field dominance problem

[Marc Hangl]

Hi everyone

I've done a 3 camera wedding video. When i created the dvds there was judder on the footage. After a bit of exploring I've come to the conclusion that field dominance is the problem. As I've only really edited in dv before its not something i've never had an issue with this before so my knowledge of this subject is fairly superficial.

The footage is hdv and avchd 50i I used pro-res setting both for importing and sequence and export the final footage as dv widescreen. I am editing on a quad core i5 imac. Now that I have looked into it I have found that both the imported clips and the sequence ( and thus final footage) has upper dominance and from what i have learnt I need the final footage in lower dominance.

From the exploring i have done so far i believe there 2 methods for correcting this I change change the general sequence settings or i can use the shift fields filter on all the clips in the filter. I was hoping that someone could confirm this or correct me if i'm wrong and tell me if there is any preference between the two.

Also a question related to the above - one of the camera operators had a nd filter on which due to the dark location meant very grainy footage which i've had to put through the neat video plugin (very time intensive) thus i know that whatever i do will lead to a very slow process. I was wondering could I just import the dv footage for the dvd back into fcp and change the field dominance as this i would imagine would speed up the process massively or am i in my ignorant state missing something obvious.

One final question if i was working with avchd and hdv 50i again would i change both the sequence and import settings to solve the problem or is it just the sequence setting?

Thanks in advance to anyone who is able to help me, you have all been so knowledgable and helpful during my video making education.

Thanks again

marc

[John Wiley]

I personally had the exact same problem when I made the jump from SD to HDV a few years ago. I tried different combinations of field dominance for the sequence settings and export settings but nothing I did ever completely removed the problem, particularly when any rescaling or speed changes were invlolved. In the end I just started de-interlacing everything when rendering the final project. It takes away a little sharpness but it's not perceptible when going to a DVD for the final output.

These days I try to only ever work with progressive footage in order to avoid the interlacing issues.

[Eric Olson]

You can use dgpulldown to change the field dominance as quickly as copying the file.

DGPulldown

Select "No change" for the framerate and check "Top field first" for the field dominance.

[John Wiley]

Eric, does this program just reverse the field order of the final rendered project? If so, I think it probably won't do the trick. Of course, it wouldn't hurt to try it out, but my experience is that most of the interlacing issues occur in the downscaling stage and the conversion from HD (which always has the same field dominance) to SD (which is different depending on PAL/NTSC). Once the issues are introduced at that point, reversing the field order further downstream will not remove it.

[Eric Olson]

I've just made a few tests and dgpulldown causes the mpeg stream to be flagged as progressive with top field first pulldown rather than interlaced with top field first dominance. This looks fine on a CRT television but when the DVD is played on a progressive system you see interlace combing. Not good, but it may be possible to modify dgpulldown so it does not force the progressive flag. Otherwise, it looks you might have to rerender or transcode to fix it.

[Marc Hangl]

If I saved a 50p version and then converted that into a sd pal with lowere dominance would that get rid of the problem
cheers
marc

[Eric Olson]

A reasonable HDV to DVD workflow first converts to 1080p50, next rescales to standard definition at 50p, and then weaves the 50p back to 50i for DVD. So, what you suggest should work.

As an aside, I just modified dgpulldown so one can switch the field dominance flags of interlaced source and so it compiles on Linux and Mac OS X. I tested it on a file with incorrect field dominance and was able to remove the playback judder without reencoding. If you are interested in trying this quick fix let me know and I can post the modified source somewhere with instructions how to use it.

[Marc Hangl]

Hi
thanks for all the answers advice etc.
Eric, I would be interested in trying dgpulldown, but as i'm using someone else's computer and can only rarely get on it might be best waiting till i get a new one, hopefully in two weeks or so - mine is a G5 imac that cannot edit AVCHD and it can do hdv but its quicker and more enjoyable letting paint dry.

Just one more question for you knowledgeable guys, if I edited in 50p in fcp ie pro-res 50p in future projects then exported as dv with lower field dominance would this solve the problem - or am i missing something

thanks again

marc

[Eric Olson]

DV is always lower field first whether PAL or NTSC, so it should work. I'm not a FCP user. My intuition would be to export as pro-res 50p from FCP. Then, use this pro-res 50p master as input for authoring the blu-ray disk and the DVD down conversion.

I'm not sure if my modified dgpullup will work on a G5 because of endian differences with the IBM processor versus Intel. I'll put up a page pretty soon, and perhaps someone can test it.

[Eric Olson]

Attached to this post is source code for a version of dgpulldown that should compile in any POSIX complaint environment such as Linux and Mac OS X. After uncompressing this archive, change to the main directory and type make to compile. If all goes well, the resulting program dgpulldown can now be used to change the field dominance of interlaced source or to add soft pulldown to progressive.

Note that the program only works on elementary mpeg2 video streams. If you are starting with a stuttering DVD you first need to concatenate the relevant VOB files into a single program stream and then demultiplex the mpeg2 elementary video stream from the program stream before running dgpulldown.