// you’re reading...


Gwilym Morris p3

The stages of computer generated animation continued from page 2...

Continued from page 2… To create a single frame of computer animation, like that of the robot in Retrofit, the stages include: matchmove, modelling, texturing …

4. Rigging – At this stage the object is a set of solid meshes like a lump of clay. It needs a method to allow it to move for the animators. Like an armature if it was a clay puppet. These joints are made in the computer. Brecht our talented rigger spent days and nights rigging a robust rig for animation. To enhance things further he was helped by Radek and Markus who re-wrote some computer code to help do things the off the shelf software couldn’t.
5. Animation – Simon and Stephanie did a great job of getting some nice clean motion capture data from Frank’s performance. This is where the character is bought to life. We had a great animation team to bring the raw mocap and plus it. Some superstar animators blow a lot of the mocap away keeping the essence of the performance. We were lucky to have a talented bunch of animators, James, Oliver, Benn, Chris and Trystan under the direction of Craig. I even got my hands dirty.
6. Lighting – This is where the integration of the robot starts to take shape and Jason and Chris did a fantastic job making it look like it was lit as if we had filmed it on set on the day. The lighting in the computer has to match the real lighting of the scene or the illusion is broken. Remember we start with a blank canvas so everything needs to be created. Pictures are taken of the set, so that tone and position of the lights can be matched. A reflective ball, grey ball and colour chart is taken before each take so that this data can be matched in each scene. The silver ball reflects where the lights are behind camera so this can be positioned correctly. Then it’s about massaging it and making it look good. But it doesn’t stop there…
If the robot casts a shadow in the filmed plate the object needs to be made because the filmed plate is just a two dimensional image with no depth and the object was never there. Even a digital Tom was made for interaction with the robot. For example the robot casts a shadow on the kitchen counter, so we took lots of photos and Tibor made the counter in the computer which acted as a shadow casting object that we could cast a shadow on. The same with the chair. At points where the robot touches the chair we found it easier to replace the chair which was made by Alex. It was lit and made so well you can’t tell which shots have a real chair in the scene and which are CG.
7. Rendering – This is where the computer crunches all the numbers and calculates reflections, shadows, and makes it all look real. This takes time and a lot of powerful computers. It took computers at www.Rendernow.com hours just to churn out a single frame at 2k resolution. Just a little interesting fact: most CG work in moves is done at 2K even though some films are shot 4K. This is because of the power needed to render these shots are so intense.
8. Rotoscoping – Again because the filmed image has no depth when the robot walks behind an object it has to be obscured by that object. The computer just places the object over the front of the movie like credits. So to integrate the robot our talented roto artist Kerensa who was a real trooper and did all the mattes herself, paints a matte around each object frame by frame basically making a white silhouette of the object. The computer looks at this object and what is black is invisible and what is white is visible. So when the image is layered over the robot image it gives the illusion the robot is behind. An example is the counter. First later is the filmed backplate with the counter. The next layer is the robot sitting at the counter but it just looks like he’s in front. The next layer is the matte. The counter is white which is seen and the room is black which is invisible. As this counter is on top of the robot layer it looks like the counter is in front of the robot and he’s sitting behind. It’s similar to layers in Photoshop.
9. Prep – This is a stage where the C-stand is painted out and replaced by the background. It’s amazing to see. We pushed the plate back with a grabber operated by Aleksandra which stood in for the robot and after Wayne cleaned up the image it looked like the plate was moving by itself.
10. Compositing – This is the last stage and the fine tweaking. Richard did a good job of integrating all the elements such as shadows, bloom, dust, glare etc. together seamlessly to create a very nice final image and a nice look.
11. Grading – Simon and Richard put a grade on the image to give it the look we wanted and this is the final version you see.

-continued next page
-copyright Simon Sandall.