Problematic Interlacing

[Pierpaolo Ferlaino]

I had a similar problem with some archive footage I used for a documentary… I tried everything… At the end JES deinterlacer did a good job for me…

[David Roth weiss]

[Heather Mahoney] “The problem is that every single frame has very visible interlacing. It looks really bad on diagonal edge lines. The interlacing problem is visible no matter if you are looking at the Digibeta playing or the files that we create.”

First, how are you monitoring Heather? Are you looking at this material on only a computer monitor?

Next, what operating system are you running presently?

I can help you, but you’ll have to help me by answering questions, because you simply haven’t provided enough information in your post for anyone to possibly conduct a forensic investigation of your issue.

The Apple deinterlace filter is almost certainly not the solution. By using it you’re most likely throwing away half of your vertical resolution.

David Roth Weiss
ProMax Systems
Burbank
DRW@ProMax.com
http://www.ProMax.com
Sales | Integration | Support

David is a Creative COW contributing editor and a forum host of the Apple Final Cut Pro forum.

[Heather Mahoney]

We are looking at the material on several professional monitors as well as a computer monitor. When I am back to work on Monday I can give you exact models. I know we have a Sony HD plasma, an older Sony tube monitor and a brand new Panasonic LCD monitor. We are running Snow Leopard on the Mac editing system. I was also concerned that the method I am using to ‘fix’ the problem is also degrading the image further. There is no doubt there is a serious issue with these transfers, as other 25 pal SD material we have received from other studios has never showed this type of interlacing issue. The studio in Russia also refuses to retransfer the material which makes me somewhat suspicious as to what’s going on. Thanks.

[David Roth weiss]

[Heather Mahoney] “We are looking at the material on several professional monitors as well as a computer monitor. “

That’s good… The exact monitor is not important, just that you’re looking on a proper monitor that displays fields.

Two thoughts occur to me:

1) I wonder, since this is coming from a progressive source, i.e. film, why did the Russians telecine as interlaced in the first place?

2) Is the material being interpreted correctly by QT and FCP, or with fields reversed?

I’ve encountered reversed fileds before when consulting for clients. There is an app I have that allows me to examine the metadata in video files, and I’ve detected errors in certain files that can cause fields to be interpreted “bassakwards.”

Try the Switch Fields filter on a clip and see what that does…

BTW, sometimes fields issues get baked-in, and the only way to fix is to re-telecine.

David Roth Weiss
ProMax Systems
Burbank
DRW@ProMax.com
http://www.ProMax.com
Sales | Integration | Support

David is a Creative COW contributing editor and a forum host of the Apple Final Cut Pro forum.

[Michael Gissing]

PAL SD on digi beta is always interlaced even from telecine. It is supposed to be upper field. When you capture the material, do a test and see if you need to set the fields to lower when you capture, in case the digi beta is wrong.

I suspect this is the likely problem if it is displaying incorrect field order on a monitor straight off tape before capture. If capturing with field order reversed or using the shift fields filter doesn’t fix the problem then deinterlacing will mostly fix the problem but a small amount of edge effects and stair stepping will remain. I have had the same issue on telecine material before and deinterlacing was the best fix as the field order issue was effectively baked in as David has suggested. The source material is progressive but captured as psf.

[John Heagy]

24fps film should have been played back at 25fps on the telecine producing no interlaced frames. Interlace means there are two distinct images making up a single frame. If you de-interlace a frame two ways: upper than lower are you seeing to different images or the same image shifted? I’d assume it’s the same frame as there’s no way to extract 50 images from 24.

If it is the same frame shifted it may be possible to shift one field over, returning it to a solid progressive frame assuming the amount of shift is constant. If you suspect it is the same frame take a short section with plenty of motion and de-interlace it upper and lower. Layer the two clips in FCP and set the transparency of the top clip to 50%. If you can shift the top layer so it matches the bottom exactly then they are the same.

Let me know if this is the case and I can tell you how to composite the two together and regain progressive frames without loss of resolution.

[Michael Gissing]

PAL SD will be interlaced from telecine. The two fields are identical

[John Heagy]

[Michael Gissing] “PAL SD will be interlaced from telecine. The two fields are identical”

Correct, but one should not be able to see any interlaced frames. They should be progressive frames created by two identical fields.

[Heather Mahoney]

Thank you all for your advice and guidance. Talking to the people in Russia it seems more evident now that they did not recently transfer the film, but they made the Digibeta from an old master. How that caused the interlacing problem I am not sure, but that also makes sense why the noise in the footage looks like analog noise. It appears the problem is ‘baked in’ and we are getting good results with deinterlacing the footage, so all is not lost. It’s just a shame that these aren’t brand new transfers, but you can’t always get what you want.