DIYbanter - View Single Post

isw

In article ,
wrote:

"Actually, 48 FPS as projected -- each frame was shown twice. "

Didn't know that. Makes sense though.

I busted "them". I believe it was a commercial for a Sharp camcorder with
good low light capabilities. They were showing what was ostensibly some kids
birthday party or something. Well back then I always had six or seven head
VCRs. I could tell when they pulled it down from the 48 Hz to 60 Hz. They
showed each from twice but every so many cycles they would show it thrice.

The video the showed was clearly on film because a camcorders does not use 24
FPS. (or 48)

Nothing happened, WTF. I didn't buy one, I didn't initiate a class action
lawsuit, in fact it is almost like who cares.

Like now, I work on "pro" audio. Know what "pro" actually means here ? It
means that the FTC and IHF standards do not apply. You got an amp the has 55
volt rails which yields about 100 WPC but they can call it 3,500 watts.
Technical Pro is one of the worst offenders but I have seen worse. I saw one
amp supposedly 1,000 watts that had a single TO-220 type chip output. There
is no way this thing was even 100 WPC.

But anyway, it is not the same with a DVD, with a good four (plus two audio)
head VCR going frame by frame you can see the effect. Two frames would be the
same and then three frames would be the same. And I am probably the only
person in the world who knows they cheated.

That's not "cheating"; it's just the standard "two-three pulldown"
technique (known as "telecine") that NTSC (but not PAL) used to convert
24 FPS movies to ~30 FPS video. It wasn't frames that were repeated,
though; it was fields (a "field" consists of either the odd-numbered
lines, or the even-numbered ones).

If you numbered fields sequentially, it went
1-1-2-2-2-3-3-4-4-4-5-5-6-6-6- ...

Worked pretty well but added a motion artifact known as "judder" because
of the irregularity. It did have the advantage of making the sound run
at the proper speed (and incidentally make the movie have the "proper"
running time).

PAL, OTOH, just overcranked the film from 24 to 25 FPS. The audio
pitches were slightly too high, people talked slightly faster, and the
film was over slightly quicker.

The 2-3 pulldown worked quite well for analog TV, but gave the MPEG
folks fits until they finally realized what was messing up their
encoding efficiency. MPEG encoding is all about finding regularities and
exploiting them, but:

If you take that field sequence
1-1-2-2-2-3-3-4-4-4-5-5-6-6-6- ...

and show it with the associated film frames:
1-1-2-2-2-3-3-4-4-4-5-5-6-6-6- ...
A-A-B-B-C-C-D-D-E-E-F-F-G-G-H ...

You see that once in a while a frame (e.g. "C") consists of two fields
*not* from the same frame, but from adjacent ones. This really messes
with the correlation between odd and even lines that makes MPEG work
well. Worse, if there is a scene change just at that point, the odd
lines and even lines are from totally different images.

The MPEG guys finally figured out that the best thing to do was just
reverse the whole process (inverse telecine) to get back to 24 "proper"
FPS with no irregular pulldown and no split frames, and then encode that.

Isaac