Mesh geometry impact on Normals behavior

Building using prims, sculpties and meshes. Texture creation techniques.
Post Reply
User avatar
Mike Lorrey
Posts: 361
Joined: Sun Sep 04, 2016 5:40 pm
Has thanked: 71 times
Been thanked: 270 times

Mesh geometry impact on Normals behavior

Post by Mike Lorrey »

As many know, I'm a bit of a perfectionist on my building. I've been working to improve many of my existing products on my marketplace store. Currently working on the Torch Tower high rise building, seeking to reduce its complexity while improving its appearance. Improving appearance obviously tends to depend on adding more geometry, either to add more detail or to create greater smoothness in the curve of non-planar faces, however this is not always the case. The images I am sharing shows the glass envelope of the Torch Tower in two geometric patterns that provide starkly different performance on smooth normals.
The four glass envelopes below are actually TWO meshes. On the left is the original mesh, slightly improved over the original. On the far left is one where the faces are all set to be shaded flat, so we see the faceting of the geometry which appears in long vertical rectangles even though the quads are formed more or less square in the mesh. The near-left, I've shaded that geometry smooth, but the normals are clearly not perfectly smooth curves as one would expect. The radius of all loops in this mesh are properly graduated so ideally this should provide a perfectly curving smooth shading, but that is not what we see.
On the right is a significantly simpler mesh, with half of number of horizontal loops of that on the left, but most importantly, the loops are staggered so that the vertical loops become spiraled when triangulated. You can see how the shaded flat normals exhibit this different geometry with this really exquisite diamond pattern that looks like its a piece of fine leaded crystal tableware. But what is really surprising is when shaded smooth, the normals reflect light much more smoothly than the MORE COMPLEX square gridded geometry to the left.
Attachments
Torch Tower Normals comparison_003.png
Torch Tower Normals comparison_003.png (230.7 KiB) Viewed 3019 times
Torch Tower Normals comparison_001.png
Torch Tower Normals comparison_001.png (406.44 KiB) Viewed 3020 times
Torch Tower Normals comparison_002.png
Torch Tower Normals comparison_002.png (503.34 KiB) Viewed 3020 times
These users thanked the author Mike Lorrey for the post (total 2):
Ilan TochnerMichael Timeless
User avatar
Ilan Tochner
Posts: 6504
Joined: Sun Dec 23, 2012 8:44 am
Has thanked: 4943 times
Been thanked: 4455 times
Contact:

Re: Mesh geometry impact on Normals behavior

Post by Ilan Tochner »

Rendering is completely viewer-side, so there is nothing we can do about improving it, but thank you for sharing this information Mike.
These users thanked the author Ilan Tochner for the post:
Mike Lorrey
User avatar
Mike Lorrey
Posts: 361
Joined: Sun Sep 04, 2016 5:40 pm
Has thanked: 71 times
Been thanked: 270 times

Re: Mesh geometry impact on Normals behavior

Post by Mike Lorrey »

I agree, the main issue with improved rendering quality is that we need viewers to add more capability to the texture/shading tab so we can have more PBR textures for things like transparency, glow, fullbright, environment and glossiness, as well as to add full reflectivity so that high-gloss objects reflect other high-gloss objects, especially including avatars. This simply requires that the lighting model add 3-4 reflections to raycast light rays off surface normals.
These users thanked the author Mike Lorrey for the post:
Ilan Tochner
User avatar
Mike Lorrey
Posts: 361
Joined: Sun Sep 04, 2016 5:40 pm
Has thanked: 71 times
Been thanked: 270 times

Re: Mesh geometry impact on Normals behavior

Post by Mike Lorrey »

I got a question asked inworld about how I make this glass look so "metallic". You apply the default "blank" texture to both the normals and specular settings. This tells the system you want both perfect "flatness" and perfect specularity on the surface (where 'flatness' is of course perfectly smooth on a surface that is non-planar). Then on the specular settings you vary the ambient and environmental values to between 50-255 depending on how reflective you want the surface to be. In this example they are both 255 which SHOULD be a perfectly mirrored surface. IMHO opensim viewer devs should code their viewers to treat this sort of setting to turn on a "mirror" code that would reflect all objects including avatars just like the system water does, but thats just my opinion.
These users thanked the author Mike Lorrey for the post (total 2):
Tess JuelMichael Timeless
Post Reply