Home of Professional Cinematography since 1996

style="margin-bottom: 0"> 

class="style8" HD Post Costs

Published : 28th March 2005

Updated : 4th April 2005

>I'm shooting a feature in Singapore with the Varicam. The producers think they might want to do post here. I'm trying to get a handle on costs for them.

>Tight budget, so I'm looking for minimal post, not the deluxe treatment. What are some typical low-end rates for DVCPro HD online? For downconverting to do DV offline? Color correction?

>Also, now that Final Cut does DVCPro HD, what are the typical post steps for this? Is it really as simple as capturing HD, then using offline in Final Cut, color correcting and then outputting to an HD deck or am I missing something?

>Thanks

>Blain Brown
DP
LA


>With FCPro HD there is no "offline". You're editing directly with the codec from the tape, bit-for-bit transfer. As long as you aren't adding any effects, etc. then you're perfectly fine compression wise. You can even do the speed ramps with the camera without worrying about compression generations.

>Now, if you going to be doing a lot of color correction, etc., my advice to you would be to edit using the native DVCProHD codec at 24p with cuts-only (so there's no generation loss), edit that locked edit back to another DVCProHD master tape (again, because it was cuts-only you didn't loose any generations going from tape to FCP and back to tape), and then take that locked edit tape and have it transferred over to D-5 at 1080/24p where the final "online" will take place, so that effects, color correction, etc. can take place on an better format with less compression that DVCProHD. If worse comes to worse, you can skip the D5 route and simply find someone with an uncompressed HD station, such as a DeckLink HD Pro, Kona 2, or Cinewave, and have them redigitize the tape at 10-bit uncompressed for the final color correction.

>These are my suggestions, anybody else have a better/different way?

>Jason Rodriguez
Post Production Artist
Virginia Beach, VA


class="style9">>my advice to you would be to edit using the native DVCProHD codec at >24p with cuts-only (so there's no generation loss...

>Why would using anything other than cuts only result in a generation loss? I assume it has something to do with how FCP renders, but what is it that is different from the raw DVCProHD footage?

>Blain Brown
DP
LA


> Blain Brown wrote:

class="style9">>Why would using anything other than cuts only result in a generation >loss? I assume it has something to do with how FCP renders, but what >is it that is different from the raw DVCProHD footage?

>As is the case with DV, material brought in via Firewire in DVCProHD format is not passed through a decode/encode cycle if it is not processed for an effect. The data comes in, and then goes out again unaltered. This results in complete transparency and no generational loss. If, however, an effect is added (ANY effect, from a color correction to a dissolve and anything else) the material must be decoded, the effect created, and the resulting image re-encoded prior to output. This is not, in the traditional videotape sense of the term actual "generational loss," but it is another pass through a lossy codec, so total transparency is lost.

>Mike Most
VFX Supervisor
IATSE Local 600
Los Angeles


class="style9">>(ANY effect, from a color correction to a dissolve and anything else) the >material must be decoded, the effect created, and the resulting image >re-encoded prior to output.

>Two questions :

>1/. How bad is it? (the producer asked me several times "Will we lose quality if we do it this way?") If it's minor and hardly noticeable, then we can probably live with it; this is a small independent feature after all.

>and

>2/. How do we avoid it? Even if the whole movie is cuts only; clearly you aren't going to make a movie that does not require some color correction, so it is a given that large portions, perhaps the entire movie will have effects applied (probably as supported by a hardware card, I know the DeckLink supports color correction, for example - does that help?)

>Does doing the color correction in a real HD tape-to-tape suite avoid this problem? My point being: if the entire movie was edited cuts only on FCP HD and then the final color correct done on a Da Vinci, is this generational loss avoided? This would not be so bad as we could edit the movie, show them the result and then let them decide if they want to go that route.

>Thanks

>Blain Brown
DP
LA


>Blain Brown wrote:

class="style9">>1/. How bad is it?

>Well, it's, uhh, "this" bad. Seriously, there is no sensible answer to that question. It depends on how much movement and detail is in the original images, the colors involved, and a lot of other factors. You need to shoot tests to determine how much of a problem it will be for you.

class="style9">>2/. How do we avoid it? Even if the whole movie is cuts only; clearly you >aren't going to make a movie that does not require some color >correction

>Final Cut, in its current form, is not the place to be doing high end color manipulations because it is bit depth limited and the color algorithms are not particularly sophisticated. Perhaps in the future there will be a third party add in solution to this, but right now if you want clean color correction and the most extensive manipulative capabilities, you'll be looking elsewhere (i.e., DaVinci, Pandora, Lustre, etc.).

class="style9">>if the entire movie was edited cuts only on FCP HD and then the final >color correct done on a Da Vinci, is this generational loss avoided?

>Yes and no. Ultimately, the material has to be decoded and then re-encoded to some format. If you use a DaVinci, you'd likely output from Final Cut back to DVCProHD (no loss), bump it over to some format that the post facility can use (either disk based or tape based), and output from DaVinci back to either disk files (if it's a true DI situation) or another tape format, incurring one compression generation, as all current HD tape formats involve some form of compression. The reality is that DaVinci’ s color processing is done in a very high bit depth environment, thus providing far more manipulative possibilities and little to no noticeable loss, provided you don't have to manipulate the look too much.

>One should keep in mind that color correction from video original is itself a destructive process, you can't change something without causing some degradation to the result, especially when you start doing things like secondary correction (basically a key) and/or severe look manipulation.

>Mike Most
VFX Supervisor
IATSE Local 600
Los Angeles


class="style9">>my advice to you would be to edit using the native DVCProHD codec at >24p with cuts-only (so there's no generation loss), edit that locked edit >back to another DVCProHD master tape

>DVCProHD native codec: is that what I get from the SDI output of the camera?

>Blain Brown
DP
LA


>Blain Brown wrote:

class="style9">>DVCProHD native codec: is that what I get from the SDI output of the >camera

>No, it's what's used for recording in the on-board recorder, and also what's used when that same data is passed via Firewire to a computer.

>This can only be done at the moment by using Panasonic's 1200A deck, because it's the only equipment they currently make that has a Firewire port. SDI is, by definition, a transport protocol that involves uncompressed video.

>Not meaning this to be flippant, but I think you really need to either read up on this on Panasonic's or Apple's web sites, talk to a Panasonic rep, or talk to your local rental house rather than continue to ask questions here. If you're going to use this stuff professionally, you really need to educate yourself on it.

>Mike Most
VFX Supervisor
IATSE Local 600
Los Angeles


>Michael Most wrote:

class="style9">>If you're going to use this stuff professionally, you really need to educate >yourself on it.

>I'm not.

>Just giving the producer some input on post possibilities. The editor will deal with it, just dealing with possible budget scenarios.

>Blain Brown


>Blain Brown wrote:

class="style9">>I'm not. Just giving the producer some input on post possibilities.

>In that case, I'm glad I said I wasn't trying to be flippant!!!

>Mike Most
VFX Supervisor
IATSE Local 600
Los Angeles


This additional update was posted on the list on 29th March 2005 :

>I am writing a response to the HD Post Costs discussion which was posted up on the website today.

>(http://cinematography.net/edited-pages/HD_Post_Costs.htm)

class="style9">>Now, if you going to be doing a lot of color correction, etc., my advice to >you would be to edit using the native DVCProHD codec at 24p with cuts->only

>This isn't strictly true. In Final Cut Pro you can 'nest' sequences within each other, a very powerful feature in managing post workflow (imnsho).

>So what I always do when I'm working with a compressed codec is to embed the final edited sequence inside a NEW sequence. That sequence is set to my output format - usually 10bit Uncompressed Blackmagic*. When you do this, you'll notice the whole sequence will need to be rendered. Before I do that, I make sure my render settings are set to 'render everything in 10bit'. I then render the final. This bypasses any generation loss because FCP converts the original sequence into 32bit float (YUV colour space)**, does all its imaging processing, and then encodes the footage to your final finishing format***.

>As Mike says FCP isn't sophisticated enough to do high end colour manipulation, so the best way to integrate this into your post path is to use this to render all your transitions, like dissolves, and basic effects, like titling, and then bump your footage to tape. Then I'd do my 'tape to tape' colour correction on whatever system I can afford. Otherwise you'll need to keep your handles when you dump the 'cuts only' edit to tape and then do a very basic conform on your high end format before doing your colour correction.

>* Although I have, for gags [aka testing purposes], rendered to 10bit RGB! I normally do this for SD down conversion, but there is no reason you can't use an uncompressed HD frame when working with the Varicam.

>** AFAIK, as of FCP4, its YUV rendering engine is 32bit float (is there any other kind of 32bit?) - but I'm happy to be shot down in flames on this point. [I'm looking at you Mike M.]

>*** Anally speaking, the mere act of transcoding introduces some kind of data loss... but hey, if you're bumping out to tape you'll be doing the same transcode anyway...

>Just my 5c.

>Stu Willis
Director + Editor
Sydney, Australia

www.biki.net