Indirect Illumination: Final Gather vs. Global Illumination
转自：https://www.pluralsight.com/blog/film-games/indirect-illumination-final-gather-vs-global-illumination
If you're familiar with rendering, you're likely familiar with Global Illumination (GI) and Final Gather (FG). These two algorithms are the key components to simulating realistic indirect illumination like you would see in the real world. Without these two techniques the lighting in the CG environments would look unrealistic and extremely flat. So if both GI and FG simulate indirect illumination what is the difference between the two and which one should you be using for your 3D scenes? Despite the fact that both methods are there to achieve relatively the same goal, there are actually differences between the two and may be times when you would want to use one over the other. Let's take a close look at each method and help you determine if one method of simulating indirect illuminating is better than the other.
The Benefits of GI
Global illumination can work in different ways depending on your rendering engine. In mental ray it works by the use of photons cast out into the scene. You can imagine these photons as little balls of energy that bounce around the scene, the more they bounce around the more energy they use and eventually die off. 
However, they don't die off until they've sampled different areas of the scene and carried that sampled energy to different parts of the scene. This is what creates the indirect illumination. If you have a bright-red desk in the middle of a white room these photons will bounce off the red desk and scatter that energy onto the white walls of the room. Giving the room a red hue which is what would occur in the real world. Global illumination gives you a lot of control over your scene and creates very realistic results that is closest to what you would see in the real world. It's great to use when you have physical lights in your scene like a light bulb, torch, etc. It's vital for architectural lighting because it produces very natural results and GI allows the light to inter-reflect between different surfaces. Global illumination uses the light source to cast the photons out into your scene so it gives you a lot of control of the results you get and let's you pick and choose where and how you want the photons cast. The photons with Global Illumination don't stop after a single bounce, but rather continue to bounce around the scene until all the energy is distributed, which is how indirect lighting works in the real world.
The Benefits of FG
Final gather works a little differently than Global Illumination in that it uses light rays to sample the scene instead of photons. It casts the light rays out based on the camera view and not a light source. You can think of it as the faked method for simulating indirect illumination. However, faked doesn't always mean bad. 
Final gather is extremely fast to set up, and while it doesn't give you the control of casting the rays from a particular light source, it makes up for it with ease of use. By simply turning on Final Gather in mental ray it will begin to calculate indirect illumination in the render and produce nice-looking results quickly. In general FG is also much faster when it comes to render time. If you're using an image-based light setup for your scene then Final Gather is a must because without Final Gather the image-based light won't be calculated in the scene. Another benefit to using FG is that you actually don't even need a light source in your scene. Final Gather will simulate indirect illumination based on the colors and materials in the scene, which is also why you need Final Gather in order for image based lighting to work.
The Downsides to GI
The biggest downside to GI is simply render time. While you get more realistic results with GI you'll have to pay for it with render time. Of course, this may not be a problem for you if you only have a still image to render. However, if you're using GI for an animation then that render time can really start to add up. GI also requires more fine tuning and tweaking to get the results you want. You may notice if you begin to emit photons from a light source in your scene it probably doesn't produce the results you want, and the rendered scene will likely be dark and splotchy. These splotches are actually the photons that are cast out into the scene. To reduce the splotches you need to increase the amount of photons, which means increasing the render time as well.
The Downsides to FG
Final Gather alone is rather limited in that the light rays only bounce once in the scene, whereas with GI the photons bounce multiple times giving you more realistic results. Now, you can increase the number of bounces in the FG settings, but increasing this number means increasing the render time significantly. Suddenly, you have a render that takes just as long as GI.
Let's All Be Friends!
While FG and GI are two different methods for rendering indirect illumination, there's really no need to use one over the other because FG and GI are best used in conjunction with each other, not alone. The question of GI vs. FG doesn't hold much precedence because they're really meant to be used together. Now, the fast results you get with FG can sometimes be enough, depending on your scene but this isn't always the case, and GI is needed for much more realistic results. You can use GI in your scene, with a relatively low photon amount, which means a short render time and then enable FG to smooth out the results. Having to get smooth results with GI simply takes too many photons. By using FG with GI it allows you to get the smooth results you want and eliminate the splotches without having to increase the photon amount. Of course, there are times when you can use just one method of simulating indirect illumination. For instance, FG may be all you need for outdoor scenes and there may be no need to include GI into the calculation. If that's the case then great! But you never know what kind of lighting you'll need until you get in there and start tweaking - if the render looks good then that's what counts. If you take a look at the image above both FG and GI were used in conjunction to get the final render. If just GI was used the render time was around 6 minutes because the photon amount had to be so high in order to reduce the splotches, and even then it was not completely eliminated. With FG enabled the photon amount could be brought down and ultimately it improved the render quality and the render time stayed the same. There are downsides and benefits to both methods; it's up to you to decide which will work best for your scenario. Next time you're rendering your scene ask yourself if using them in conjunction will be the best route to take, because more likely than not you'll be happy with the results. Be sure to check out the hundreds of rendering tutorials in the Digital-Tutors library for more great tips and techniques.
