I started a project which I had in my mind for quite some time now. Triggered by an article about product visualization, which uses fake lighting, I decided to finally give it a try myself.

I created a thread in blenderartists to show the various stages of my progress and will probably post a tutorial, if anything useful come out of this.

Right now this technique seems only applicable for studio setups and animations, where no camera movement is involved, but maybe I can come up with extended applications.

Here is a quick animation to show the technique

Fake Lighting Test from loramel on Vimeo.

For this material breakdown I chose a gold material based on the material I used in my crown project.

The goal is to make a gold material, which can be seen on old jewellery exhibiting some worn look and is not the typically clean shiny gold look you get with modern jewellery.

One remark before I start: Its important when developing a material, that you use a model for preview, which resembles the final object. Just using a simple sphere or a plane will most certainly lead to longer development cycles. The lighting shouldn’t be too simple too. Just using a single point or meshlight will not suffice for the most materials, which exhibit some sort of reflection.

My sample scene for this breakdown uses a very simple broad bracelet and I use 3 meshlights to light the scene.

Selecting the base material

The immediate idea is of course to use luxrender’s metal material with the built-in gold setting. If we use it we get the following image

Besides setting the roughness, there are no parameters to fine tune the actual color texture. The metal material gives too light a result for our intended purpose. The roughness is quite ok though.

This leaves us with the next basic material type – shiny metal. This material has two channels: reflection and specularity, both fully configurable with textures. Lets assign a quick yellow color for both channels, set the roughness to 25  and see what we get.

Now that’s not really what we are aiming for. Its too shiny (as the material’s name already suggests). What we have here is in effect a compound material. Reflection as well as specularity is a reflection of light, where specularity has the additional benefit of a roughness parameter. Lets see a close-up of the above render

Using both reflection and specularity results in a rough material which seems to be coated by some shiny substance. The reflection channel is always perfectly smooth, so as long as we use it, we will always end up with a smooth shiny surface. The solution is to turn off the reflection by setting it to 0. With the roughness value  we should have all the control we need to describe the material.

See below for the same shiny metal material with reflection turned off and using different roughness settings.

To be honest, I cannot image a situation, where I could make use of the compound nature the shiny metal is delivering, because in the end it will collide with the material setup I have in mind, but who knows what time will bring …

Defining the basic color

Now that the reflection issue is solved we have to concentrate on the basic color for our gold. This sounds trivial but I will spend quite some time on this topic.

The simple approach is of course to directly assign a color to the specular channel. For this project I won’t use this approach, but will rather concentrate on the gaussian texture. So its good to look at what this gaussian really is.

A gaussian offers the following controls

You have a wavelength slider, which is conveniently placed under a spectrum image, so you won’t have to guess which color belongs to a wavelength. The whole visible spectrum is available.

This texture produces a wavelength distribution in a gaussian bell curve, the maximum being the selected wavelength. The width parameter sets the standard deviation of the bell curve, thus actually telling luxrender how many wavelengths left and right from the maximum are to be used. This now sounds very theoretical, so the best is to actually visualize it

This series was rendered with a gaussian texture set to a wavelength of 586 nm, which is an orange hue. With a width setting of 20 the orange hue is quite visible. The broader the width the more yellowish, then slightly green and at the end almost white the reflected color gets. The broader the width, the more frequencies from the visible spectrum gets reflected. With a width of 150 we have already a very high percentage of the visible spectrum covered, hence the almost white color.

Width 35 and 55 are more interesting. From our main wavelength to the right, only red hues are available, but to the left there is yellow and green in the immediate neighbourhood. Broadening the width  from 20 to higher values will add more different hues from the left then it does from the right. Yellow and green in this case, therefore we see the shift from the orange to the yellow/green.

I like to use the gaussian texture, as I think it gives a richer appearance to the final render ( but that is maybe my conception only )

What the gaussian texture lacks in comparison the the constant color, is a control for the saturation of the color. You have the energy slider, but this only acts as gain and not as saturation control. To get this, we have to build it by our own and thanks to lux, this isn’t all too difficult.

We use the mixing feature for textures. I setup a mixing like this

With this we are mixing a constant white color with the gaussian and using the amount parameter to blend these two textures. Lets have a look at a series of renders for different values for the mixing amount.

This produces quite acceptable results, especially the render with the amount of 0.85 looks promising for our gold material. As a side note: the mix amount for textures works just the other way round as it does for mixing materials. A value of 1 in the mix amount will use 100% of texture 2 when mixing textures, but will use 100% of material 1 when mixing materials. This may be resolved in future releases ( maybe even 0,6 final), but for the time being this can cause some headache.

For the basic gold color I settled on a wavelength of 586 and a mixing amount of 0.86 giving the following starting point

Adding some bumps

This gold material is ( besides its specularity roughness) perfectly smooth, and with this perfectly recognizable as being cg. We need some bump mapping to counter the perfect look and give it a more hand crafted look.

A word of caution before we begin:  What I observed in my projects so far is, that the real look of a material, when it comes to subtle bump mapping is only fully visible at quite high samples/px rates. The render above was intentionally rendered up to 2000 samples/px to show its too perfect surface. The noise introduced with lower sample ratios tend to suggest some noisy bump mapping which in the end is not there. So to get the bump mapping right, you have to render it to high samples/px ratios to be sure its ok.

A popular approach to achieve something like a hammered look is to use a voronoi texture. So lets start with it

Not so bad, but still too regular. To counter this I use the blender distorted noise texture and distort the voronoi texture with an improved perlin, adjust the bump amount and get the next render

Next I give the bump a specific direction and end up with this setup and result

Again a word of caution: I am using here a non uniform transformation to have the texture stretched along the x-axis. Now the x-axis in discussion is the global x-axis. If you copy the object, rotate it and assign the same material to it, you will get the stretching in the same global x-axis, making the object look weird. As long as there is no object coordinate system for the texture transformations, all these transformations have to be done very carefully and sometimes cannot be done at all. Needless to say, that this disqualifies its use in animations …

The material looks better now, not so smooth, but lets have a closer look

Again, although having a bump map, we have too clean a look. Real world noise, dirt and bumps tend to have a fractal character. The closer you look, you will always see similar patterns. Here is where you have to decide how this material is going to be displayed. Is it low res or seen from far away, you will get away with the setup we have right now. But rendering at higher resolutions where the material takes some area of the render will demand for more work on the bump mapping.

I take a simple approach and aim at adding some small scale noise. But before actually adding the it, we design it as single noise to better see its appearance.

I found the blender musgrave texture to be quite flexible in creating noise, which isn’t too even in itself.  Lets start with the noise texture itself.

The important controls ( besides the musgrave ones)  are the contrast and and tex1/tex2 sliders. The contrast lets you control the amplitude for your bump map while the tex1/tex2 sliders define the limits in which your bump values will vary. You could also set the amplitude with the tex1/tex2 but I prefer it this way, especially as these parameters are quite useful when combining textures.

Now that the single noise bump is defined lets combine it with the previously developed voronoi texture. The simplest approach for this is to mix these two textures. Doing so we get the following.

The two textures are now mixed with a ratio of 1:1, the small noise still too big. What bugs me though is that the small noise is uniformly distributed over the voronoi texture.

Its probably more plausible, if the most elevated portions of the voronoi texture are more smooth than the lower regions. The higher ridges should have more contact with the environment and exhibit a more polished look.

To accomplish this, I restructure the texture mixing in a way, where I use the voronoi texture as amount to mix the small noise texture with a constant bump value. See here

The effect is now more or less what I intended, but lets have a look at the texture setup to analyse why this is so.

The voronoi texture acts now as modulator (amount value) between two textures. The extreme values for this texture control the amount of each of the two textures. The range is set to [0..1] but the contrast is lowered, so the maximum value won’t reach 1, which translates to texture 1 (the small noise) not being used as pure texture.

The musgrave texture has a very low contrast (bump amplitude) and the range is set to [0.18..0.23]. Texture 2 is a constant value of 0.3. With this setup the maximum amplitude will be from 0.18 to 0.3.

When the value for the voronoi texture (the modulator) reaches a value of 0.0, only a constant of 0.3 will be the result of the mix i.e. no noise is mixed in. This are the elevations of the voronoi texture. Going gradually to higher values will mix more amount from the noise texture to the result, thus adding noise and lowering the average bump value.

This means, to control the overall amplitude of the resulting texture you have to adjust the levels of the two textures.

Now lets have a look at the whole object

For this material breakdown I will leave the bump map as it is now, but you could spend days here adding subtle tweaks to it.

Color revisited

I will use the created bump texture as base for a slight color variation. The idea is again, that the lower areas are less shiny ( especially the small holes ) than the elevated ones.

To do this, I will use the same technique as with the bump map, but now I will use the complete bump texture as amount texture for mixing the basic gold material with a darker one.

First lets find a good variation of the basic gold material to act as darker less shiny one.

I use a checker texture to emphasize the different materials. Now we setup a more elaborate texture for the basic gold material. Instead of using the simple checker texture we use a variation of the bump map texture.

This render uses no bump mapping to clearly see the effect of the color variation.

You can see here, that the size and composition of the texture is the same as the bump texture, but has other values regarding its contrast and limit values.

The important values to have full control over the balance of dark/light areas and their distribution are the last 3 constant values (now at 0.4, 0.2 and 0.0 resp ). Its not immediately obvious, that this is the case, and its now starting to get very hard to see the structure of the material setup.

A graphical representation would be a real blessing, but alas, for the time being we have to work very concentrated. It also helps to actually print out the setting and make notes and groupings on the paper and emphasize the important parameters.

Now lets add the bump map again and have a look at the result.

That should do for the basic material, but I have still another bit planned for the gold material -  some additional dirt independent of any bumps.

Adding dirt

For this I set up the following dirt texture

This a mix texture of 2 musgrave textures scaled and rotated and the mixing is controlled by just another musgrave texture. Maybe overkill, but I like it

Now I setup a mix material with the actual gold material in slot 1 and a very simple gold material in slot 2. The simple material is basically the material we started from, a shinymetal with a gaussian assigned to the specular channel, but the specularity is now lower and the roughness is set to 2.0.

Now just by varying the specularity of the dirt material and the bump map on the first material we can create a wide variety of gold materials.

Well, thats it for a start.

The important point is to keep yourself organized. This is tedious with increasing complexity of the material, so know your building blocks, know how they contribute to the final result and keep notes about them, for otherwise you will have a hard time deciphering the material setup after only 2 weeks.

Hope this is helpful for someone.

Next material breakdown will highlight composite materials.

I will start a series of material breakdowns, where I show  step by step the principles of how I created the materials from the Austrian Imperial Crown.

This is by no means an absolute guide to the lux material system, but rather an insight into how I use the options given to me by lux.

Lets start with a pearl material.

Glossy Material

Lux offers some basic material types from which you can choose. In our case with the pearl, its a good start to use the glossy type, which offers a diffuse and specular aspect for the reflected light.

Using default values for a glossy material results in something like this.

Lets just concentrate for a moment on the specular part alone, to see how its parameters affect the appearance.

The basic parameters for you to change are the color, the amount and the roughness (exponent value) of the specular part.

Specular Amount Variation

See here for a variation on the amount. Diffuse is fixed at 1 (pure white) roughness at 500.

You can see the character of the sphere starts as a matte ball, then changes to a plastic ball and ends up being almost metallic in appearance. It is important to keep in mind that the specular part is indeed a reflection, not just a specular highlight trick. This means, that when turning on specularity, you also turn on real reflection, best seen with the example on the right. To speak in blender terms, glossy specularity is comparable to the he glossy reflection available in blender internal.

Specular Roughness Variation

Now lets vary the roughness part. This is done by manipulating the exponent value.

For this  example  I’ve set the amount to 0.5 pure white, diffuse is at 1.0 pure white.

Again it starts at a matte material and gets more and more polished until it could be used as starting point for a billiard ball. Keep in mind, that we used an amount setting of 0.5 so we are more on a plastic side, thus the billiard ball effect. What we already see, is that at a setting of 50 we already have a pearl like effect.

Specular Color Variation

Now for the last available specular parameter, the color. I use a value of 0.75 for the amount and 1000 for the exponent while varying the saturation of a yellow hue.

Again, the specularity is a real reflection, so the reflected environment now gets a tint. Colored reflections are typically a property of metallic surfaces, so the character of the ball now moves more to some metallic aspect, though not really metallic in itself.

Whats also interesting to note, is that we also get quite a pronounced blueish tint.  Just using color for the specular seems to effect the diffuse color as well. The reason for this is the following ( a good educated guess from my side) :

An object can only reflect the amount of light it receives. The two channels diffuse/specular have to use/share this amount of light. If light gets reflected by the specular channel it will not reach the diffuse part.  Setting the amount value of the specular channel to 1 for any hue results in no light for the diffuse reflection in that color. For a full spectrum of colors, this means if you steal a color (specular color) the diffuse part will get a tint in the complementary color.  In our example the complementary to yellow is blue, thus the blueish tint.

Diffuse Amount Variation

So much for the specular part. Now lets have a look at the diffuse one. I set the specularity to 0.5 and 500 roughness and just vary the diffuse from 0 to 1.

What we see here, is again a shift from a metallic character to a plastic one. Combining this with the observations made before, I think its valid to state that the ratio in intensities between diffuse and specular control the materials character.

diffuse < specular : metallic

diffuse > specular : plastic

Also keep in mind, that the specular channel gets the light first. All light reflected by the specular part wont reach the diffuse part.

Basic Pearl Considerations

Now back to the pearl. Pearls have certainly some plastic aspects in it, but have also some pronounced reflections with a hint of some metallic character to it. Thus by applying the previous observations, I would say a ratio of 1:1 with not too smooth a surface should provide a good start.

Basic Pearl Setup

That already has some pearl character to it. Still has no color, so lets add some. The basic color is given by the diffuse part. So for a starter lets just add some slightly yellow/orange color to the diffuse part.

We remember, that using a colored specularity enhances the metallic effect, so lets add a faint blue to the specular channel. This not only gives a slightly more metallic appearance, but also should enhance the yellow part of the diffuse channel, since blue is the complementary color to yellow.

Adding thin film interference

Pearls are made of countless very thin layers of mother of pearl, produced by the pearl oyster. These thin layers create the same effect as a very thin film of oil on water, a colorful interference pattern. The effect isn’t too pronounced on pearls but its there.

Luxrender offers a thin film property for materials, but unfortunately only for completely smooth surfaces (glass, mirror). Mixing such a material with the actual glossy one can bring in the thin film effect, but destroys the materials character by adding a perfectly smooth reflection. So we have to come up with some way to fake these thin film effect.

I decided to simulate such an effect by adding a color pattern consisting of slightly blue and pinkish colors. Now the immediate idea to do this, would be to add a texture to the diffuse part with the desired colors. I decided to use another approach.

The glossy material offers an absorption attribute, which lets you specify colors to be absorbed by a certain amount. If a color gets absorbed you will see the complementary color.

Lets setup an absorption texture for our pearl.

I use a blender marble texture to modulate between green and orange at a relatively low contrast and very low turbulence. The basic color is set to a dark grey. This main color gets multiplied with the texture, so as long as there is no hue in the the main color, it can act as gain for the texture.

The absorption depth is set to 1, which in itself is a gain value for the amount of absorption done. The setting shown above is too strong of course, so we will lower the absorption depth and add it to our pearl material. This is another reason I chose the absorption technique. I have one adjustment parameter to control the amount of the thin film effect. To achieve the same control using a texture for the diffuse part would involve some elaborate texture mixing.

_________________________________________________________________________

Adding another material

The current setup is already quite ok, but the specular, while good at the highlights is too pronounced at the other areas. I imagine having a sphere inside the pearl with a much smoother specularity, something like this.

I just took the previous material setup without the absorption and decreased the roughness value to 5, giving a very smooth specularity, which adds to the diffuse part, but not in a completely diffuse way.

To have those both aspects together in one material, lux offers the mix material. The mix material has two slots for input materials and an amount parameter, which controls the blending of these two materials. Input materials can be any material, except light or portal.

Now lets mix these two materials.

Using the mix amount parameter we can now look for the best balance for these two materials. See below how the material behaves when going from low to high values ( 0.25, 0.5, 0.75 ). A value of 1 has 100% of material 1 and a value of 0 100% of material 2.

Lets settle on a value of ~0.65.

What I also did in my crown project, was to add a bump texture to the second ( more diffuse) material. The idea was to simulate a slightly irregular layer beneath the top reflecting one. So I use a relatively big texture size to have the modulation comparable to the size of the pearl itself ( the dimension of the pearl in this example is 0.7 blender units ).

I use a blender clouds texture with smooth noise and a noise depth of 2. Below is a variation of the texture noise size to find a good value for the bump’s size. I exaggerated the bump setting (0.1) to clearly see the bump distribution.

Lets settle here on the 0.5 texture size. Now lets add this bump map to the second material slot and have a look at the result.

One should experiment with the bump settings to find a good value for the bump amount, but I have now chosen a value of 0.01 for this setup.

What about SSS

Real pearls exhibit a certain amount of SSS (subsurface scattering), which enhances the glow of a pearl and makes it appear softer. Luxrender as of version 0.6 has not yet any SSS to offer, it’s planned for a later release. So again, we have to find a way to fake SSS.

A way to do this is by using a mattetranslucent material type and mixing it with the existing material at hand. A mattetranlucent is a matte material with an additional translucent part to it. The translucent light shines diffusely through the object, but no real scattering is applied.

We want the translucency to have the color of creamy white, very similar to the yellow hue we used before. A pure mattetranslucent looks like this

We have now to mix this material with the already created one. As you see, there is no real limit to the stacking of mixed materials, which on the one hand is extremely powerful, but on the other hand very confusing. The multitude of parameters added with each mix material is enormous and cannot really be overseen. So the advise here is to adhere to the KISS ( Keep It Simple Stupid) strategy, and always test the individual building blocks separately against the attributes you want to achieve.

Ok, so lets mix the translucent part and make a test series to get a good mixing amount value.

TThe above series use 0.5, 0.7 and 0.9 resp.

I choose a value of 0.79 and have arrived at a final pearl material.

with the following setup

link to lbm file

Conclusion

To recapitulate: this pearl material consists of 3 major building blocks, each having it own attributes, which add to the overall pearl appearance. This is a good starting point for further adapting this material to an actual scene, light setup etc. It is not finished, but should act as solid base. Once all the elements are defined and understood you can create a whole spectrum of pearl materials, suitable for your project.

This concludes this material breakdown. I hope someone can get something useful out of this. Comments and suggestions are of course welcome.

Next material breakdown will be based on the gold material used in the crown project.

Its finally done.

This project took me the better part of the last 3 weeks to finish in my spare time. It was an exercise in modeling and an  exploration into the possibilities of luxrender, especially now as the new version 0.6 is almost there.

This project will form now the base for some upcoming breakdowns of several aspects mainly aimed at luxrenders material system. So stay tuned …

As always the project files are all downloadable from my 3D showroom. But one word of caution. This is a huge project in terms of needed resources during rendering. It will use up to 3.7 Gb of memory when rendered in full resolution ( 1600×2000 ). The mesh setup alone takes 1.2 Gb. This means that you must have a 64bit version of lux running to be able to actually rerender it. Just browsing it in blender should be fine with a 32 bit version as well.

So I postpone the final part and use my current project ( luxrender forum wip, blender artists forum wip) to learn more about the material system and render options lux has to offer. By then version 0.6 should be ready.

So stay tuned …

This is the second part of my obervations on luxrender made during my last project.

Materials

This is the area where I spent most of my time in getting the blender scene transferred to lux. The material system is quite different from blender’s and I had to try various approaches to get the look I had in mind. This process is by far not finished, as I am currently just scraping at the surface. But some general findings can be shared.

Basic Material Types

Instead of having one base material where you then specify all the diffuse, specular, reflection etc properties, lux offers some basic material types to start with, covering totally rough (matte, mattetranslucent), metallic ( metal, shinymetal), shiny (glossy) and refractive (glass,) types.

Depending on the base type you may specify further attributes. All attributes can be defined by using textures, giving a wide range of possibilities.

Further on you can mix any two materials with a fixed amount or again controlled by a texture. Doing this recursively gives you an impressive tool to create really complex materials.

Here an simple example of mixing a matte with a shinymetal material modulated by a checkerboard texture.

Textures

As with any material system, textures play an important role in defining several surface properties. In blender you can have dedicated maps by assigning a texture to influence a certain material property.

Using these mappings results in diffuse, normal/bump, specular and transparency maps. All these maps have their counterpart in lux as well, but are achieved a little bit different:

• diffuse maps: those base materials with a diffuse element (matte, mattetranslucent, glossy) can have a texture assigned to modulate this property, thus acting as diffuse map.

• normal/bump maps: Each material has a separate bump property, where you can assign a texture to define the bump map. Lux currently does not support normal maps, so you have to use bump maps instead. I had the problem to convert the normal maps I had to bump maps, which the gimp normal map plugin supposedly should be able to generate, but the quality leaves a lot to be desired. After some search on the internet I found a very handy tool ShaderMap, which in its CL version is free to use in non commercial projects. Its Windows only but runs quite well under wine.
• specular maps: Those materials with a reflective property (glossy, glass, roughglass, shyinymetal) can have a texture assigned either to the specular or reflective channel. This would be the direct approach to specifying specular maps. The materials glossy, roughglass and shinymetal offer the additional possibility to use texture maps to control the roughness of the reflection, thus creating a roughness map.typical specular maproughness map
• transparency maps: all materials with a transparency property (glass, roughglass, mattetranslucent) may have a texture assigned to this property.

As with blender, textures may be images or procedurals. Luckily lux has all the blender procedural textures with all their parameters implemented as well. There are pitfalls to observe though:

The mapping mode in lux is always global, so you have to do a lot of fiddling, if your original blender mapping was other than global. Sometimes its plain impossible, e.g. I have not found a decent way to use the blend procedural.

Also the result of lux and blender with the same parameters differs slightly.

It looks like lux is producing finer structures but with less contrast.

Images can be used similar to blender. Here you have the possibility to map using different methods but I prefer uv mapping. Be aware that paths to images are stored as absolute paths, so once you move your project, you have to adjust all the external references by hand.

Similar to mixing materials, you can also mix textures in the same unlimited way. This again gives you a great control over your texture system.

Mixing marble and voronoi, amount controlled by checkerboard texture.

Note: in the current version the semantic of the amount value for mixing materials and textures are reversed. Whereas when mixing materials the amount controls the amount of the first material for textures it controls the amount of the second input texture. ( reported as bug and may already be fixed by the time you read this … )

Multiple Materials/Material Nodes

Both are not directly supported with lux at the current time. Being a completely different material system, the blender node system has no place here. If it is just mixing different materials, you were doing with the node system, you can easily achieve this with the mixing feature in lux.

[edit 13.02.2009] Multiple Materials are supported within lux without any restrictions.

Material Nodes

This is not supported. Being a completely different material system, the blender node system has no place here. If it is just mixing different materials, you were doing with the node system, you can easily achieve this with the mixing feature in lux.

The stencil feature in blender can also be accomplished using a stencil map as input to the amount of mixing materials or textures.

using a marble texture as stencil for a noise bump texture.

This concludes part II of my luxrender observations. In the next (and last) installment I will look at the render settings, tone mapping and analyze the shell material of the snail from my project.

This is more a description of the things noticed and done during the process than a real tutorial. I have come to recognize, that I still have a lot lot to learn and explore before I am entitled to do a real howto/tutorial.

Using the forest project, I will try to highlight some of the areas and give some general thoughts and examples.

At the time of this writing I a referring to the actual CVS version both of luxrender and luxblend, which will closely resemble the upcoming 0.6 release.

So lets start with some introductory statements.

As lux is a standalone renderer with its own scene description language, BDRF based material system and light setup, it obvious, that you need an converter/exporter to be able to use the blender generated scene and render it with lux. In this case that’s what LuxBlend is doing.

The one thing I learned, was (although luxblend tries hard to convert your materials and lights) to completely ignore the automatic conversion done and build all the materials from scratch. Blender’s and lux’s materials are fundamentally different, so a conversion can be a compromise at best, with a lot of manual tweaking, which in the end gives you more work to adapt the converted material than if you built it from scratch.

Data Management

External Libraries

What is not immediately obvious (and will eventually be fixed in upcoming releases, both blender and lux) are some obstacles concerning the way you decide to organize your project.

My intention was to explore the possibilities external libraries provide. So I had the main objects – snail, mushrooms and moss – modelled and textured in separate blend files. The main project file linked to those objects. All references to external data in the libraries and the main file itself were set to use relative paths, so I was able to move the whole project without hassle.

Now Luxblend stores its settings inside the blend file itself, using something called an ID system. As I am not really familiar with this system, my guess is, that you are able to attach extra data to already exitsing data sets using separate IDs and fetch them again later. So e.g. luxblend can store the lux material settings for the blender material ‘mushroom’ in the data set for mushroom.

Unfortunately this doesn’t work with materials linked from external libraries. You can setup all the material, but after loading the file again you have lost all the work done. I have already reported this as bug.

This forced me to make all the external references (objects/objects data) local copies, which triggered another bug – this time in blender. Making data local from an external library, which uses relative paths, results in the same relative paths used in the local copy, which woks if the library file is in the smae directory, but fails miserably if not – which was the case in my project. (This bug is reported too – and already confirmed)

Conclusion: to use lux with a distributed project, you first have to make all external data references absolute and than make complete copies of the referenced objects/data in the main project file.

Luxblend and relative paths

In the current version luxblend has no support for the blender type relative paths where ‘//’ denotes the blend file’s directory. All paths have to be made absolute. The automatic conversion of materials will use the relative pathname and thus invalidate the material setting.

Conclusion: convert all external references to absolute paths.

Modelling

Scale

Being a physically correct renderer, lux has to know about the scale of objects. And it has to know it in physical units. To accomplish this Luxblend maps 1 blender unit to 1 m. This is important if you plan to use lamps with IES description files and want a real lighting setup.

There is a global scaling parameter which can be used to get the scale to a physically correct dimension if you did not pay attention in the beginning.

Subdivision / Displacement

I think it is better to let lux handle the subdivision and displacement, as this significantly reduces the size of the resulting mesh data file and thus the export and load process. If you want to have more control over the shape – using the Median Crease value in the edit properties box – you still have to use the blender subdivision.

Particle System

In my project I did not use hair, but only used a particle system to duplicate tree needle objects around the mossy floor. As I know, hair is not supported in lux for the moment.

My findings with a particle system for duplicating objects are, that luxblend does not generate all the instances but only the source object itself.

A simple particle system setup ( duplicating a sphere across a plane)

results in the following

Conclusion: Convert the particle system to real objects before exporting with luxblend.

Lights

Lux supports 3 types of blender lights – point light spot and sun. But what is true for materials is maybe more so with the light setup. Taking a tweaked blender light setup will surely not result in a working setup with lux. Even if you adopted a linear workflow, blender’s light handling is so different, that you cannot have a working conversion.

Additionally tweaks like the ambient light type or (A)AO is used inside blender to get GI like lighting, which is inherent to how lux is handling light.

Any mesh object in lux can be a light source, either by using the base material type light, or setting a emmissive property on the mesh’s material. This gives a tremendous advantage in defining complex lighting scenarios. Complex light objects come with a price though – rendertime.

Conclusion: you are far better off, setting the lighting up completely from scratch.

In my forest project I used one sun in combination with an angular environment map as single light setup. Lux uses a sun modell where you additionally can specify the turbidity of the sky. My pure blender setup involved one spot light, 2 ambient lights and I used AAO in sub mode.

this concludes the first part and will be continued soon ….

I just finished my project to learn the different workflows using blender or luxrender.

The blend files including the textures (low res version as the high res would take ~60MB) are available from my showroom.

A document describing all the things learned and encountered during this project will follow soon.

I have started a new project with the goal to have the same scene rendered in blender internal and luxrender.

I hope to get some insight how much additional work has to be done to get the luxrender image done. Furthermore it should be interesting to see how the workflow differs and how to map certain blender goodies ( SSS, multiple materials etc ) to the luxrender world ( if at all possible )

I am still in the process of setting up the scene in blender and will open a WIP thread at blenderartists as well as on the luxrender forum. You can see above a very preliminary preview of how the scene will be composed.

stay tuned …

The entries are not many, but very diverse. I will be interesting to see, which one will be used as splash.

My final entry can be seen in my 3d showroom.