www.si-community.com

I'm trying to create a rendermap texture from a procedural bump map, but am failing miserably.
I've tried plugging the bump.out into an image node and plugging that into the surface slot of the material,
creating a rendermap to plug into a bump map node in a new shader,
but the resulting map doesn't resemble the originally procedurally generated bump.

So your rendermapping a scalar image?

Is the procedural very high frequency?
If so is it perhaps just a resolution thing?

Or do procedural values go above 1?
If so perhaps baking out as EXR?

Otherwise can you visually describe the difference?

I am trying to recreate this EXSI preset: So I plug the second branch into an image node and plug it into the surface slot of the material to send it to rendermap.
The Combiner node gives me the following "shaderball": But the resulting (albedo) rendermap made with the image node looks like this: So using the image node doesn't give the right result...

edit: if I plug the Combiner into a VectorToColor node and plug this into the surface slot of the material I get the correct "shaderball" coming from the VectorToColor node
but once plugged into the material I get a similar (wrong) result after rendermapping...

Generally you'll want to bake the results of bump inputs (these two ending with exsi paving1 and fbm), into two textures, then to replace the procedural inputs of bump map generators with rendermaps, or, to re-assemble exsi bump combiner if possible and plug the combined result into bump map generator node (possible = if some simple math operation is used, add or something).

Output of bump node is vector, that is, changed surface normal, so as another option, theoretically it should work to plug the result you got, directly into bump map slot (you'll need to convert color to vector using appropriate node), but this is really theoretically - not sure is there a way for particular renderer, to properly sample antialias in this case, what happens about local and global object coordinates. By the way, results are looking like properly rendered vectors.

Indeed what Mathaeus said,

This looks like a bent normals (in world or object space?) RGB representing XYZ normal orientation.

So the shader ball (or shader grid) actually probably yields the same result, except you're viewing it in one axis therefore showing one color with some variations from the pattern,

What does it show when you select 'apple' as shaderball geo?

& BTW, what are these Hitpoint, Rounded, Agate, Paving ... 'EXSI' nodes?

Cheers,

So how do I feed this as a correct bump map back to the material? I'm completely stuck...

BTW: http://www.shaders.co.uk/enhance_xsi/

Since it seems to yeld the same result, doesnt swapping the procedural with the rendermapped output using the same UV used to bake out the map work? if not what does it do?



And thanks for reminding me about Enhance textures,

For anyone interested, these are part of the demo samples
http://www.shaders.co.uk/enhance_xsi/demo.htm

The original: After rendermapping: Result:

I think I know (at least partly) why, the rendermap is generated with the camera always pointing towards the normal which yields something different than what comes in from the procedural.

There is probably a way to convert from world to tangent space (which is like a "normal offset" from base normals) in the rendertree to use it as a normal map (anyone with insight on that?),
to then plug it into a 'XSINormalMap3'.


Or you can convert your map from world space to tangent space using this ::
http://www.handplane3d.com/

found here ::
http://wiki.polycount.com/wiki/Normal_M ... al_Details
(also contains desciptive info about NormalMaps and various space)

.. and then plug it into a 'XSINormalMap3'

Normal maps can have their elements stored in different channels, but often uses the same standard channels, so Let us know how it goes.

Or perhaps you can try this for using it directly..

If result is close but incorrect, you can swap what channel XYZ uses until it works.

@FXDude your network is wrong you have to introduce the Normal from the Shading State.

@Hirazi
You have to bake exsi paving1 first, then fbm in second, plugged on the Surface port directly.
Then in your New shading network using thoses baked maps you have to reproduce the Conbinaison of the Two

NNois wrote:@FXDude your network is wrong you have to introduce the Normal from the Shading State.

As all this appears to be "way out of my league"
what would the correct way to plug this in the render tree?

Hirazi Blue wrote:

NNois wrote:@FXDude your network is wrong you have to introduce the Normal from the Shading State.

As all this appears to be "way out of my league"
what would the correct way to plug this in the render tree?

(in case of you want to do the math that's the shading state node http://download.autodesk.com/global/docs/softimage2014/en_us/userguide/index.html?url=files/shaderef_state.htm,topicNumber=d30e525091)

But don't do that , just rendermap your bump as color map (not bump map) separatly then Mix2Color your 2 rendermaps and plug that to the bump node (or daisy chain two bump map node, same same)

@FXDude - I can't attach files to PMs, so I attach the file here: Trying a completely new scene (with a Phong, not a Constant shader... ) gets me better results as to the bump,
but the end result still differs a lot from the original:

Original: Rendermapped:

NNois wrote:just rendermap your bump as color map (not bump map) separatly then Mix2Color your 2 rendermaps and plug that to the bump node (or daisy chain two bump map node, same same)

I assume you already did, but indeed you can just use that scalar output as a bump.
Nevertheless, thanks Nnois for the info about shading state (and Hirazi for the opportunity), it helped understand normal vector workings.