Eg. renderers like AIR using the Reyes architecture support micropolygons. But such engines cannot be compared to todays alembic engines which do not require any background knowledge to spit out images.
_
c.
1 Like
Quality of the rendering, or of the geometry? In Rhino the NURBS geometry is independent of meshes used for display and rendering. The ârenderâ mesh obtained using ExtractRenderMesh can be altered by changing the mesh options. The NURBS geometry is unaffected. Rendering software can create itâs own mesh based on the NURBS geometry.
Those engines cannot be compared either to the amazing quality and speed you can get today with a GPU rendering like Octane. Have you seen the metal shaders in Octane 4? Not a renderer that spits out images without any knowledge either, but the quality is worth it iIMO. We live is a much better world rendering-wise than we have ever lived.
Of course, as long as youâre dealing with data which fits on your GPU all is smooth and easy and faster to get the desired look in realtime. The funny thing about the old engines is that their demands did not increase over time, with nowadays CPUs itâs a whole new experience.
_
c.
Wow! I didnât know itâŚ
I meant it. The output image completely depends on a grid.
Look, like the commenters above explained to the n-th degree, NURBS surfaces and polygons are used for different purposes. Polygons donât look âbetterâ than a NURBS surface, like a pineaple does not look âbetterâ than a steak. They are totally different things.
1 Like
Actually, the pineapple - in the right context - does look better than a steak. Then again the steak looks better in other cases. Same for polygons and NURBS surfaces. The latter perhaps a bit artificial, but stillâŚ
1 Like
The subject is closed.
Thanks!
Thatâs mathematecally impossible. The 3rd law of noshdynamics postulates that red meat tops yellow fruit in the known universe, possibly even behind the event horizon.
4 Likes
http://render.ru/gallery/work/135301#work
Look at a car grid, it seems excellent quality. Why the model canât be used in production? In what a problem?
By the way, if to simulate similar in SubD and NURBS. Where it will be simpler where is more qualitative?
turn on flat shading on this model and youâll see the problem of this model for production. The curvature look between each vertex is a âcheatâ. Its a colour interpolation which simulates curvature.
Between each vertex there is no mathematical representation, which is fundamental different to a nurbs patch⌠⌠I believe Iâve already said it in previous posts.
1 Like
Well. But somebody can attach the file with comparison of NURBS and SubD on the example of some object?
I didnât use SubD yetâŚ
have a look at this mesh sphere - left with normal shading - right with flat shading. Now imagine you want to mill this sphere. It will look (more) like the right one in real life.
It is true for most production the model ends up as mesh again. But this âproduction meshâ can be much more precise, usually more dense at higher curvature and very different to a ârender meshâ.
2 Likes
Unless you use custom render mesh settings that are the same as your production mesh settingsâŚ
1 Like
But SubD, after all, can also be subdivided (mesh). Is not it?
actually I meant not ârender meshâ but âmodelling meshâ.A render mesh does already what a production mesh wants as well, maybe just less precise due to performance issues.
So this explains the difference between a nurbs converted mesh and âcleanâ modelled meshes from an polygon artist.
That makes more sense, agreed (:
1 Like
yes, but a.) complex models are never 100% sub-d only b.) design capabilities and curvature control are limited with sub-d (both meshes and t-spline) c.) subdivision polygons are divided equally not curvature related, which produces unnecessary polygons. d.) its always good to be able to produce a point anywhere on your shape â mathematical description
1 Like
It is possible here in more detail?
It seems I have understood therefore also a grid at surfaces often non-uniform (Rhino).