DI Workflow Questions

[Patrick Barry]

Hello all,

 

I'm in the early stages of pre-production for a feature that combines S16 and HD video footage. I've scoured the old posts and have found answers to alot of my questions but still had a few and would appreciate any help.

 

A little background - the film is about 2/3 super 16mm reversal and 1/3 HD video. We are recording audio on a DAT with plans to sync the audio ourselves in post, scanning the film in HD for post in FCP. For the HD video segments, we might use a RED camera, but if money is tight we might have to settle for an HVX (and money will probably be tight. ha.)

 

Here are my questions.

 

1. Is a 2k scan the same thing as an HD telecine? And what is a datacine?

2. What is the difference between using HDCAM-SR tapes as backup versus LTO tapes?

3. Since we're mixing media and have audio from a DAT without timecode, is the best approach getting uncompressed HD scan, then down rezzing for an offline edit of both the film and HD video? Or should we get the lab to give us ProRes and mix the HD video in?

4. Can ProRes HQ be edited in the FCP timeline like SD footage? Or is there some downrezzing involved as well?

5. Does anyone have any suggestions as far as mixing media, like what not to do. Would I use 24 P, non-drop frame?

 

Thanks all!

[Phil Rhodes]

1. Is a 2k scan the same thing as an HD telecine? And what is a datacine?

 

Yes and no.

 

A "scan" tends to refer to a non-realtime process using something like a Northlight or Arriscan, which may take a significant fraction of a second per frame on modern equipment and usually gives you data output, a series of still image frames on a computer storage medium of some kind. It's obviously a slow process, but the quality tends to be very high. A "telecine" tends to refer to something that happens in realtime and is recorded to video tape. I say "tends" in each case because these terms are much abused. A datacine, a term often used to refer to one of the Thompson Spirit range, is a hybrid device which generally works either in realtime (or a significant portion of realtime) but which gives you data output. The general setup is generally more likely to be telecine-like than scanner-like.

 

2. What is the difference between using HDCAM-SR tapes as backup versus LTO tapes?

 

HDCAM-SR is a video tape format to which you record things in a similar manner to any other. LTO is a data storage medium which is used to archive computer data. In short, you'd connect HDCAM-SR to a telecine, but you might use LTO to store the output from a scanner. LTO is considerably slower than realtime, but it allows you to store uncompressed images and is massively cheaper.

 

3. Since we're mixing media and have audio from a DAT without timecode, is the best approach getting uncompressed HD scan, then down rezzing for an offline edit of both the film and HD video? Or should we get the lab to give us ProRes and mix the HD video in?

 

Depends on a lot of things. If you have a lot of film footage and you're unlikely to use all of it, you could telecine everything as uncompressed and produce a downconversion for the edit, as a data operation on a computer. Scanning is expensive and it'd be unusual to get everything scanned, so if you want to scan rather than telecine, or if you have enough film rushes that it isn't worth doing an HD transfer of everything, you might want to use the more traditional technique of having a rough, low-cost transfer done of your film, do your edit, then get only the material you need re-transferred (either on a scanner or telecine) later on.

 

I'm not an audio expert but I suspect that you may find that the DAT has timecode on it, but that it just doesn't mean anything. This isn't necessarily the end of the world, since when you eyeball-sync your sound during the edit, you should be able to create an audio EDL which refers to whatever timecode you have. If you really don't have any timecode, you may need to dub your DATs onto new ones which do have timecode.

 

4. Can ProRes HQ be edited in the FCP timeline like SD footage? Or is there some downrezzing involved as well?

 

Not my area, sorry.

 

5. Does anyone have any suggestions as far as mixing media, like what not to do. Would I use 24 P, non-drop frame?

 

You need to be a lot more specific about what you're trying to achieve. Everything I've said here has been in the broadest possible sense, outlining general approaches and standard techniques because I don't really know what you've got and what you're trying to achieve. If you provide some more info, we'll be able to help more.

 

P

[David Mullen ASC]

1. Is a 2k scan the same thing as an HD telecine? And what is a datacine?

2. What is the difference between using HDCAM-SR tapes as backup versus LTO tapes?

 

Phil answered these pretty thoroughly, but to provide further clarification (or not)... Yes and no, as Phil said, is correct, the difference is both major and minor at the same time. "2K" tends to refer to a data file, as in 2K RGB DPX files, for example. "HD" tends to refer to a video format, as in 1920 x 1080 4:4:4 (RGB) or 4:2:2, usually recorded to videotape, though not always. The general concept is that telecines give you a video signal and scanners give you data files.

 

The Spirit Datacine is a telecine that also acts like a scanner, hence the name "datacine".

 

Now, practically speaking, there is not a whole lot different in quality between 2K RGB DPX and 1080P 4:4:4 HD video, if both are 10-bit LOG (if the HD version is Rec 709 broadcast gamma, then it's optimized for monitor viewing not a film-out). The 2K data version would be uncompressed, but the HDCAM-SR 4:4:4 version would be very minimally compressed.

 

Which approach you take all depends on the post workflow going out to film, basically do you want to work in an HD video post workflow that ends with an HD videotape master that you give to a company to put out to film, or if you want to work with 2K DPX files in a D.I. environment and create a 2K master that goes out to film, and then make an HD Rec 709 broadcast video master from the 2K LOG master.

 

The one thing I'd keep in mind is that if you choose the HD video route, I'd transfer to 10-bit LOG, and do the final color-correction in LOG. But this means you'll probably want to make a Rec 709 gamma copy of the LOG tapes for dailies and offline editing just so you won't be looking at pastel, low-con images through post. Then go back to the LOG originals with your EDL to color-correct and create a final master. Later, you'll make a second master from that final LOG master that is converted to Rec 709 gamma (often incorrectly called a LOG to Linear conversion.)

[Mike Nichols]

Phil answered these pretty thoroughly, but to provide further clarification (or not)... Yes and no, as Phil said, is correct, the difference is both major and minor at the same time. "2K" tends to refer to a data file, as in 2K RGB DPX files, for example. "HD" tends to refer to a video format, as in 1920 x 1080 4:4:4 (RGB) or 4:2:2, usually recorded to videotape, though not always. The general concept is that telecines give you a video signal and scanners give you data files.

 

The Spirit Datacine is a telecine that also acts like a scanner, hence the name "datacine".

 

Now, practically speaking, there is not a whole lot different in quality between 2K RGB DPX and 1080P 4:4:4 HD video, if both are 10-bit LOG (if the HD version is Rec 709 broadcast gamma, then it's optimized for monitor viewing not a film-out). The 2K data version would be uncompressed, but the HDCAM-SR 4:4:4 version would be very minimally compressed.

 

Which approach you take all depends on the post workflow going out to film, basically do you want to work in an HD video post workflow that ends with an HD videotape master that you give to a company to put out to film, or if you want to work with 2K DPX files in a D.I. environment and create a 2K master that goes out to film, and then make an HD Rec 709 broadcast video master from the 2K LOG master.

 

The one thing I'd keep in mind is that if you choose the HD video route, I'd transfer to 10-bit LOG, and do the final color-correction in LOG. But this means you'll probably want to make a Rec 709 gamma copy of the LOG tapes for dailies and offline editing just so you won't be looking at pastel, low-con images through post. Then go back to the LOG originals with your EDL to color-correct and create a final master. Later, you'll make a second master from that final LOG master that is converted to Rec 709 gamma (often incorrectly called a LOG to Linear conversion.)

 

 

David,

 

Maybe you can help me solve this situation then:

 

I have 2K DPX scans (10bit RGB LOG) that I am grading in Apple Color. The images are low contrast and pastel like they should be. In Color, I set the printing density to FILM 95-680 to reset the black/white points and grade from there. I bake that grade into the final output. My ultimate finish is going to be 4:4:4 HDCAM-SR. I will never be going back to film, so is this the correct workflow? It certainly LOOKS right.

[David Mullen ASC]

David,

 

Maybe you can help me solve this situation then:

 

I have 2K DPX scans (10bit RGB LOG) that I am grading in Apple Color. The images are low contrast and pastel like they should be. In Color, I set the printing density to FILM 95-680 to reset the black/white points and grade from there. I bake that grade into the final output. My ultimate finish is going to be 4:4:4 HDCAM-SR. I will never be going back to film, so is this the correct workflow? It certainly LOOKS right.

 

Sounds like a question for a colorist or post person, but the question is... what is this 4:4:4 master for? For DCP or Rec 709 broadcast video?

[Mike Nichols]

Sounds like a question for a colorist or post person, but the question is... what is this 4:4:4 master for? For DCP or Rec 709 broadcast video?

 

 

The 4:4:4 master is going to be the exactly what it sounds like! The master "digineg" if you will, from which Broadcast and other deliverables (Blu-Ray, DVD, MPEG_TS) will be produced from. I actually will probably skip the HDCAM-SR hard master and just back up the Digineg to LTO and make SR's only upon request. I don't think DCP in the works, but if there is, that's all the better for me!

[David Mullen ASC]

The 4:4:4 master is going to be the exactly what it sounds like! The master "digineg" if you will, from which Broadcast and other deliverables (Blu-Ray, DVD, MPEG_TS) will be produced from. I actually will probably skip the HDCAM-SR hard master and just back up the Digineg to LTO and make SR's only upon request. I don't think DCP in the works, but if there is, that's all the better for me!

 

Well, a "digineg" and a master for video deliverables are sort of two different things -- a "digineg" implies some sort of archival copy that preserves all the information of the original (in a low-contrast LOG form) and a master for video deliverables implies that it will be in Rec 709 video gamma / color space. And DCP is another color space altogether.

[Mike Nichols]

Well, a "digineg" and a master for video deliverables are sort of two different things -- a "digineg" implies some sort of archival copy that preserves all the information of the original (in a low-contrast LOG form) and a master for video deliverables implies that it will be in Rec 709 video gamma / color space. And DCP is another color space altogether.

 

Thanks! Didn't realize digineg was clearly defined.

 

I guess whatever my requested deliverable turns out to be will dictate the real answer.