5 Reasons why your 3D renders look fake
Technology will only get you so far
All the technology in the world will not help you achieve photoreal imagery if you don’t follow some basic principles. And whilst 3D software has made huge strides in the realm of photorealism, it will at best only get you eighty percent there. The remaining twenty percent is the difference between an image that looks “kinda” real, and one that truly convinces.
For the creator and consumer
If you’re a creator of 3D imagery, you know that many aspects of a scene need to come together to create a truly photoreal image.
It becomes a little harder if you’ve commissioned 3D work, and you feel it’s just not hitting the mark. What is it that isn’t working? Trying to pinpoint the problems can be frustrating and time-consuming. This list is aimed at both the creator and consumer of 3D imagery. Use it as a guide as you evaluate your work, and in doing so you’ll avoid some common mistakes.
Want to read this story later? Save it in Journal.
The good news is that increasing your quality levels come down to some simple observations. Following these basic rules will go a long way in helping you achieve jaw-dropping results.
1. Let’s talk about sharp edges
Computers are great at perfection. The real world is imperfect.
This is especially true when it comes to the edges of objects. Make a cube in any 3D package and it will be mathematically perfect. Every surface will be exactly at a ninety-degree angle. All the edges match perfectly.
Real-world objects are not mathematically perfect. It’s physically impossible for two surfaces to meet at perfect right angles and remain completely sharp. If you look at the meeting of any planes, there is rounding, damage, and unevenness where they meet.
This natural (or manufactured) rounding is so important in creating visually accuracy. Why? Because these edges often create highlights. And it’s these highlights that provide crucial visual cues. They help define planes, divide surfaces, whilst adding volume and creating visual interest.
So make sure to add bevels to create a slight radius to even the sharpest of edges. This becomes especially important if your objects and surfaces are seen up close.
The problem with CAD data
Here at The Image Faculty, we deal with a lot of CAD design data. This often lacks fillets and radii. Most render engines have a function to create the appearance of a rounded edge at render time. This can be a huge time saver, as it will massively increase realism without the need for extra modeling.
You can take it a step further by adding subtle damage to an edge. Popular render engines such as Arnold and Vray have their own techniques for achieving this.
Sweat the detail. Always pay attention to your edges, and make sure to introduce rounding.
2. Pay attention to shadows
So often the type of shadow and specifically the edge quality of a shadow is ignored by 3D creators. With a wide range of CG lights available to the user, it becomes easy to make a reality breaking choice.
It’s worth noting that these principles apply to both pre-rendered and realtime lighting scenarios. However, due to technical limitations in realtime engines, the user may not have the same lighting opportunities or level of control.
A common tell-tale sign of a CG light source is razor-sharp edges. Another giveaway, are shadows that have no gradient or change in tone.
Real-life shadows are rarely sharp. A shadows edge will get softer the further away it is from the shadow-casting object. As a shadow travels away from a surface it will often become partially lit by the surrounding ambient light. This will affect its colour and intensity. Generally, a shadows intensity (it’s darkness) decreases with distance from the shadow-casting object.
So how can you avoid these common errors? Well firstly, some of the blame can be laid at the door of history. Due to previous legacy issues, most 3D packages offer types of lights that just don’t exist in the real world.
Avoid these light sources
These are two light sources that we can safely say you should avoid, or use with extreme caution.
Point light / Omni lights — emit light in all directions from a non-physical point in space. This means they can produce not only unrealistic lighting effects but also unconvincing shadows.
Ambient light — this light source or render setting has the visual effect of lifting the darks and mid-tones of your image. It’s intended effect is to give the appearance of more light bouncing around the scene. But this often leads to a desaturated, washed-out look. It’s better to increase the number of global illumination bounces or add more lights to the scene.
Stay rooted in reality
These light types are more representative of what we experience in the real world. As such, they instantly produce more believable results.
Area Lights — are surfaces that emit light. Much like a real-world window or photographic soft-box. Area light sources create natural-looking shadows with soft edges and realistic fall-off. They also offer the benefit of being visible in reflective surfaces. This is often a desirable effect and will help add realism.
Dome Lights / Skylights — simulate large light sources such as the sky. When used in combination with High Dynamic Range Images (image-based lighting), they introduce a global light source that goes a long way in establishing realistic lighting conditions. This is a great base to start from, offering you the opportunity to take the lighting (and therefore shadow quality), in any direction you choose.
Remember the simple rule that large light sources such as Area lights and Dome Lights will create soft shadows. Smaller light sources, such as spotlights, create sharper shadows.
By observing reality and making wise lighting choices, you’ll increase the realism of your shadows and boost your images believability.
3. Adding a pinch of chaos
Just as your 3D software will generate the perfect 3D cube, it also produces perfectly flat surfaces. These are the enemy of realism. We need to add imperfection to the surfaces in our scene, even those surfaces that are relatively clean, such as smooth painted interior walls.
Surfaces that seem relatively clean and unweathered still have visual variety. This can be due to lighting, dirt and scuff marks and uneven surfaces that catch highlight and self-shadow.
Make some noise
The aim of the game is to quickly and easily add subtle imperfection to your surfaces. This is most easily done using noise patterns. Applied as either colour/tone or a bump map, they are a flexible way to break up a uniform surface.
Your render engine of choice will provide noise patterns that can be applied in a variety of ways. Try mixing in noise as a part of your diffuse colour or texture. Adjust the scale so that the effect is not immediately apparent. The aim here is to subtly break up the surface so you don’t have large areas of flat colour or tone. If your wall is being lit at a glancing angle, then applying a noise as a bump will introduce undulations to the surface. This will introduce shadow and highlight, helping to add subtle depth and interest.
Easy does it
This is an effect that you want to “feel” rather than see. So when adding noise patterns be careful not to overdo the effect. Push it to the point where the pattern becomes obvious, then dial it back twenty percent.
It will surprise you what a huge difference this delicate but impactful change can have on the realism of your images, especially interior scenes that have large flat surfaces.
4. Focus on the focal length
Here’s a quick question. What’s the default focal length of the camera in your 3D package?
If you don’t know, it’s not a crime. However, if don’t understand the affect focal length has on your images, that’s a big barrier in creating photoreal imagery.
Photographers and directors-of-photography (DOPs), take great care in choosing the right focal length and framing when they compose a shot. This is an aspect often overlooked by 3D artists. It’s too easy to start framing with the default camera, giving no thought to what focal length is being used.
There are rules to this game
It’s extremely important to have a basic understanding of the principles of composition and focal length. They have a dramatic effect on how your scene is depicted and appreciated by the viewer.
Too often we see examples of 3D renderings were the lensing deviates from how a real-world camera behaves. Your 3D virtual camera is mathematically perfect, whilst physical glass lenses produce imperfect distorted imagery.
Common mistakes are super-wide shots that feature no distortion or close-up, macro style shots that have no depth-of-field.
All of these controls are available in 3D packages. But it’s important to appreciate that distortion and depth-of-field are switched off by default. With a physical camera, there is no choice, these features are naturally occurring.
Be aware of these fundamental differences. Take a real camera out (preferably a DSLR) and experience for yourself how a physical lens works. Study your results and appreciate the dramatic difference a lens choice can have. Your output will be far richer and photorealistic for the experience.
5. Adding a layer of photoreal goodness
OK, let’s make an important distinction. When we discuss photorealism it’s easy to assume that we’re aiming for an image that’s based on reality.
Whilst that’s a correct assumption, there are two realities at play here. First is what we see with our eyes and brain, and then what a camera sees. The vast majority of media we visually consume has been taken through the lens of a camera. If we can replicate some of the lens characteristics in our 3D renderings, it’s going to make a massive difference in their believability.
Boosting highlights and blooming the brights
The brightest highlights in your scene will be visible light sources and their reflections in shiny surfaces. Cameras and their lenses cope with these bright areas as best they can. And cope is the right word to use in this instance, as lenses create certain artifacts when dealing with these bright areas. The side-effects of these intense patches of light are often burn-out, flaring and blooming. These visual imperfections have become part of our visual vocabulary, so by replicating them, you will add a good measure of photorealism to an image.
It’s recommended that these effects are added in post-production as part of the colour grading or retouch phase. Here is an example of how to apply such an effect to 3D renders. The same principles can be applied to 3D animations.
Adding photographic grain
Traditional photographic film for stills and motion cameras produce image grain. As indeed do modern digital cameras, but it’s generally thought to be less visually pleasing. 3D cameras do not produce grain. Once again we’re going to counter the mathematical perfection of your 3D output by adding some real-world, analog chaos.
There are two main sources of grain that can be added to your images in post-production. It can be generated procedurally in a program like Photoshop or Nuke. The advantage of this approach is that it’s flexible, highly adjustable and can be fine-tuned based on the RGB channels of the image.
If you want to be truly authentic, then it’s possible to use scanned film which contains grain. This can be applied in a mode such as overlay. Whilst this method isn’t as adjustable as procedural techniques, it’s quick and effective.
Don’t make the mistake of applying a single frame of grain over the top of a 3D animation, it will look totally fake. Instead, use a sequence with changing grain.
The power of optical effects
These types of effects can be considered optical effects, caused by the physical interaction of light with the mechanics of a camera and it’s lense. Whilst these aren’t the only characteristics produced by cameras, including these features will immediately add a layer of photorealism to your work.
It’s about fighting perfection
As you’ve probably noticed there is a common thread running through all these points. The real world is messy, unlike the precise digital world of 3D and CGI. Achieving that extra level of believability in your 3D images and animations requires introducing imperfection.
By making real-world observations and being aware of the sometimes subtle differences you see, it’s possible to bridge the reality-gap.
📝 Read this story later in Journal.
👩💻 Wake up every Sunday morning to the week’s most noteworthy stories in Tech waiting in your inbox. Read the Noteworthy in Tech newsletter.
