analogue video interlacing - does BFF/TFF have any meaning?

[intracube]

Ok, this has been puzzling me for some time and I'd like to plug the hole in my knowledge

The concept of TFF/BFF is easy to understand for digital video, but less so for analogue. I've read that the field order only applies to digital as analogue is just a continuing stream of fields - one after the other. The 'first' or 'second', 'top or 'bottom' have no meaning seems to be the gist of it.

But! For that to be true, if a CRT didn't distinguish between the pairs of fields it receives it would have to scan both fields over the same area of the screen (no slight vertical offset). If that were the case, then the lines on the test image below shouldn't have any vertical jitter as both fields are the same:


But this doesn't seem to be the case. What am I missing?

[jagabo]

It's not that the CRT doesn't distinguish between top and bottom fields, but that they are presented to it sequentially, as an alternating series of top and bottom fields. The CRT then displays them sequentially, in the correct location (ie, the top field is drawn on scanlines 0,2,4,6... the bottom on scanlines 1,3,5,7... It's only when those fields are digitized and stored as frames that they become endowed with a "field order". Ie, the the player has to know which of the two fields was captured first in order to play them back in the same order.

In your case, you are starting with a digital frame. If it is sent to an interlaced device it will be peeled apart into two fields. Since you have sharp, 2 pixel high horizontal lines, each of those fields will contain half of those lines. On a true interlaced display, whether you send the files TFF or BFF, there will be a noticeable bounce of those lines.

Here's an example. I've changed your source to grey and white to make it more obvious what's going on. Here is a portion of your image enlarged 4x (nearest neighbor):



When broadcast interlaced it is split into two fields, top, bottom:





The black lines represent the lines of the other field -- that aren't displayed as the field is being drawn on the CRT. Ie, only the non-black lines are drawn on the face of the CRT, the others are left blank. Notice how one field contains all the even scan lines (top) and the other contains all the odd scan lines (bottom). This is how the TV displays them:



In practice the electron beam is blurred so the black lines between the lines of each field are barely visible.

[intracube]

Thanks for the diagrams

That confirms what I was seeing. So with that said, the question should be; how do analogue displays distinguish between the two fields?

This page has the answer to that - field synchronisation (half way down):
http://martin.hinner.info/vga/pal.html

^That's the final missing part and is something that hasn't been discussed on the forums - at least nowhere near as much as digital TFF/BFF errors. I guess if it were possible for an analogue CRT to get confused between the two fields, the effect would be something like this:

- the field sequence (order) would be correct, but the position relative to each other wouldn't be.

[jagabo]

I guess if it were possible for an analogue CRT to get confused between the two fields, the effect would be something like this:

[Attachment 13192 - Click to enlarge]
- the field sequence (order) would be correct, but the position relative to each other wouldn't be.

Yes.

[intracube]

Thanks for clearing that up jagabo.

[jagabo]

By the way, that (field position reversal) sometimes happens with video captures, especially with multi-generation VHS tapes. And some capture cards capture that way all the time (driver error). The Field Swap filter in VirtualDub and some MJEG codecs is used to fix it.