low fidelity rendering guide

Tutorial / 07 August 2022

wow! 

Lighting for Asset Presentation

Tutorial / 11 January 2021

In this short walkthrough I’m going to show my asset lighting process using a gun I made a while ago. This isnt the only way to light assets, there are many different styles of presentation, but it's a way that I find works well.  In a nutshell the idea is to use more lights than you might be used to, and use each one intentionally to draw out certain parts of your asset.

Since the focus here is on lighting I wont talk much about compositional choices, which could be its own subject. I assume we have our asset ready, posed the way we want it, the camera looking at it, filling the frame. The only detail I'll go into here is that your camera's FOV will affect how light looks on your model. A wider-angle lens will produce tighter visible highlights. A tighter-angle lens will produce broader, flatter highlights across surfaces. My example is using a 50mm lens, which is a typical case for shooting portraits. A telephoto (very tight perspective) would be something like 90mm. Replicating a firstperson view from a shooter game will require something wide-angle like 20mm. 

Since this is about lighting aesthetics, it shouldnt matter much which software you're using. I’m using Marmoset Toolbag 3, which is now slightly outdated, as it doesnt have raytracing; but these principles should apply forward to any rendering software, as well as to real life photography.

====================

We start off with basic ambient illumination. For my example, this is based on an HDRi image, but it could also be based on a flat color, or a dim soft light in the scene somewhere. 

I prefer to keep the ambient illumination fairly neutral in color, so as not to conflict with the texture. It's also important not to make it too bright, because then things will get washed out. Make sure there are still visible dark recesses.

Think of this as the base layer on which we will add lights to paint areas of interest.


The next step is dropping in the main light, usually called a key light. This will be fairly bright and also tight in size, meaning it casts hard shadows. Place this in a way that it pulls out some of the most interesting and camera-facing forms on the model. In my case the cylinder and barrel of the gun are dominant large shapes, so my key light is placed to highlight these. It also picks up some hard edges without flooding entire surfaces with light.


Next we add what's called a fill light. This is more dull than the key light, and also broader, casting indistinct soft shadows. The purpose of the fill light is to assist the key light by filling in some of the larger planes that are facing the camera. The reason to keep it duller and broader is so that it wont compete with the key light for intensity. If you have lots of tight highlights all over the place, they're going to start looking like racing stripes, which isnt very naturalistic.


Notice how the edges of the model tend to blur indistinctly into the background? For example the top of the frame, the face of the hammer, or the area under the barrel. These silhouette areas are very close in tone to the background because they are still fed only by the ambient lighting. They're kind of muddy and ill-defined. For this reason we add another light, located behind and above the asset. This one's quite bright, since we arent facing the side of the asset that it's illuminating, and we're only using it to pull out the silhouette. We call this a rim light because it highlights the rim of the object.
  


See the hard shadow on the trigger and inside the trigger guard? That's caused by the rim light we just placed. Those shadows are pretty distracting, mostly because they contradict the hard shadows we already have from our front-oriented key light. In order to soften them up we can broaden the rim light so that its energy is being transmitted from across a larger surface, blurring out the shadows. Below this next image is a .gif of before/after this change so we can see the difference it makes more easily.


At this point our basic lighting is all set up: we have a key light, a fill light, and a rim light. This is what's called a three point lighting scenario. In order to push our work further we should start asking ourselves which parts of the asset are still muddy, flat, or otherwise indistinct. We spent a long time making this thing after all—why let any of that go to waste?

In this case I drop three more lights in, focusing on adding volume to the grip as well as the contours of the cylinder, barrel, and hinge surfaces. Some of these act like rim lights, and others like fill lights. In general none of them are anywhere near as overt as the three main lights we started with. The further we go in lighting, the more subtle the changes we're trying to make. We don't want to overdo it, just accentuate the forms that are already there.

Once again there's a .gif after this next image, since we are now dealing in more subtle changes.


Next we continue the process we just started—looking over the model, finding areas that either blend into the background or lack discernible form, and putting in lights designed to add interest to specifically those areas. At this point each light we add is typically affecting a very targeted part of the model to call out precise details or shapes. The main consideration is not overdoing it, not competing with the main lights, just supplementing them with additional support. Generally, all of these lights will be fairly dim and broad so as not to conflict with the existing setup.


That's about enough lights I think—eleven of them, in this case. A three-point lighting setup, with eight additional supports to highlight specific details of our asset.

Next comes what I think of as finaling, or tuning. All the components of our shot are in place, we just need to tweak values and optimize the appearance. To start with I will take the background and dim it down (without changing the actual ambient lighting effect) so as not to visually compete with the asset that we're presenting. By keeping the background darker, the work pops out.


Finally we adjust the tone across the entire shot using a curves edit. Assuming you have an option, the reason to use a curves edit rather than a levels edit is for an extra dimension of control. Here I'm setting a subtle version of what's called an S-curve, where you darken the dark areas and lighten the light areas. This is an extremely common pattern in post-processing across games, films, and photography. The purpose is to add contrast without blowing out either extreme end of the value scale. Sometimes you'll see this applied far more dramatically than I'm doing it here—my main focus is on displaying the asset, not creating an overly stylistic effect.  One thing you might try on your own is messing around with raising the level of the darkest darks, or lowering the brightest brights. It can create a very filmic effect replicating darkroom processing or exposure errors.


Here's the final light setup showing where all 11 lights are relative to the HDRi. This may be mostly relevant to TB3 users, but should still give an idea of how dispersed the lights are, how many there are, and how dim some are. Remember that most of these are only placed to do one or two specific things.


And finally a .gif walking through the entire process.