Silly questions about interlacing

[qwerty1983]

Sorry, i don't seem to understand a few things about interlacing.
Also my English may not be the best. Can you help me?

Let's say we have a purely interlaced source. For example a tv-series shot with a digital camera at a resolution of 720x576.
The video camera is capturing 50 photos/fields per second and resizes them automatically to 720x288?
I've noticed that field-no1 misses the last line and field-no2 misses the first line. So the 720x576 picture is cropped to 720x575 and then resized to 720x288?

I don't get it. When i use separatefields() in avisynth i notice that the fields contain all the information (except one line). They are just squeezed to 720x288, they don't miss every second line.

What about analog-captured material? Field-no1 consists only of lines 0,2,4,......,574 . Lines 1,3,... of field-no1 don't exist? There's a space over there?
And field-no2 consists only of lines 1,3,....,575 ???? What's in between? Nothing?
If there's nothing in between how do you fill the spaces when you digitize analog material?

WOW!! Lot's of silly questions.

[Asmodian]

Let's say we have a purely interlaced source. For example a tv-series shot with a digital camera at a resolution of 720x576.
The video camera is capturing 50 photos/fields per second and resizes them automatically to 720x288?
I've noticed that field-no1 misses the last line and field-no2 misses the first line. So the 720x576 picture is cropped to 720x575 and then resized to 720x288?

This is not how it works, the camera captures all the odd lines or all the even lines in the first field, and the other lines in the next field.

I don't get it. When i use separatefields() in avisynth i notice that the fields contain all the information (except one line). They are just squeezed to 720x288, they don't miss every second line.

They do miss every second line, the information is not squeezed to 288, only displayed without spacing them out.

What about analog-captured material? Field-no1 consists only of lines 0,2,4,......,574 . Lines 1,3,... of field-no1 don't exist? There's a space over there?
And field-no2 consists only of lines 1,3,....,575 ???? What's in between? Nothing?
If there's nothing in between how do you fill the spaces when you digitize analog material?

Interlacing is the same with analog as it is with digital. Yes, lines 1,3 of field-no1 do not exist. When digitizing you can keep it interlaced or run a deinterlacing filter on it. A deinterlacing filter can be as simple as resizing field-no1 to the full height and dropping field-no2 or as complicated as QTGMC (an Avisynth script, amazing quality).

Here is a good explanation of interlacing in video. (http://www.neuron2.net/LVG/interlacing.html)

edit: the guide is a bit old, progressive video is much more common now than it used to be, but it explains how interlaced video works very well.

[qwerty1983]

Thank you for your answer.
--------------------------------------------------------------
This is not how it works, the camera captures all the odd lines or all the even lines in the first field, and the other lines in the next field.
I got confused because i read on this site (http://www.100fps.com/) that : "Here is an example of what your digital camcorder does:
Capture field1 (captures at half the height, or full height and then resizes down):"
When he says 'full height' doesn't he mean that the camcorder captures all the lines (odd and even) ?

They do miss every second line, the information is not squeezed to 288, only displayed without spacing them out.
How is that? When i use a script like this in avisynth :
mysource
assumetff()
separatefields()
spline36resize(720,576)

i don't see any missing lines (except the top or bottom ones). Are the lines blurred together to fill the gap? When i use separatefields() i see two 'perfect' images/fields. They are just squeezed. I see no 'steps' if you know what i mean. If the odd or even lines were missing the picture would be messed up with stairs/steps i don't know how to express it in English.

[Gavino]

I remember being confused by this stuff too when I first got involved with video, especially the question of what cameras actually do when interlacing.

The answer seems to be "it depends on the camera".
Way back then, I found this post quite helpful.
Also, this article, especially the bit about 'dual-row' and 'single-row' scanning.
http://www.adamwilt.com/TechDiffs/FieldsAndFrames.html

[Asmodian]

i don't see any missing lines (except the top or bottom ones). Are the lines blurred together to fill the gap? When i use separatefields() i see two 'perfect' images/fields. They are just squeezed. I see no 'steps' if you know what i mean. If the odd or even lines were missing the picture would be messed up with stairs/steps i don't know how to express it in English.

This is true but what I said is also true. :p

Gavino's links probably already explained it better, the article on adamwilt.com is very good, but I'll put it in my words too. :)

While the camcorder CCD may capture all lines it is then "mixed" into one line. This isn't like a resize but more like a sensor with 1/2 the vertical resolution that moves up and down 1/2 a line for each field. You will notice that after doing a good motion compensated deinterlace the details look "sharper" due to higher vertical resolution from including the information from both fields. BTW that post at 100fps.com is also a bit out of date, QTGMC is a good motion compensation based deinterlacer and it is free. ;)

Also try running that script on a native progressive source, I do not think you will see the missing lines, only a softer picture.

[qwerty1983]

BTW that post at 100fps.com is also a bit out of date, QTGMC is a good motion compensation based deinterlacer and it is free. ;)
I know of qtgmc, i'm using it to deinterlace my dvb-t recordings.

I have one more question. When i see a half black line on the top of my video, and another one on the bottom of it, does this mean that i should be 100% sure that the content was originally shot using an analog camera?

When i'm deinterlacing with qtgmc (doubling the frame rate) i see a slight up and down movement which is more obvious when i notice the half black lines (one is slightly thinner than the other).
I know that the two fields are not at the same position and probably qtgmc tries to correct this. But perhaps the result is not absolutely perfect. Is this right?

[ChiDragon]

Both PAL and NTSC utilize half lines because each format is actually an odd number of lines (625 & 525). In (50Hz) PAL you have 312.5 lines in each field and in NTSC we have 262.5 lines/field. In NTSC we usually don't see them because several of the active lines are cropped in order to create the 480-pixel dimension divisible by 16 for MPEG encoding. European PAL's 576 active picture lines are already mod16, so you get to see everything for better or worse.

To answer more directly, any content including film transfers or HD downscales can have the half-lines introduced by being passed through equipment adhering to the legacy standards. Whereas in a fully file-based workflow you can make a 720x576 DVD without half-lines if you're starting from a higher res source.