Interlaced PAL captures are useless ?

[hypercube]

with my V7700, interlacing is performed by driver beyond 288 lines.
so:

capture A: 288 lines 25fps => not interlaced
capture B: 576 lines 25fps => interlaced

I read www.100fps.com, and I understand this:
desinterlacing without resizing capture B,
will produce 576/2 = 288 lines at 50 fps.

50 fps is unusable for SVCD.

so, why everybody talking about this stuff ?
Interlacing is useless for those, like me, who want
to make SVCD ? (I mean at 25fps)

I'm not talking about resizing because as a read in www.100fps.com,
576 interlaced frame contain two frames with 288 lines. No more.
So, none resizing will help you to reconstruct the 576 lines.

SO, in this world, there is no way to produce a capture
with 576 lines ? In fact, full PAL resolution is not 576
progressive lines, but 576 interlaced lines.
SO, there is only 288 real lines in PAL resolution.

SO, Interlacing is useless !

[flaninacupboard]

a) you're horribly confused.

b) interlacing serves a very good pupose (see a above)

c) what is your point?

an interlaced SVCD will have two fields of 288 lines, giving you a total of 25 frames a second, or 50 fields a second.

if you're interested in learning, you have all the info you need at your fingertips. quick google search "interlace progressive" comes up with a load of pages. try here http://nickyguides.digital-digest.com/interlace.htm

[hypercube]

a)

>an interlaced SVCD will have two fields of 288 lines, giving you a total of >25 frames a second, or 50 fields a second.
so what ?

if we are talking about full image information:
1 image = 1 field of 288 lines.
there is no way to found image information beyond 288 lines.
beyond 288 lines, information is for the next image (at 50fps) !!

I mean, a lot of people thinks it is better to capture beyond 288 lines
because image will be bigger. In fact, this is false, because of
interlacing ! For me:

288 progressive capture, then resize to 576
produce the same image as:
576 interlaced capture, then desinterlace (with resize)

or

288 progressive capture
produce the same image as:
576 interlaced capture, then desinterlace (without resize)

if not, why ?

[RoopeT]

If you capture progressive 288 lines, you´ll throw away half of the original information. Just capture full screen (720x576), resize to 480x576 and encode as interlaced.

[hypercube]

>you´ll throw away half of the original information
we have to precise what "information" mean.

- I am talking about 1 image information.

- you are talking about 2 image information.


half of the original information = information for image 1 (field 1)
isn't it ?

if I capture progressive 288 lines, I lose 1 complete image because
there is 2 complete image of 288 lines in 1 frame of 576 lines.

BUT I don't lose any information about the image itself !!

what I want to say, is some people say

capturing at 576 lines
=> there is more image information
=> best picture quality

I think it is wrong ! there is only 288 lines of image information in all
cases !!

[Amontillado]

you're wrong,

Consider a picture of 576 lines
if you spilt this picture up in two fields, the first field will contain the first, third, fifth etc line of the picture, the second will contain the second, forth etc. So although these two fields by them self only contain 288 lines the combined picture equals 576, and has more information then one single field. (since every single field misses half of the lines)

First try to get this, then see if you can figure out what happens wenn you record the first field in 1/50 sec, then the second 1/50 later and play them back on a medium that refreshes the first one half of lines (1 field)(in 1/50 sec) and then the other half of the lines.

And please read the links better, interlacing has been here ever since the first tv's were made, there is(was) a good reason for it and 576 interlaced lines are different the 288 progressive !

[hypercube]

And please read the links better, interlacing has been here ever since the first tv's were made, there is(was) a good reason for it and 576 interlaced lines are different the 288 progressive !

please, be carefull, I only want to talk about Interlaced PAL capture
to make SVCD. We won't talk about tv history, here. I know interlace
processing is closely related to physical properties of phosphor dots.
So yes, there is a good reason for it. BUT this is not the subject here.

Consider a picture of 576 lines
if you spilt this picture up in two fields, the first field will contain the first, third, fifth etc line of the picture, the second will contain the second, forth etc. So although these two fields by them self only contain 288 lines the combined picture equals 576, and has more information then one single field. (since every single field misses half of the lines)

your example is not realistic. a picture of 576 lines does not exist when
you capture TV signal ! There is only a frame of 576 lines containing
2 pictures of 288 lines.
if not, why ?!

[quote]First try to get this, then see if you can figure out what happens wenn you record the first field in 1/50 sec, then the second 1/50 later and play them back on a medium that refreshes the first one half of lines (1 field)(in 1/50 sec) and then the other half of the lines.[/code]
Sorry, I don't understant, my english is not so good.


The only solution I could imagine is that second field does not start a the same line than the first one.

****** this is my idea, tell me if it is true *******

So, I take a picture of 576 lines, as you said. I take the even lines to
produce field 1. The ODD lines are LOST.
I take the next pciture of 576 lines (at 50fps), I take the odd lines to
produce field 2. The EVEN lines are LOST.
I produce interlaced frame with field 1 and field 2. My framerate,
becomes 25fps.

To desinterlace, the field 1 is the ODD lines of picture and the field 2
is the EVEN lines of the NEXT picture. The EVEN lines of first picture
does not exist any more ! The ODD lines of next picture does not exist any more !

So The information is a video stream with 288 lines at 50 fps in this order:
frame 0: even lines - odd lost
frame 1: even lost - odd lines
frame 2: even lines - odd lost
frame 3: even lost - odd lines
...

playing such video, will produce a shaking image.

but it is possible to reconstruct 576 lines with 2 frame. So
framerate will be 25 fps:

frame 0: even lines - odd lost
frame 1: even lost - odd lines
=> frame 0 at 25 fps: even lines - odd lines

frame 2: even lines - odd lost
frame 3: even lost - odd lines
=> frame 1 at 25 fps: even lines - odd lines

because even and odd are not on the same time, we must apply
some algorythm like Blend,Bob or Weaves

********** end of my idea ************

am I completly crazy, or this is why deinterlacing can produce
real 576 lines ?

[Amontillado]

You seem to understood it right this time.

The picture example is indeed not completly realistic, i tried to keep it simple first cause you gave no indication what so ever that you understood even the slightest thing about tv signal/capturing.

But know, please explain again why you should want to deinterlace the svcd.
You burn a svcd interlaced so you can view it normal (interlaced) on a tv set and if nessecary deinterlace it if you plan on viewing it on a pc screen or eg projector.

There is stil a difference between a video wich is caputered at 25 frame per sec with 288 lines wich is played back with 576 lines (so the lines are simply doubled)
and with interlaced video wich is captured at 50 fields per second (288 lines per 1/50 sec) and then played back with 576 at 25 frames per second. Although there is some detail lost in the deinterlacing proces the video will be of better quality, since there was a lot more info to begin with.

[Gazorgan]

Wow, I didn't think the concept was that hard.

PAL
Pal is 25 fps, or 25 frames per second. Let's stop at this point. The concept of a frame is the sticking point. A frame can be 1 field (Progressive) or 2 fields (Interlaced). This applies to PAL, SECAM, or NTSC. There are not 50 pictures, there are 50 1/2-pictures. This is why when the field order is off the picture looks so bad, you are looking at 2 different 1/2-pictures put togehter.

You can have a 720x576 25 FPS video that is progressive ( say a DivX ). You can have the same video as an Interlaced MPEG2. What's the difference? Well the DivX really is 720x576, 30 FPS, 30 Fields/second. The MPEG2 is 30 fps, 60 Fields/second, and 720x288 per Field in resolution. They should look identical on you computer (barring artifacting from bitrate/conversion). You can't look at them on your TV, a TV out card will convert interlaced/non-interlaced as needed.

Basically it boils down to what the final product should be. For highest PAL resolution, capture in 720x576. It doesn't matter if you capture in Interlaced format, you can de-interlace during encoding. VCD's are progressive while SVCD's are Interlaced, but you can make both from the same capture.

[hypercube]

@flaninacupboard

You seem to understood it right this time.

yes. it become clear now. In fact, it is very very important
to say "1st field is only EVEN lines and 2nd field is only ODD lines".
because, in my mind, it was like "1st and 2nd fields was only EVEN lines"
In this case, it is impossible to recontsuct 576 lines with the two fields
of 288 lines.
that's it.

[/quote]But know, please explain again why you should want to deinterlace the svcd. [/quote]
I had never tried to make interlaced PAL SVCD.

If I capture at 576 interlaced lines, 25fps.
Can I make a SVCD without desinterlace my capture and
watch it perfectly on my standalone DVD player ?

@Gazorgan

Wow, I didn't think the concept was that hard.

in fact, most information about interlace is not complete. This is
why I post this thread.
www.100fps.com does not seems to talk about the EVEN/ODD tricks...

you are looking at 2 different 1/2-pictures put togehter.

yes, and theses 2 1/2-pictures are shifted with 1 line.

you can de-interlace during encoding

can I bypass this step ? to produce a interlaced MPEG 2 stream
and to produce interlaced SVCD ?

[Xesdeeni]

Your input video is interlaced.

Your SVCD can and should be encoded as interlaced.

Your DVD player and TV will show the SVCD interlaced.

You should never have to deinterlace at all.

Xesdeeni

[hypercube]

so, desinterlacing is only usefull for those who watch their
movies on PC with monitor.

In my case, I watch SVCD on TV with standalone DVD player.
So I don't have to desinterlace.

:P everything becomes clear now.

If I want to make VCD, I will have to desinterlace because
VCD does not support interlacing content ?

[flaninacupboard]

if you want to make a standard VCD (mpeg 1 352X288) then you do not need to de-interlace, as obviously being only 288 vertical resolution it only supports one field, so one field is discarded.

[hypercube]

arf,arf ! you are absolutly right ! I should better to sleep a little bit
before reply.