In this post it contains a very quick reference of the book Digital Lighting & Rendering and Author Jeremy Birn shares ten tips adapted from his new book Digital Lighting & Rendering, Third Edition on creating more convincing lighting and rendering.
To help you make better 3D renderings, here are some tips for creating more engaging and believable lighting. Some of these steps are frequently forgotten or skipped by beginners, but all of them are necessary and useful techniques in crafting professionally lit scenes.
#10. Collect References
It’s not cheating to begin a lighting and rendering project by collecting reference images. Download any photographs you find on the internet showing similar scenes to what you want to create. Grab still frames from movies you rent. Studying another artist’s work in order to better inform your own is not a violation of copyright law; it’s a fair use of the material.
Look at a scene and analyze the lighting by asking yourself questions such as : Which are the brightest parts of this scene? Which are the darkest? How saturated are the colors in the bright areas or the dark areas? What directionality of light is shown by one side of objects being brighter than another? Is there any haze or atmosphere changing the tone or saturation in more distant parts of the scene?
Reference images are also great conversation starters to help you communicate with your client, director, or art director. You can start discussion reference pictures together before there are any test renderings to review, so you can hear what they like or don’t like about the images and what kinds of effects they want you to achieve in your own lighting. The earlier you start communicating about these issues, the easier it will be to create a final lighting design that they will approve.
#9. Optimize Your Models
How your models are built can have a big impact on how good your lighting looks. There are several ways that improving your modeling can help you render more believable looking scenes.
It’s a good start to your lighting to bevel everything. Corners in real-life aren’t perfectly sharp angles. The edge of your desk is probably rounded or beveled in some way. Walls don’t meet at infinitely sharp angles. Even a small bevel can catch highlights from a range of angles that would be missing if you left a corner unrealistically sharp. You can go beyond a basic bevel by extruding other shapes to run along corners or adding some other variation to the corner areas, to make them more convincing. You sometimes see a strip of caulk or glue between surfaces, or a molding or trim between the wall and the floor.
Build thicker geometry in your architecture instead of using infinitely thin surfaces. Walls of real houses have a thickness, and yours should too. Giving walls a realistic thickness, so that a wall is more like a cube than a plane, helps prevent light leaks in many situations. Also, group your walls or ceilings together with everything that is mounted on or hanging from them. It should be easy to remove the ceiling and ceiling fixtures from the shadows of lights above the ceiling, or hide a wall from primary visibility by your camera.
While it’s always a good idea to prune unneeded models out of a shot, don’t go too far by pruning out models or surfaces that could contribute to shadows, reflections, or global illumination.
8. Choose Natural Shader and Material Colors
Beginners in 3D graphics often err on the side of choosing surface colors that are too saturated or too close to pure black or pure white, and thereby they create surfaces that don’t respond realistically and consistently to light. As a rule of thumb, keep most of the red, green, and blue values on surfaces roughly between 0.2 and 0.8 (when represented on a 0 to 1 scale). This way, you leave room to use your lighting to determine most of the brightness values in the scene, instead of having some surfaces that always appear much bolder or respond to light differently from others.
Avoid using surface colors that are so saturated that any of the individual color channels (red, green, or blue) drops down to zero. If a surface color had a zero in its red channel, this would mean that it reflected zero percent of the red light illuminating it, so that if it were lit by a purely red light the object would never get any brighter no matter how bright you made the light. You can avoid this kind of unrealistic response by not making surface colors too saturated.
#7. Don’t Forget the Spill Light
A spill light is one of the lights that’s most often missing from 3D renderings. A spill light is just a light coming from roughly the same direction or position as your key, but made more soft so that it covers a broader area than the key itself. A spill light around your key can make your rendering look much more natural and believable.
If you add a sunbeam to an interior rendering, the sunbeam will look better with a soft spill around it. After you have your main sunlight set up, and you have test-rendered it to make sure you like what it illuminates and where it casts shadows, you can duplicate the sunlight and rename the copy to become a spill light. Leave the spill light aimed the same direction as the sunlight, but set it to have much softer shadows, so that it spills out beyond the edge of the sunbeam. Make the spill dimmer than the sunlight itself. In some scenes, it is useful to give the spill a richly saturated color. For example, around sunset, the sun itself might be an ordinary yellow, but the spill could be a rich red or deep orange.
Figure 1: Sunlight entering a room is shown in isolation on the left. On the right, a spill light is added around the sun.
Figure 2: Shown in context with other lights, you can see the spill light surrounding the sunbeam, adding warmth, richness, and color variety to the scene.
If a character’s key is only lighting one side of her, or if it is back-lighting her and leaving her face dark, then you can move the spill light away from the key, more towards the front of the character. This way, the spill light extends the key illumination further around her face, providing more shaping and visibility for a backlit character. Spill light doesn’t light the whole face, so you still need some fill and bounce from other angles, but it does make the facial expressions more visible.
Figure 3: On the left, the key light on a character is shown in isolation. In this example, the key is so far behind her that it provides only a little more than a kick. On the right, a spill light is added around the key, extended the key illumination onto more of her face.
#6. Divide Up Your Space
To create more engaging lighting, try to divide-up your 3D scene into different areas that each have different qualities of light.
Many 3D scenes let you look through a door or stairway so you can view more than one room, area, or floor of a building. Each area of a scene like this can have different tones, softness, or colors of light. If one room has a window that lets in daylight, and another room is lit by a lamp, then you’d expect the room lit by daylight to be brighter and have cooler, more blue-colored light, whereas the room lit by the lamp would appear less bright, but have warmer illumination. Light can flow through doors, windows, and stairways within your scene, but the walls around these openings should always cast shadows, and the lights should decay with distance so that they only light part of a room.
Even within a room, corners can receive different lighting than the middle of the room. There can also be an overall transition of color and brightness between the area around a window and the parts of the room further from the window.
There are many ways to divide up space based on distance. You can put more light near the camera, so the scene moves from a bright foreground to a dark background, you could have a dark foreground and a bright background, or you could alternate dark-light-dark. Viewers expect more distant parts of a scene to have reduced contrast and saturation due to smoke, fog, dust, or haze in the air, so you can use atmospheric perspective as a part of how you divide the scene up by distance. Whichever approach you choose, to convince your audience that they are looking into three dimensional space, there should be some difference visible between nearby objects and distant objects.
#5. Enrich Your Scene with Extra Bounce Colors
You may think that using global illumination means that you don’t need to add any extra bounce lights to your scene. Global illumination makes lighting more similar to live-action cinematography, in that each surface of your set can reflect indirect light back into your scene, providing natural bounce lighting. However, even live-action cinematographers still frequently bounce extra light onto actors and areas of the set. Choosing the right places to add extra bounce light, and choosing good colors for the extra light, can help make your renderings look more natural and engaging.
When you add bounce light to characters, tint the bounce lights to a warm tone. Adding some pink or red helps create the feeling that the skin is tinting indirect light. Especially if you aren’t rendering with subsurface scattering, or if you feel that the subsurface scattering by itself looks imperfect and could use a little help, a warm fill or bounce light can be a great way to keep your character’s skin from looking too gray and lifeless.
When you study reference images of real scenes, often the darkest areas of the scenes have the richest, most saturated colors. In computer graphics, if you aren’t careful, the shadow tones in your scenes can become unnaturally gray and de-saturated. You can find motivations for many bounce colors in a scene. For example, if a room has a red carpet, then a red bounce light could look well motivated and believable. Around plants or vegetation, a green bounce looks natural. On a clear day, blue light from the sky can bounce into many cracks and corners. Add extra bounce lights to enrich dark areas with dim but saturated colors. This not only helps you avoid the CG cliché of gray looking shadow areas, splashes of color in dark corners of your scene can also help better divide up your space into differently colored regions.
#4. Nail the Eye Lighting
Your character’s eyes are places where the audience is likely to focus a great deal of attention, and are central to how an animated character expresses herself. This means it’s worth the effort to get the character’s eyes lit perfectly, even if they are a relatively small part of your scene.
The iris doesn’t just provide color to the eyes; it also provides depth. Set deep into the eye, behind the cornea, the iris often receives a bright gleam on the opposite side of the eye from where the light enters. If the eyes are brightest on the right, with highlights on the right side, then the iris gleam (sometimes called scooping) will appear on the left, the opposite of the shading of the overall eye.
Figure 4: This cartoon character shows how the iris gleam (or scooping) appears on the opposite side of the eyes from the key light. Notice how the eye highlights are on the right side of the eyes, and the scleras are brighter on the right, but the iris gleam is on the left side of the irises.
Real iris gleam is actually a caustic pattern, caused by light that has refracted through the lens of the eye and comes out the opposite side of the lens to illuminate part of the iris. In computer graphics, one of the simplest ways to achieve iris gleam is to model the irises so that they slope inward into the eyeball, instead of bulging out into the lens.
Figure 5: This wireframe shows how the eyeballs are modeled. Shown in green, the inner surface of the eye slopes inwards in the central area that is textured to represent the iris. This catches the iris gleam on the side opposite from the key.
Generally, the pupil should remain black. You can model the pupils as holes cut through the center of the iris, showing through to an all-black surface behind the irises. Even if the pupil itself is black, the cornea (the transparent outer part of the eye)can still receive highlights and reflections, so you have to watch each shot to make sure no bright reflections or highlights land right over the pupil. Too much brightness over the pupil could make it look milky and reduce the contrast and readability of the eye, or even make the character look less clever.
Eye highlights help make a character look alive. Without highlights, a character’s eyes look unnaturally dry. Even when eye highlights are only a few pixels in size, they are often some of the most important pixels in the image. Reflections and highlights can happen anywhere on the cornea, but you generally don’t want bright highlights right along the upper or lower edge, because in that area many reflections are blocked at least partially by eyelashes and the reflection of the eyelid. Highlights stuck right in the lower edge of the eye can make a character look as if he is crying, with highlights appearing on the welled-up tears. Often the most convincing place for a highlight is along the key-side edge of the iris. To improve the odds that a highlight will land in this spot, you can model the lens shape into the cornea as a bulge in front of the iris to help it collect highlights from more angles. If your key light doesn’t provide a good enough highlight, you can always add a specular light specifically for eye highlights.
#3. Name Your Lights
To work professionally, you need to give every light in your scene a descriptive name. Clear names help you avoid getting one light confused with another. Naming becomes twice as important if you are installing lights that other people will use or edit later. If you expect other people to be able to make sense of your lighting design, or if you want other people to find the light rigs that you install useful, don’t leave the default names on any lights.
The most informative names refer to the type of light, its motivation, and what it is illuminating. For example, a light named “Spec_fromMatch_onJohnEyes” tells you that it is designed to create specular highlights, is motivated by a match, and illuminates the character John’s eyes. “Bounce_fromRedCarpet_onSet” describes light bouncing off the red carpet onto the rest of the set. Most studios have much more exacting naming conventions, to make sure that everyone follows the same set of rules and consistently creates helpful names for each light.
#2. Solo Your Lights
When you begin adjusting a light, the first step is to solo the light. This means you hide all the other lights in the scene and render the scene one light at a time. When you isolate each light, you know exactly how it contributes to the lighting and shadows in the scene, and you can accurately adjust the controls and options on the light.
You might be surprised how often you can have a light in your scene that is not actually contributing anything to the lighting. Perhaps an object around it is casting a shadow. Perhaps the light is not bright enough to see. In a scene with many lights visible the best way to avoid a chance of confusion is to solo each of your lights to test and adjust each of them one at a time.
If your scene is going to contain a row of similar lights, such as fixtures running down a hallway, it’s best to create just one light first, solo it, and adjust and test-render it until you are happy with every setting. After you are happy with the way your light looks in one fixture, then you can duplicate it into all the other fixtures.
#1. Use a Linear Workflow
A linear workflow is an approach to the entire process of preparing texture maps, choosing surface colors, lighting, rendering, and compositing that allows all software calculations to maintain a direct, uniform relationship between digital color values and actual light intensities.
Rendering software and compositing programs work internally as if they are dealing with linear data. But, in reality, the images that you give them as input (such as texture maps and background images) are not linear; they have built-in gamma correction (usually a gamma of 2.2, the common gamma used with the sRGB color standard that makes images look correct on most monitors). Without a linear workflow, rendering and compositing programs work with this data as if it were linear, and then they produce output that is not gamma corrected. If your output is not gamma corrected overall, but it displays on your monitor at a gamma of 2.2, it’s easy to mistake this output for a scene that’s just under-lit. Adding to the confusion, although your lighting is displayed at the wrong gamma, the texture maps within the scene were gamma corrected, so the textures within the scene are at the correct gamma. This can fool you into thinking that the only problem is that your lights aren’t bright enough and cause you to adjust your lights to compensate for the incorrect output. Especially in the 1990s and 2000s, a lot of people worked without a linear workflow in this way, and developed some bad habits while they attempted to compensate for the lack of a linear workflow in how they lit and composited scenes.
A Quadratic decay (also known as Inverse Square, or an exponent setting of 2) is the most realistic setting for how a light should decay with distance. With a linear workflow, Quadratic decay accurately recreates how the light from real light sources radiates out through space. However, one of the symptoms of working without a linear workflow is that Quadratic decay looks too abrupt: the area near the light source can be badly over-exposed while more distant objects barely get lit at all.
Another symptom of working without a linear workflow is that global illumination never looks completely convincing. Global illumination simulates light transfer between surfaces, and the natural quadratic decay of light between one surface and another is an essential part of what global illumination adds to your scene.
In compositing, the most natural way to combine two illumination passes is an Add operation, also called Linear Dodge in Photoshop or Merge (Plus) in Nuke. Only the Add blending mode combines lighting passes to match the output you would have gotten if you had rendered with two lights both visible at once in the same pass. However, artists trying to composite without a linear workflow find that adding together two dark-looking images jumps much too quickly towards pure white. As a work-around, some artists started using the less-realistic Screen blending mode instead of Add, when all they really needed to do was adopt a linear workflow.
The good news is that many artists and software companies have been taking linear workflows seriously in recent years. Some 3D programs, such as Cinema 4D, now work in a linear workflow by default for new scenes. Other programs, such as Maya, support a linear workflow after you adjust a few settings that are documented in the manual. Search the Web for the name of almost any graphics program followed by the words “linear workflow,” and you’ll find 3D artists talking about what settings to use and how to take advantage of this process.
To maintain a linear workflow, your software can convert material colors and texture maps into linear data, even if they were created with a built-in gamma of 2.2. Then, you can render linear data, stored with precision as half floats (half precision floating point data uses only 16 bits per channel) in an .exr (OpenEXR format) file without gamma correction. Your render view window can be color managed to show you your scenes corrected for your monitor, even though you are rendering linear data. Finally, during compositing, you can work with linear data all the way through the process until you convert to the gamma of your monitor as the very last step in the process. These steps are covered in detail in Chapter 8 of Digital Lighting & Rendering, Third Edition.
Figure 6: This split image shows how a scene looks without a linear workflow (left half) and with a linear workflow (right half). The scene is lit entirely by one area light positioned at the florescent panel. There are no changes to the light or the global illumination settings between the left and right halves of the image, the only difference is the linear workflow.
The linear workflow truly is number one in my tips list. If you’re not using it, then it is the single change to your lighting and rendering that will best help move a beginner’s work up towards a more convincing and professional level.
Jeremy Birn is the author of Peachpit’s new book Digital Lighting & Rendering, Third Edition. He has been a Lighting Technical Director at Pixar since 2002, and has lit shots in movies including Monsters University, Brave, Cars 2, Toy Story 3, Up, WALL-E, Ratatouille, Cars, and The Incredibles. He has previously done visual effects and lighting work at companies such as Tippett Studios, CBS Television, Wild Brain, and Palomar Pictures. Jeremy has taught advanced lighting and rendering at the Academy of Art University and the California Institute of the Arts, and has taught master classes for Autodesk.
.jpg "3D modeling")
