Preview, even hardware accelerated, does not employ all of the sophisticated shading algorithms that the renderer uses. The preview specular, in particular, is overly simplistic and does not do anything like what the specific specular nodes, such as Blinn or Anisotropic, actually do.

If you're working on a shader, the preview is not at all relevant. Look at the Poser Surface preview instead. But even that is simplifying the outcome. The only thing that confirms what a parameter or node change really did is to render.

Too many users become frustrated by setting up slow rendering, thus they avoid the proper level of experimentation that yields insight into how to configure shaders and lights.

I do not generally test shaders with shadows on, for exmample. That's a big speedup. For experimenting with lighting and shader reaction to lighting, I also don't bother with a lot of sub-pixel sampling, or a low min shading rate. Usually, the use of high quality render settings doesn't tell you anything you wouldn't learn by test rendering with low quality settings. Only when you are tweaking really tiny details does it matter.

When adjusting something like specular settings, my renders of a figure are usually around 10 seconds. I can try 20 different settings in 10 minutes. I don't need hair or shadows or anti-aliasing to learn what I need to know.

On the other hand, if my render times are 10 minutes, then I learn practically nothing in 10 minutes. Many people marvel at how much I know about the nodes. The reason is simple. On any given day, I get to try a couple hundred setups, while others who have less effective workflows only get to try 10.

---

Renderosity forum reply notifications are wonky. If I read a follow-up in a thread, but I don't myself reply, then notifications no longer happen AT ALL on that thread. So if I seem to be ignoring a question, that's why. (Updated September 23, 2019)