What is Multi-Pass rendering?
Examples of Michael Cauchi’s render pass breakdowns.
The term used when rendering a CG shot into separate passes. These passes may relate to the physical attributes of the object or additional utility passes. The beauty pass is broken into further passes.
There are two main ways we can break down CG renders. The first being lights, if we have three separate lights in a scene, we can render three beauty passes for each and then composite them back together. This is a simple technique giving some control over the brightness and colour of the lights.
The second is through splitting down the beauty pass into separate passes based on physical attributes. This means can have a pass for physically any attribute in the scene. If there is no information in certain passes, for example reflectivity, the pass will have no information and appear black.
Benefits of Multi Pass Rendering
The two main reasons to split into two passes is due to two reasons; speed and control.
We always want to create the best renders we can directly in our 3D packages, however, they are rarely a correct match to our background plate. If we were to just use a beauty pass, any changes must be done back in 3D- taking time to re-render.
Using multi-pass rendering, however, we have control of the attributes separately. Overly bright highlights can be brought down without effecting other parts of the shot and when the Visual Effects Super Visor wants changes done, these can be done instantaneously, with him watching over the shoulder if necessary.
The process allows the 3D artist to be left alone to work on the next shot, the compositor can work through shots more quickly and the Visual Effects Supervisor can make quick changes to the data as they like.
Different Types of Passes
Next we are going to explore the different pass types. First the light passes, visible passes which lead to the creation of our final image.
This is the default render image, when creating all our render passes we want to include this to give a basis to start our composite.
This does no cost any additional render times, as we are computing the information once, just saving it into different passes. The beauty pass takes all the information and processes it together as an image.
This is the diffuse shading of the scene, no reflections, refractions or any other effect, just the diffuse from the shaders/ textures.
It is affected by the direct lighting in a scene, so highlights and shadows are cast into geometry.
This pass can be used to alter surface colour of objects without affecting other attributes. If any other additional textures are needed within the composite, we can do this alongside the diffuse pass when applying the other surface layers- such as specular and reflection over the top.
The RAW diffuse pass is similar to the diffuse, but has no light contributions. It is much brighter and flatter than the standard diffuse pass as it has no light information, it just shows the texture and colour information on the objects, therefore no dark or highlight areas are shown.
The RAW diffuse pass can be multiplied but the direct irradiance pass to create a standard diffuse pass in which the textures or colours of the geo are effected by the scene lighting. We use this process if we want to change the texture or colour of our object without effecting the highlight or shadow look.
The DI pass records the direct illumination and shadows cast from all the lights in the scene.
The pass records the direct light in our scene- no global illumination of bounce light. It records the colour and light intensity and shadow information.
Indirect/ Global Illumination
The GI pass records the indirect light bounce light in the scene from all light sources.
This is similar to the diffuse pass, but instead of recording the diffuse surface lit from the direct light source, it records the diffuse surface lit from indirect bounce light.
This pass can only be created if using a global illumination render engine. Although more time consuming to create, it gives a lot more accurate lighting due to the bounce lights.
This pass can be used easily to brighten or darken areas of the shot that only receive bounce light.
An example of when this pass could be used is a cave. Outside of the cave the brightness and exposure is good, but we need to be able to see the inner cave. When we apply our GI pass we can grade just the bounce light entering the cave, without effecting our exterior lighting we are happy with.
The AO pass is not an attribute of the final beauty render which has been saved out on its own, if you broke down a beauty pass into all its parts, there would actually be no AO pass present.
AO is created and used as a way to fake a global illumination render, to give a flat surfaces graduation and additional contrast in corners and recesses.
AO can be used very effectively in shots without GI or used to subtly add extra contrast in details. However, the pass should be used very lightly if at all.
The specular pass contains just the bright highlight information in our scene. We will only have information in the specular pass if we have an object in our scene with the specular highlights on it.
With the specular information separated out, we can get better control the material appearance of our objects. The 3D artists want to get as accurate a material as possible, but sometimes tweaking the specular then re-rendering can take too long.
When we use the specular pass, we are limited to changing the amount of specular highlights we want, but we are also able to apply a wide range of effects. We could commonly make hue changes to better match the plate lighting colours and add blurring if the number of specular samples isn’t sufficient.
This contains the bounce reflection information from any reflective material in the scene.
The reflection pass contains all of the reflections of the environment, the objects self-reflection, and the secondary bounce reflections if there are any. The amount of reflection information will depend on the raytracing settings in the render engine. If the ray depth is high enough, we could get information of the reflection from a reflection.
This pass records all the information passing through transparent or semi-transparent objects.
In the similar way to the reflection pass, this pass records information taken from the environment, the object itself and other objects in the scene (if we can see through the material).
Ray limit is taken into conisderation here, controlling the number of surfaces the light can pass through before it stops.
Sub Surface Scattering
The SSS records the informations from objects where light penetrates the surface of that object.
A good example of SSS is human skin.The final appearance of the skin is a combination of light bouncing back from different layers within it – it is not a single surface colour. Without the SSS, skin tones appear flat and plastic. The way light reacts to the layers of skin is what gives the appearance of depth. SSS is the most noticeable around the ears with strong lighting behind, or in translucent liquids like milk. Marble stone is also a similar texture.
The shadow pass records all the raw shadow information in the scene.
The information can be calculated using ray traced shadows. Depending on the types of lighting in the scene, the settings may need to change to get the best out of the lighting. Lights with a clear penumbra, or area lights which cast soft shadows, should have their settings increased as otherwise they may appear grainy.
The main use of the shadow pass should be to cast shadows int he CG element onto the background place of the shot. We never use them to alter or adjust the shadows of the objects themselves.
This uses a single channel to record the z-depth away from the camera.
It can be compared to a radar image which finds out and records how far away objects are. It is not a light pass, or one we use to create our final image, but a utility pass to make out compositing work easier.
The depth pass enables us to grade parts of our image according to depth by using the depth pass as a matter for grades. We can use depth passes for hair and fog in our shots and for cutting out objects in depth.
This layer can also allow depth blur in shots.
Render Layers– allow you to render out separate objects within a scene (specific objects and lights to be rendered in each layer)
Render Passes-physical properties or utilities mentioned earlier.