FCP 6, Color, Sheer, HDV

[Amos Kim]

Anybody know a way to incorporate Sheer codec into a HDV 720p/FCP/Color workflow? Or would I need to use prores? I've been racking my brain and reading posts to find a solution, but can't find a solid answer. And I capture as HDV, but what do I use as sequence setting? And under Apple Color, do I set output as prores or original? I'm just looking for the least lossy workflow.

thanks

[Robert Lane]

According to Sheer's own website, it is still not fully compatible with FCP filters and effects. Your best bet is ProRes422 which was specifically designed to capture and edit HDV in the best fidelity possible.

[Amos Kim]

I know I should use DVHSCap, capture the m2t files and then use MPEG Streamclip to convert the files to QT .mov and select Sheer. Take these to final cut but how do I get apple color to use sheer? Or do I?

[Robert Lane]

I have no idea where you've come up with such a convoluted workflow or why you think you need it, but suffice it to say you're going way, way overboard for a codec that doesn't have anywhere close to pristine color information to begin with.

All you need to do is ingest/capture your footage directly into ProRes422, edit, then output or encode into your final. That's it.

If you were working with a telecine process or uncompressed then I could see using an alternative like Sheer to reduce file sizes and retain as much as possible the original color information, but as stated before you can't even apply filters and effects to the Sheer codec (read the info on their website) to begin with, so you're starting off with a dysfunctional setup.

Read the various HDV workflow white-papers on the Apple website and learn how simple and easy it is to get superb results with the built-in tools in FCP.

[Andrew Kimery]

I'd stick w/ProRes since you want to use Color. Color is finicky about a lot of things, including codec, and hoping Sheer will work okay just seems like an unnecessary headache waiting to happen.


-A

[David Knaggs]

Hi Amos.

This is an area I am currently looking at (very intensely) myself. (And I'm also trying to determine the best codec to use for digital cinema projection, besides post-production.) So, first of all, I'll answer part of your original question.

Do ALL of your capturing in Native HDV. If you are working with 24p footage, I would advise against MPEG Streamclip (it does not give perfect frame-for-frame conversion, i.e. gives occasional repeat frames) and also HDVxDV (gives you out-of-sync audio). MPEG Streamclip is, however, PERFECT with 25p footage (720p25). I'm not sure about 30p footage though. Maybe MPEG Streamclip is okay with 720p30 (?).

If you capture with DVHSCap and convert to a final codec before importing into FCP you have cut off all future codec options, as well as losing all of your timecode.

Whereas, if you capture natively, you will have full timecode, use the least possible disc space for your original footage and, most importantly, have your footage at the highest possible quality (FCP simply captures the .m2t file from the camera without transcoding and puts it in a QuickTime wrapper).

You can cut (assemble) your footage into its final sequence in a native HDV sequence and then either export as a Sheer QuickTime movie:

(Select File>Export>QuickTime Movie, then from the pop-up window under "Setting:" scroll down to "Custom" and select it. From the new window, under "QuickTime Video Settings", scroll down the pop-up window next to "Compressor" and select "Sheer" or any of the other 7 variations of the Sheer codec)

or copy and paste your sequence into a new Sheer sequence:

(to make a new Sheer sequence, simply make a new HDV sequence, then select Sequence>Settings. From the new window, under "QuickTime Video Settings", scroll down the pop-up window next to "Compressor" and select "Sheer" or any of the other 7 variations of the Sheer codec).

Of course, you may decide to go with ProRes (as Robert and Andrew have suggested). Or you might decide to go with Uncompressed 4:2:2 (either 8-bit or 10-bit). Again, you'd simply scroll down the QuickTime Compressor settings and select that instead of "Sheer".

But the great thing is that your options will be totally open if you change your mind later and want to go from Sheer to Uncompressed or whatever. So long as you capture natively to start off with.

Now, up until now, I've simply covered HOW you can convert your sequence to Sheer (and others might know better ways to do that) rather than WHETHER you go with Sheer or not.

Of course, as Robert has implied, you are dealing with footage which has initially been sampled at 4:2:0. The question which I'm currently pondering is, with successive generations (Greenscreen, color correction, etc.) during post-production, is it really acceptable to have ANY sort of further loss of your color information? Especially if your destination is a cinema screening (rather than an SD DVD).

Paolo (Ciccone) is, of course, a big advocate of not losing ANY further information from HDV 720p footage after the camera's initial compression, is a fan of SheerVideo and has advised against ProRes if you want best results:

http://www.dvinfo.net/conf/showpost....08&postcount=7

The president of Bitjazz (makers of the SheerVideo codec) has earlier posted the following in DV Info:

http://www.dvinfo.net/conf/showthrea...t=85526&page=2

Originally Posted by Andreas Wittenstein

Even converting to a lossless format such as SheerVideo won't give you a better picture - at least, not right away. But after the initial conversion, sticking with a lossless format will give you a better picture than you would otherwise end up with, because you'll avoid all the generational loss from subsequent recompression cycles. In other words, if you're starting and ending with a lossy format, using SheerVideo in between won't give you a better picture than the original, but it will preserve the original quality from getting ever worse.

I was just checking the Bitjazz (SheerVideo) website and there doesn't seem a lot of info after mid-2007 (that I could find). So I have no idea whether SheerVideo will currently work with Color.

But, if you capture and edit natively, you could try exporting with Sheer and, if it proves too buggy with Color, etc., just go back to the native sequence and export it as Uncompressed (or ProRes or whatever). Uncompressed 8-bit 4:2:2 runs at approximately 3 GB per minute, so even a 120 minute movie will only take up about 360 GB. And these days, you can probably pick up a 750 GB hard drive for about $150, so Uncompressed 4:2:2 is a much more viable option than it used to be. Remember, you don't have to capture (or transcode with MPEG Streamclip) all of your footage Uncompressed (which would take a vast amount of hard drive space). Capture and edit natively through FCP and only send your final cut out (to Color, Motion, etc.) as Uncompressed 4:2:2.

[Amos Kim]

Great reply David. I, too, am looking for the best workflow for digital cinema projection for my jvc 720 24p footage. I tend to lean towards the mindset that hdv is already heavily compressed and should not go through any more compression or render. I agree with paolo and would love to utilize sheer video to it's fullest. I'm learning to use apple color and am amazed at the power of this application and would love to use sheer for the roundtrip from final cut 6. Too bad DVHSCap and Mpeg Steamclip have those limitations!

You mention editing sequence in native HDV and converting to uncompressed 4:2:2 later, but correct me if I'm wrong, I thought the only way to go uncompressed 4:2:2 is to capture directly from cam onto a kona card directly into a computer. I'm thinking once you record to tape or DTS, it's HDV 4:2:0.

Anyways, I'm glad to see that I'm not alone in my quest. Keep me updated on your progress David, as I will too. Because, there's gotta be an ideal workflow for us!

mrak1979@yahoo.com

[David Knaggs]

Originally Posted by Amos Kim

You mention editing sequence in native HDV and converting to uncompressed 4:2:2 later, but correct me if I'm wrong, I thought the only way to go uncompressed 4:2:2 is to capture directly from cam onto a kona card directly into a computer. I'm thinking once you record to tape or DTS, it's HDV 4:2:0.

Hi Amos.

Yep, it's definitely HDV 4:2:0. The only reason to then use SheerVideo or Uncompressed 4:2:2 is to prevent any FURTHER generational loss. Your image has already taken one hit (compression) when recording to tape or DR-HD100. No need to add to it. The "advantage" of SheerVideo over Uncompressed is the smaller file size (about 45% the file size of Uncompressed according to a very rough test I made of it).

I gave a link in my post above to a post made earlier this year by Paolo. I'll quote a bit of it here in the hope that this helps fully answer your query:

Originally Posted by Paolo Ciccone

But, if you capture to tape you already have a generation loss as the HDV compression squeezes color information out. If you convert to any lossy codec than you add a generation loss. You might not be able to see it but it's there and it might show up later on when you try to color correct and grade your footage.
Converting from .m2t to uncompressed, PNG-encoded QT or SheerVideo-QT is perfectly safe and I do it on a regular basis. This is because those codecs compress data without discarding anything, similarly to what .zip does.

And finally, for your further info, a good current thread on projection of ProHD footage is:

http://www.dvinfo.net/conf/showthread.php?t=125086

[David Knaggs]

Hi Amos.

I just started my tutorials on Color and discovered a very interesting fact!

Apparently Color WILL accept HDV footage sent to it from FCP. But Color won't render and output in HDV (avoiding the quality degradation of an extra MPEG-2 compression, thank goodness!). It will output and render the footage in Uncompressed.

So it looks like:

1/ Capture and edit in native HDV.
2/ Send the final cut over to Color in HDV.
3/ Receive back the rendered footage from Color in Uncompressed. (A 2 hour movie would be about 360 GB.)
4/ Send the audio to Soundtrack Pro for mixing and receive back the "sweetened" audio.
5/ Output your final 360 GB Uncompressed QuickTime movie.
6/ Project the Uncompressed QuickTime.

Well, that's a possible workflow, anyway.

I only hope they were talking about BOTH HDV1 (720p) and HDV2 when they said that HDV works with Color!

[Amos Kim]

I'm wondering if color works with sheer... I have my editing machine in storage right now, so I can't experiment right now. When I get it back, I'm gonna see if color does work with sheer. Thanks for the update, David! Have you downloaded final cut 6.0.4 yet? I'm hoping there are no kinks with 720p hdv footage.

A

[David Knaggs]

Yes, I upgraded from 6.0.3 to 6.0.4 and I'd rather I hadn't! The update info said that it handled a problem in capturing at the end of an HDV tape. Well, I wasn't having any problems with this in 6.0.3, but thought, "Maybe there are some other benefits for HDV capture." So, I upgraded to 6.0.4 and NOW I'm having problems escaping capture at the end of the tape. It seems to have CAUSED the problem it said it was trying to handle! Oh well.

Another current "gotcha" depends on how you are intending to monitor. I'm thinking of getting a Matrox MXO in a few weeks (with either a Dell Ultrasharp LCD screen or an Apple Cinema Display). But I noticed in this forum last week that Matrox MXO users said that QuickTime 7.5 in combination with FCP 6.0.4 had rendered them UNABLE to monitor at all with their Matrox MXO! So these upgrades seem to be a bit unstable right now.

[Amos Kim]

Looks like apple color does not work with sheer codec, even though it's a QT codec. That's unfortunate. I guess I'll be cutting natively in HDV, sending it to color, sending it back to FCP as uncompressed, 8bit or 10?

[David Knaggs]

Originally Posted by Amos Kim

Looks like apple color does not work with sheer codec, even though it's a QT codec. That's unfortunate.

Thanks for that info.

Originally Posted by Amos Kim

sending it back to FCP as uncompressed, 8bit or 10?

That's the big question, isn't it?

The camera manual (for JVC GY-HD100/110) says, "Video signal recording format: HDV720p format, 8-bit, 19.7 Mbps." So the footage is compressed at 8 bits. But I've started my tutorials on Color, and it goes over the difference between 8-bit and 10-bit and how it's easier to get banding in 8-bits. Which has me a bit concerned about banding during dissolves or when you fade to black. So, even though the original footage has been encoded in 8-bit, is it better to work in a 10-bit sequence? That's what I'm currently pondering.

[Noah Kadner]

Never even heard of Sheer but I can assure it's got no place in a Final Cut Pro>Color workflow. As others have stated your best bet is edit HDV and send either that or better ProResHQ or even uncompressed to Color. I don't think you'll see Sheer supported in Color in the foreseeable future.

-Noah

[David Knaggs]

Originally Posted by David Knaggs

So, even though the original footage has been encoded in 8-bit, is it better to work in a 10-bit sequence? That's what I'm currently pondering.

Well, I had the above question answered emphatically yesterday at the Final Cut World Tour event in Melbourne.

They showed a slide of 8-bit with banding, then one with 10-bit and no banding.

Then the speaker said that ProRes was a 10-bit codec and pointed out the advantage for HDV shooters of taking their 8-bit HDV footage and working with 10-bit ProRes because they can push things a lot further with color grading due to the 10-bit format.

He was also stressing the visually lossless aspect of ProRes and said that you'd be unable to visually discern any difference, even after 10 generations!

There was also an excellent rundown of Soundtrack Pro. I saw one neat trick (previously unknown to me) where he removed a loud noise made WHILE a woman was talking (I only knew how to replace an unwanted noise made BETWEEN words by replacing with ambient noise). He brought up the Frequency Spectrum view (which I knew about), but then he went View>File Editor Display>Show Spectrum Controls, then on the Spectrum View window he clicked Logarithmic (rather than Linear) in the Frequency Scale section.

Switching to the Logarithmic view then revealed the loud noise area as being UNDERNEATH the woman's voice. So he then Option-dragged to select the noise area and then hit the delete key. Voila. Gone. And without affecting the woman's voice.

That info (about the Logarithmic view) might be well known to others on this forum. But it was news to me!