Nasty interlace in 24p, what am I doing wrong??

[David Jimerson]

From the illustration on Adam Wilts site your frame grab is perfectly normal for the third frame in 24p which is shown as interlaced.

24pa will give you interlace free video. I think that is what you are after.

http://e-motionsmedia.com/K&J mix2.mov

That's not what it's showing -- the C frame isn't "interlaced"; the illustration shows that it's reconstructed from fields of what would be thought of as two different "frames" of 60i video. The thing is, in 60i video, there ARE no frames, only fields. The fields brought together to reconstruct the C frame are the correct two fields needed to make the progressive frame.

I thik it would be better for all if the whole concept of "frames" when referring to interlaced video would just be done away with.

If you're using an NLE which can remove standard 2:3 pulldown properly, you won't be seeing any interlaced frames at all once you're in 24p.

 

[CritUnitB]

From the illustration on Adam Wilts site your frame grab is perfectly normal for the third frame in 24p which is shown as interlaced.

The thing is that every single "frame" looks like this, especially with motion involved. I noticed when I watch the captured clips in quicktime, they look fine. It's only in fcp and after export that they become this way. It's very frustrating.

If you're using an NLE which can remove standard 2:3 pulldown properly, you won't be seeing any interlaced frames at all once you're in 24p.

So what do I need to do to have it removed in fcp? Reverse telecine as was said earlier? Does 24pa come with any hidden hassles or should I just use it for my web stuff?

 

[David Jimerson]

With FCP, you'd have to use Cinema Tools to do a reverse telecine; FCP doesn't remove 2:3 pulldown by itself.

24pA offers no "hassles" unless you try to edit it on a 29.97 timeline. FCP will remove the pulldown during capture, so you'll be working in native 24p -- then you'll have a full range of options for rendering for the web (except WMV).

 

[novelt]

okay if you shot standard and as i said the interlacing is normal...but you want to remove it then deinterlace. why worry about pulldowns?

 

[David Jimerson]

That'll make it WORSE. You do not want to deinterlace 24p footage; yiou want to remove the pulldown properly, or leave it as is an edit as 60i. If you try to deinterlace it as if you're deinterlacing 60i footage, you'll end up with a horrible mess.

Proper removal of pulldown (from either 24p standard or 24pA) will give you the original 24 progressive frames. It's the only way to do it correctly.

 

[novelt]

worse? what's worse than losing the "c" frame? sure it goes by fast enough to where someone can't tell you've lost a generation on the "c" frame...which is why i said might as well deinterlace. it seems everyone has spent 3 pages trying to explain pulldown when the easy-out is deinterlace. the footage is going to be compressed for web anyway...though i saw a mention of film-out.

Sorry, I should have mentioned that this is for web only

but here's the real question...

Can anyone tell me what I'm doing wrong?

shooting in standard. :grin: no seriously if you don't understand pulldown or reverse telecine...read up on it. someone here could probably give you the ins-and-outs or tke the short-cut and deinterlace. depending on your compression method...it'll look better than the stuff on youtube for sure even with deinterlacing.

 

[David Jimerson]

"Worse" is this -- if you merely deinterlace it as if it's 60i, all you're doing is shoving fields together with no regard as to what fields SHOULD go together -- or, worse still, you're interpolating fields, but several of the WRONG fields, and you're giving up half your vertical resolution to boot.

By correctly removing 2:3 pulldown, you won't lose a generation in your C frame. Maybe Cinema Tools isn't very good at it, but other software is.

But even if you do lose a single generation every four frames, simply deinterlacing is far worse.

 

[novelt]

By correctly removing 2:3 pulldown, you won't lose a generation in your C frame.

so adam wilt is wrong? if the first frame is composed of twp "a" fields...how are you shoving 2 fields together? all you're doing is removing a field which the NLE interpolates and it's not as bad as you're making it sound. but you are the one with the yellow name :grin:

 

[David Jimerson]

And by interpolating the field, you sacrifice up to half your vertical resolution -- when it's not necessary.

The efficiency of the pulldown removal software will determine the level of, if any, degradation when reconstructing the C frame. Like I said, maybe Cinema Tools isn't the best tool for it, but other software, like Vegas, does it without any quality loss. Adam Wilt wrote that section of his DVX page several years ane several generations of software ago, and it hasn't been revised since.

But I'll repeat what I said -- a generational loss in a single frame, if you see one, will be far less than losing half your vertical resolution in that frame, and certainly far less than losing half your vertical resolution in your footage in general.

As for "shoving 2 fields together," there are two methods of deinterlacing, and I covered them both. One is to blend fields, i.e., "shove them together," and the other is to interpolate fields. Both have their own cans of worms involved.

 

[novelt]

well actually it might not matter if it's placed on a site like youtube no matter the method the OP uses there's going to be sacrifice.

do you really think dvx footage looks that bad when it's deinterlaced? because just as using the right app for pulldown...if you use the right "deinterlace method" you won't throw away a field so you don't lose vertical resolution. question is how many NLE's even offer that. for the record "Weave" keeps the fields intact no interpolation.

 

[David Jimerson]

If it's on YouTube, it's gonna look like crap no matter what . . .

Not many NLEs can deinterlace footage. Vegas can. Premiere Pro 2.0 can, though it only offers blending fields as its method.

As for it looking "that bad," my original post was comparing it to removing 2:3 pulldown, in which case yes, it looks worse. "Bad" is relative term.

Blending fields will save resolution instead of throwing away fields, but you'll get ghost-like auras around moving objects when you mix fields taken from different frames.

 

[Barry_Green]

so adam wilt is wrong?

No, but I think you're interpreting what he said incorrectly.

if the first frame is composed of twp "a" fields...how are you shoving 2 fields together?

Because the fields come from the same "a" frame,but the fields are not the same! They're from the same instant in time, but contain very different data. De-interlacing means basically discarding one field, and interpolating a new field from the remaining field. There's no valid reason to do that if you're working with progressive-scan footage.

De-interlacing is an archaic technique held over from the days when people would shoot interlaced video and then wanted to try to simulate progressive-scan with it. De-interlacing, simply put, should NEVER be done on progressive-scan footage. All de-interlacing will do is ruin the integrity of your progressive-scan footage.

 

[novelt]

weave doesn't blend or interpolate. it as the name says weaves, but not many NLEs give you that option. so the footage isn't ruined...i'm using it right now and i must say before i changed the setting...the deinterlaced footage wasn't "ruined"...main reason i could tell it was deinterlacing was because i was looking for the 5th frame (shooting advanced) and it wasn't there. now that i set it to weave i'm getting the full resolution frame with the 5th frame intact.

but all of that is moot...there's nothing wrong with being a stickler...but saying NEVER...depending on the final format i doubt the average person will be able to tell that the source video was deinterlaced or reversed telecined after it gets compressed for web distribution especially if it's youtube.

 

[Noel Evans]

You can edit 24p on a 29.97 timeline...

Please dont confuse progressive / interlace with FPS.

EDIT: 24p on a 29.97 timeline. From what I read on Barry Greens post you CAN do this in a 24p timeline also.
EDIT: 24pa on a 23.97 timeline.
EDIT: 30p on a 29.97 timeline.
EDIT: 60i on a 29.97 timeline.

 

[elsmacko]

"OMGeez".

I was confident that I understood when to be shooting which Before I read this thread, but now I have no idea.

So, I'm just going to ask this question.

Lets say (because I am), that I'm shooting progressive footage with a DVX that is meant for a student television-production group. This means that it's destine to be both shown on TV as well as streamed and downloaded over the internet.

Under shooting for these circumstances, do you (the royal you) think we should be shooting on 24p or 24PA, to then be edited in Pro 2?

 

[Noel Evans]

"OMGeez".

I was confident that I understood when to be shooting which Before I read this thread, but now I have no idea.

So, I'm just going to ask this question.

Lets say (because I am), that I'm shooting progressive footage with a DVX that is meant for a student television-production group. This means that it's destine to be both shown on TV as well as streamed and downloaded over the internet.

Under shooting for these circumstances, do you (the royal you) think we should be shooting on 24p or 24PA, to then be edited in Pro 2?

OK well I think part of my post is incorrect... I may have confused some parts. SO I have removed a great deal of it as to not to try and confuse anymore... sorry but I feel I have some missing info on this.

 

[elsmacko]

Oh, well sorry if it seemed as though I meant Your posts in particular, I really mean the whole thread. So many conflicting things have been said (and as people have stated, conflicting things are Always said on this topic), that it's kind of bewildering overall.

In my case, someone said near the beginning that one is better for TV product, and the other better for on-computer product. Since we're shoooting for both and Not for film, I'm really not sure now which one we should be using.

Also, it's been said that one is better when editing in ~30, and the other better when editing in ~24, but that doesn't answer as to When you would be wanting to edit which way.

 

[David Jimerson]

Your output isn't relevant; it's all about how you plan to edit.

If you have a 24p-capable editor, you should edit your footage in 24p. You will be able to output any way you want.

Because you should edit in 24p, you should shoot in 24pA.

The only time you should ever edit 24p footage on a 29.97 timeline is if you have 29.97 footage you want to mix in -- and keep looking like it's 29.97.

If all you have is 24p footage, you should never NOT edit in 24p.

 

[novelt]

Your output isn't relevant...

throw out that section in the DVX user manual...

seriously i get your point, but to me that's kinda starting it off wrong, but you're right about editing and output.

which to shoot? shoot 24p standard that way you can output interlaced footage without any problems and then for your progressive footage you reverse telecine. because if you shoot advanced though you'll be able to output progressive...you might have audio sync problems using reverse telecine. can't truly vouch for that...i've read articles on that subject.

 

[Scottdvx100]

Reverse telecine won't cause sync issues.