6. BRIEF OUTLINE
Conceptual and technological differences
CONVERGENCE!
Why the idea excites people, why it hasn’t happened
Game tech applications in film
Case studies
CryENGINE Cinema Sandbox (CineBox)
Issues that we have run into, solutions
9. THE DETAILS
PRESENTATION
Single view, 2D images
Interactive view, 3D assets
TIME
Static, based on the current cut
Dynamic, based on player progression
10. THE DETAILS
IMAGE GENERATION
Not time dependent
Time dependent (‘interactive framerate’)
ANIMATION
Hand tailored to one specific view
Interactively blended from library, any view
11. THE DETAILS
PHYSICS
Static. Bake the hits, ignore the misses
Dynamic, based on physical attrs
LIGHTING / MATERIALS
Lit before render time, mats often for one env
Lit real-time, need to work in any env
12. THE DETAILS
AUDIO
To the cut
Generated by physical impacts and triggers
VFX / PARTICLES
As many attempts as you’d like. Bake. Record.
All live. Based on physics, physical materials
13. THE GAP IN UNDERSTANDING
Film is linear, static, 2D
Hand-crafted, directed, optimal
Games are non-linear, dynamic, 3D
Interactive, chaotic, user-controlled
19. WHY IS THERE INTEREST?
REALTIME creative iteration, brainstorming
30-40% is ‘just finding your feet’
Fast iteration = More time experimenting
Everything realtime means realtime inter-
disiplinary collaboration
Game engines visualize complexity well
Game engines are world simulators
20. GAME TECH APPLICATIONS IN FILM
Rendering
Pre-vis, Post-vis, Pitch-vis
Blocking / layout
Lighting
Virtual Production
Performance capture
Live collaboration
Puppeting or ‘playing’ characters
Utilization of same assets
22. COMMON CONVERGENCE MYTHS
Many games are created by the
same studios that do the films!
Game assets are derived from
original film assets
Game engines are used for film
previs
26. PREVIOUS CASE STUDIES : THE ABYSS
David A. Smith built ‘Deep Core’
in Colony Editor
Team realized one wing of set
would not appear on camera
Smith went on to create Virtus
Walkthrough
29. PREVIOUS CASE STUDIES : AI
Modded engine to support
lenses
Used for Rouge City location
scouting
ILM went on to create virtual set
tracking/previs with Softimage
Realtime Viewer
38. ISSUES THAT WE HAD TO SOLVE
Rendering
Simple lights/shaders (one shading model)
Area Lighting, Improved AO, SSS, physically accurate rendering,
multiple shading models (Ward anisotropic, Cook-Torrance specular, etc)
All color and lighting calculations in linear space
Support for custom IBLs (import/export)
43. ISSUES THAT WE HAD TO SOLVE
Rendering
Simple lights/shaders (one shading model)
Area Lighting, Improved AO, SSS, physically accurate rendering,
multiple shading models (Ward anisotropic, Cook-Torrance specular, etc)
All color and rendering in linear space
Support for custom IBLs (import/export)
No support for standard rendering features
45. ISSUES THAT WE HAD TO SOLVE
Rendering
Simple lights/shaders (one shading model)
Area Lighting, Improved AO, SSS, physically accurate rendering,
multiple shading models (Ward anisotropic, Cook-Torrance specular, etc)
All color and rendering in linear space
Support for custom IBLs (import/export)
No support for standard rendering features
Floating point render targets to channels in EXR
Shot batching, rendering to system codecs
Offscreen rendering, scene pre-roll
In-viewport stereo
46. ISSUES THAT WE HAD TO SOLVE
Cameras
Game engines have no concept of ‘lenses’ or ‘filmback’
Not an acurrate camera model
51. ISSUES THAT WE HAD TO SOLVE
Cameras
Game engines have no concept of ‘lenses’ or ‘filmback’
Not an accurate camera model
Re-wrote camera model - filmback, shutter, aperture
All attrs exposed to liveSync interface (Maya/Mocap sync)
User editable tonemapper – OpenColorIO support
52. ISSUES THAT WE HAD TO SOLVE
Characters / Animation
Final anim comes from AI-pathfinding/state-
machine/blended clips/physics/etc
No external data input (motion capture, DCC apps)
Support for Vicon, Motion Analysis, Organic Motion,
PhaseSpace
Display/record, piece together clips, shoot cameras on clips
SDK / API
[Vicon vid, 2009]
53. ISSUES THAT WE HAD TO SOLVE
Characters / Animation
No external data input (motion capture, DCC apps)
Support for LiveSync to DCC apps (Maya)
Characters, cameras, lights, etc…
[liveSync Video]
54. ISSUES THAT WE HAD TO SOLVE
Difficult to integrate into a pipeline
Engines cannot handle mesh complexity and
displacement
New formats are a pain, scene translation
56. ISSUES THAT WE HAD TO SOLVE
Difficult to integrate into a pipeline
Engines cannot handle mesh complexity and
displacement
32bit mesh support, Catmull-Clark subd [vid1, vid2]
Displacement
New formats are a pain, scene translation
Alembic cache support [alembic_video]
58. BIGGEST ISSUES REMAINING
Pixel-Accurate Playback and Determinism
Physics/AI/Particles/Sound
CryENGINE is a dynamic world simulation
Still not ideal to get complex things working (crowds,
etc)
Playing nice with other devices
HD SDI video, timecode, thin clients, cameras, etc..
60. OUR VISION
Paradigm shift in how linear content is made
The power of games is REALTIME is ITERATION
Live inter-departmental collaboration
Physics/AI/Particles/Sound
CryENGINE is a dynamic ‘multiplayer’ world simulation.
Power experiment, mock up ideas fast
Less risk
61. DO YOU BELIEVE?
Do you believe what we’re doing is important? Do you
want to be part of our team?
www.crytek.com
Or speak with me.
Want to be involved in the beta?
62. THANKS:
Interviews: David A Smith (The Colony), Wilson Tang (ILM/AI)
Code Monkeys
Stephen Clement
Serhat Eser Erdem
Vaclav Kyba
Minghao Pan
Xiaomao Wu
Support
Christopher Evans
Kirthy Iyer
Carl Jones
Leon Zeer
TEAM
Hinweis der Redaktion
I researched this, but I don’t know everything
7 years Crytek
Define backlog, priorities
ILM
Avatar – I saw need for better templates
How was the week?
Has anyone seen anyone up here talking about using game engines?
Seen game engines?
Don’t have much time.
Start with something scary
TO: THE GAP
THE GAP – between film and videogames
I know this is very dramatic…
Many people mention casually that they are thinking about using a game engine.
– film isnt different every time you watch
Film from 70s doesn’t change today
TO: DETAILS
Time – this is something I will come back to over and over
Physics – crysis1, throw all props in water
Prop collisions in database
Car tire example
I know I said film was linear, static, 2D – I like that, best possible scenario.. No one wants a tire in the face
You can have the baddest ass looking game ever, but if it’s not running at an interactive framerate…
Crytek background
Real time all the time
Who has seen demos of the CryENGINE editor?
Crytek made it’s name with industry-leading interactive tools
This is not even CryEngine3, this is a youtube video from 2007
Explain the detail in Crysis1 physical world simulation.
Can fail USK for this!
TO: CONVERGENCE!!!!!
This is the slide where I convince you this is important
Sometimes %30 of the time I spent on shots was just figuring out how to make the scene ‘cool’
==more time trying out ideas
DP CAN LIGHT A SHOT
COLLAB = don’t need to wait for a process or department, everyone can work together, iterate and brainstorm together
RENDERING – at Crytek this is our legacy, people usually ignore all the things I mentioned and want a renderer…
THIS IS ABOUT MORE THAN RENDERING
Talk was not in virtual production track
WHY ISNT IT HAPPENING?
TO: THE GAP
TO: CASE STUDIES AND MYTHS
TO: CASE STUDIES
So now I have talked about some of the past attempts, and I would like to talk about our software
Not trying to sell you anything!
HISTORY OF INNOVATION – WHAT YOU SEE IS WHAT YOU PLAY
Not trying to sell you anything!
HISTORY OF INNOVATION – WHAT YOU SEE IS WHAT YOU PLAY
Not trying to sell you anything! Just trying to get people excites.
If you’re in the beta and see a feature you don’t have yet: it’s coming.
TO: AREA LIGHTS
All bells n whistles off, ‘preview’ quality, only shadows and light, no other sampling
All bells n whistles on
NOT UNBIASED RENDERER
All bells n whistles on
More pretty pictures!
Floating point render targets to channels in EXR
Shot batching, rendering to system codecs
Offscreen rendering, scene pre-roll
In-viewport stereo
Preroll cameras
HISTORY OF STEREO
C2 - FIRST MULTIPLAT S3D GAME
ALL S3D ATTRS KEYABLE
ONE THING TO GET RIGHT
Re-wrote camera model - filmback, shutter, aperture
All attrs exposed to liveSync interface (Maya/Mocap sync)
User editable tonemapper – OpenColorIO support
Re-wrote camera model - filmback, shutter, aperture
All attrs exposed to liveSync interface (Maya/Mocap sync)
User editable tonemapper – OpenColorIO support
Re-wrote camera model - filmback, shutter, aperture
All attrs exposed to liveSync interface (Maya/Mocap sync)
User editable tonemapper – OpenColorIO support
Re-wrote camera model - filmback, shutter, aperture
All attrs exposed to liveSync interface (Maya/Mocap sync)
User editable tonemapper – OpenColorIO support
All attrs live on the cameras
Can load film stocks
Record and piece together mocap
SHOOT CAMERAS on existing clips or FINAL CAMERA
Camera able to be attached to diff spaces
14 year old system
Record and piece together mocap
SHOOT CAMERAS on existing clips or FINAL CAMERA
Camera able to be attached to diff spaces
14 year old system