detail maps, alpha masks, emissive maps, etc.) (i.e. Red, Green, Blue or the Alpha channel.)Ī physically-based material which neglects to leverage a Pack Map (opting for multiple image files, instead) will unnecessarily bloat the memory overhead of the material, as much as 400% in many cases, even more if additional maps are in-play (e.g. The term Pack Map describes a composite image which contains multiple single-sample (i.e. grayscale) texture maps, each of which is isolated to a unique color channel. Ever thought a game looked like it was “dipped in vaseline”?Īs rendering fidelity approaches physical-accuracy humans are increasingly sensitive to the (often small) disparities between what is and isn't consistent within a environment. As members of the human race we have the remarkable ability to perceive and immediately identify visual novelties, regardless of art style or fidelity goals. Now obviously 256 degrees of interpolation is a far cry from being considered countless, but the spectrum of surface characteristics represented by each degree is substantial so much so that deviations from physical reality can be realized at run time by the average pair of eyes. There are "countless" possibilities between the roughest surface (0) and the smoothest surface (255). Whether it's as smooth as a cue ball or rough as chalk, it can be approximated with this single channel. Well for starters, the Smoothness Map has the great responsibility of approximating all the subtle, nuanced characteristics which light energy exhibits upon interacting with a (micro)surface.
But it is also the most enjoyable to author! Why is that? Upon adopting a PBR pipeline it is often the Smoothness Map which gives artists a difficult time. Level adjustments have the propensity to break the histogram, especially when clamping down the global value range. * If you're using level adjustments rather than curve adjustments, the time is now to adopt a curves-based workflow. To remedy this, try reducing the contrast of the photo reference, adjust output curves*, gamma and/or exposure as necessary in order to prepare the photo for a clean conversion. One of the drawbacks of this process is the increased likelihood of finding artifacts and noise in the generated normals. Many artists (self-included) enjoy the efficiency of generating normal data from their preferred bitmap-to-normal converter. Photo-Normal ConversionĪpplications like Substance Designer, Crazy Bump, and Quixel's nDo still have a place in a PBR pipeline, make no mistake. This is especially important when transitioning to (or from) a metal/non-metal surface due to the (obvious) disparity in surface properties.
You'll want to remove unnecessary noise as much as possible from your Normal Maps and focus your efforts on defining distinct shapes and forms. These micro-surface details are best realized in the Smoothness Map. (We'll go into detail about the Smoothness Map below.) Tactile descriptors like smooth, wet, scuffed, worn, and rough are all considered micro-surface details which don't merit additional normal information. As we'll see below, this is a no longer a good idea - in fact it's a terrible idea unless you're literally authoring a 40-grit sandpaper material. An example of this would be adding a noise filter to your Normal Map to simulate a "rough" surface. Micro-Surface Details, as the name implies, are far subtler surface details which, until recently, were usually found in the Specular Map, or sometimes the Normal Map. In PBR, Normal Maps serve to provide directional lighting information for Macro surface details which (for one reason or another) don't merit the overhead of additional geometry.