The material seems fine. It seems much more complex than it needs to be, you can achieve this result with fewer channels. A couple of odd things:
1. There is the Diffuse Channel color located in the top third of the Mat lab, and then there is the Diffuse Intensity Slider in the second third of the lab. In the past, any time I applied an image to the Diffuse Intensity slider, it would override the setting applied with the slider. You applied this image as slot A in the Diffuse intensity, yet unlike in the past, adjusting the intensity slider seems to have an impact. Very odd. I must be missing something here. Please elaborate.
2. You have some complicated looking stuff going on with specular and bump as well, I am convinced that not all of it is necessary. But to know for certain I would need to know your intended goal.
Some quick observations about photo textures and bump:
1. It is best to slightly blur a bump map. Sharp transitions fail to produce bump.
2. World Cubic is very useful for seamlessly tiling images, such as the two samples you have provided. Parametric gets distorted.
Join Date: 01.03.2004 Comments: 2227
Thanks for the feedback Rashad.
1 - if you look in texture mapping modes, check alpha scaling and the slider will then scale the input of the alpha channel rather than be overridden by it.
2 - the occlusion map, it is true, could be combined with the diffuse but keeping it as a separate entity allows experienced users to optimise the material for the particular lighting of their scene. The specular map in alpha is intended to give greater highlights on exposed (and hence worn) parts of the surface and the normal map is controlling anisotropy so that the specular is not only changing in intensity in response to the virtual surface but also the direction of the response (this embellishment is as you say, convincingly superfluous). But I'd rather provide more rather than less for people to play with.
Thanks for the tip about World Cubic I will give that another look. I must have chosen the wrong scale when I tried it last time since it didn't seem to map seamlessly on my test sphere.
The bump controlled by pict I've been investigating. And found some curious things. Horo is presently peer reviewing my findings before I come up with something formal. Here is the story so far, just bear in mind, these tests are ongoing.
Alpha channels do not get interpolated.
Alpha channels only operate at 8 bit resolution (cf. procedural alpha).
Surface sampling for the generation of normal information is done on a per pixel basis. The practical upshot of which is that the smaller the alpha image supplied the lower the angle of the surface normal generated. (Sharp transitions will always fail to produce good bump on procedural textures - but it does not have to be the case for picture textures depending on resolution) - which is clearly a bit odd.
The alpha response for bump generation is reversed between procedural and picture texture sources. Eg, in procedural textures alpha white corresponds to a high point, while the reverse is true for picts.
From my tests - the bump material as you suggest needs to be filtered. Blurring is good up to a point... but on subtle transitions stepping will still appear because the blur radius will not encompass the entire transitional area - if it did... all the smaller bump detail will be lost - catch 22. As you might guess, from my inclusion of a normal map, using this and a 16 bit grayscale stage an an intermediary is what I'm currently toying with.
Also... while it is possible to get away with 512x512 textures for the colour, occlusion and specular, the bump is clearly degraded at anything lower than 2048x2048 for the size provided in these tests. This of course is also due to my desire to capture high quality information (fine detail) in the bump. More blurring would allow a smaller map, but at the cost of the detail I wished to convey.
Join Date: 10.15.2010 Comments: 824
I have tried to make a hexagonal texture in DTE by means of noise "triangle". Alas, while at me it has turned out nothing...
How Rashad hasn't noticed such thing as "Alpha Scaling"? :))
Join Date: 01.03.2004 Comments: 2227
Yes I too have made many attempts to get a hexagon out of the DTE and to no effect. I also put in a feature request for an appropriate procedural function to be added - also to no effect. If anyone manages in the DTE I would like to see how it was done. Indeed, I may have another little go myself later on - but I can't help but feel that the available functions just do not lend themselves to this pattern.