field order when using DV

[kwmarriott]

Hi: I know that if I grab analogue video with my Marvel G400 I keep everything in field order B (for pal). This works fine. However, when I grab Digital Video the field order (according to Ulead Mediastudio Pro) is A. Now: everytime I render a dv projects and output it to video tape via the G400 'BOB' I get artifacts. If the marvel outputs field order B do I convert the DV footage to B, or do I render the video in field order A (same as the capture). Indeed: is the most important thing maintaining the same field order as the capture, or having the correct output field ordr?...Phew.

[Phil Pugh]

By this I am going to assume that you are rendering your captured digital video (DV CODEC) to analogue (MJPEG CODEC).


The important thing is to maintain correct field order when output to an interlaced display system (e.g. TV).

This probably means (and I have never tried this) that you need to render your DV video to Field Order B.

Try it and see what it looks like.

[colinbr]

> This probably means (and I have never tried this) that you
> need to render your [Field Order A] DV video to Field Order B.
> Try it and see what it looks like.

Did this work? I'd be interested to know, as I'm not so sure.

I've been thinking a lot about this, and I'd like to suggest how I think this works. I am no expert so I'd be interested if anyone disagrees with any of this.

As far as I am aware there is nothing in avi files which states the field order of the file contents (which is why TMPGEnc tries to deduce the field order by scanning the file contents).

So, assuming your DV captured material really is field order A, then it will be represented thus in your video editing program:

1U1L/2U2L/3U3L/4U4L ....

that is, each frame in the editing program will have an upper field temporally first.

Unless you have synthesised effects in your editing program it will not matter whether you export as Upper or Lower Field First. The Upper or Lower setting just designates the field dominance of the effects/transitions generated in much the same way as the capture card defines the field dominance of captured video footage.

So when the exported file is output to the capture card the same

1U1L/2U2L/3U3L/4U4L ....

sequence is loaded into the capture card drivers buffers frame by frame. If the capture card is field order A then each frame will be output as an analog sequence of fields like this:

1U,1L,2U,2L,3U,3L,4U,4L ....

which is correct, but if the capture card is field order B each frame will be output as an analog sequence of fields like this:

1L,1U,2L,2U,3L,3U,4L,4U ....

and because we know that the original footage has upper field temporally before lower field, this will display as jitter as it is zig-zagging through time.

It seems to me that if your captured material is field order A and your output capture card is field order B, you need to process the footage so that the fields of subsequent frames are combined to form new frames, i.e.

1U1L/2U2L/3U3L/4U4L ....
becomes
1L2U/2L3U/3L4U ....

This is effectively a single FIELD shift of the entire footage. It is
still in the correct temporal order (but means you discard the first and last field).

When this is output frame by frame to your capture card which is field order B, you will get the following stream of fields output:

1L,2U,2L,3U,3L,4U

As the original footage has the upper frame temporally first, the analogue stream of fields is now in the correct temporal order.

I'm not sure what filter or setting in your editing program would make this field-shift happen. I think that "swap fields" just swaps the upper and lower field lines within each frame.

Does anyone know?

Feel free to shoot holes in the above - if I've made any mistakes I'd be happy to learn by them!

Colin

[colinbr]

I've been experimenting with Avisynth, and here are a couple of scripts which do the field dominance conversion from upper-field-first to lower-field-first and vice-versa.

If all your material is field A dominant (upper first) you could convert it to field B dominant (lower first) before working with it. Alternatively work with it as field A dominant and export it then convert it.

Colin

----

# LOWER to UPPER field dominance conversion
#
# This Avisynth script changes the field dominance of the input
# clip from lower-field-first to upper-field-first. In doing so
# it discards the first and last fields of the clip.

# Open the file and declare that it is frame based.
# Avisynth assumes it is bottom-field-first by default.
DirectShowSource("L:\test.avi")
AssumeFrameBased

# Split into fields
SeparateFields

# Remove 1st and last field of clip
Trim(1,framecount-2)

# Re-create frames from the fields. Because the first (lower)
# field is now missing, the new first frame will be comprised
# of the upper field of the old first frame interlaced with
# the lower field of the old second frame. Thus the output clip
# will be upper-field-first.
Weave

----

# UPPER to LOWER field dominance conversion
#
# This Avisynth script changes the field dominance of the input
# clip from upper-field-first to lower-field-first. In doing so
# it discards the first and last fields of the clip.

# Open the file and declare that it is frame based
DirectShowSource("L:\test.avi")
AssumeFrameBased

# Avisynth assumes bottom-field-first by default, so
# declare that the clip is upper-field-first
ComplementParity

# Split into fields
SeparateFields

# Remove 1st and last field of clip
Trim(1,framecount-2)

# Re-create frames from the fields. Because the first (upper)
# field is now missing, the new first frame will be comprised
# of the lower field of the old first frame interlaced with
# the upper field of the old second frame. Thus the output clip
# will be lower-field-first.
Weave

[colinbr]

Sorry, forgot to add ....

Does anyone know how to do this sort of conversion in Premiere or Media Studio?