3D Industry

Compositing and Assembling Rendered Images for 3D

>> Last Updated on

Home » 3D Industry » Tutorials » Compositing and Assembling Rendered Images for 3D

What is Compositing?

source: pixabay

Compositing is a step after rendering in the CG production workflow.  The process serves to assemble all the different render passes and layers produced by the lighting deptarment.  They are the 3D artists who finalize the look of every shot in a film.  It takes a tremendous amount of people to do this step on a feature film.  Is is also often associated with the lighting process, because they are closely interlinked tasks.  Often times, 3D artists do both  the lighting and the compositing on a same project

Compositing allows for image adjustments without requiring another long time of rendering, as long as all the base material is properly exported.  This step is only done on animated and VFX films, as the game medium doesn't allow for such a workflow.

 

Foreground vs Background

In compositing, it is important to separate the render of the foreground character with the background (the environment, also called a set in).  This is done in order to give distinct light treatments to each and to enable quick changes to the look of a character, to make it contrast with its background.  After all, the eye first looks for contrasts, and characters need to 'pop' in order to be registered easily in each shot of a film.

Main Render Passes

Render passes are related to the different material properties.  They are separated in similar categories, because each material property affects different render passes, to be controlled at compositing time, after the final images have been rendered (calculated).

Diffuse Color

The diffuse color is the main visible color of a material.  It is applied under every other type of pass, such as the specular gloss and reflectivity.  The diffuse will go against the transparency level, meaning the more an object is transparent, the less of the specular color will be seen.  It can also be entierly overridden if an object is 100% reflective or transparent.

 

Refraction / transparency

source: pixabay

The refraction or transparency is the amount of light transmitted thought the object.  What is visible behind the object will therefore be visible in front, with a distorsion equivalent to the refraction index.

 

 

Reflectivity

The reflectivity is a percentage value (0 to 1) indicating how much the material will reflect the environment around it.  It is applied over the diffuse color, and overrides it only if it is 100%.

 

Specular / glossiness

The specular or glossiness is the amount of reflection of direct light sources onto the surface of a material.  It is applied over the reflectivity and diffuse color.

 

Shadows

The shadow pass is the darkening of certain zones of the object, corresponding to the areas that are not hit by some lights.  Shadows will often appear colored inversely to the lights, as the shadows of a yellow light appear bluer, and vice-versa.  This is often also because lights are used in pairs of opposites, for contrast purposes.

 

Ambiant Occlusion

shadows light scattering subsurface occlusion
source: pixabay

The ambian occlusion pass is an automatic calculation of darkened zones, according to the folds in the geometry.  For example, a folds under a 3D model will always appear darker, as it is likely to be lit by a lot less light than the top of an object.  This pass can be rendered with the standard shadows, if it doesn't need to be adjusted separately.

 

Sub-Surface Scattering (SSS)

The sub-surface scattering is a method to calculate light entering the 3d volument and bouncing inside to create a luminous light-emitting area inside the model.  For example, a candle or empty halloween pumpkin (lit) always has subsurface scattering, and so does a human hand with a flashlight behind it.

 

Luminosity / Self-Illumination

The luminosity of an object is its ability to emit light from itself.  This is usually reserved for mesh lights, such as neons, light bulbs of LEDs.  Objects that are self-illuminated actually act like light sources like area lights.

 

Compositing Masks

Alpha

source: pixabay

An alpha is a black and white mask outlining the shape of an object.   It can be customized to isolate any object that need separate treatment in compositing (such as color correction).  It is most frequently used for separating the characters from the backgrounds.  When you need to do a lot of different alpha masks, it may be better to thing about creating puzz mattes or ID mattes instead, to minimize render times.

 

Mattes / AOV / ID / Puzz

Puzzle mattes, AOVs (Arbitrary Output Variables) and Mask IDs are all methods of creating multiple cutout masks in order to isolate sections of a rendered image in order to separate it's light treatment in compositing.  Alpha mattes or alpha masks, are black and white images, and Puzzle Mattes, AOVs or Mask IDs are flat colored masks.

 

Pre-Multiply Problem

 

There is an important thing to note when using alpha mattes, in order to avoid a frequent mistake, even professionnals make.  When you're cutting a color image with an alpha, there must not be any antialiasing or transparent softening of edges on the color pass.  Only the alpha matte should have tones of grey to fade edges.  Otherwise, this will lead to a white border on all edges, and to fix this, we will need to degrade the edges farther inwards that it was planned at render time.  This may in turn look like an aggressive greescreen keying, leading to chopped off hairs and other thin parts.

 

Technical Passes

Depth of Field (zDepth)

The zDepth is a greyscale image indicating the depth of evey pixel from the camera.  This can  be used to recreate depth-of-field after rendering (during compositing).  It isn't always a  perfect way to recreate depth, but it is much less time consuming that calculating depth inside the render engine.  Additionnally, depth calculated in the renders cannot be adjusted, reduced or removed afterwards, which then require a new render to fix.

 

deep depth of field
source: pixabay

Point Position (P)

The Pixel Position pass effectively renders pixels in 3D space, creating a point-cloud of data that can be used to insert realistic fog, in the real depth of the rendered image once in compositing.  This can also be used to create a better depth for out-of-focus areas of the image.

 

Full depth EXR

EXR files can render different layers or slices, depending on the depth of 3d objects.  This is nice, but rarely used because it is very slow to render and creates extremely big files (approx. 1Gb per 2K image)

 

UVs / Normals

The UVs or normals can be exported as an RGB pass, if your compositing software can use this to recreate some 3D effects at compositing time.

Compositing Render Passes

Linear Workflow

A linear workflow is essential for correct color management on large scale CG productions, especially when aimed for mass distribution such as the big screen.  It requires computer equipment, mostly monitors, that support it and a frequent calibration to assure consistency throughout the production process.

color range spectrum linear
source: pixabay

The default behavior of renderers and monitors is to apply a gamma correction to rendered images, in order to display it correctly on your computer screen.   This adjusts for the specific parameters of the monitor's color space.  However, this can introduce a bias in the image, as other image corrections are applied on top of it.  Furthermore, it reduces the effective color range of the images, limiting some compositing possibilities.

The concept of linear workflow is to work in full sRGB color range and to never apply any image adjustments to rendered images, throughout the whole pipeline, until the final moment of compositing.  This means that hardware and softwares have to be configured accordingly.  This is especially confusing since monitors apply color corrections to images according to color space without the user's involvement.  This can be tricky to set up, but is well worth the trouble for the control and exactitude of the colors, when reproduced on other screen types.

 

Order of Compositing Passes

Passes should be composited in this order, and with the overlay type shown below to best recreate the composite from the 3D renderer'S own beauty pass.  The beauty pass is the auto-composite that the renderer produce by default when you render a test frame without any passes.  It needs to follow this in order to keep respecting the rule of energy conservation, which creates lighting conditions that are convincingly realist to the eye, as does the real world.

  1. Alpha Matte (Luma or Alpha)
  2. SSS
  3. Radiosity Scatter
  4. Luminosity (Add)
  5. Specular (Add)
  6. Reflection (Add)
  7. Shadows (Multiply)
  8. Ambiant Occlusion (Multiply)
  9. Refraction Color (Standard)
  10. Diffuse Color (Standard)

 

 

FX passes compositing layers
source: pixabay

Technical passes to consider, connecting to other attributes.

 

  • Depth of Field (For depth blurring)
  • Motion vectors (For the motion blur)
  • Normals (Displays the lighting direction)
  • P (Used as the pixel depth)
  • Mattes (Used as masks)

 

Careers in 3D Compositing

Compositing Software

  • Nuke Foundry (Currently the best)
  • After Effects (Old)

 

Career Levels

  • Junior Compositor
  • Compositing Artist
  • Light / Comp Artist
  • Senior / Lead Compositor

 

Career Paths

  • Art Director
  • Director of Photography
  • Film Director

Recommended Books

Compositing for Live-action and CG films (4th edition)

Digital and VFX Compositing

Color script reference from Pixar's first 25 years

Nuke 101 Compositing

 


Back to the careers Table of Contents
blue curtain show stage Careers in the 3D Animation & Visual FX Industry


 

Featured Sections

 

Artist Hubs:

www.artstation.comwww.cgsociety.comwww.awn.comwww.highend3d.com

 

Tell us what you think, we welcome your comments!  Did we miss something?

- www.3Dwombat.com -

 

Join the Wombats

Opt-in our occasionnal newsletter to be the first to be informed of our new tutorials series, original content updates, special offers and important announcements!   It’s free and spam-free, plus you can unsubscribe at any time.

Follow @3Dwombat on Social Media