Motion Blur - was Re: Pixar mystery...
Cathy Beier
>In article <Cww70...@unx.sas.com> sas...@caneel.unx.sas.com (Andy Barnhart) writes:
>>>
>>>A question (from a relative neophyte to quality animation):
>>>
>>>Is motion blur used a lot in animation? I have seen it in stills as a good
>>>effect to indicate motion. But in a moving picture you have motion. Doesn't
>>>the blur result naturally if the frame count is high enough (perhaps a big
>>>if) ? Do you still have to artificially produce it? The reason for this
>>>question is that it is a feature that some but not all packages offer, and
>>>I would like to gauge its importance. Most of my animation experience is
>>>with arcade style sprites, which are becoming more photorealistic all the
>>>time.
>>>--
Motion blur is critical to any high-quality animation of quickly moving
objects rendered at low frame rates. Real film cameras typically use
a 180 degree shutter, that is, the shutter is open for one half of the
frame time. This is because projectors flash each frame twice (to
reduce flicker.) Using a 180 degree shutter allows the blurred objects
to blend into a smooth (well, as smooth as possible) image with a minimum
of stobing.
For video, there are two alternatives. At 30 fps, (on frames) you have
to motion blur. This is similar to film, in that time sample is shown
twice, on the even and odd field. Or, you can render at 60 fps, (on fields)
and get incredibly smooth motion without motion blur. You can combine
fields rendering with motion blur, and it looks even better.
Try it!
thad
Pacific Data Images
(reply to th...@pdi.com)
Cathy Beier
>
>Anyway, field-rendering makes use of what some might consider an "artifact"
>or limitation of today's video standards. For my example I'll use NTSC,
>though field-rendering could be applied to PAL animations as well. To put
>it simply, NTSC has a framerate of ~30frames/sec. Each of these thirty
>frames for every second is made up of two fields, even and odd scanlines, a
>process called interlacing. The entire frame is not drawn at one time, in
>one sweep, but by scanning out all even scanlines and then all odd (or vice
>versa, I forget). By the time your TV gets around to painting out the second
>set of fields the first set have already started to fade. This is why video
>flickers.
>
Well, I think that fields are a feature, not a bug. Displaying a new
frame every 60th of a second is objectively better than only every
30th of a second.
Video flickers, really, because it is _only_ 60 fps, and most eyes
can detect flicker up to 70 Hz or so, at least anywhere but the
center of you vision (the rods are more sensitive to quick changes
than the cones)
>Field-rendering exploits the above effect by having every frame of video
>reference what ammounts to two frames of video. The even scanlines show
>one image, the odd scanlines show another. A still would look very
>bizzare, but in motion the effect is that of watching 60 frames/sec motion
>even though you're actually only seeing 30 frames/sec (this is refered
>to as 60 fields/sec). The most common way to achieve 60 fields/sec motion
>is to render your animation for 60 frames/sec. Odd and even frames are then
>interlaced together (use even scanlines from frame 1, odd from 2, even from 3,
>odd from 4, etc.) The downside is that you end up rendering twice the ammount
>of information you normally would, and then throwing half away.
It works much much better if you only render half-height fields, not full
height ones. The anti-aliasing works much better that way too. This way,
too, you only render the same number of pixels as if you were doing it on
frames. Of course, you still have the expense of doing the 3D transformations
twice, and the shuffling fields together at the end.
>
>Field-rendering and motion-blur are two means to a similar end, reduced
>temporal aliasing. I prefer motion-blur because it's more film-like, works
>best for 24fps, and just looks cool. Field-rendering makes for some nice,
>smooth animation but the end effect looks like video, and video (in my eyes)
>looks fake and cheap.
>
>: Cya,
>: Patrick Collins -- pat...@cs.jcu.edu.au
>: _ _ _
>:
Well, I said above that 60 fps is objectively better than 24 fps, but, you
know, objectivity is not what the animation business is all about. Every
single animator at PDI agrees that 24 fps looks better than 60 fps, because
it looks more like film. As a (mostly) programmer, it sort of gives me
the willies.
Thad Beier
Pacific Data Images
(send replies to th...@pdi.com)
Chris Trimble
: [field rendering] works much much better if you only render half-height
: fields, not full height ones.
Yup, this is how SI does it. This turns out to be much more efficient
than rendering motion blur with SI.
: >I prefer motion-blur because it's more film-like, works best for 24fps,
: >and just looks cool.
: Well, I said above that 60 fps is objectively better than 24 fps, but, you
: know, objectivity is not what the animation business is all about. Every
: single animator at PDI agrees that 24 fps looks better than 60 fps, because
: it looks more like film. As a (mostly) programmer, it sort of gives me
: the willies.
So, at PDI and DD, what do you generally do for television? 30fps
motion blurred, or 60fps field rendered, or 60fps both [wow that would
be a long render]?
BTW - has anyone else noticed that the quality of discussion on this group
has gone up quite a bit with this thread and the one about renderman?
- Chris
who only renders for video right now but wants to do film someday
[and is graduating in december, hint hint]
Sean C. Cunningham
: >
: >Anyway, field-rendering makes use of what some might consider an "artifact"
: >or limitation of today's video standards. For my example I'll use NTSC,
: >though field-rendering could be applied to PAL animations as well. To put
: >it simply, NTSC has a framerate of ~30frames/sec. Each of these thirty
: >frames for every second is made up of two fields, even and odd scanlines, a
: >process called interlacing. The entire frame is not drawn at one time, in
: >one sweep, but by scanning out all even scanlines and then all odd (or vice
: >versa, I forget). By the time your TV gets around to painting out the second
: >set of fields the first set have already started to fade. This is why video
: >flickers.
: >
: Well, I think that fields are a feature, not a bug. Displaying a new
: frame every 60th of a second is objectively better than only every
: 30th of a second.
It wasn't a feature when the standard was being developed, it was the best
they could do at the time. If a standard were being developed today it would
_NOT_ be interlaced (witness all the hoopla over MIT's non-interlaced, digital
HDTV standard =vs= Japan's 1125/60 standard). Interlacing was a way of
shoehorning more resolution into an established method of image broadcast,
similar to the way they shoehorned in the color subcarrier. Sorry if this
sounds kurt, but praising NTSC's interlace signal seems akin to praising
the weightloss associated with something like intestinal flu :)
: Video flickers, really, because it is _only_ 60 fps, and most eyes
: can detect flicker up to 70 Hz or so, at least anywhere but the
: center of you vision (the rods are more sensitive to quick changes
: than the cones)
NTSC's interlaced flicker is due to its horizontal frequency of ~15.75kHz, not
its vertical of 60Hz. Progressively scanning an NTSC signal up to about
31.5kHz (while retaining a 60Hz refresh) will almost eliminate the flicker
completely, along with removing noticeable scanlines (some high-end AV gear
can do just this). 60Hz x 31.5kHz computer displays (like VGA) still have
noticeable flicker, but the reason this is noticeable is due to large areas
of similar color and usually very high contrast color elements. You would
almost never see this type of content on a TV broadcast, unless it is a
graphic screen instructing you how to order Time/Life's Sounds of the 70's.
: >Field-rendering exploits the above effect by having every frame of video
: >reference what ammounts to two frames of video. The even scanlines show
: >one image, the odd scanlines show another. A still would look very
: >bizzare, but in motion the effect is that of watching 60 frames/sec motion
: >even though you're actually only seeing 30 frames/sec (this is refered
: >to as 60 fields/sec). The most common way to achieve 60 fields/sec motion
: >is to render your animation for 60 frames/sec. Odd and even frames are then
: >interlaced together (use even scanlines from frame 1, odd from 2, even from 3,
: >odd from 4, etc.) The downside is that you end up rendering twice the ammount
: >of information you normally would, and then throwing half away.
: It works much much better if you only render half-height fields, not full
: height ones. The anti-aliasing works much better that way too. This way,
: too, you only render the same number of pixels as if you were doing it on
: frames. Of course, you still have the expense of doing the 3D transformations
: twice, and the shuffling fields together at the end.
: >
Right, it would be much better. But, what I said, more than once, was that
your particular software package will determine how you accomplish field-
rendering. Not everyone has a renderer that will render only half-height
fields. So I would imagine the most common means to the end would be throwing
out half your work. In fact, I've encountered spatial aliasing problems
associated with half-height field rendering...your mileage may vary.
: >Field-rendering and motion-blur are two means to a similar end, reduced
: >temporal aliasing. I prefer motion-blur because it's more film-like, works
: >best for 24fps, and just looks cool. Field-rendering makes for some nice,
: >smooth animation but the end effect looks like video, and video (in my eyes)
: >looks fake and cheap.
: >
: >: Cya,
: >: Patrick Collins -- pat...@cs.jcu.edu.au
: >: _ _ _
: >:
: Well, I said above that 60 fps is objectively better than 24 fps, but, you
: know, objectivity is not what the animation business is all about. Every
: single animator at PDI agrees that 24 fps looks better than 60 fps, because
: it looks more like film. As a (mostly) programmer, it sort of gives me
: the willies.
Well, as a programmer or as an animator this is my my rule of thumb:
1) watch a good movie, Bladerunner, Close Encounters, Lawrence, whatever.
2) watch your local news program, your home videos, any Showscan demo
Now, which looks better??? Baring the contrast differences and color
differences between video and film (though Showscan is shot on film) there
is a noticeable difference in the quality of motion. Video is objectively
more true to life than film, in terms of its accuracy in representing the
way we percieve motion in life. But if I want real life I can look at it
all day long. Movies are not supposed to be real life but abstractions of
life, and film's look and its representation of movement gives it an
otherwordly feeling that allows the audience to detach from real life
for a short period. Sorry, it just gets me all emotional inside thinking
about it, (sniff!) :)
Besides, this would not be the first example of something more true to
life being unacceptable on the big screen (real physics in space, human
movement in toons rather than "Disney" or "toon physics"...one of the
reasons rotoscope animation looks so bad, forget that its real).
: Thad Beier
: Pacific Data Images
: (send replies to th...@pdi.com)
--
//
Digital Domain: sean.cu...@d2.com | -- \X/ / LT1 -- | There's no
N e t c o m: poc...@netcom.com | | substitute
R e a l World: Sean C. Cunningham | -* Pixel Cowboy *- | for cubes.
Cathy Beier
> So, at PDI and DD, what do you generally do for television? 30fps
>motion blurred, or 60fps field rendered, or 60fps both [wow that would
>be a long render]?
>
We do broadcast work (show openings, stuff like that) typically at 60Hz,
because it is usually moving text which has to be readable. Everything
else is mostly mixed CG with live action, which is always shot on film
at 24Hz, and so we do it at 24Hz.
An interesting motion-blur tidbit, is that for animation that is only
flashed on the screen once (that would be field rendering, or ShowScan)
you needn't motion blur the part of the screen that the viewer is focusing
on, if you know what that is.
To explain, let's look at normal film, which is flashed on the screen
twice. It is illuminated for 1/96 sec, dark for 1/96 sec, then the same
frame is illuminated again for 1/96 sec, then in the next dark interval
the film is pulled down to the next frame. Say you are looking at a moving starfield
(I first noticed this watching the film version of Battlestar Galactica, so
you know how old I am). What do you see? Believe it or not, ever star appears
as two stars flying in formation, one right ahead of the other.
What is happening is that your eye really really wants, more than anything, to
track moving things. So, your eyes start following a star. It tracks it as
it moves. The first time frame 0 is flashed, say your eye sees it in the center
of the retina. Now, the eye is moving exactly the same speed as the stars are moving,
so that the first time frame 1 is flashed it is also in the center of the retina.
But, what happens the second time frame 0 is flashed? The eye (maybe the brain,
but it's not really important) has already moved 1/2 of the distance to the
next frame, so the image appears in a different place. This is the second
image of the star that you see. Try it!
This indicates the amount of motion blur that should be used for film, 1/2 the
frame time, because that will cause the image of the stars to be blurred into
a continuous line.
But, and this is the interesting part, for 60 fps images that are shown once,
your eye can track the object, and each image on the retina will line up
exactly with the others, and you will get the appearance of perfectly crisp
motion, with no strobing or any other affect. So, if you have the hero
object moving across a background, and you can be pretty sure that your
viewer (client, whatever) will be watching the hero, you might motion
blur the background and not the hero.
Again, this is only if you want it to look as crisp and sharp as possible.
Reality dictates that people really want all the film artifacts in their
animation.
One last thing. We were doing animation for a Japanese company, and there
was a character that was speaking Japanese. We were given an English track
to do the lip-synch to. Now, to us, this seemed wrong, because when it was
dubbed into Japanese the lips wouldn't move with the sound. We were upbraided
by the client who said that "All good animation is done that way, the lips
move in English and the words are in Japanese." Cool.
Thad Beier
Pacific Data Images
(reply to th...@pdi.com)
David Ruigh O7-257 x68713
<<<stuff deleted>>>
|> Now, which looks better??? Baring the contrast differences and color
|> differences between video and film (though Showscan is shot on film) there
|> is a noticeable difference in the quality of motion. Video is objectively
|> more true to life than film, in terms of its accuracy in representing the
|> way we percieve motion in life. But if I want real life I can look at it
|> all day long. Movies are not supposed to be real life but abstractions of
|> life, and film's look and its representation of movement gives it an
|> otherwordly feeling that allows the audience to detach from real life
|> for a short period. Sorry, it just gets me all emotional inside thinking
|> about it, (sniff!) :)
|>
|> Besides, this would not be the first example of something more true to
|> life being unacceptable on the big screen (real physics in space, human
|> movement in toons rather than "Disney" or "toon physics"...one of the
|> reasons rotoscope animation looks so bad, forget that its real).
|>
|> : Thad Beier
|> : Pacific Data Images
|> : (send replies to th...@pdi.com)
|>
|> --
|> //
|> Digital Domain: sean.cu...@d2.com | -- \X/ / LT1 -- | There's no
|> N e t c o m: poc...@netcom.com | | substitute
|> R e a l World: Sean C. Cunningham | -* Pixel Cowboy *- | for cubes.
|>
Wow! one post with responses from anim. folk from PDI and DD, cool.
My emotional/visual response is that film is a superior format. 2 weeks ago
I rented Koioniscotsi (sorry about the sp.) on video, and the impact of the
film was lost (note also no dig. FX, although it is considered a landmark
effort from a visual standpoint). Another interesting facet ofthe comparison
of film to video is that film looks better even after it's transferred to
video! Even 16mm. This is, of course why most big budget TV series are still originated on film. (Though in the case of Bab 5, Sea-Quest, etc, there are certainly lots of dig. vid FX). Is it the contrast ratio, 3:2 pulldown,
grain, subtle color distortions? Recent threads in this and other groups
indicate that there is a lot of interest in transfering animations to film.
Even many music "videos" are produced on film for the "look". What a subtle
Irony that all the laboriously crafted digital images are only considered finished art when transfered to that ancient medium. What will happen when
That big Panavision is replaced by a little box with a 2000 X 3000 CCD array inside, and those photons strike not a chemical emulsion, but are transferred, megabytes in an eyeblink, to some mass digital storage medium. An era will
have ended.
Dave
Sean C. Cunningham
...snip!...
: So, at PDI and DD, what do you generally do for television? 30fps
: motion blurred, or 60fps field rendered, or 60fps both [wow that would
: be a long render]?
At DD the route taken would depend on the source material. If a series of
BG plates were shot on film at 24fps it would be best to transfer this
footage to our disks with no 3:2 pulldown or other 24->30fps conversion.
We would then animate for 24fps, matching the BG plates, and the conversion
to 30fps would be the last step in the process.
But it's very common for commercial directors to shoot film at 30fps, negating
the need for any conversion. This is somewhat of a compromise however since
you get most of the benefits of film but motion that looks more like video.
In any event, we would animate at 30fps, matching the BG plates, and the
end result would need no conversion of any kind.
In both instances motion-blur would be part of the rendering process, but
the only motion on fields would be due to the 24->30fps conversion process
necessary if the BG plates were shot at 24fps. Hopefully that all makes sense.
Also, while we're talking about motion-blur, we would attempt to match our
motion-blur to any reference motion-blur in our BG plates.
As far as I'm concerned, the only time we would ever intentionally animate
anything on fields (do 60 fields/sec animation) would be if we were adding
effects to something shot on video, which is not very likely, or if we were
asked to create something that looked like video, which is also not very
likely. Because most of the work we deal with involves matching CG elements
to some type of film footage I doubt the issue will ever be raised. You've
got to match the look of what you're given.
: BTW - has anyone else noticed that the quality of discussion on this group
: has gone up quite a bit with this thread and the one about renderman?
It's nice seeing topics that are more interesting than what seems to have
become the norm (latest FLC players, software upgrades, call for papers,etc.).
: - Chris
: who only renders for video right now but wants to do film someday
: [and is graduating in december, hint hint]
--
Sean C. Cunningham
: Wow! one post with responses from anim. folk from PDI and DD, cool.
: My emotional/visual response is that film is a superior format. 2 weeks ago
: I rented Koioniscotsi (sorry about the sp.) on video, and the impact of the
: film was lost (note also no dig. FX, although it is considered a landmark
: effort from a visual standpoint). Another interesting facet ofthe comparison
: of film to video is that film looks better even after it's transferred to
: video! Even 16mm. This is, of course why most big budget TV series are still originated on film. (Though in the case of Bab 5, Sea-Quest, etc, there are certainly lots of dig. vid FX). Is it the contrast ratio, 3:2 pulldown,
: grain, subtle color distortions? Recent threads in this and other groups
: indicate that there is a lot of interest in transfering animations to film.
: Even many music "videos" are produced on film for the "look". What a subtle
: Irony that all the laboriously crafted digital images are only considered finished art when transfered to that ancient medium. What will happen when
: That big Panavision is replaced by a little box with a 2000 X 3000 CCD array inside, and those photons strike not a chemical emulsion, but are transferred, megabytes in an eyeblink, to some mass digital storage medium. An era will
: have ended.
When directors start shooting "films" in HDTV or some other nonesense, we'll
be looking at the beginning of the end for cinema as we know it today. I hope
this doesn't happen in my lifetime...and I've got a long time to live being
only 23 :)
: Dave
Andrew M. Gough
> My emotional/visual response is that film is a superior format. 2 weeks ago
> I rented Koioniscotsi (sorry about the sp.) on video, and the impact of the
> film was lost (note also no dig. FX, although it is considered a landmark
> effort from a visual standpoint).
Koyaanisqatsi.
The biggest problem with the video, IMHO, is that it's panned &
scanned! If ever a film called for letterboxing . . .. A lot is
lost in the P&S of this film. For example, an interesting backwards-
travelling pressure wave on an LA freeway.
> Another interesting facet ofthe comparison
> of film to video is that film looks better even after it's transferred to
> video! Even 16mm.
A closely related point is that for computer animation to look "real"
or "good", one has to reproduce artifacts from the film process. The
best example of this, I think, is lens flare. On a recent episode of
Bablyon 5, they had a sequence in which a star came into view from
behind a planet--and as soon as the sun appeared, one could see lens flare.
It works for me--I thought it looked more "real" (or even "better")
with the lens flare than without it. [Although one is quite aware
that they can't be filming on location! :-) ]
Now, if every film could be shot in IMAX (or better yet, OMNIMAX) . . ..
Regards,
Andy
Disclaimer:
All views expressed are my own opinions, and not necessarily
those of Intel Corporation.
_______________________________________________________________________________
Andy Gough | Internet: andrew_...@ccm.ch.intel.com
Intel Corporation CH5-131 | ICBM : 111^55'42" W; 33^18'23" N
5000 W. Chandler Blvd. | "Knowledge is power."
Chandler, AZ 85226 | -- Francis Bacon
-------------------------------------------------------------------------------
Daniel Robert Goldman
>When directors start shooting "films" in HDTV or some other nonesense, we'll
>be looking at the beginning of the end for cinema as we know it today. I hope
>this doesn't happen in my lifetime...and I've got a long time to live being
>only 23 :)
Oh, Pockets, you're so negative! Where's that passion for digital
filmmaking? ;-)
IMHO, it *will* happen in our lifetime. Probably not HDTV, but a
higher-res digital format which can be piped directly to theaters
equipped with light valves. (Have you seen these? They're amazing.)
Much of the "film look" can be attributed to the nonlinear response
curve, the film grain, and other effects which are easy to implement
in hardware. It will look the same, feel the same, and taste the same
as film, it will simply be much easier and cheaper to record and
distribute. (Not to mention the boon for archival and redistribution
purposes.)
No-one cried when reel-to-reels and vinyl gave way to digital
mastering and CD's, because in fact they created a quantum leap in
profit margins, allowing independent distributors to compete in local
and global markets. This is why the selection at your local CD store
is roughly 10 times the selection you might have had with tapes and
LPs. Similarly, I reckon digital filmmaking has the potential to
vastly widen the distribution of independent and/or low-budget
pictures, freeing the intelligent moviegoing public from the tyranny
of the Hollywood shlockbuster(tm).
Well, I hope so.
David Ruigh O7-257 x68713
|> : Wow! one post with responses from anim. folk from PDI and DD, cool.
|>
|> : My emotional/visual response is that film is a superior format. 2 weeks ago
|> : I rented Koioniscotsi (sorry about the sp.) on video, and the impact of the
|> : film was lost (note also no dig. FX, although it is considered a landmark
<<<<my original gibberish (gibberish as in gif) deleted>>>>
|> : That big Panavision is replaced by a little box with a 2000 X 3000 CCD array inside, and those photons strike not a chemical emulsion, but are transferred, megabytes in an eyeblink, to some mass digital storage medium. An era will
|> : have ended.
|>
|> When directors start shooting "films" in HDTV or some other nonesense, we'll
|> be looking at the beginning of the end for cinema as we know it today. I hope
|> this doesn't happen in my lifetime...and I've got a long time to live being
|> only 23 :)
|>
|> : Dave
|> --
|> //
|> Digital Domain: sean.cu...@d2.com | -- \X/ / LT1 -- | There's no
|> N e t c o m: poc...@netcom.com | | substitute
|> R e a l World: Sean C. Cunningham | -* Pixel Cowboy *- | for cubes.
|>
I'am sorry (or am I?) to report (really only wild ass guessing on my part) that
it may be closer than we think, Kodak has the 2k X 3k color ccd chip available
today. It is designed for moving image capture, I believe max data rate is 25Mhz
though, which would not do 24 fps, faster ones are forthcoming. Raid racks are
available with the terrabytes that would be required to store it (we're talking
720Mb/sec at 2k*3k*32 bits color at 30fps). As soon as it becomes economicaly
advantageous to replace film, you can bet someone will do it. Think of the immense
costs of producing film, certainly the film stock cost savings alone would not
justify it ($1.5k - $2k/1000ft for camera negative, developed?), but think of the
advantages of doing the editing and post work in a completely digital domain
(sorry, no pun int.). Dallies?, heck the scene could be previewed on the set!,
and in full resolution, no need to wait for processing (concurrent editing?)
Every copy would be 1st generation quality. Release "prints" could be distributed
by fiber or satelite to a digital projection system at the theater (or in the home!).
Some respondents to my "Digital feature film" post a few months ago thought it
would happen inas little as 5 years, no one thought it would take more than 15.
Anyone want to give it a try?
Dave
Michael B. Johnson
>>poc...@netcom.com (Sean C. Cunningham) writes:
>>
>>>When directors start shooting "films" in HDTV or some other nonesense, we'll
>>>be looking at the beginning of the end for cinema as we know it today. I hope
>>>this doesn't happen in my lifetime...and I've got a long time to live being
>>>only 23 :)
>>
>>Oh, Pockets, you're so negative! Where's that passion for digital
>>filmmaking? ;-)
>>
>>IMHO, it *will* happen in our lifetime. Probably not HDTV, but a
>>higher-res digital format which can be piped directly to theaters
>>equipped with light valves. (Have you seen these? They're amazing.)
>>Much of the "film look" can be attributed to the nonlinear response
>>curve, the film grain, and other effects which are easy to implement
>>in hardware. It will look the same, feel the same, and taste the same
>>as film, it will simply be much easier and cheaper to record and
>>distribute. (Not to mention the boon for archival and redistribution
>>purposes.)
>>
While I think we're certainly hurtling towards a variety of digital
media distribution mechanisms, I've yet to be convinced (maybe Kodak's
Cineon propaganda is working on me...) that we're going to come up
with a better origination format than film for awhile. Heft that
Arriflex camera and think on how many bits of information it can
capture... It will be awhile before we have the digital equivalent
to that in anywhere near as sturdy and portable a form. Digital for
production, distribution, and delivery yes, but for live origination...
I'm unconvinced. Again, this is for the high end. For many puposes,
an MPEG2 handycam will be fine, and that might be pretty affordable
for an independent producer soon.
As for all of this digital convergence helping out the little guy; well,
I like to look at the comic book field for clues of how things might turn out,
and the lesson to be learned there (as is being learned in CD ROM based
games now) is that publishers/studios wield enormous visible
power, and the folks that hold the distribution channel (the Capitol
Cities of the comics world and the Merisels of the software world)
have an enormous amount of invisible power. Unless we see changes
in those structures, it's not going to matter an enormous amount if
there are orders of magnitudes more independent, quality creators
out there.
Well, back to writing up my thesis to see if I can do anything about
it... :-)
--
--> Michael B. Johnson -- wa...@media.mit.edu
--> MIT Media Lab -- Computer Graphics & Animation Group
--> 20 Ames St. E15-023G -- (617) 666-4119 (day office)
--> Cambridge, MA 02139 -- (617) 253-0663 (night office)
Sean C. Cunningham
: it may be closer than we think, Kodak has the 2k X 3k color ccd chip available
: today. It is designed for moving image capture, I believe max data rate is 25Mhz
: though, which would not do 24 fps, faster ones are forthcoming. Raid racks are
: available with the terrabytes that would be required to store it (we're talking
: 720Mb/sec at 2k*3k*32 bits color at 30fps). As soon as it becomes economicaly
Make that 24fps, we are talking motion-pictures here. If they, or should I
say "we" start shooting both digitally _AND_ at 30fps, I'll just stay home and
watch my laserdisc collection, thank-you-very-much.
: advantageous to replace film, you can bet someone will do it. Think of the immense
: costs of producing film, certainly the film stock cost savings alone would not
: justify it ($1.5k - $2k/1000ft for camera negative, developed?), but think of the
I think it's got a long way to go before it's cost-effective. Film is the
cheapest part of making a movie. Film-stock and development costs may be a
big concern for the indies, but not for major studios.
I think a more realistic scenario over the next couple years will be for
directors to continue to shoot on film, BUT, once shot the negative would be
scanned and all copies would be made digitally. This way you get very clean
copies, a method of reliable and safe archival, and automatic plates for
effects work.
The point being, you're not going to get any of today's great directors to
shoot with anything but film. The Scott brothers, DePalma, Scrocesse (sp),
Eastwood, etc., etc. won't do it I would imagine. Copolla has threatened to
shoot his next "film" in HDTV, but we'll see. I think you would have to
irradicate just about every director who's somebody and hire a whole new
wave of MTV-generation "filmmakers" to start shooting with glorified
camcorders.
: advantages of doing the editing and post work in a completely digital domain
: (sorry, no pun int.). Dallies?, heck the scene could be previewed on the set!,
Editing and post take place, largely, digital today. The scenario I outlined
above would make it even easier without sacrificing image quality for
technology.
Scenes are already previewed on-set. Just about everyone shoots with a
video-feed to a tape deck these days. They see the image masked off and
everything.
: and in full resolution, no need to wait for processing (concurrent editing?)
: Every copy would be 1st generation quality. Release "prints" could be distributed
: by fiber or satelite to a digital projection system at the theater (or in the home!).
: Some respondents to my "Digital feature film" post a few months ago thought it
: would happen inas little as 5 years, no one thought it would take more than 15.
You've still got to shoot on film though, digitize from here, but film has
still got to be where we start. Technology for the sake of technology is
dogsquat. A first-gen copy of dogsquat is still dogsquat.
Like I said, it's more likely we'll scan right after shooting on real film.
Projecting digital material rather than film, I'll buy this, though I'm not
for it (resolution and projection still too weak). In fact I believe that
this is already happening on a small scale (AMC).
: Anyone want to give it a try?
Not I.
Dave Ruigh
of the responses to this thread (Though, who am I to judge :).
In article <pocketsC...@netcom.com>, poc...@netcom.com (Sean C.
Cunningham) writes:
> David Ruigh O7-257 x68713 (dz...@eng.amdahl.com) wrote:
<<<<<my idle ramblings deleted>>>>>
> : it may be closer than we think, Kodak has the 2k X 3k color ccd chip
> available
<<<<<more stuffa mine chopped >>>>>
> More likely, this chip is prob'ly intended for film scanning.
The device to which I am referring is the KAF-6300 image sensor: "The 10MHz data
rate of each of it's two output channels enables the device to capture multiple
frames per second." quoting from Kodak's data sheet. Well OK, so you'd have a
little
flicker, (grin), 3 fps on a good day. We've seen nearly an order of magnitude
increase
in the speed of microprocessors in about 5 years, so it's not unreasonable to
suppose that the next generation of these devices will improve as well. The
KAI-2090CM,
however, is designed for video frame rates at 1928 X 1084 (HDTV).
>
> : available with the terrabytes that would be required to store it (we're
> talking
> : 720Mb/sec at 2k*3k*32 bits color at 30fps). As soon as it becomes
> economicaly
>
> Make that 24fps, we are talking motion-pictures here. If they, or should I
> say "we" start shooting both digitally _AND_ at 30fps, I'll just stay home
> and
> watch my laserdisc collection, thank-you-very-much.
I intend to shoot my digital movies at 30 fps. Why? That AC line freq./2 is just
so darn convenient =:+} Yeah, ok 24 looks better (though I've been told each frame
is projected twice by most modern projectors to reduce flicker).
>
> : advantageous to replace film, you can bet someone will do it. Think of the
> immense
> : costs of producing film, certainly the film stock cost savings alone would
> not
> : justify it ($1.5k - $2k/1000ft for camera negative, developed?), but think
> of the
>
> I think it's got a long way to go before it's cost-effective. Film is the
> cheapest part of making a movie. Film-stock and development costs may be a
> big concern for the indies, but not for major studios.
Scary, but true, 1000'= 11min, avg feature 90min, avg film used in origination,
perhaps 10 times as much as the cut length, say 80,000'= $160k. More if you
include workprints, release copies, mattes, etc. Many low budget films are
much more frugal with film usage, and can get by using reel ends, outdated
stock, and, ahem, using what they shot, no matter how bad. Still, that Kodak
price schedule continues to escalate, as do most petro-chem based products.
>
> I think a more realistic scenario over the next couple years will be for
> directors to continue to shoot on film, BUT, once shot the negative would be
> scanned and all copies would be made digitally. This way you get very clean
> copies, a method of reliable and safe archival, and automatic plates for
> effects work.
Agreed, for now, but in 5?
>
> The point being, you're not going to get any of today's great directors to
> shoot with anything but film. The Scott brothers, DePalma, Scrocesse (sp),
> Eastwood, etc., etc. won't do it I would imagine. Copolla has threatened to
> shoot his next "film" in HDTV, but we'll see. I think you would have to
> irradicate just about every director who's somebody and hire a whole new
> wave of MTV-generation "filmmakers" to start shooting with glorified
> camcorders.
"Throw the rascals out" I say. <;+}
Ironicaly, Many of the MTV-X'er-music-video-wannabee-directors are the ones
snapping up all that used 16mm gear that shoulda been dirt
cheap by now, so I could afford it :-{
>
> : advantages of doing the editing and post work in a completely digital domain
> : (sorry, no pun int.). Dallies?, heck the scene could be previewed on the set!,
>
> Editing and post take place, largely, digital today. The scenario I outlined
> above would make it even easier without sacrificing image quality for
> technology.
>
> Scenes are already previewed on-set. Just about everyone shoots with a
> video-feed to a tape deck these days. They see the image masked off and
> everything.
>
<<<<<more of my pedantic babble cut>>>>>
> : Some respondents to my "Digital feature film" post a few months ago thought it
> : would happen inas little as 5 years, no one thought it would take more than 15.
>
> You've still got to shoot on film though, digitize from here, but film has
> still got to be where we start. Technology for the sake of technology is
> dogsquat. A first-gen copy of dogsquat is still dogsquat.
Ah, sadly dogmerde is all around us, why I've got megabytes of my digital images,
and reels film to prove it, but I stillthink completely digital origination will
soon be possible, though the power,inertia, and tradition of film will be
difficult
to overcome. I detestoversimplified analogies (I'll use them anyway), but one
has
only to look at the long history of the phonograph (That Eddison guy again!),
and
observe how completely this medium has been purged from our society by cd's and
(still analog, dammit!) cassetes. I didn't say I liked it, and I may have no
part, other than a rhetorical one in shaping the future of this emerging digital
medium, but it will happen.
>
> Like I said, it's more likely we'll scan right after shooting on real film.
> Projecting digital material rather than film, I'll buy this, though I'm not
> for it (resolution and projection still too weak). In fact I believe that
> this is already happening on a small scale (AMC).
Yes, projection technology will probably take the longest to reach the quality
standards set by film. No theater-sized projection system I know of can approach
the 2k X 3k minimum that would be required, and it is likely that even HIGHER
resolution would be needed to eliminate pixel artifacts visiblein the front rows.
These have a rather annoying character compared with themore random, dynamic
nature of grain in film. (No pain no grain?) The questionis, will we be forced
to endure some marginal electronic substitute for film based on some
marketing/focus(haa!) group's analyses of the minimum acceptablestandard for
profitable operation?
>
> : Anyone want to give it a try?
>
> Not I.
"C'mon Francis, 3 fps isn't so bad, we'll do this time lapse thing of...
clouds, yeah, clouds man, it'll be digital...what's that? the RAID cabinet
fried it's controller board AGAIN, darn those union 'lecrishuns anyway...
Dave
Kevin Bjorke
== STNG and DS9 which use real film.
I assume you define "better" as "more projected-film-like." Perhaps you
are merely perceiving that ST shows all shoot at 30fps, to accomodate their
heavy video post requirements. Part of a thorough "film look" process
would be to remove frames and slide their dominant fields around, to
create a fake 3-2 pulldown. Because ST shoots at 30, the effect is of
greater overall smoothness -- the motion is truly in sync with the
display, and thus individual grain (which is on for slightly less time)
is also reduced. In addition, the art direction of ST is significantly
contrastier than TP, being both glossy scifi and shot almost entirely
on-stage. Look at the "vineyard scene" of the last TNG, shot out on location
in Sonoma or somewhere like that. Shockingly narrow in its color dynamics,
compared to most of the rest of the show....
--
---- Kevin Bjorke ---- Animation Scientist ---- Hi Tech Toons ----
Paul Bruggeman
Any comments on video processors/color correctors that claim "Film look". They
would lose some resolution for sure, but I am curious how much truth is behind
the advertising. One box is actually called The Film Look.
Paul
brug...@netcom.com
Sean C. Cunningham
: < discussion on Film vs Video/digital deleted >
: Any comments on video processors/color correctors that claim "Film look". They
: would lose some resolution for sure, but I am curious how much truth is behind
: the advertising. One box is actually called The Film Look.
I believe I read several years ago that this was used for the season opener
of TWIN PEAKS, and possibly the DARK SHADOWS series remake. I'm sure it was
used in TP but not so sure about the later.
I believe it adds grain and bumps up contrast, among other things. It doesn't
look all that bad, though not as good as the real thing. I'll give it this
much though, TWIN PEAKS using "Film Look" looks alot better than shows like
STNG and DS9 which use real film.
: Paul
: brug...@netcom.com
Tom Branham
The goals of the two shows were so drastically different, that
comparing them for "better" looks might be the wrong approach.
ST is really simple story telling, where media effects are more
of an intrusion than help. TP was much more deliberately visual
in nature, and very carefull use was made of lighting and the media
( ex. I doubt that the Flickering Lights so prevelant in the show
were slapped-together-in-5-minutes-scenes, I expect a good deal
of planning went into making the scene just right visually :-)
--
tom
tbra...@mm.wellesley.edu