25PF question

[Digidave]

Hi all,

I shoot 1920 x 1080 25pf because I am happy that it gives me a better image quality than 50i.

As I understand it, the HF S100 shoots this progressively, but stores it as an interlaced signal and the output is interlaced, but both fields are identical. I convert the output to Cineform AVI using NeoScene, then edit in Sony Vegas Pro 9e.

My question is, should I set NeoScene to Maintain Source Frame Format and if so should I select Progressive Source, or should I set it to Deinterlace? The manual doesn't mention pf, just progressive and interlaced.

I ask because, having done side-by-side comparisons, it doesn't seem to make any difference what I select. I am guessing here that as the fields are identical the interlaced frame appears like a progressive frame so acts like one. Could that be right?

As yet I have found no conclusive answer on the 'net and that includes the Cineform page.

Anyone here have any ideas?

Dave

[Guy Bruner]

Dave,  there are two ways two record progressive frames -- as one continuous frame or as two interlaced fields that are recorded at the same time.  However, continuous progressive frames are not compatible with many TV sets that were designed for interlaced media and over the air transmission.  So, most consumer camcorder manufacturers store the frames as two interlaced fields.  Since that is the way your camcorder records, I would select maintain source frame format and interlaced.  Whether you do this or decide to deinterlace on the timeline to get a true continuous progressive clip, the video will look the same in preview and on the final display.  IMO, there is no advantage to convert to continuous progressive for editing since you will be cutting on the frame boundaries, not on field boundaries.

[Chris Harding]

Hey Dave
I certainly wouldn't argue with Guy on this one so you have your answer!
Just for interest I find that 25P is a bit stuttery when there is motion...even when I pan the camera the image stutters a bit in the EVF.....My cameras have 1080 24P and 25P  but no 50P so I tend to shoot everything that need high quality at 720 50P ..it's also neat for slomos ... In the USA they have 30P which is just a bit higher than our 25P so it seems to be a lot more useable!!!
Now, I thought that you were one of the first to get a super-duper high speed computer? Why do you have to go with Cineform? Surely your machine can handle dropping the native files directly onto the Vegas timeline and editing from there. My preview on long sequences does drop down to half resolution but it's still smooth on the i7 processor!!
Chris
 

[Digidave]

Thanks Guy, I have done exactly what you recommend on the last couple of projects. Now I know this is correct, I'm happy.

Chris, I shoot 25p now primarily so it better matches the output from the 550D, which only shoots progressively. By way of a bonus, it gives me slightly better image quality.

Yeah, the upgraded computer improved my editing experience and cut my DV-AVI render times down dramatically, but when I first tried AVCHD it didn't like it and nor did I. I burned a lot of midnight oil trying to find a solution and finally settled on Cineform because it worked. Now, using different cameras with different file types, it makes me happier to convert them all to Cineform AVIs and put just one type into Vegas.

Yes, I know Vegas will handle .mts and .mov files natively as long as you install Quicktime Pro, which I neither have nor want, I also know it can handle more than one file type at a time, but my way works well for me. If it ain't bust, don't fix it!

Dave

[peterphelan]

Chris,
Like Dave I take everything over to Cineform for consistency; I have also standardised on 720p fo everything. Furthermore, to quote from the Cineform site:
 
>>Chroma Interpolation: 4:2:0 → 4:2:2. Background: Most DSLR, HDV and AVCHD camcorders record chroma (color) in a format known as 4:2:0. Without diving too deep, 4:2:0 chroma is half the color resolution of more professional 4:2:2 formats. When Neo Scene detects 4:2:0 chroma it properly interpolates the source chroma to 4:2:2 for more accurate color processing during editing and effects work. And if you ever "key" your material, CineForm’s chroma interpolation will substantially improve your resulting visual fidelity.<<
 
Peter

[Digidave]

Peter,

To add to that, if you delve around a bit, you find a lot of pros insist that it is wrong to try to natively edit high compression files and that you should always transcode to guard against further quality loss.

Dave

[peterphelan]

Hi Dave,
 
yes I would agree with that. I remember the debates I had with fellow photographers in the early days of digital imaging, over shooting raw versus jpegs. I just could not see why anyone would from choice, throw away data at the moment of capture. It seems a bit like this with video to me. We know that the camera compression introduces artifacts, along with editing issues, so to me it makes sense to try and smooth all that out as much as possible, and even if it's via interpolation, increase the perceived colour spectrum, before we start work on the production.
 
This kind of thing, along with using say the Technicolor Cinestyle picture style, all adds to the amount of work in post production, but in many cases, I feel this is certainly worthwhile.
 
Peter

[Digidave]

Hi Peter,

I couldn't agree more.

While I can't use Technicolor Cinestyle with my HF S100, I did some video tests and found that by applying similar logic to the camcorder and setting in-camera processing to the minimum, I can produce a substantially better quality image by doing the processing in post. Of course, as you pointed out, the images I am processing are 4:2:2, so the starting point is better.

Ever since I got my first digital camera I have shot in raw mode, as my wife has with hers and still does with the EOS, for exactly the same reason as you.

Dave

[Guy Bruner]

As I recall, Dave's initial question was about quality differences between importing or transcoding video that was shot in PF mode versus fully frame progressive.  However, the thread morphed a bit.  And I agree totally with the expansion comments.  It is always better to do color grading, green screen and video overlays in non-compressed mode or using a high quality lossless codec like the one Cineform sells.  You can also use Lagarith or HUFFYUV which are free codecs.  With today's horsepower and cheap storage, using non-compressed video is not a significant load on the computer.  If you are going to do green screen or overlays, it would be better to have the video in progressive mode.  In Vegas, the overlay media is progressive so for best results put progressive video on the timeline.

[Phil_L]

Hi

Peter,

To add to that, if you delve around a bit, you find a lot of pros insist that it is wrong to try to natively edit high compression files and that you should always transcode to guard against further quality loss.

Dave

This is only true if you are exporting out using the same lossy compression and then bringing that back into another package for further work, then exporting again using the same lossy compression etc.

The best output quality comes from using the original footage, unless you use a lossless codec as an intermediate, it's just a simply matter of county up the lossy transcodes:

Use original AVCHD footage on the time line to produce Blu-ray output:

AVCHD ->Timeline render out -> final output to Blu-ray, lossy, = 1 lossy re-compression of the original footage

Using a popular intermediate like Cineform:

AVCHD ->Transcode to Cineform which is lossy -> Timeline render out -> final output to Blu-ray, lossy = 2 lossy re-compressions. 

Of course Cineform is very good and the losses are visually hard to spot, but a loss there is.  If you wanted to use the footage in another package, you wouldn't of course render it back out as AVCHD or Blu-ray, as that is quite lossy, so at that point you would render out to something much less lossy like Cineform, or even better lossless like Lagarith.

It is always better to do color grading, green screen and video overlays in non-compressed mode or using a high quality lossless codec like the one Cineform sells.

This again isn't always true.  You an still use the original the footage on the time line.  What is overlooked is video editors don't colour grade or chroma key the compressed footage, they always decompress the footage into uncompress video in memory with a big colour space then apply the effects there.  You don't need to start out with something like Cineform in 4.4.4 for the best quality.  Cineform the company like people thinking this of course, but it isn't the case.

Cineform or uncompressed is great if the output is going to go on for further grading in order to keep the large colour space the editor decompressed the video into in order to apply the effects, but if all you are doing is staying within one package, it gains you little (gains Cineform more  ) unless your PC is so slow it helps to use Cineform on the timeline.

As I recall, Dave's initial question was about quality differences between importing or transcoding video that was shot in PF mode versus fully frame progressive

If it's PsF, just leave all settings as interlaced in the editor, that is the point of PsF is that it can be mixed with interlaced footage, you don't need to treat it differently.  Also Blu-ray doesn't officially support progressive 25fps, so it has to be PsF to go on Blu-ray if that is the target.  Modern HD TVs will recognise it is PsF and it will be displayed no differently to the viewer as if it were 25 progressive frames native. 

As for judder with progressive 25fps (or 24fps) you need to always keep the shutter speed locked at 1/50th of second (ND filters might be needed in bright daylight), this helps blur movement in each shot which helps fool the brain into still seeing motion at such slow frame rates.  If shutter speed is 1/100th or higher, our brains lose the illusion of motion, and we start to perceive each individual frame as a picture, this is why it judders or strobes.  Camera pans should be avoided, in film they will use a dolly and track rather than a camera panning on the spot as this avoids judder.  The rule with film is nothing should cross the frame quicker than 7 seconds.

Hope that helps.

Regards

Phil