G
Guest
Guest
Archived from groups: alt.tv.tech.hdtv (More info?)
> Frank Provasek wrote:
>
>> >> > If that were the only source of the problem, then everything
>> >> > would become crystal clear in stationary scenes.
>> >>
>> >> True, and that is very fatiguing to the eye, so a gaussian
>> >> blur
>> >> is applied to stationary scenes to match the motion blur...
>> >
>> > *cough* The motion blur depends on the SPEED of motion, so
>> > there
>> > is no such thing as a gaussian blur that "matches" it. Oh, and
>> > 24
>> > FPS film has a huge amount of motion blur just from the long
>> > frame
>> > exposure times, and I never noticed that fatiguing anyone's
>> > eye.
>>
>> Motion blur in video standards conversion refers to synthesized
>> frames to fill in where
>> frames are deleted or extra frames supplied. The algorithms
>> sometimes create a frame from
>> two frames 1/25 sec apart, giving an "exposure" of 1/12 sec.
>>
>> Normal 35mm motion picture film uses an exposure of 1/50 sec or
> faster.
>
> Have you ever looked at the individual frames in a fast moving
> film
> scene? The blur of one frame very commonly ends near where the
> blur of
> the next frame starts. That's actually the better way to do it,
> because using shorter exposures in fast moving scenes makes judder
> more
> noticeable. Shorter exposures, as have to be used in strongly lit
> scenes, highlight the inadequacies of 24 FPS frame rates pretty
> obviously sometimes.
>
> ...Which is not the main issue I called bullshit on: that there is
> no
> such thing as a gaussian blur that "matches" motion blur. The
> width of
> a motion blur, in pixels, depends entirely on how fast the objects
> in
> the picture are moving. The artificial blur applied to pal/ntsc
> conversion is INDEPENDENT of motion blur!
>
You have been wasting my time for a week that denying that
conversion looks bad
because of the frame rate problems, because "if that were the only
source of the problem, then everything would
become crystal clear in stationary scenes." Which if you look at a
raw output IT DOES, and since a picture that resharpens
and blurs when motion stops and starts is MORE annoying, a gaussian
blur is applied to the motionless
scenes to SUBJECTIVELY even out the "look." Then you argue about
THAT, but later say
" The artificial blur applied to pal/ntsc conversion is INDEPENDENT
of motion blur!"
Yes ..that is exactly what I said...it's an artificial GAUSSIAN blur
to SUBJECTIVELY match the
motion blur.
Definition of Gaussian Blur
http://www.maths.abdn.ac.uk/~igc/tch/mx4002/notes/node99.html
Then you argue that motion pictures blur more than video due to the
long exposure time. And claim
"the blur of one frame very commonly ends near where the blur of
the next frame starts"
IMPOSSIBLE (on a standard film camera) which typically uses a 170 or
180 degree shutter.
The film is blocked half the time so that the next frame can be
positioned. The long motion blurs you
describe are common (and physically possible) on electronic systems
ONLY, especially on material
produced with tube cameras, which have lag not applicable to film
EVER.
> Frank Provasek wrote:
>
>> >> > If that were the only source of the problem, then everything
>> >> > would become crystal clear in stationary scenes.
>> >>
>> >> True, and that is very fatiguing to the eye, so a gaussian
>> >> blur
>> >> is applied to stationary scenes to match the motion blur...
>> >
>> > *cough* The motion blur depends on the SPEED of motion, so
>> > there
>> > is no such thing as a gaussian blur that "matches" it. Oh, and
>> > 24
>> > FPS film has a huge amount of motion blur just from the long
>> > frame
>> > exposure times, and I never noticed that fatiguing anyone's
>> > eye.
>>
>> Motion blur in video standards conversion refers to synthesized
>> frames to fill in where
>> frames are deleted or extra frames supplied. The algorithms
>> sometimes create a frame from
>> two frames 1/25 sec apart, giving an "exposure" of 1/12 sec.
>>
>> Normal 35mm motion picture film uses an exposure of 1/50 sec or
> faster.
>
> Have you ever looked at the individual frames in a fast moving
> film
> scene? The blur of one frame very commonly ends near where the
> blur of
> the next frame starts. That's actually the better way to do it,
> because using shorter exposures in fast moving scenes makes judder
> more
> noticeable. Shorter exposures, as have to be used in strongly lit
> scenes, highlight the inadequacies of 24 FPS frame rates pretty
> obviously sometimes.
>
> ...Which is not the main issue I called bullshit on: that there is
> no
> such thing as a gaussian blur that "matches" motion blur. The
> width of
> a motion blur, in pixels, depends entirely on how fast the objects
> in
> the picture are moving. The artificial blur applied to pal/ntsc
> conversion is INDEPENDENT of motion blur!
>
You have been wasting my time for a week that denying that
conversion looks bad
because of the frame rate problems, because "if that were the only
source of the problem, then everything would
become crystal clear in stationary scenes." Which if you look at a
raw output IT DOES, and since a picture that resharpens
and blurs when motion stops and starts is MORE annoying, a gaussian
blur is applied to the motionless
scenes to SUBJECTIVELY even out the "look." Then you argue about
THAT, but later say
" The artificial blur applied to pal/ntsc conversion is INDEPENDENT
of motion blur!"
Yes ..that is exactly what I said...it's an artificial GAUSSIAN blur
to SUBJECTIVELY match the
motion blur.
Definition of Gaussian Blur
http://www.maths.abdn.ac.uk/~igc/tch/mx4002/notes/node99.html
Then you argue that motion pictures blur more than video due to the
long exposure time. And claim
"the blur of one frame very commonly ends near where the blur of
the next frame starts"
IMPOSSIBLE (on a standard film camera) which typically uses a 170 or
180 degree shutter.
The film is blocked half the time so that the next frame can be
positioned. The long motion blurs you
describe are common (and physically possible) on electronic systems
ONLY, especially on material
produced with tube cameras, which have lag not applicable to film
EVER.