indirect calculation still operates correctly, but the direct calculation (shadowtesting) cannot work properly anymore when rays through a 'dielectric' medium (see chapt. 1.3). Dependent on the geometric circumstances this can be more or less noticable. Perforated metal Being a material mix rather than one single material, perforated metal of course is also one of the semi-transparent materials. Two forms (with round and square holes) are provided by the material editor as part of the raytracer materials library. In the true sense of the word, shooting through the holes with the calulation rays becomes a problematic aspect, the more, the smaller the holes are. As rule of thumb, © Relux Informatik AG, Fit for Raytracing, 27.07.2011

MATERIALS

36/52

one can state that from hole-diameters roughly about 1 cm onward, the direct light algorithm is capable to consider the degree of transparency of the material correctly. So modeling the hole diameter greater than it is in reality (by keeping the hole/surface ratio unchanged) may be a necessary workaround to produce a correct simulation. For pure value calculation, replacing the material with a 'trans' with transparency according to the hole/surface ratio is another way, especially also, if accurate indirect contribution through the holes is to be determined. Luckily this is seldomly the case, though. Of course also a single surface (ReluxPro workplane) is recommended as geometry base, thereby ignoring the thickness, which is justified for usual sheet thicknesses in the mm range. Dependent on the geometric circumstances, scenes with perforated metal objects are one of the few cases, where a high luminaire subdivision (cf. chapt. 4) and high image quality settings in the expert mode render parameters dialog may be necessary to minimize artifacts.

7.4

Mirror material

The raytracing method has received much attention because of its capabilities to simulate mirror-like reflections by arbitrarily shaped objects in a very realistic manner. Thinking of this, it is astonishing at the first glance that the mirroring of illumination in contrast is a very difficult mathematical task indeed, and not at all easy to solve with raytracing software. Mirrored illumination is often neglectible and only plays a role in certain cases (e.g. redirected sunlight for daylight scenarios or special lighting designs). One should be aware that, unless one points directly onto a mirror with a narrow beam spot, visible effects of mirrored illumination are often very subtle and become apparent only at the second glance (see Fig. 22 )

Note: The mirroring effect for the illumination is only calculated for mirrors made

from ReluxPro workplane and cube objects, it does not work for arbitrary 3D objects carrying the mirror material, even if they internally contain only even faces. In this case, the material acts as a mirror only for the surrounding scene objects. The additional consideration of the mirrored illumination may quickly result in dramatically higher calculation times (esp. in scenes with a lot of luminaires). So the material definitely should be used with care. Multiple mirror reflections By default, the number of mirroring relays is limited to 1. In cases, where multiple reflections need to be simulated, i.e. in geometric configurations where one mirror reflects the light from a source onto another one, a special commandline parameter: –dr 2 needs to be entered in the commandline text field at the bottom of the expert mode dialog to activate 2 mirror relays.


37/52

Fig. 22 Scene showing effects of mirrored illumination, caused by the small elongated mirror panes at the walls.


SELECTED IMPORTANT TOPICS

8

38/52


This chapter draws attention on some selected topics, which can be of importance in individual cases, and which demand some consideration to avoid unwanted simulation results.

8.1

Modelling glass panes with volume or face primitives

Fig. 23 Glass pane as single polygon (left) and volume object resp. flat box (right) For simulations, it is generally important to adjust the degree of abstraction of the models according to the circumstances, the desired results and the used calculation method. In other words, one should work neither too fine, nor too coarse. A prominent example where this aspect comes into play is the modeling of glass panes for windows and skylights or other installations (e.g. separation walls made of glass). Of course each pane in reality has a finite thickness, and the light gets partly reflected multiple times on ing through both sides resp. inbetween the front and back 'interface' of the glass and surrounding medium. But in the context of a lighting simulation in architecture and interior design, calculating these detailed steps of light propagation is not appropriate. It is rather advisable to use a single face with an special material type, which correctly simulates the mentioned effects in total. So normally, in lighting simulations even multiple glazing is treated with a single face polygon carrying (in conjunction with the used material) the properties of the complete system. The standard ReluxPro windows and skylights use this approach. Now special care has to be taken when whole building models are imported into ReluxPro from external CAD programs (e.g. via the 3ds import). Here often all surfaces resp. objects appear as volume primitives. If a glass material is then assigned to those objects which represent the windows of the building, erroneous results are produced, as the material properties effectively act two times on each ing ray, one time on the front and one time on the back side. The situation is even worse when the special milky glass material 'trans' is used, because of the complex separation into direct and diffuse transmission. Moreover, it does not only lead to wrong results, but also to considerably higher calculation times. A check of the used model for this detail is recommended. When the model is established in ReluxPro, always the 'workplane' primitive should be used for glass and other transparent or partly transparent panes. © Relux Informatik AG, Fit for Raytracing, 27.07.2011

SELECTED IMPORTANT TOPICS 8.2

39/52

Invisible material

A principle problem of every simulations tool is, that certain features are offered, which may be configurable for a wide variety of tasks, but nevertheless contain limitations and cannot cover every possible scenario equally well. So in certain situations it can be helpful to tweak the tool a bit, or to use unorthodox approaches to tackle a tricky problem and thus extend the range of application of the tool. A nice example for this approach is the use of an invisible material. Objects which otherwise cannot be deleted from the scene may get assigned this material to vanish de facto in the rendering. (Take for example an open room geometry, which is nevertheless modeled based on the ReluxPro interior scene to be able to use the window precalculation feature.) The invisible material is configurable with the basic material 'trans' and the following parameters: colour: 1.0 (for red, green and blue)

specularity: 0.0

transmission: 100%

roughness: 0.0

specular Transmission: 1.0 Of course using this material is not 100% optimal, because the objects are still present in the scene, ray intersections occur only to continue the ray unmodified afterwards. But if it is used with care, the performance loss is tolerable, and the invisible material can be a quite helpful trick.

8.3

Modeling objects with curvy surfaces

The effect of the new option 'smooth edges' (see chapt. 3.2) depends strongly on the type of 3D model used. If the 3D object is present as volume object, the option may result in unwanted 'rounding effects' in the rendering.

Fig. 24 Influence of the modeling type on the appearance of 3D objects with activated 'smooth edges' option. On the left, the models are depicted in the wire frame display, on the right a rendering shows two luminaires made from these cylinder models. Only the right one displays the correct effect. © Relux Informatik AG, Fit for Raytracing, 27.07.2011


40/52

Take for example the 2 cylinder models shown in the wire frame display on Fig. 24. One is modeled as volume object with inner and outer surface plus the ring-like caps, the other is modeled solely with one surface. The rendering on the right shows two design luminaires made out of these models, the left one with the volume-type cylinder shows an unnatural 'rounding effect', whereas in the single face model, the smoothing only occurs radially, generating the expected appearance.

Note: The smooth edges option is included in the actual version as a first implementation and will be improved in future versions, providing more convenience and automatically appropriate smoothing effects also for objects containing both 'hard edges' and curvy surfaces like in the above example.

8.4

Luminous faces

Actually, the luminaire models don't come which so much detail as to model the complete geometry including the lamps. A lot of them are in contrary rather minimalistic. So often the problem arises, that the luminous face also has to represent the missing geometry detail in its effect in the scene. The luminance of the face generally scales with the LDC. Take for example a spotlight: no light emerges outside the cut-off angle, and this should be shown also in the appearance of the luminaire. But in reality, some part of the fixture often reflects some light, so this scaling with the LDC is limited to avoid completely black luminous faces which would look equally unreal. This minimum luminance can be adjusted by an entry in the initialization file within a [Raytracing] section. (See chapt. 2.5 for general information about setting these entries). Add the line: LumFaceConstFrac=

with a factor between 0.0 and 1.0 for .

The default setting (which is active if no entry is set by the ) is 0.01. This generally produces a good, realistic impression. In special cases an adjustment might be necessary (see Fig. 25).

Fig. 25 For luminaires with very narrow-beam LDCs, the effect of scaling the visible luminance is noticeable strongly (left image). If a brighter appearance even for view directions outside the cutoff angle is wanted, the mentioned threshhold value needs to be increased (right image, calculated with a setting of LumFaceConstFrac = 0.1)


SIMULATION RESULTS

41/52

9 SIMULATION RESULTS 9.1

Illuminance values

For all measurement planes of the scene, illuminance values are generated and displayed in various forms: tables, isocontour lines, falsecolor images and 3D mountain displays. For daylight scenes with CIE overcast sky also tables with daylight factors are produced. The resolution of the calculation grid is adjustable in the 'raster measuring areas' tab of the raytracing dialog. The raytracing module currently does not calculate cylindrical illuminance values. Results are accessible in the result tree overview in the node raytracing results.

9.2

Visualisations in the HDR format

The visualisations generated by the ReluxPro raytracing module are saved in a so called 'High Dynamic Range' (HDR) format, i.e. in contrast to normal images (JPG, TIFF, etc.) the files contain the full luminance information of the visualised scene view. Displaying these luminance information on the screen is of course limited by the dynamic range of the display medium. During the filtering of the images after the rendering, already an exposure adjustment is done, the exposure of the image display can of course also be further adjusted with a slider afterwards (accessible by a context menu from the image, click on it with the right mouse button). The same menu also offers the option of white balance adjustment (see the ReluxPro documentation). Displaying luminance information The luminance information contained in the image can be displayed interactively by checking the corresponding entry in the mentioned context menu. Luminance values appear then for the pixel at the current mouse cursor position. Saving/exporting of images The context menu finally offers the option to save the images. One has the option to either save the current view (i.e. the image with the actual exposure setting) in a standard JPG or BMP format, or as original HDR file in the RADIANCE HDR format. A lot of todays common image manipulating software is already capable of reading images in the RADIANCE format. Producing HDR visualisations for external applications A common use of HDR images is the interactive simulation of time-varying illumination scenarios and the effect of lighting controls. This is possible, because the illumination adds up linearly. So a complex lighting scenario can also be simulated by rendering several images with only selected luminaires switched on, and later mixing together the images into one, with different scaling factors for each one. This is analog to sound mixing in music. Evidently, one needs HDR images with the full, original luminance information for that purpose. An example of such a lighting control software is VIVALDI by Zumtobel. The ReluxPro raytracer HDR images can be imported into the VIVALDI tool.


SIMULATION RESULTS

42/52

Fig. 26 Idea of interactive simulation of lighting controls by mixing several HDR images.


SIMULATION RESULTS

43/52

Note: When several images with partial illumination are generated for later mixing,

the automatic exposure adjustment should be switched off in the expert mode dialog, to avoid having images with different prescaling (exposure) factors.


QUICK REFERENCE FOR EXPERT MODE PARAMETERS

44/52

10 QUICK REFERENCE FOR EXPERT MODE PARAMETERS As a guideline for the daily work, a short description is given for each of the raytracing expert mode parameters. Hints for value settings cover the most usual cases. One may get away with lesser settings, special tasks may need further attention. References to the relevant chapters are given for looking up further details. Scene Export Octree resolution

Cryptic parameter controlling the raytracers internal space subdivision structure. No need to set this manually, except when scene conversion failes, i.e. when an error message appaars. Then try augmenting stepwise by factors of 2.

Memory Optimization

No need to set this manually. Only try to deactivate it for the spec. object type if export error messages occur. See chapt. 0.

Smooth edges

Activates smoothing for 3D mesh facets for selected object types (needs above Mem. Optimization feature). The corresponding option has to be checked also in the properties of the individual object. Use only if 3D objects really are made up out of curvy surfaces. Artifacts may occur for rectangular geoemtry. See chapt. 3.2.

Image Quality

Influences a.o. image filtering (anti-aliasing and pixel noise smoothing), but also some internal second-line calculation parameters, e.g. ray continuation thresholds etc. Should be set at least to 'medium' for value calculation.

Autoexposure

Performs automatic exposure adjustment on images. Switch off in case that several images are produced for use in later analysis software, to avoid different prescaling factors. See Fig. .

Direct Calculation Soft shadows

Keep this checked to avoid multiple shadow boundaries.

Luminaire subdivision

Augment this to produce softer penumbra areas with lesser pixel noise. See chapt. 4.

Commandline option –ds

If the lum. subdivision is set to 'high', this code enforces an even finer subdivision, for generating maximally soft penumbras. Chose a value between 0.05 and 0.01 for .

Commandline option –dr 2

Has only an effect in scenes with objects carrying the 'mirror' material: activate multiple relays for the lighting calculation. (needed only in very special geometric configurations. Use with extreme care. See chapt. 7.4).



45/52

Indirect Calculation Generally, more than one parameter needs to be adjusted to overcome problems. Most often the best point to start with is augmenting the initial number of sampling rays, followed by raising the interreflections. Interreflections

Number of ray bounces followed for analysing the indirect illumination. Set at least to 2-3 for normal visualizations. Augment if image appears splotchy. Complex indirect scenarios may need 4-7. Set at least to 4 for acccurate value calculations. Settings higher than 10 don't make much sense. See chapt. 5.2.

Background illumination

Sets a global value for a background radiance throughout the whole scene (in three different color channels). Makes only sense in scenes with more or less uniform lighting. Typical values are 0-0.001 for exterior night/artficial light, 0.01-0.1 for interior artif. light, 0.01-1 for interior daylight and 1-10 for exterior daylight scenes. If the scene contains many unlit areas, better use a zero value avoid erronous 'adding' of light. See chapt. 5.2.

Spatial resolution

Controls the interpolation distance between calculated values of indirect illumination. Default is 0.4 m, refine this if images show splotches concentrated locally around geometry detail. Has a high influence on calculation time. Values below 0.05 m should be avoided. See chapt. 5.2.

Number of initial Controls how detailed the surrounding gets sampled for sampling rays determining the incoming indirect illumination. Augment if images appear splotchy. Minimum is around 600-1000, complex scenes may need between 2000 and 6000. Extreme settings above 15000 should be avoided. See chapt. 5.2. Soft interpolation

Activate this if all other indirect parameters are set accordingly, and images still appear splotchy. Option still in a test state. May be helpful in image smoothing in pure indirect light scenarios, or scenes with sunlight patches or otherwise highly variant illumination. Intended for visualizations only. See chapt. 5.2.

Commandline option –am 0

If huge external obstructions are modeled, so that the area of interest (room, measurement plane) is small compared to the whole scene, set this option to ensure sufficient accuracy. May also be helpful in scenes with very scarce, locally concentrated illumination. See chapt. 5.3.



46/52

Window precalculation Window precalculation parameters are only in effect for ReluxPro interior scenes. See chapt. 6.4 for all parameters. External interreflections

Determines, how much indirect light enters the generated window LDC. In daylight scenes without sun, a setting of 0 means external buildings act as pure obstructors.

Number of rays

Determines the resolution of the window LDC. Augment this if complex external geometry is present.

Number of sample Controls if window LDC is calculated based for the window points center only or for points distributed over the whole window pane. Augment this for complex external geometry, thick walls, etc. (Maximum 64) Subdivision threshhold

Window panes get subdivided when their sidelengths exceed this value, to avoid too big areas for the virtual window sources. Set to a reasonable value dependent on the whole scene. If segments are big, shadows will appear unnaturally. E.g. in normal interior scenes, patch sizes with more than 1 m side length should be avoided.

View clipping planes Commandline option –vo

Sets a foreground clipping plane perpendicular to the view direction at the given distance (in m) from the current point of view. Set a positive value for . See chapt. 2.4.

Illuminance image calculation Commandline option –gi

Calculates a greyscale image where each pixel contains the illuminance value at this point rather than the luminance. Works correctly only on the opaque surfaces in the scene, not on transparent ones, light sources and the sky. !Caution! Somewhat unorthodox, the Relux interface still writes the luminance unit (cd/m2) in the 3D falsecolor image legend or the values picked with the mouse cursor, but with this option values mean Lux of course.


TROUBLESHOOTING

47/52

11 TROUBLESHOOTING A raytracing calculation is a process with several stages. Thus errors can be subdivided into three different categories: export errors, problems during the geometry conversion and errors during the calculation. During a raytracing run, the little information window always displays the current state of the process.

11.1 Export errors As mentioned in chapt. 3 , the whole scene geometry has to be exported into the raytracers input format before starting the calculation. In the following section, some error messages are covered which may appaer during this phase, together with hints for fixing the problem. In general, these export errors appear in conjunction with a specific object. If the message is simply acknowledged, the export continues, however, the object in question either is missing then in the rendering or will appear not completely functional in some other way. (e.g. a luminaire without light emission). Alternatively, the has the choice of cancelling the whole export by clicking on the 'Cancel' button in the error message. Error: Too many triangles locally accumulated Too many triangles In this case the conversion of the named 3D object fails because it either contains too many triangles or is modelled in such a way that it is unsuitable for the raytracers internal mesh format. This can happen for very big 3D objects. The best way then is to divide the geometry and import the parts as separate 3D objects into ReluxPro again. Alternatively, you may try to deactivate the memory-optimized export (see chapt. 3.1) for 3D objects. Then the internal storage format is different and the conversion might work. Another possibility is to delete some partial objects in the 3D object within ReluxPro, preferrably those which are not important for the simulation. Of course the object needs to be present in a suitable way for that method. Editing the object is best done within the library itself, befor placing it into the scene. Error: No triangles in the 3D object or triangles below a threshold size The 3D object is either corrupted and contains no usable data anymore, or it is modelled with an inadequately high level of detail, resulting in all triangles of the mesh being of sizes much lower than 1 mm side length, failing to meet the tolerance criteria of the converter. Here also the object needs some editing in an external modeler prior to a re-import into ReluxPro. Of course, deletion of partial objects (s.a.) is also an option if the problematic parts of the model are localized in these parts.


TROUBLESHOOTING

48/52

Error: Invalid luminaire model The raytracer has higher demands on data integrity for luminaires, compared to normal geometry, because for the former objects, the correct light emission is important for the simulation, too. Beneath the above mentioned geometry export errors, further problems may appear, leading to the corresponding message. Of course, the message does not always mean a real error, often it is more like a kind of warning message. There are luminaire models which do not fulfil the specifications exactly, but nevertheless can be used without restrictions in the simulation. A common example for that are models which contain some empty partial geometry (or 'node'). Here the empty node can be deleted by the by editing the luminaire in the products library. Another case which may appaear is the one of a correct geometry, which appars fully in the rendering, but without light emission due to an invalid model assembly.

11.2 Errors during geometry conversion The somewhat cryptic term denotes the preparation step performed by the raytracer to set up the necessary internal data structure for the final calculation. If this step should fail exceptionally, the calculation is not possible. But changing the model setup can resolve the problems in almost all cases. Error: Too many objects in the scene Too many objects locally accumulated in the scene Unfortunately, an exact assignment of the error messages to the real origin of the problem is not always possible. If one of both message above appear, generally one of the three possibilities below. or a mixture of them, is the case: 1) The scene really is too big, containing too many objects (in relationship to the available memory on the system) 2) There is a resolution problem, i.e. for example, in a very big scene of several 100m side, also very tiny objects like self-drawn ReluxPro cubes of some mm sidelength appear clustered resp. locally grouped in heaps. 3) Several 3D objects with almost identical bounding boxes overlap at the same point in space. In the second case, stepwise doubling the octree resolution parameter (cf. chapt. 3.1) can solve the problem. In severe cases, the tiny objects have to be collected into one bigger object with the option 'generate furniture' In the third case, it is not always easy to identify the problematic overlapping objects, because what counts for the algorithm is their bounding box. Generally, it pays off to divide huge objects into several parts, and also to avoid placing many imported 3D geometry objects onto each other at the same position in space. Beneath the error message, a stopping progress bar during the process stage 'geometry conversion' is a further hint for the overlapping case.


TROUBLESHOOTING

49/52

It is important to note that general intersection or partial overlapping of objects is no problem for the raytracer at all. The real trouble starts when almost identical objects (at least judged from their bounding boxes) appear clustered, i.e. more than 5 times, at the same spot in space. One might have erroneously duplicated objects in the same location, for example. Another scenario, where this overlapping problem might appear, is the import of whole building models from external CAD modelers in parts, cf. Fig. 27.

Fig. 27 Common strategy for building modelling with CAD software: identical building parts are collected in layers. If these layers then are exported as separate 3D objects and reassembled in ReluxPro after import, in extreme cases a situation can occur, where several objects with exactly equally-sized bounding boxes overlap at the same position in space.

11.3 Errors during the calculation After the first two hurdles are taken, the real calculation starts. In this phase, the following messages might appear: Error: No calculation values Defaultwise, in ReluxPro an option is active which leads to ignoring calculation points if they happen to lie within objects. The reason is, that design objects like furniture do not compromise the uniformity results of a lighting analysis. Within opaque objects, no illumination can happen anyway, so the zero values have no significance at all. If whole building parts are imported as 3D objects, or if measurement planes lie within transparent objects, the ignoring of points does not make sense anymore, so it has to be deactivated in the property page of the measurement plane in question. In extrem cases the ignore option may otherwise lead to the removal of all calculation points from the list. Error: Invalid commandline option Spelling mistake during use of special commandline options within the expert mode dialog (cf. chapt. 2.3) © Relux Informatik AG, Fit for Raytracing, 27.07.2011

TROUBLESHOOTING

50/52

Error: Consistency error Error in process communication a.o. These errors, if they might occur during a calculation, are generally of a severe kind and, with e.g. the exception of a program abort due to lack of memory, cannot be influenced by the . Evidently, even despite of thorough tests, program faults never can be excluded completely. Also, because of the several hundreds of thousands of products within the databases, or the option of arbitrary import of 3D models, occasional inconsitencies in geometry or material description might occur. In such cases, one should not hesitate to the Relux technical . Sending the complete error desription including the additional cryptic information appearing in the message box (if present), e.g. with a screenshot, greatly facilitates the analysis.

11.4 Erroneous results More exactly, this chapter should be named 'unexpected results', as in most cases the real cause of the error lies in wrong input and/or parameter settings. Prominent examples of such unexpected results are of course image artifacts. In the following, some of these are mentioned. Effect: Shadow artifacts, 'high pixel noise' Normal intersection and partial overlapping of 3D objects in the scene is no problem for a raytracing engine. But for luminaire placement, such overlapping must not occur, otherwise the shadow test cannot operate correctly anymore. This results in characteristic artifacts shown in Fig. 28 for the example of a downlight which is partly shifted into the ceiling.

Fig. 28 Pixel noise resp. shadow-artifacts occuring when luminaire geometry intersects with other scene geometry.


TROUBLESHOOTING

51/52

Effect: Unnaturally bright image with low contrast If one has chosen a background illumination setting (see chapt. 5.2) which is too high for the current scenario, unnaturally bright images will be generated. Chose a lower value then. This topic is important in all scenes which contain many areas with sparse or even no illumination, (e.g. exterior scenes, cf. Fig 29, left) . If all luminaires are switched off, or the models are faulty, producing no light output, or no luminares are present at all (this means also daylight calculations in rooms without windows) , the result looks very similar (Fig. 29, right)

Fig. 29 left: unnaturally bright image resulting from an inappropriate background illumination setting in an exterior scene, right: scene with completely switched off luminaires. Normally, the image should be pitch black then, only the background value causes the objects to be displayed at all. Effect: Luminaires appearing black / with dark luminous faces If a luminaires visible luminous face appears pitch black or partially black in the image, the model is faulty (Fig.30). Here only a corrected data set provided by the manufacturer can solve the problem.

Fig. 30 Faulty luminaire model with black luminous face. The model still can act as lightemitter in the scene, because the lighting calculation is performed separately. © Relux Informatik AG, Fit for Raytracing, 27.07.2011

TROUBLESHOOTING

52/52

Note: If the luminous face appears relatively dark, resp. not as bright as expected,

the reason is that one looks into it from an angle in which there is only low emission due to the LDC. In these cases, however, the displayed luminance can be adjusted with a further parameter, see chapt. 8.4.

Effect: Light leaks / interpolation errors As explained in chapt. 5, the indirect calculation process is a complex issue. Occasionally, image errors can occur, which most often result from insufficient parameter settings, but in rare cases also can be triggered by suboptimal models, which are difficult to handle for the algorithm. Chapt. 5.2 and 10 state, that the number of rays, and also the interreflection setting, should be risen in case of generally splotchy images, whereas a finer resolution value is needed in case of splotches appearing locally around geometry detail. Fig. 31 now shows an example of inappropriate modeling, where even this setting cannot remedy all image errors.

Fig. 31 Example for suboptimal modelling of a daylight scenario. The walls are only made up out of infinitely thin polygons, some surfaces stick out, too (at the outer edge of ceiling and walls, also the inner floor extends into the outside area). In such situations, the indirect algorithm runs into difficulties in separating the relevant areas for a correct illumination calculation. This is noticeable by splotches at the edges, as can be seen in the image on the right. So providing good, 'light-tight' models is also important for achieving reliable and correct results. The issue is especially important for daylight projects. Also, building models imported as 3D objects should be checked thoroughly for this criterium. Of course, the presented example is only one of a lot of possibilities. So gaining experience in this matter is vital to quickly identify possible problem points in each new project.


Raytracing Manual 4gi5e

Overview 26281t

More details 6y5l6z

Related Documents 3h463d

Raytracing Manual 4gi5e

Manual 425y5x

Manual 425y5x

Manual 425y5x

Manual 425y5x

Manual 425y5x

More Documents from "Ber Salazar Jr" 5q3w63

How To Become A Pee 335m2i

Kitchen Spot Audit j5d1p

Caliraya Published Rate.pdf 3fy

Simplex - Truealertproductselection 4n1i6c

Avoid_the_migration_minefield.pdf 2u1618

Raytracing Manual 4gi5e