#"Save 80% on render"?
7 messages in this thread
This is a long shot, so humor me if go off the deep end.
I have a suggestion for a IPAS (if they can do it) or a 3DS enhancment
that may cut down rendering time by 80% and give rendered videos a
more "film" feel. Here goes…
Being involved in film and video, I have found that people
differantiate between the two mediums by look (video is higher
contrast than film) and the feel of motion. That "feel of motion"
comes from video refeshing at 30fps and film at 24fps. That 6fps
difference gives video a more live feel. Especially if you consider
that the illusion of motion starts at only 12 to 13 fps.
When film is transferred to video, that 6fps difference is handled by
step-printing the video. On your VCR you can see this by playing it
slow – every forth frame or so is a duplicate of the previous. I'm
sure that if a IPSA routine could be made to mimick this stop
printing, you would have two benefits: A) you'ld only have to render
24 frames for a second of video; B) you'ld get a more film "feel" to
it.
What do you guys think?
There is a lot more that make up the differences between video and
film. The grain, color linearity (a ccd "sees" colors differently
than the silver nitrate cristals in films) and some other odd ball
physical characteristics.
This is food for thought. There is no clean way of doing this from an
IPAS but… there are some interesting hacks to do that. It all
depends on how you're layering the images to tape/disk.
Greg, what do you think?
I remember a panel discussion the ILM gave about the effects in T2. They said
something to the effect about spending a long time perfecting a good film grain
effect. They obviously did something right.
I wish they would have written a paper about it. Sigh.
I've done some work on this myself. I have developed a "software"
version of a process to make a video image look like film but the
amount of calculations involved is ridiculous. At last count it was
taking almost 4 minutes to process each frame on a 486/66. I've
reduced a little the number of interations to a point where, at 30
fps, it can't realy be noticeable and sent the stuff to Greg [Pyros]
so he could try this in large scale. I can rotoscope out of my laser
disk but the interface is rather crude. I have to do it all by hand!
Greg should be able to run a batch and a have a short sequence to
experiment. So far a single image looks pretty interesting.
Actually, Martin, the way that film is transferred to video is referred to as a
3:2 pulldown. Every frame of video has two fields. During the transfer, every
other film frame is given a third field. The 12 additional fields per second
compensate for the missing 6 frames. There are processes available which give
video this "film look." I think it would be very interesting to see the
technique applied to animation. –Rick
I think that it would be an interesting avenue to explore, but it's probably
quite a bit more complicated than just 24fps. Film grain, chromaticity, gamma,
and other physical properties are just as important to its "look" as the fps.
Since I'm not an expert on film physics, I can't really make any specific
comments.
– G
Martin,
I rather agree with Gus' assessment of the many differences between video and
film and their "look and feel". Film is actually made of layers of _emulsion_,
which is an organic substance that holds the silver nitrate grains that react
to the light during exposure. I get the feeling that a large part of what we
consider "film quality" comes from the way these grains are suspended in the
emulsion, and then react to the light while filming. (Not to mention all the
different resolutions, timings, color temps, etc.) This may sound a little
simplistic, but I would love to see the result of, instead of recording to
video tape, burning each frame of an animation to a film recorder at
approximately video resolution. The apparent detail would be the same as VTR
recording but may aquire the characteristics of film that we may, at some
times, be looking for. In Babylon 5, there was a bit of a dicontinuity when
they cut to video effects from filmed live action that you don't notice in,
say, Star Trek, although I have read that many of thier effects originate on
video. Maybe this technique could be used to better integrate desktop 3D
animation with network-level television?
Just some thoughts…
-Alan Iglesias