I am relatively new to Poser (Pro 2010).

I am confused about 'displacement' and what limitations there are with regard to JPG files which I apply to a 'Square Hi-Res' Prop.
The Preview and Firefly renders both appear to blur (anti-alias?) pixels, and 'displacement' appears to calculate an average. Looks excellent from afar, but indistint with the regard to the original image's quality.

I have read many parts of the manual and various related Forum and web pages; but I am not getting confused! I realise that Poser is not intended to replace other CAD products, but I am keen to understand how far the 'displacement' shader system can be used.

And whether there is ever any reason to use 'displacement' and 'bump' in the same shader?

Thank you.

There are poser users who will have more technical answers than me but I use displacement in place of a bump map if I want a detailed texture that looks more 3D.  Great for fur, grass, stucco, etc. 

I think of bump and displacement in terms of bump = texture, displacement = details.

I hope that makes sense and is somewhat helpful.

JPG only has 256 levels for grey (or any single hue).  For highly detailed or extreme displacement, this isn't enough.  I'd suggest trying a 16 bit tif displacement map using compression for detailed or very large scale work.

You might also look at the texture filtering on your map(s), the shading rate you're using in rendering, your pixel samples, and other rendering settings.  If you are using render gamma correction (GC), you should make sure that your displacement map(s) have a custom gamma of 1.

Displacement should be used for large scale effects that need visible volume, like piping or large wrinkles on clothing.  Bump should be used for small scale effects that don't need visible volume, like fine seams and stitching, cloth or leather textures, and other surface variations that shouldn't affect the mesh.  Bump renders faster than displacement, so it's best to use displacement only when bump won't suffice.

Difficult to answer without an example but if your pixels blurr?

In the displacement map image_map set texture filtering to none.
For Poser9 / PP2012 set is to crisp

In the firefly render settings set MSR to 0.2

Also check image dimentions:

512x512
1024x1024
2048X2048
4096x4096 (the Max Poser can handle for a single texture)

Anything in between will cause additional blurring.( interpollation of the pixels)

Displacement maps can be VERY sharp in a rendering.

Also remember that jpg files are compressed and the detail may be blurred.

Use one of the lossless formats for displacement when you need close up work

Bill

 

> Quote - > I am relatively new to Poser (Pro 2010). > > I am confused about 'displacement' and what limitations there are with regard to JPG files which I apply to a 'Square Hi-Res' Prop. > The Preview and Firefly renders both appear to blur (anti-alias?) pixels, and 'displacement' appears to calculate an average. Looks excellent from afar, but indistint with the regard to the original image's quality.

A displacement map and bump map, both store the same information, they can be interpreted as height maps storing sample values of the height (displacement) of the surface of a (real) object relative to the surface of a model. But because it stores only sample values, not a continuous surface, but the renderer renders surfaces (at least in theory) and not sample values, the renderer has to interpolate these sample values to get a surface. It does so by using linear interpolation, and because of that the object always gets blurred. You can somewhat minimize the blurring by disabling texture filtering, but cannot completely get rid of it. You can however use a higher resolution texture or (in the case of bump maps) use a tangent space normal map; the latter one explicitly contains normal values, not height values, and therefore no interpolation is needed to render them, so there is usually less blur.

Quote -
[...]

And whether there is ever any reason to use 'displacement' and 'bump' in the same shader?

I guess the usual reason (at least for me) is, that you can disable the displacement in the render settings (because you want to get a fast preview), but still get a similar impression how your surface looks like with displacement enabled. But your doubt is justified as the illustration shows; i used a white rectangular spot on a black background for a texture, and rendered it as bump, displacement and both (lit at a shallow angle from the left). As you can see, the displacement and bump are indistinguishable when looked straight on, but when both displacement and bump are used with the same map, the bump map gets applied on top of the already displaced surface, which is wrong. (On the other hand, Poser rendering is not exactly rocket science, so anything that looks good, is allowed :-)

The big important thing about displacement is that it is the only feature that actually changes the profile of a model at render time; bump and normal maps do not.  Depending on how you're using it, this may or may not matter, e.g. if the profile of your model is not visible in the render, or if the changes made by displacement are so small that you can't see that the profile of the model is changed.

Bump/normal will self-shadow correctly (that is, the bump effect will cast shadows on the model itself) but will not change the model's shadow cast by lights on OTHER parts of the scene.  I.e. if you render a sphere, no matter what the bump map is, the shadow will be that of a sphere.  Displacement WILL change the shadow that the model casts though.

attached is an example of a sphere with bump ....

and here's the exact same sphere and lighting and material, only it's plugged into displacement instead of bump.  See that the shadow is now irregular, and the profile of the sphere is very different?

there is only so much you can do with pure displacement. start to push it too much and youll get cracks and other anomalies.  its meant for an enhancement of details on a model, not to replace modeling altogether.

My thanks to all the responses to my post. There were many helpful suggestions and considerations.

I am grateful to Vilters. I had misunderstood the purpose of MSR. Setting a low value has now produce a vastly more 'accurate' render. Also, I was unaware that Poser's accuracy would be affected if the Image was not in those 4 particular sizes.

I think Millighost's explanation of the sampling and 'linear interpolation' (I need to research a little more , to understand that) well explains how the inaccuracies occur. I don't mind Poser having limitation, but I like to know what they are.

As an example, I attach a 512*512 uncompressed TIF file.

http://www.tima.uk.com/img/simple_lines.tif

It is entirely black, other than some 1 pixel wide lines (on 2 of the edges, and a cross near the centre).  When adding it as a displacement image, no displacement occurs.
I would be grateful if someone might check, as I am a bit new to using Poser. And may easily have done the wrong thing.

Well 1 pxel is pushing it a bit if you use any texture filtering.  Use none(pre poser 9) or chrisp.

 

Bill

I took your tif file, added soms circles and a box shape. See screengrab. The X is yours. Remember, displacement is not visible in preview, only in a render. And you have to put displacementy On in render settings.

My Poser units are inches. !

Thank you again.

In both examples, the cross is clearly visible - but not the 1 pixel wide lines which run down the entire length of 2 of the edges of the TIF.

Are those areas of the UV space actually placed on the model?  It's pretty common for people making models to leave the edges of the UV space alone.

I have raised this issue with SM. They agree that something unexpected is occuring; and have passed the matter to the developers for their comment or explanation.

Their developers have now accepted there is a bug, possibly within the rendering code. I guess I shall now wait for any service pack.