Question regarding de-interlacing.. at DVinfo.net

Jim Justice · January 26th, 2007, 07:28 AM

So I have been digging back and reading up on de-interlacing footage to get that progressive look.
I have come to the conclusion that it is as simple as making sure that the field order must be set to NONE (progressive) in the render settings...This helps me obtain the look I need.

Now this may be a dumb question....but if I was to render it out with the field order set to NONE, and in .avi, and then put this new rendered avi clip in a new timeline and do all my text work...do I have to make sure that I render it out again in the progressive mode? Can a progressive scan clip revert back to an interlaced clip? Basically what I am asking is how do you handle a file that was rendered as "progressive"....DOes it continue to degrade the reso each time you have to render it out...kinda like a gen'd out VHS tape, Seems confusing and I think I am making things more complicated then what they really are.

Mauritius Seeger · January 26th, 2007, 06:49 PM

Quote:

Originally Posted by Jim Justice

So I have been digging back and reading up on de-interlacing footage to get that progressive look.
I have come to the conclusion that it is as simple as making sure that the field order must be set to NONE (progressive) in the render settings...This helps me obtain the look I need.

Now this may be a dumb question....but if I was to render it out with the field order set to NONE, and in .avi, and then put this new rendered avi clip in a new timeline and do all my text work...do I have to make sure that I render it out again in the progressive mode? Can a progressive scan clip revert back to an interlaced clip? Basically what I am asking is how do you handle a file that was rendered as "progressive"....DOes it continue to degrade the reso each time you have to render it out...kinda like a gen'd out VHS tape, Seems confusing and I think I am making things more complicated then what they really are.

no it's a good point actually, the way i've always thought of it is as follows

the setting for the clip in the timeline (project settings and the clip propeties) determine what you have. the setting in the render option what you want. vegas then figures out how to get from a) what you have to b) what you want. one would hope it figures it out in a smart way.

i think if you want progressive output you should always set the render option to NONE, but if you don't i think there are two things that could happen (i.e. i don't really know either):

1) the new output clip is identical apart from a flag that sais it's interlaced - even though of course both fields are still originating from one coherent frame.

2) vegas converts adjacent frames to fields. i.e. it takes frame one and throws away all the odd lines, then takes frame two and throws away all the even lines. then puts them together - i remember seeing an option like that somewhere. it's definitly possible if you so desire.

David Jimerson · January 28th, 2007, 02:10 PM

Looks like there are a couple of things going on here, and some missing info.

Jim, in another thread you were talking about converting 60i to 24p -- is that the kind of "deinterlacing" you're talking about here? Or converting 60i to 30p?

It's a slightly different process depending on which you're doing, because converting to 24p involves a more complicated formula.

Converting to 30p can be easier, because the 60i "frame" rate of 59.94 is easily divided by 30p's 29.97 frame rate. So, you can "deinterlace" straight to 30p in two main ways (and you set these as your Deinterlace Method at the Project Properties level) --

1) "Blend Fields": this is (more or less) taking Field 1 and Field 2, pushing them together, and blurring out any interlace artifact (the comb effect) caused by the different positions of any moving objects. If there is little motion, this can be an effective method, because it will preserve more vertical resolution. But if there's a lot of motion, it will produce undesirable results, with funky movement and ghost-like auras around anything moving.

2) "Interpolate Fields": this is (again, more or less) taking every other field (Fields 1, 3, 5, 7, 9, etc.), throwing away the rest, and then making (interpolating) full frames out of the remaining fields. Because the full frames will be comprised of only single moments in time, instead of two moments in time, there's no motion to reconcile within the individual frames, and you won't get the motion artifacting or ghosting. But because the frame is a whole picture being made out of the information from only half a frame, there can be a loss of up to half of the vertical resolution. But, other than that, it's good 30p, because it will be full pictures taken every 1/30th of a second, just as 30p should be.

You can export an uncompressed AVI which contains only the 30p frames -- a true frame rate of 30 (29.97) fps. But after you've made the conversion, it's just as easy to export an NTSC DV AVI. What it will then do is take each 30p frame and split it into two fields -- frame A becomes fields 1 and 2; frame B becomes fields 3 and 4, and so on. This will NOT then look the same as the original 60i footage, because you've already discarded half the fields -- half the moments in time -- from the video. The fields made from the 30p frame are of the same moment. So, if you view them on a TV, field 1 is the same moment as field 2, field 3 is the same moment as field 4, and so on. In other words, even though a frame has to be split into 1/60 second fields, two of those fields in a row are from the same picture, so the picture is on screen for a net of 1/30th of a second, just as 30p should be. The motion and the entire aesthetic is preserved.

For 24p, it's more complicated; 24 (23.976) doesn't fit into 60 (59.94), so there's some more complicated math involved in deciding which fields are converted into which frames.

The same general principles apply, though -- if you use "Blend Fields," you preserve more resolution, but it doesn't look good if there's a lot of motion; if you use "Interpolate Fields," the motion is smooth and looks good but you sacrifice up to half your vertical resolution.

Then, going back to tape, and for display on a 60i source, pulldown is introduced. Remember, you've thrown away 60% of your fields, so you have only 40% of the motion information left, and you need to have your 60 fields per second in order to be shown on a TV.

With 30p, as I said above, it's easy, because all you have to do is split every frame into fields and just show them in order. Everything works out.

But 24 isn't half of 60. If you just split the 24 frames into fields, you only get 48, so if you tried to just play those fields in order, you'd speed up your footage by 25%. You need 12 extra fields to fill the gap.

Forutnately, 12 is half of 24, which makes it fairly straightforward. You take an extra field from every other frame and play it. So, you run 2 fields from the first frame, 3 fields from the next frame, 2 fields from the third frame, 3 fields from the fourth frame, and so on -- 2 3 2 3 2 3 2 3 2 3 2 3 2 3 -- and the fields even out to 60 per second. That's 2:3 pulldown, and it's how 24 fps movies have been shown on TV since the beginning of NTSC.

That may have been more info than you wanted, but I was trying to answer the obvious follow-up questions as well.

In short, you can do your text work and export any way you need to, and not worry that your progressive aesthetic will be lost. If it's appropriate to export as progressive -- for the web, for DVD, etc. -- then do so. But if you NEED to export as interlaced, for going to tape, you've got no worries.

Jim Justice · January 28th, 2007, 08:26 PM

Just the info I was looking for...
David....I originally was looking for an efficient way to convert 60i to 24p...but thru trial and error...I wasnt having much luck,...so then I figured it was easier to just go 60i to 30p thru the progressive conversion.
I just wanted to eliminate the wretched interlaced look. However,...I would prefer 24p so I will have to do more trial and error work!
Thanks for the 411 guys!

Marcus van Bavel · January 29th, 2007, 08:14 PM

Quote:

Originally Posted by David Jimerson

The same general principles apply, though -- if you use "Blend Fields," you preserve more resolution, but it doesn't look good if there's a lot of motion; if you use "Interpolate Fields," the motion is smooth and looks good but you sacrifice up to half your vertical resolution.

What he really needs is DVFilm Maker (dvfilm.com/maker) which applies different methods depending on the amount of motion present in each part of the screen: blend fields where there is motion, progressive fields where there is no motion. So when something is moving quickly, it's at half resolution, but smooth, and when something is still, it has the full resolution of the original frame.

Here is a sample 60i demo avi: http://dvfilm.com/maker/60idemo.avi (in Explorer right-click Save Target As)

If you convert it to 24P with Maker you will see smooth, even steps of the green ball, and the diagonal line and text will be unchanged and just as sharp as the original. The ball will be at half resolution, but since it's blurred anyway from motion, it's not apparent.

If you convert this to 24P in Vegas you will get different results, most likely a blurring of the line and text.

January 26th, 2007, 07:28 AM	#1
Jim Justice Regular Crew Join Date: Sep 2005 Location: Cincinnati, OH Posts: 38	Question regarding de-interlacing.. So I have been digging back and reading up on de-interlacing footage to get that progressive look. I have come to the conclusion that it is as simple as making sure that the field order must be set to NONE (progressive) in the render settings...This helps me obtain the look I need. Now this may be a dumb question....but if I was to render it out with the field order set to NONE, and in .avi, and then put this new rendered avi clip in a new timeline and do all my text work...do I have to make sure that I render it out again in the progressive mode? Can a progressive scan clip revert back to an interlaced clip? Basically what I am asking is how do you handle a file that was rendered as "progressive"....DOes it continue to degrade the reso each time you have to render it out...kinda like a gen'd out VHS tape, Seems confusing and I think I am making things more complicated then what they really are.

January 28th, 2007, 02:10 PM	#3
David Jimerson Jubal 28 Join Date: Mar 2005 Location: Wilmington, NC Posts: 872	Looks like there are a couple of things going on here, and some missing info. Jim, in another thread you were talking about converting 60i to 24p -- is that the kind of "deinterlacing" you're talking about here? Or converting 60i to 30p? It's a slightly different process depending on which you're doing, because converting to 24p involves a more complicated formula. Converting to 30p can be easier, because the 60i "frame" rate of 59.94 is easily divided by 30p's 29.97 frame rate. So, you can "deinterlace" straight to 30p in two main ways (and you set these as your Deinterlace Method at the Project Properties level) -- 1) "Blend Fields": this is (more or less) taking Field 1 and Field 2, pushing them together, and blurring out any interlace artifact (the comb effect) caused by the different positions of any moving objects. If there is little motion, this can be an effective method, because it will preserve more vertical resolution. But if there's a lot of motion, it will produce undesirable results, with funky movement and ghost-like auras around anything moving. 2) "Interpolate Fields": this is (again, more or less) taking every other field (Fields 1, 3, 5, 7, 9, etc.), throwing away the rest, and then making (interpolating) full frames out of the remaining fields. Because the full frames will be comprised of only single moments in time, instead of two moments in time, there's no motion to reconcile within the individual frames, and you won't get the motion artifacting or ghosting. But because the frame is a whole picture being made out of the information from only half a frame, there can be a loss of up to half of the vertical resolution. But, other than that, it's good 30p, because it will be full pictures taken every 1/30th of a second, just as 30p should be. You can export an uncompressed AVI which contains only the 30p frames -- a true frame rate of 30 (29.97) fps. But after you've made the conversion, it's just as easy to export an NTSC DV AVI. What it will then do is take each 30p frame and split it into two fields -- frame A becomes fields 1 and 2; frame B becomes fields 3 and 4, and so on. This will NOT then look the same as the original 60i footage, because you've already discarded half the fields -- half the moments in time -- from the video. The fields made from the 30p frame are of the same moment. So, if you view them on a TV, field 1 is the same moment as field 2, field 3 is the same moment as field 4, and so on. In other words, even though a frame has to be split into 1/60 second fields, two of those fields in a row are from the same picture, so the picture is on screen for a net of 1/30th of a second, just as 30p should be. The motion and the entire aesthetic is preserved. For 24p, it's more complicated; 24 (23.976) doesn't fit into 60 (59.94), so there's some more complicated math involved in deciding which fields are converted into which frames. The same general principles apply, though -- if you use "Blend Fields," you preserve more resolution, but it doesn't look good if there's a lot of motion; if you use "Interpolate Fields," the motion is smooth and looks good but you sacrifice up to half your vertical resolution. Then, going back to tape, and for display on a 60i source, pulldown is introduced. Remember, you've thrown away 60% of your fields, so you have only 40% of the motion information left, and you need to have your 60 fields per second in order to be shown on a TV. With 30p, as I said above, it's easy, because all you have to do is split every frame into fields and just show them in order. Everything works out. But 24 isn't half of 60. If you just split the 24 frames into fields, you only get 48, so if you tried to just play those fields in order, you'd speed up your footage by 25%. You need 12 extra fields to fill the gap. Forutnately, 12 is half of 24, which makes it fairly straightforward. You take an extra field from every other frame and play it. So, you run 2 fields from the first frame, 3 fields from the next frame, 2 fields from the third frame, 3 fields from the fourth frame, and so on -- 2 3 2 3 2 3 2 3 2 3 2 3 2 3 -- and the fields even out to 60 per second. That's 2:3 pulldown, and it's how 24 fps movies have been shown on TV since the beginning of NTSC. That may have been more info than you wanted, but I was trying to answer the obvious follow-up questions as well. In short, you can do your text work and export any way you need to, and not worry that your progressive aesthetic will be lost. If it's appropriate to export as progressive -- for the web, for DVD, etc. -- then do so. But if you NEED to export as interlaced, for going to tape, you've got no worries. __________________ www.wrightsvillebeachstudios.com

January 28th, 2007, 08:26 PM	#4
Jim Justice Regular Crew Join Date: Sep 2005 Location: Cincinnati, OH Posts: 38	Just the info I was looking for... David....I originally was looking for an efficient way to convert 60i to 24p...but thru trial and error...I wasnt having much luck,...so then I figured it was easier to just go 60i to 30p thru the progressive conversion. I just wanted to eliminate the wretched interlaced look. However,...I would prefer 24p so I will have to do more trial and error work! Thanks for the 411 guys!