Hello Steven,
have a nice holiday!
Please don't mind these (and the following) words, I know, that you have
a slightly different opinion towards interlacing than I have... That's
perfectly OK! So enjoy that little short-story below... :-) I could not
resist :-)
How it came to that interlaced-video-bug (as far as I could get hands to
the facts...)?
In the 1920s all frames were progressive. Paintings, photographies and
the young art of making movies. The people of that time used film to
record everything. Film however is progressive and not interlaced. No
one at that time thought about interlaced displays...
Some time later some clever geeks thought that it would be a great idea
to transmit film-frames hundreds of kilometers over the air via some new
technology: radio transmission. As facsimile was existent, why not
making it use VHF instead of HF, making it faster, using it for movies?
The first "products" in that field of investigation were all
progressive. No exception. Film was recorded with conventional
16fps/20fps/24fps/25fps film-cameras, scanned, transmitted over the air
and recorded back to *film*... This way TV-transmissions were to be seen
in cinema, only. This still was progressive....
A little later that geeks thought that it would be a fine idea to not
convert the transmission back to film. It would be lots "cooler" to view
it on a cathode-ray-tube, directly. There was only one problem: The
phosphorescent materials which were good enough for "painting" some
image back on film had to short decay times. The frame was transmitted
line by line, and when viewed with the human eye directly, it was
awfully flickering. Longer decay times were not the solution as bad
smearing would have been the result. So what now? Forget about that fine
idea? No way... Film projectors used a doubled shutter to avoid flicker,
so why not use this to transmit the image? But wait... Transmitting all
images twice?!? Why not half by half? This would allow for
back-to-film-conversion and for direct viewing. A little modification to
the circuit and here we go: Double the refresh-rate reducing that
annoying flicker, without the hassle of wasting more bandwidth (yes,
704x576x25p has the same bandwidth as 704x576x50i...) than before...
this still was (guess what?) progressive... (Oh, come on, call it 2:2
pulldown/pullup, if you like ;-)...
So, when did it happen? When did the fields drift temporally away from
each other? What was the cause for this BUG we suffer from for decades,
now? Easy: the specification was not precise enough. As it had been
designed to carry one single frame (just split into to half-frames) and
no one could imagine that someday someone would (just because it was not
forbidden by the specification) misuse it sending different information
in the two fields/half-images, it happened: The Vidcon-tube was
invented. Now, a film-camera was not required any more. Now video could
be recorded directly and sent over the air... But wait... There was no
frame-buffer (possible). The fields for the first time ever showed
different points in time... Interlacing was born. Transformation back to
film? Impossible. But cinemas were outdated at that time, anyways...
Everyone watched "at home"...
So, believe it or not, interlacing is a bug. Interlacing was invented
because the specification of a/the TV-signal did not clearly state that
the fields of a frame must originate from the same point in time. The
higher temporal resolution which is always referred to be, was clearly
*not* the reason...
This is why I love to get rid of it... I *hate* interlacing... If he
were not dead already, the one who invented it were on my Zap-O-Kill
Top-10-List *VBG*...
BTW: *I* see interlacing defects (reduced resolution, alias ...) even on
an interlaced TV-Screen. It is unnatural high frequencies in the image,
it is reduced spatial resolution, it is just ugly. Even with sports. (I
was to a public presentation of a soccer game lately... SDTV on the big
screen... deinterlaced by a very simple VT deinterlacer... *Not* a
single person there identified it being a 25fps progressive image
originating from 50fields/sec. None moaned about the "jerky" image. And
even I did not notice it until they turned the signal off and the
features of that beamer were presented... It looked bad, OK, but not
worse than I expected interlaced video to be...)
>> clearly depends on what he tries to do... I (at least) prefer good
>> deinterlaced material to interlaced material as I can not tell the
>>
> It's your (and his) video.
>
As I said, it depends on what he tries to achieve. If he will only
present it on interlaced screens, than, fine, I agree, do *not*
deinterlace it. But if he ever wants to see it on a progressive screen,
then by all means deinterlace it *before* encoding. Why before? All
encoders which can encode interlaced do a bad job in terms of the
bitrate... So if limited to a DVD the 7200-9000 Mbps are not enough to
avoid visible blocks. Because all deinterlacers which can do this job
realtime in an comparable quality cost gazillions of bucks...
> NOTE: reversing a 2:2 or 2:3 pulldown IS NOT deinterlacing. Un-doing
> a film->video pulldown is LOSSLESS when correctly done.
I was referring to deinterlacing, not to pull-down reversal. I know the
difference, see above...
> Deinterlacing is, as you might guess, not lossless.
>
I know, and you know. I don't care as it still looks better on
progressive displays and it does not look worse than before on an
interlaced display (tested this extensively).
>> BTW I do record progressive only, now... *g*
>>
> Silly to record progressive if the source is interlaced. Or is
> broadcast TV in Germany progressive?
>
Very often it is 25fps/progressive and not 50fields/s... but not always.
And I really tend to like it better than the interlaced broadcasts as I
can not see alias in them... flicker at best...
> I think part of the problem is that '-s' was being accepted as valid
> even though it is ignored (does nothing). I have removed and checked
> in the change to completely remove -s.
>
Seen, and thanks! If you hadn't done it, I would have done it right
now... ;-)
> True, there should not be any green screen (which is a symptom of
> 0 filled buffers since Y'CbCr of 0/0/0 is a bright green).
>
See the other mail... Sometimes it is very easy to overlook simple
things. Esp. if you do not use 'em...
> Deinterlacing can be done at playback time if needed. So for TV
>
Yes, it can be done then. But not good enough...
> ('xine' I believe has a choice of methods to use when deinterlacing at
> playback time).
>
Yes, and all of these are very simple and look ugly...
cu
Stefan
PS: I will try hard to not react on your next interlaced vs.
deinterlaced mail ;-))))))))))))))))))))) Or, better: next time I throw
with used Pampers (my 2-weeks-old girl has lots of ammunition *gggg*) if
you tell me the "I"-word again... Where the heck is my psychologist?
As said above: If you want to deinterlace something or not, clearly
depends on what you want to achieve...
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
Mjpeg-users mailing list
Mjpeg-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mjpeg-users
