What are they?
Renderers such as MentalRay, VRay, Keyshot and many others are capable of out putting render passes that each contain separate information about the scene. When an image is rendered from a 3D package the renderer will usually display the “Beauty Pass” which is a composite of the passes together. This however, is not an efficient way of producing correct results as in many cases minor tweaks will need to be made which without the use of passes would require a full re-render. This can take a lot of time and many hours would be lost producing test renders.
In many cases 3D imagery will need to be combined with live-action or other imagery, this process of called compositing. Photoshop, After Effects, Smoke and NUKE/NUKEX are all examples of compositing packages that allow you to seamlessly combine images together using layers and channels.
Each render pass may contain different information in the image such as, diffuse, specular, reflection, motion, Z-depth. These passes (separate images) can be combined together to produce a final beauty pass within a compositor. With each pass containing unique image data minor tweaks such as; colour correction, shadows, glossiness etc can be modified post-render, therefore saving time.
Common Workflow – 3D with live-action
Compositing 3D with live action footage is an art in itself which is why compositors working a major post-houses get paid the wage they do. The us of passes is essential in this scenario because there is not option to keep re rendering files at 2hours+/frame and with passes containing shadows, diffuse, spec etc there is a lot of flexibility when producing the final image.
- Beauty – usually automatically output by most renderer
- Diffuse – this contains the basica colour information
- Specularity – a grey scale image showing points of specular intensity
- Reflection – relflections with colour
- Shadow – often used with render layers to separate shadows for easier compositing. Seperating the shadows means that they can be colour matched to the other shadows with in the live-action scene.
- Matte/Mask/Alpha – if not contained in the beauty pass this pass will contain all the data required for the alpha channel (transparency) in the image
- Motion Vector (2D/3D) – motion blur is expensive at the render stage, therefore, creating an image that uses a colour ramp to determine the motion within a scene allows the compositor to add this effect with much less computing required.
- Z-depth – this contains data on the depth of the scene, useful for adding lens effects
Using Autodesk Maya I have created a scene using simple primitive objects and I have set up the lighting using IBL (Image-based-lighting) and a directional light with ray-traced shadows.
In this setup I am using multiple layers to set up the render passes on.
Default: everything is rendered as a beauty pass.
Colour Layers (Ground and Objects): colour related passes are rendered including; diffuse, spec, indirect lighting and reflection.
AO Layers (Ground and Objects): Separate layers for the ambient occlusion passes. Ground and objects are separate.
Shadow Layers (Ground and Objects): Separate layers for ground shadows and object shadows so that they can be independently manipulated in the compositor.
With all of these layers combined the final object beauty pass can be created in a compositor such as Nuke or Photoshop. This same process can be followed for the ground layer and then all of the scene components can be edited individually of each other.
Below is an example of what the ground shadow and AO passes would look like. These can be used if the objects were to be placed in another scene where the shadow data would be desirable.
With all these passes rendered out as separate images can they are now ready for compositing.