TV motion smoothing algorithms

At 04:10:29 Wed, 14 Sep 2011, davidrobin
wrote:
On Sep 13, 2:12*pm, Mark Myers wrote:
On Mon, 12 Sep 2011 21:23:36 +0100, Steve Thackery said...

The [Top Gear] pans and general movements showed really clear judder. It's
almost
as if the TV had no motion smoothing at all. *But, every other clip in
the preview sequence was as smooth as silk.

I thought they occasionally used effects on Top Gear, to make it look
like Saving Private Ryan. I forget the name of the technique, but it
introduces a deliberate juddering effect.

That's due to using a fast shutter speed, which makes the picture
strobe - i.e. look like you're watching someone move when illuminated
by a strobe light, so you see a rapid series of individual "pictures"
rather than normal motion.

I think this explains the juddering effect seen with the Top Gear clip
on BBC HD. For the panning shots, where you might expect to see some
motion blur, there is little if any. Frame by frame, it's like looking
at a succession of still photos. A good example is when the vehicles
are blown up, throwing out fragments at high velocity. The pieces would
have appeared considerably blurred if the shutter had been open for more
than a fraction of the frame at 25 fps, but this isn't the case.

John L

On Sep 14, 4:58*pm, John Legon wrote:
At 04:10:29 Wed, 14 Sep 2011, davidrobin









wrote:
On Sep 13, 2:12*pm, Mark Myers wrote:
On Mon, 12 Sep 2011 21:23:36 +0100, Steve Thackery said...

The [Top Gear] pans and general movements showed really clear judder.. It's
almost
as if the TV had no motion smoothing at all. *But, every other clip in
the preview sequence was as smooth as silk.

I thought they occasionally used effects on Top Gear, to make it look
like Saving Private Ryan. I forget the name of the technique, but it
introduces a deliberate juddering effect.

That's due to using a fast shutter speed, which makes the picture
strobe - i.e. look like you're watching someone move when illuminated
by a strobe light, so you see a rapid series of individual "pictures"
rather than normal motion.

I think this explains the juddering effect seen with the Top Gear clip
on BBC HD. *For the panning shots, where you might expect to see some
motion blur, there is little if any. *Frame by frame, it's like looking
at a succession of still photos. *A good example is when the vehicles
are blown up, throwing out fragments at high velocity. *The pieces would
have appeared considerably blurred if the shutter had been open for more
than a fraction of the frame at 25 fps, but this isn't the case.

I've watched it again, and of course you're right.

Cheers,
David.

At 07:14:25 Thu, 15 Sep 2011,
wrote:
On Sep 14, 4:58*pm, John Legon wrote:
At 04:10:29 Wed, 14 Sep 2011, davidrobin

I think this explains the juddering effect seen with the Top Gear clip
on BBC HD. *For the panning shots, where you might expect to see some
motion blur, there is little if any. *Frame by frame, it's like looking
at a succession of still photos. *A good example is when the vehicles
are blown up, throwing out fragments at high velocity. *The pieces would
have appeared considerably blurred if the shutter had been open for more
than a fraction of the frame at 25 fps, but this isn't the case.

I've watched it again, and of course you're right.

Thanks for that. Looking at the distances travelled by the individual
flying vehicle fragments in the 1/25 sec between frames and the slight
motion blur, it seems to me that the effective shutter speed could have
been 1/1000 sec or higher. In conventional photography, I would have
associated that with a wide open lens aperture and consequent shallow
depth of the field. Yet the blades of grass in the foreground directly
in front of the lens in these successive frames are in focus:

http://www.john-legon.co.uk/temp/TG04.JPG
http://www.john-legon.co.uk/temp/TG05.JPG

I've cropped these and saved as jpg to reduce the file size - about 90k.

--
John L

I think this has been really interesting and I've learned a lot.
Thanks, guys.

Whatever, I still hate juddery motion, even if it is done deliberately.
:-)

--
SteveT

Steve Thackery wrote:

I still hate juddery motion,
You should see me in a morning.

Bill

On Tue, 13 Sep 2011 14:12:44 +0100, Mark Myers wrote:

On Mon, 12 Sep 2011 21:23:36 +0100, Steve Thackery said...

The [Top Gear] pans and general movements showed really clear judder.
It's almost
as if the TV had no motion smoothing at all. But, every other clip in
the preview sequence was as smooth as silk.

I thought they occasionally used effects on Top Gear, to make it look
like Saving Private Ryan. I forget the name of the technique, but it
introduces a deliberate juddering effect.


The effect is called "Filmic". I can't believe that no one else has
answered your question.

And, to save a separate post. For the same bandwidth costs, 50i (50
interlaced fields per second) gives superior motion rendering over what
you get with 25p (25 progressive frames per second - a slightly faster
version of the 24fps cinema film standard) provided the camera _and_ the
TV receiver both use interlaced scanning in real time as per the purely
analogue system that was extant prior to flat panel display technology
displacing CRT based TV sets.

The big downside of the 50i standard, even in a totally analogue system
with CRT based TV sets, is the horrendous flicker of horizontal edge
detail on static (or almost static) images since such edge detail is only
refreshed at an effective rate of 25 fps.

If you've ever tried to use a low spec PC monitor at a refresh rate of
60Hz (60p frame rate), you'll appreciate why the flicker effect on edge
detail can become ever so excruciating with very slow moving or static
images (which the visual cortex can quite happily ignore when it is
pre-occupied with the higher priority task of interpreting movement within
the transmitted scenes - the lack of motion creates a situation for the
visual cortex that can be accurately described by the phrase "The Devil
makes work for idle hands").

Unfortunately, the flicker effect with the now obsolescent analogue 50i
system was further aggravated by the need to use phosphors with fast decay
times in order to avoid motion smearing in CRT based displays. The effect
of constant illumination in the image relies almost totally upon an
optical illusion known as "flicker fusion", a physiological effect within
the human visual system whereby a scene or source of light displayed as
successive brief flashes starts to look like one continuously lit scene or
light source when the flashing rate goes above a certain threshold
frequency which varies from individual to individual but is typically
around the 50 to 60Hz mark when said scene is being observed directly
rather than out of the corner of the eye where a much higher refresh rate
is required to achieve such "flicker Fusion"[1].

The downsides to the old analogue system with regard to static images
were not major ones in the context of the prime function of a television
broadcast system being that of transmitting _moving_ images.
Unfortunately, as soon as flat panel display technology became the
retailers' wet dream, a whole host of issues arose, initially with
analogue broadcasts 50i being represented on displays that buffered either
a full frame or a full field between screen refreshes which then
introduced the first of the motion artefact issues that curse the modern
LCD based displays. Sadly, the situation with the advent of DVB-T and
other digital moving picture systems has become even more problematical.

For a given bit rate limit, the choice lies between 50i where detail is
sacrificed for the sake of better motion fidelity or 25p where detail is
improved at the cost of motion fidelity.

Whichever of those two options is selected will result in a compromise in
either detail or motion rendering in a broadcasting system that is
constrained by a need to keep latency minimised to a mere second or so for
the sake of 'Live' broadcasting of realtime events.

As things stand with the current DVB-T standard, getting a TV receiver to
reduce motion artefacts in a reasonably competent way is like asking
someone to turn a pig's ear into a silk purse. IOW, with the current bit
rate limits being used, the DVB-T standard is pretty well a pig's ear
proposition.

If enough processing grunt is thrown at the problem in the TV receiver
end of the chain, we might eventually be able to enjoy better quality
reception at a price the broadcasters' bean counters will accept. I
believe that the optimum solution is to tweak the DVB-T standard so that
such TV receiver processing is assisted by additional 'hint data'
contained within the broadcast stream itself.

[1] Our enhanced sensitivity to flicker in our peripheral vision is thanks
to the evolution of visual perception that pre-dates our species existence
by some five hundred million years or so. The sensitivity arose out of the
need to be able to rapidly detect low resolution changes at the periphery
of a sighted organism's field of view so as to both detect prey and
predators alike in order to improve its chances of passing on its genes
before starving to death or becoming lunch.

--
Regards JB Good

On Mon, 19 Sep 2011 02:16:35 +0100, Johny B Good
wrote:



For a given bit rate limit, the choice lies between 50i where detail
is sacrificed for the sake of better motion fidelity or 25p where detail
is improved at the cost of motion fidelity.


Oops! I just noticed that I had omitted a very important adjective, the
word "edge" when using the word "detail" in the above paragraph.


--
Regards JB Good

In article [email protected], Johny B Good wrote:
The big downside of the 50i standard, even in a totally analogue system
with CRT based TV sets, is the horrendous flicker of horizontal edge
detail on static (or almost static) images since such edge detail is only
refreshed at an effective rate of 25 fps.

I wouldn't call it "big". It's a real effect of course, but quite rare for
enough horizontal detail to be present in a typical TV image to cause a
problem, as long as programme makers avoid things like venetian blinds or
take care to photgraph them at an angle. The occasional twittering on
horizontal edge detail is nowhere near as annoying as the result of throwing
away picture information from alternate fields in the interests of imitating
the behaviour of cine film, which causes entire objects to judder every time
they move, or the entire picture to judder if the camera moves.

I've never seen a rational explanation for doing this. My understanding of
photography is that the primary aim is to achieve the most convincing
depiction of the scene being photographed, not to imitate the technical
failings of another photographic system. Good TV production should draw the
viewers' attention *away* from the technology so they can enjoy the
programme.

Rod.
--
Virtual Access V6.3 free usenet/email software from
http://sourceforge.net/projects/virtual-access/

On Mon, 19 Sep 2011 09:37:24 +0100, Roderick Stewart
wrote:

In article [email protected], Johny B Good wrote:
The big downside of the 50i standard, even in a totally analogue system
with CRT based TV sets, is the horrendous flicker of horizontal edge
detail on static (or almost static) images since such edge detail is
only
refreshed at an effective rate of 25 fps.

I wouldn't call it "big". It's a real effect of course, but quite rare
for
enough horizontal detail to be present in a typical TV image to cause a
problem, as long as programme makers avoid things like venetian blinds or
take care to photgraph them at an angle. The occasional twittering on
horizontal edge detail is nowhere near as annoying as the result of
throwing
away picture information from alternate fields in the interests of
imitating
the behaviour of cine film, which causes entire objects to judder every
time
they move, or the entire picture to judder if the camera moves.

The UK and European 50i TV broadcast systems solve the problem of
transmitting standard 24 fps movie footage by subjecting each frame of the
movie to two fields worth of scan when they're run through the telecine
machine. Although this results in a 104 minute movie being broadcast in
only 100 minutes (along with the associated 4% sharpening of pitch in the
sound track), all of the spatial information that it is possible to
capture from the film is retained with nothing 'thrown away'. The
resulting 'filmic effect' in this case is simply that inherent in the
source material.

Since movie broadcasts don't exhibit this 'filmic effect' to any
noticable degree as far as I'm concerned, I suspect you're referring to a
deliberate 'artsy fartsy' effect applied by the "meeja studies" generation
of production 'talent' in the very mistaken belief that the production
will be enhanced in some way. Since this genus of 'filmic' effect is very
much more noticeable than the filmic effect inherent in real 24 fps movie
footage when properly broadcast, I think your assessment of the 'filmic
effect' is probably spot on.

What worries me about your reasonable righteous indignation over the
misuse of the 'filmic effect' in TV program productions is that it seems
to be tarring the cinematic movie standard with the same brush. Whilst
movement isn't as well rendered by a 24 fps movie as in a 50i TV
broadcast, it does at least offer much higher picture detail when 35mm SD
format or 70mm wide screen format film stock has been used.

The movie industry has been as equally afflicted by "Bean Counteritis" as
the TV broadcasting industry. AFAIAA, there was no obvious incentive to
"Up The Bit Rate" and produce 48 fps movies where each frame would only
need to be shown once on its trip through the projector as opposed to the
current practice of showing each frame twice in order to up the flicker
rate to a less objectionable 48 Hz.

I've never seen a rational explanation for doing this. My understanding
of
photography is that the primary aim is to achieve the most convincing
depiction of the scene being photographed, not to imitate the technical
failings of another photographic system. Good TV production should draw
the
viewers' attention *away* from the technology so they can enjoy the
programme.

Well, the effect you're describing goes way beyond a simple emulation of
the motion rendering limitation of 24 fps movie footage. Do you find
conventionally broadcast classic film footage just as annoying or is it
just this artsy fartsy filmic effect that gets you wound up?


--
Regards JB Good

On Monday, September 19th, 2011 at 23:40:44h +0100, Johny B Good wrote:

The UK and European 50i TV broadcast systems solve the problem of
^^^^^^^^^^^^^^^

Tautology -- UK is European.