TITLE: Anthrosphinx - Escape 2

NAME: Markus Altendorff
COUNTRY: Germany

EMAIL: maal-irtc20030115@anthrosphinx.de
WEBPAGE: http://www.markus-altendorff.de/

TOPIC: Escape
COPYRIGHT: I SUBMIT TO THE STANDARD RAYTRACING COMPETITION COPYRIGHT.
MPGFILE: maal_esc.mpg
RENDERER USED: 
    Cinema 4D R8.2


TOOLS USED: 
    Cinema 4D, Cinema 4D NET, Adobe Photoshop, Adobe
Premiere, Lemkesoft GraphicConverter, Apple Quicktime,
M.Pack 2


CREATION TIME: 
    Not sure, started with hand-drawn storybook
pretty soon, but got "distracted". Real work started around
Sept. 7th and got intense over the last three weeks (planned
on taking two weeks off work for this, but ended up with
prolonged weekends instead)


HARDWARE USED: 
    Apple G4 800 Dual for all kinds of things,
AMD Athlon 2400 as render slave


ANIMATION DESCRIPTION: 


From the Diary of Questioning, Volume 3:
"The last night, We were roused by an unearthly sound of
howling from the skies and found, right outside the olde
forest of Ogead, the evil Creature that now so stubbornly
defies all our efforts to make it speak.
It was coated in a monstrous armour and rested in a deep
hole in the ground. Our brave men could enchain the unholy
evil before it awoke."

Well, now the "unholy evil" is awake and bent on making it
out of the dungeons alive... with a little help from above.



VIEWING RECOMMENDATIONS: 


Any MPEG player should do (tested with Windows Media Player,
ATI Player, Quicktime Mac 9/X), including VideoLANclient
0.5.2 or higher on MacOS X which doesn't stutter on the
audio any more, but seems to introduce more distortion than,
say, Quicktime Player. Best results with Quicktime Player
and ATI Player, Windows Media Player seemed to emphasize the
compression artefacts.

Side note for Quicktime Player: At least on my machines,
the video is much darker than with any other player i've
tried. May be some gamma table effect or whatever.

Players that do "soft rescaling", i.e. pixel interpolation,
give best results.

Play it at fullscreen and step back to the distance you
would use when watching TV, e.g. 2-3 m on a 19" screen. Or
play it at 2x size. Or leave it at original size. Your
choice...

Play it twice. 1. Read the dialogue to get the story. 2.
Look at the video. Sorry for the lots of text, but there's a
limit to what story you can tell with gestures alone...

General recommendation: Make sure to set up the brightness
right. By the way, regarding one comment "it's the artist's
job to balance the lighting" - yes, i try. Really. But it's
your job to balance your screen. The videos look just fine
on mine...

Listen to the soundtrack. I hope you're into electric guitars
and stuff. And you could leave a note about the music
in the comments. My composer really likes that... :-)


DESCRIPTION OF HOW THIS ANIMATION WAS CREATED: 


+ Credits / External sources:
-----------------------------
Musical score once again composed and recorded by
T. "JoJojohn" Wessely.
Used with permission. Thanks a lot, JoJo!

Textures of planet earth: NASA Blue Marble Project
http://earthobservatory.nasa.gov/ 
"These images are freely available to educators, scientists,
museums, and the public. NASA images are generally not
copyrighted." (NASA Website).

+ Alterations to the raw renderings:
------------------------------------

Depth of field, lensflares etc. are post-process FX built
into the Cinema4D render engine, and are already part of the
rendered TIFF files.

The scenes atop the tower are composed from two render
passes: Characters and architecture. Both are modelled
together, but due to some odd side effect that i don't quite
understand, if i render all-at-once, it takes almost an hour
per picture, but if i "turn off" characters in pass A and
turn off architecture in pass B, then it's down to less than
1 minute per picture... so, i've decided to go the way of
post-processing on this one, until somebody can tell me what
i'm doing wrong in my setup here... a hint would be the
manual note: "Don't mix very large and very small things in
one scene" - looks like size relations of 0.5 to 60000 don't
mix too well...

The rendered sequences have been cut and edited in Adobe
Premiere for fade-overs between scenes and adding of
subtitles. All images are "contrast-softened" (adding a
layer of 128/128/128 grey at 8 percent opacity on top of the
tracks sequence) in Premiere, to increase the MPEG quality
(the algorithm of M.Pack 2 cranks up the contrast, so you've
got to soften before).

No further enhancements or compositings werde made.


+ Things i hate about it
------------------------
(So you critics can't say that i hadn't noticed :-)

Compression artefacts. Bah. Horrbile. Doesn't do justice to
the effort. Couldn't avoid it, though. Squeezing 3:45 in 10
MB is not for those that value image clarity above all. Look
at my homepage for a less-compressed version (broadband
recommended). Original is/was 720 x 360 for DVD :-)

0:30 the book: Had no time to make a monk-like figure
writing the text...

0:30 the inkstand is too bright. Looks like added afterwards,
but isn't. It's just my bad glass material.

at 1min35secs, the chain link that connects to the right
hand's manacle twists into an impossible position. Should've
added an angle limit there.

Made the chain links too big... (14 x 7 cm)

at 1:37, the chain link that cracks and bends open has a
"slippery" texture - it's an animated extrusion, and the
"fix texture to morphing object" tag doesn't work there.

at 1:50 etc., still no good walking cycle, and the
"climbing the stairs" cycle isn't too good, either.

Depth blur is a bit too strong here, too.

2:07 the torch enters the floor - collision detection was
too sloppy here ;-)

2:14 really missing the smeared blood around the muzzle
and up the arms here ;-)

2:20 NOBODY would walk a stairway up like this...

2:29 the flagpole has a very short "guest" appearance of
3 frames, didn't check the camera path.


+ Things i like about it
------------------------
- Everything. Man, am i proud of this video. For the first
time, it's 80 0f what i imagined...
- It finished in time! Thanks to Steve Schaneville for the
hint about compositing.
- LIP SYNCING! (hope i won't have to see my own face on
single-step-forward recording any time soon...)
- physical chain model :-) abusing the Cinema bone dynamics

+ Technical:
------------

++ Special F/X:

Lip syncing:
I've spoken the sentences on video, digitized them and put
them on a laptop (so i can keep the 3D editing on my desktop
and see both at once). Then i've tried to find the "key points"
of the lip and jaw motion and transferred them to my models
(as far as possible). And that's about it. Just lots of
step-by-step adjusting...
Usually, i set up the editor camera to show a frontal view of
the face. Then i undock and lock the controls for face expressions
(a bunch of sliders) and jaw angle (numeric field) into two
floating windows. And then it's click-drag-"press record"-"go to
next frame" for the next half an hour.

The cloaking device (2:36-2:39):
The sequence is 275 frames long.
1. build complete scene
2. animate complete scene
3. group scene into "distant" and "local" objects
4. hide "local" group
5. render "distant" group without depth blur
6. hide "distant" group
7. use video from 5. as "frontal projection texture" on
background object
8. hide sphere and girl in frame 0-199
9. render fr. 0-199, make sure camera depth blur doesn't
affect the sphere area at the last stop
10. activate sphere and girl
11. use fr. 199 as "frontal projection texture" on sphere,
no shadow calculation, texture as lighting texture
12. "fixate texture", convert it to UVW
13. explode and shatter sphere over time
14. render fr. 200-275
Simple, isn't it?

Compositing video:
I wish i could have avoided it, but (as said above) time was
of the essence.

++ The models:

+++ Right out of the archive:

"Space Eye": the large spacecraft.
First appearance in "Alien speed freak" (Jan. 2003)

+++ Out of the archive and modified:

+ "Amurel": the cat-creature is a ongoing "work-in-progress"
since Aug. 2000.
New this time:
- separated head mesh from body mesh
- facial animation done with PoseMixer
  (sort of morph targets, but with multiple channels and
mixing control)
- first try with decal textures and textures limited to
polygon selections to paint the wounds
- first try with real lip sync... need to rebuild the
muzzle, vocals like "u" look ... well... strange.
- disabled upper and lower arm twisting algorithm - it
works much better without it.

+ "Wyngz": the "flight assistant" ;-)
New this time:
- disabled upper and lower arm twisting algorithm - it
works much better without it.
- Headset added

+++ Imported / pre-made models:

+ The Monk
- the "Otto" mesh that comes with Cinema 4D was used for the
visible anatomy (i was running out of time here...)
- Cloth is a self-made mesh over splines
- of course the cloth is red - you do remember what happens
to the guys in red shirts in Star Trek Classic? ;-)

+++ Textures:

- Wounds: Alpha mask drawn in photoshop, bump map done using
gaussian blur, smeared blood by using "wind" filter and
gaussian blur, inside flesh from some photos of ham (really)

- Stone walls: Actual photos of medieval walls,
photoshopped and bodypainted to create seamless look.

- Copper bowl: bodypainted from photo of eaves gutter

- Glowing embers photo from summer barbecue party :-)

- sky texture from my panorama #275

++ The animation:

(see the SPEED description file of the Jan. 2003 round for
additional details, i'm still working the same way ;-)
This time, facial expressions and hands are controlled with
the new PoseMixer control of Cinema4D, which greatly
simplifies the setup and animation. Have just started to
scratch the surface on that one alone.

+ My personal comments:
-----------------------
This was better prepared than the "journey" video... if i
were a bit less reserved, i'd be running around shouting
"Yes! YES! YESSS!!!! BOOYAAH!" for the next few days ;-)

+ Things i've learned:
----------------------
- Save. Save often. Save NOW! Or else...
- Never ever forget about doing a preview render of any
animation!
- The last sequences always take longest to render...
- The screening on the last day before the deadline reveals
that my friends find connotations to certain scenes
that i really don't wanna talk about... :-) so i had to
add a few shots to clarify...

+ Things on my to-do list:
--------------------------
- Add more details on "Wyngz".
- Try modelling convincing clothing.
- Add a SoftIK chain with gravity dynamics to "Amurel"'s
tail.
- Maybe add gravity dynamics to other body parts, too... ;-)
- Learn how to really use 3D software.
- Build a real panoramic camera (at least that's going
along...)
- Finish composing all these panoramic pictures in the
slide films hanging from my locker
- Get a life.
;-)

