PDA

View Full Version : RGB and deliverables



Stuart English
08-02-2007, 08:12 AM
Time for a new thread here I think. RGB v's RAW is a great subject for discussion.

Part one is about deliverables and NLE compatibility, part two about sensor usage and dynamic range / color space.

So what is a deliverable? Is it the thing you hand off to your client at the end of a days shoot, or the final product you deliver after post production and on the target delivery medium?

An obvious answer is both. But lets explore the end of the day's shoot deliverable. What is your expectation about what you will show to the client, give the client, and what they will do with that?

Also when we describe RGB we really do mean RGB 4:4:4, not Y,Pb,Pr 4:2:2, so some transcoding to "television color space" is still required.

Antoine Baumann
08-02-2007, 08:22 AM
Also when we describe RGB we really do mean RGB 4:4:4, not Y,Pb,Pr 4:2:2, so some transcoding to "television color space" is still required.

But Jim wrote that


720P and 1080P are RGB recording formats- 10bit 4:2:2.

antoine.

Jannard
08-02-2007, 08:28 AM
Stuart is right (as usual), Jim is wrong... oops

I have fixed my other post. Sorry about that.

Jim

Antoine Baumann
08-02-2007, 08:29 AM
ah that's good news as long as I am concern.

thanks for the quick update.

antoine.

I Bloom
08-02-2007, 08:33 AM
An obvious answer is both. But lets explore the end of the day's shoot deliverable. What is your expectation about what you will show to the client, give the client, and what they will do with that?

Stuart,

Thanks for starting this thread.

It seems clear that REDCine can deliver most types of deliverables needed for any Final Cut System. Final Cut may support the format natively or REDCine can transcode it to another type of quicktime. I understand that all RED formats will be RGB.

The primary economic issue is how much transcoding is needed prior to editing. What does it take to support a client who doesn't want 4K or doesn't use Final Cut Pro or worse isn't upgrading to FCP6.

So here are my questions for RED:

1. How labor and time intensive is REDCine per hour of footage? Give us two standard apple setups, a Macbook Pro and a MacPro tower, and some popular transcodes like 4K to DVCProHD (a question you may not be able to answer at this time. But I think we established in the "losers" thread this is really a deciding factor.)
2. Does shooting in a smaller format in camera reduce the time and labor or increase support in alternative NLE's or FCP less than 6?
3. How does REDCine support other NLE's such as AVID, Premiere or Vegas etc.?
4. What about RED support on PC's where every quicktime codec is not available?
5. Also what about Red's quicktime components on non-intel systems like G5's.

90 percent of what I do is handed off to clients straight from camera. In no circumstances will I dictate to them their post production pipeline. Many of them have fast turnarounds. I need to be able to say "yes" to them... hand off a drive... and the next day I only hear about how good the footage looks.

Warm thanks,
IBloom

6. One more question, will you compile REDCine for Linux. Sounds like a wierd question but it makes a lot of sense if you are looking for a stable robust and cheap system to do some grunt work.

Paul Hazlett
08-02-2007, 08:43 AM
I am not too sure about handing drives to a client yet.

unless they bring there own....good luck. or you give them one, good luck
gettting it back. But I sure there are some who will have better luck than others.

I dont know about other ENF/EFP guys but I am hoping to do tape on the scene so they can walk with that, end of story. if its run and gun, not sure.

Keith Brust
08-02-2007, 08:48 AM
Stuart,

Like Ibloom, unless I'm shooting something for myself, I will be handing off a drive to the client at the end of the day. I shoot nature docs for TV. They will decide which format I will be shooting in and they will get the files from the camera as shot.

Keith

Paul Hazlett
08-02-2007, 08:52 AM
Also when we describe RGB we really do mean RGB 4:4:4, not Y,Pb,Pr 4:2:2, so some transcoding to "television color space" is still required.

So are you saying recording to tape is not possible unless we have an hdcam
sr or an outboard converter?

Stephen Webb
08-02-2007, 09:03 AM
iBloom & Keith, you do realise that you'll still have to transcode your footage before you hand it to your client don't you? You can't just hand them a drive with REDCODE RGB files (regardless of the resolution) unless they have a workflow themselves that supports REDCODE. So either they can support REDCODE (in which case the resolution doesn't matter) or you'll have to convert it for them.

Steve Freebairn
08-02-2007, 09:12 AM
I'm planning on making tape (archival data tape) backups on set and then either selling them a drive with their data on it or transferring the data to a drive they have provided.

Chris Gearhart
08-02-2007, 09:17 AM
Yeah, there seems to be the need for some sort of laptop intermediate step for the standard hand-to-client-at-end-of-day deliverable. Could you run REDQUICK from, say, a PDA or a palmtop?

David Battistella
08-02-2007, 09:27 AM
So what is a deliverable? Is it the thing you hand off to your client at the end of a days shoot, or the final product you deliver after post production and on the target delivery medium?

Stuart,

Here is how I would handle POST deliverables. 24P HD universal mastering is becoming the current standard for broadcast. This format allows the most flexibility and range of markets to sell into. With a 24P master you can sell to virtually any broadcaster via, most likely, your distributer. As a producer you want this format ad it relieves a huge burden of multiple HD SD formats and frame rates.

Theatrical releases currently rely on 2K digital projection or a film print. In the future affordable 4K projection may be the answer.

So a 2k timeline in FCP could go to a file, a film print or a 24P master tape. With that same 2K timeline you also use compressor to make SD, or web deliverables. At this stage who cares, is just encoding times.

If you know that you are not going to deliver in anything but SD then work with 1K proxies as you masters. You always have the 4K to go back to.

As a small content producer, creating, short film, corporate, tv or web it is still very worthwhile to shoot 4K at all times and over cranked 2K sometimes. This allows me to produce at the highest quality and scale the best possible image down the pipeline all the way to youtube. So for these, in house productions, this really is the only workflow that makes sense. The key is to PLAN for it.




An obvious answer is both. But lets explore the end of the day's shoot deliverable. What is your expectation about what you will show to the client, give the client, and what they will do with that?


The deliverable you hand your client could be the same thing, a 4K RED file. The question you ask your client is not, so you want 720P or 1080P because with a 4K RED file you are giving them that.

The thing to ask your client whether they are using FCP. If your client is an FCP house then you should provide them with the 4K files that can be scaled back to what ever format they are editing or finsihing in. Why would you limit their potential to go bigger, etc.?

I would suggest making REDCINE a free download and allowing us to put it on the drives we hand the client to limit the amount of initial confusion. Acceptance of a new format will partially be driven by how easy we make it for people to use.

In my experience people do not like closed systems, or ones which require a lot of effort on their part. If it's as simple as "have your editor go here and get the download and you are set to go in FCP" or "everything you need is on the drives" then I think people will quickly warm to this new file system.

At the end of the day if the client wants a tape, the client wants a tape and there is no real way to address that situation without warming them to this new standard of working. Once people understand that it is actually a time saver and you are giving them an image you could project on the side of a building, they will eventually see how much sense it makes.

David

I Bloom
08-02-2007, 09:42 AM
I am not too sure about handing drives to a client yet.

unless they bring there own....good luck. or you give them one, good luck
gettting it back. But I sure there are some who will have better luck than others.

I dont know about other ENF/EFP guys but I am hoping to do tape on the scene so they can walk with that, end of story. if its run and gun, not sure.

I'm not talking about a RED drive. But yes after I copy the footage, I often hand drives to clients, sometimes they bought it, sometimes its a loaner. I also have a standard letter that I give to editors with information on the codecs and framerates on the drive. Why would you give them a tape and make them re-import footage that is already on a drive and seperated into clips???

IBloom

Stuart English
08-02-2007, 09:44 AM
It seems clear that REDCine can deliver most types of deliverables needed for any Final Cut System. Final Cut may support the format natively or REDCine can transcode it to another type of quicktime. I understand that all RED formats will be RGB.

The primary economic issue is how much transcoding is needed prior to editing. What does it take to support a client who doesn't want 4K or doesn't use Final Cut Pro or worse isn't upgrading to FCP6.


Bottom line is that unless the NLE has native support for Wavelets, there is no ability to immediately edit a RED ONE recordng made at 4K RAW, 1080p RGB or 720p 4:2:2 (even if we offered the latter) without a transcode.

The landscape is changing rapidly here, and the new data point is multi-resolution timelines and ability to edit / decode directly from RAW data. As I mentioned in another thread, I have been playing Quicktime reference movies on my 15" laptop in Quicktime Player - the codec is extracting proxy data in real time from the 4K REDCODE RAW data set. - That's pretty sweet.

If you need to deliver to an older NLE, recording REDCODE RGB or REDCODE RAW is the same or very similar render time, because Wavelets compression is a different animal than DCT based codecs that these NLEs been using. The advantage in staying RAW is wider color space and higher dynamic range. Its a purer record, with little or no downsides (except for maximum recordable frame rate at 4K RAW v's 2K RGB)

Again, just information, please to engage in the discussion.

I Bloom
08-02-2007, 09:52 AM
iBloom & Keith, you do realise that you'll still have to transcode your footage before you hand it to your client don't you? You can't just hand them a drive with REDCODE RGB files (regardless of the resolution) unless they have a workflow themselves that supports REDCODE. So either they can support REDCODE (in which case the resolution doesn't matter) or you'll have to convert it for them.

I think in some cases you can. If they are cutting on FCP6 you will be able to say, here you go, you can use this and its better. FCP6 is format independant, so they don't even need to adjust their timeline settings.

However a 4K file is still 4K. Its compressed but still about 4 times the size of HDCAM. That means 4 times the storage and 4 times the throughput on the transcoder.

Remember REDCine doesn't have a FPGA to accelerate compression, it runs off and is limited to the processing power of your computer. A computer you need to have near you at the end of a day of shooting if you expect to deliver the next morning.

IBloom

Stuart English
08-02-2007, 10:00 AM
You can't just hand them a drive with REDCODE RGB files (regardless of the resolution) unless they have a workflow themselves that supports REDCODE. .

True, but from the perspective of being able to instantly review the footage that is on the drive, the Quicktime reference (proxy) files that the camera or secondary RED software can rapidly generate, will provide your client a viewable video clip decodeable on the majority of Macs or PCs. So the infrastructure needed for this our RED Quicktime plug in component.

That's what I have been doing on this 15" laptop - reading a 510K resolution file which extracts directly from the native 4K REDCODE RAW data recorded on CF or RED-DRIVE - b.t.w we always reserve 200MB of space on the media to permit Quicktime reference files to be written.

Steven Parker
08-02-2007, 10:03 AM
I would suggest making REDCINE a free download and allowing us to put it on the drives we hand the client to limit the amount of initial confusion. Acceptance of a new format will partially be driven by how easy we make it for people to use.

David

Well said and I agree - this is something I've been wanting since the get-go. Allowing RedCine to be downloadable will make the RedCode viable immediately.

It also takes the pressure off us as shooters to turnaround footage so quickly.

A client version of RedCine doesn't need to be fully-featured, either - just useable enough for them to edit in their NLE of choice and transcode out their desired format(s). Or is that full functionality anyway?:huh:

Steven Parker
08-02-2007, 10:19 AM
True, but from the perspective of being able to instantly review the footage that is on the drive, the Quicktime reference (proxy) files that the camera or secondary RED software can rapidly generate, will provide your client a viewable video clip decodeable on the majority of Macs or PCs. So the infrastructure needed for this our RED Quicktime plug in component.


Just saw this on Jim's ENG thread

http://www.reduser.net/forum/showthread.php?t=3501&page=2...

so RedQuick sounds like the thing we've been wanting. Now, can a client edit with RedQuick QTs? Seems like they could... which means... no transcode times?:ohmy:

Zach Hilton
08-02-2007, 10:32 AM
True, but from the perspective of being able to instantly review the footage that is on the drive, the Quicktime reference (proxy) files that the camera or secondary RED software can rapidly generate, will provide your client a viewable video clip decodeable on the majority of Macs or PCs. So the infrastructure needed for this our RED Quicktime plug in component.

That's what I have been doing on this 15" laptop - reading a 510K resolution file which extracts directly from the native 4K REDCODE RAW data recorded on CF or RED-DRIVE - b.t.w we always reserve 200MB of space on the media to permit Quicktime reference files to be written.

So much new information...so many new questions. You mentioned that it always reserves 200MB of space, so I take it that is for the entire media? So in essence I would have 200 MB less of formatted space on my CF's and RED DRIVE? I know it will probably be minimal when it comes down to REDCODE data rates, just curious.

Gavin Greenwalt
08-02-2007, 10:50 AM
Workflow to get footage into an AVID or non REDCode supporting pipeline/system

from 1080p:
Copy to RAID from recording medium.
Transcode to DVCPro.
Copy to deliverable drive.

from 2k or 4k RAW:
Copy to Raid from recording medium.
Transcode to DVCPro
Copy to deliverable drive.

Workflow to get footage into a REDCode supported workflow (REDcined client or compatible NLE)

from 1080p:
Copy to RAID from recording medium.
Copy to deliverable drive.

from 2k or 4k RAW:
Copy to Raid from recording medium.
Copy to deliverable drive.

So what's the problem with RAW again?

Stuart English
08-02-2007, 10:58 AM
So much new information...so many new questions. You mentioned that it always reserves 200MB of space, so I take it that is for the entire media? So in essence I would have 200 MB less of formatted space on my CF's and RED DRIVE? I know it will probably be minimal when it comes down to REDCODE data rates, just curious.

Yes. That's space we know exists so when we come to write Quicktimes reference files. Its a fixed 200GB at the moment, but it really should depend on the capacity of your media and be a % of that. These reference files are pretty small, they point to the RAW data files and contain enough data to tell Quicktime Player how to render this information.

Michael Ragen
08-02-2007, 11:06 AM
Hey Stuart,
I remember hearing at NAB that you can transcode to Prores on import to FCP. Is it possible to import and transcode 2k or 4k Redcode Raw directly to 1080p Prores?

Steve Freebairn
08-02-2007, 11:12 AM
Can Adobe software on PC and Mac read the reference files and play them back ok? How is the workflow for clients who are using Adobe?

Zach Hilton
08-02-2007, 11:12 AM
Yes. That's space we know exists so when we come to write Quicktimes reference files. Its a fixed 200GB at the moment, but it really should depend on the capacity of your media and be a % of that. These reference files are pretty small, they point to the RAW data files and contain enough data to tell Quicktime Player how to render this information.

Thanks. That makes sense.

Thomas Mathai
08-02-2007, 12:56 PM
I think it all depends on the deal between the client and the Red owner.

It can be just shoot it and hand over a drive (hopefully a cloned drive) with the RAW camera source.

This would be like a camera man shooting film and giving the exposed film cans to the producer to have someone else worry about.

On the other hand, it can be beneficial for the Red owner to provide additional services to the producer like pretiming and transcoding Raw media.

In any case, I think all Red owners have to be somewhat post saavy.

I would guess what HVX200 users have been doing for their clients would be similar, since many clients want just MXF files.

stephenlnoe
08-02-2007, 01:39 PM
Do we actually have any of the raw red media to experiment with (downloadable)? RGB only creates havoc for TV production which requires fast turn around.

REDHKSC
08-02-2007, 01:50 PM
Stuart,

besides REDCODE w/ RED one camera, will RED One camera on board also support PRO-RES 10bit 4:2:2 as well onto either CF or RED drives ?

Stewart



Yes. That's space we know exists so when we come to write Quicktimes reference files. Its a fixed 200GB at the moment, but it really should depend on the capacity of your media and be a % of that. These reference files are pretty small, they point to the RAW data files and contain enough data to tell Quicktime Player how to render this information.

Häakon
08-02-2007, 01:51 PM
The advantage in staying RAW is wider color space and higher dynamic range. Its a purer record, with little or no downsides (except for maximum recordable frame rate at 4K RAW v's 2K RGB)
Unfortunately, for many of us, that's a huge downside. For my own personal uses, I could see myself shooting 4K almost exclusively (gives me the best picture with a way to deliver any resolution to a client), but as it tops out at 30p, it's just not going to cut it for everything. 2K RAW allows more flexibility with the framerates (and has the added benefits of being 4x smaller - which means longer record times), but the downside is that it is windowed. I'm really not sure how I want to deal with that one.

I am not as worried about the deliverables as some - I think that clients will adapt just as they quickly have with P2, especially if they are ASKING for RED for a specific project - but I'm just concerned about what it means we're actually losing if we dump the RGB formats.

David Battistella
08-02-2007, 05:09 PM
Stuart,

besides REDCODE w/ RED one camera, will RED One camera on board also support PRO-RES 10bit 4:2:2 as well onto either CF or RED drives ?

Stewart

Red,

I seriously doubt this. Pro Res is a hardware dependent codec that really needs the full power of an INTEL mac to even use it in FCP. The camera does not shoot in 4:2:2 apple proprietary codecs.

David

Joe Carney
08-02-2007, 05:18 PM
Can Adobe software on PC and Mac read the reference files and play them back ok? How is the workflow for clients who are using Adobe?

I've asked that question in one form or another several times with no answer.
What will it take to get support for Premier and After Effects on Windows and Mac?
Has Red agreed to some exclusive deal with Apple?

Paul Hazlett
08-02-2007, 05:19 PM
I am not as worried about the deliverables as some - I think that clients will adapt just as they quickly have with P2, especially if they are ASKING for RED for a specific project - but I'm just concerned about what it means we're actually losing if we dump the RGB formats.

The problem here is the asking part. Most clients would still be shooting
3/4 inch, well maybe not that bad, but I think we will need to do a few dog
and pony shows first to sell them on the idea of red since it will cost more
but there are benefits, sell sell sell.

David Battistella
08-02-2007, 05:24 PM
However a 4K file is still 4K. Its compressed but still about 4 times the size of HDCAM. That means 4 times the storage and 4 times the throughput on the transcoder.

IBloom


Wait . Wait. Wait.

A REDCODE 4K file can not be compared to HDCAM unless you are only comparing the FRAME size. The data size of the 4K file is only about 27MB/sec which is about the size of 10 bit uncompressed SD. SO IT IS NO WHERE NEAR 4TIME THE SIZE OF HDCAM.

When you capture from an HD CAM tape you have the choice of capturing that to an uncompressed 10 BIT HD signal which is about 165MB/sec. This means that there is actually about six time more number crunching when you are working in a HDCAM uncompressed workflow. If you captured the HDCAM to DVCPRO HD it might be about 14MB/sec.

So working with REDCODE is going to be more like working with DVCPRO HD than it would be working with HDCAM. **although the quality of REDCODE will be far superior**

You have to think of it this way.

You use REDQUICK to put a quicktime wrapper on the file. Your computer sees this media as it does any other QT media. Now you can choose to tell that file to play back at a frame size of your choice. 4K, 2K, 1, K or .5K
Let's say you are editing a proxy of the 1K file. This means you are asking the REDCODE to DECODE the footage at a lower data rate, it does this on the fly, making it more comparable to DVCPRO HD.

The best part of the scenario is that you have shot with all of the latitude and benefits of 4K and the codec does the scaling work for you.

With REdcode enables in FCP 6.0's open timeline, you could actually intercut this footage witha multitude of formats.

David

Poi Boy
08-02-2007, 05:26 PM
My market is television commercials particularly those shot on film. I think red will be an easy sell to them because they will be able to get as good and better quality with a smaller budget.
Aloha
-A

Blair S. Paulsen
08-02-2007, 05:29 PM
Throughout these discussions the Red Team has been careful not to quote render times for RedCine - my comments are based on a completely unsupported assumption that by the end of 2007 the rendering will happen in close to real time.

In the world of end of the day deliverables from a data based camera I see three options:

1) the client provides CF cards or RedDrives and has a QuickTime savvy post pipeline. Probably a small portion of the market but you get immediate turnaround, full image quality available plus no need for a data wrangler on set with gear.

2) Tether a deck or DVR to the HD-SDI outs

3) Hire a data wrangler to clone footage onto client supplied drives (or invoice them for the drives you brought just like tape stock) either in RedCode if they can handle it which means standard hard disk transfer times or use RedCine to render out the file types the client specs out. Assuming you pull the CF, RedDrive, etc off the rig every hour or so and are not rolling continuously, I am hoping I can clone the last few takes of the day during the strike.
The two ways I would consider doing this would be either laptop based or full tower depending on the field situation.

Portable: Laptop with eSATA adapter & FW800, 2 disc RAID in a compact enclosure, several RedDrives or CF cards. For power either car DC via inverter or AB bricks, gold mount plate with d-tap to voltage regulators. All packed in a Pelican type case. Could use a "lunch pail" type power supply also. Obviously available AC would make this even easier.

On wheeled cart or in nearby vehicle: Laptop or Tower computer with bigger RAID.

What about the resistance to the cost of the data wrangler? The short answer is that they will save it in post since there is no ingest time but the way many budgets are written that may not help your stressed producer. I will try to sell the data wrangler as someone who can help set up in the morning, lend hands during big moves, repos, turnarounds, etc and can manage video village - YMMV.

Ultimately, unless the RAW to codec of choice render rate is much slower than the RGB to codec of choice render rate why would I give up the advantages of RAW even in fast turnaround shoots?

David Battistella
08-02-2007, 05:38 PM
but I think we will need to do a few dog
and pony shows first to sell them on the idea of red since it will cost more
but there are benefits, sell sell sell.

How is RED going to cost more than a P2 workflow?

I do not think you should be saying that it is RED, but rather "tapeless" workflows that added many benefits, but also the data transfer and storage problems associated with moving large amounts of DATA.

This is why I think that eventually the CF card will be to RED what the tape is to HDCAM, BETACAM, DIGIBETA, BETA SP, DVCPRO, MINI DV, the disk is to XDCAM, and the E is to Etc, ENG, EFP and L M N O P. :)

P2 is proving to be the M2 of the digital age. It is cost prohibitive to purchase. It is time consuming to both archive and deal with once it has been shot. Having an external, reliable source of media on the shelf is necessary. Film has this great benefit and so does tape.

If RED wants to extend the camera revolution into POST production then I think it is esential for them to get around this part of tapeless workflows.

RED's biggest push should be for affordable, reliable, inexpensive compact flash media that can be used as the archive. They could really revolutionize post by creating a large demand for fast, inexpensive solid state media that is affordable enough to keep on the shelf.

Shoot ONCE
transfer ONCE
archive ONCE

If the CF media becomes the archive, like tape or film, then we have solved MANY interm problems. We are so close to that now. HOW ABOUT A four BAY CF card reader that attaches to the side of the camera instead of just one card, and the ability to roll over from one card to the other, that is one good thing about P2.

David

Gavin Greenwalt
08-02-2007, 07:07 PM
Tapes you have to capture.
Film you have to develop (and telecine)
Data you have to copy (and or transcode).

When you hand of a tape at the end of the day, unless the editor is still using decks and EDLs exclusively (highly unlikely) there is still going to be a step where the data becomes "useful".

If you can get to disposable solid state then you've eliminated copying. And if you can convert enough pieces of software to work natively with REDCode you can also dispose with all of the transcoding. At which point of the three options Data Aquisition is the only option which doesn't have an intermediate step.

It would be absolutely fantastic if REDCine came with a wizard mode where you just hand the editor a copy of REDCine along with the footage and he can choose:

Framerate [24]
reverse-Pulldown and interlace [ ] Advanced [ ]

Resolution:
[ ] SD
[ ] SD (Anamorphic)
[X] 720 HD
[ ] 1080 HD

Codec:
[ ] DV
[ ] DVCPro
[X] DVCPro HD
[ ] Apple ProRes
[ ] DNxHD

------------------
You have chosen to encode the source material (4k RAW @ 24fps) to (24fps 720p DVCPro HD)

[Process]


I think there is going to be a strong desire for a brain dead conversion process.

Häakon
08-02-2007, 07:11 PM
Fantastic point, Gavin.

Simon Dean
08-03-2007, 01:56 AM
Am I missing something, because I'm not understanding why you couldn't edit the footage in an NLE directly - given enough power. As someone mentioned, cineform is wavelet based and the codec lets you edit directly in the NLE - that it's whole purpose. So why is there a question over editing the footage directly?

I know on the whole it's more to do with what people are used to, but is it not technically possible, isn't that what we should aim for?

Perhaps I'm being dumb here.

David Battistella
08-03-2007, 06:15 AM
Am I missing something, because I'm not understanding why you couldn't edit the footage in an NLE directly - given enough power. As someone mentioned, cineform is wavelet based and the codec lets you edit directly in the NLE - that it's whole purpose. So why is there a question over editing the footage directly?


I think the big question might be:

How exactly does RED footage work in FInal Cut Pro?

We know that it REDCODE will be supported in a future release so this has me assuming that ANY red footage with a QT wrapper file can be used as any other CODEC in FCP.

What I am curious about is if FCP can edit the proxy files or whether those files need to be transcoded before being able to edit them?

This would obviously be a huge time saver.

Stuart?

Could you lay out a brief FCP workflow if I shot 4K REDCODE.

David

Johnny Friday
08-03-2007, 06:39 AM
It would be absolutely fantastic if REDCine came with a wizard mode where you just hand the editor a copy of REDCine along with the footage and he can choose:

Framerate [24]
reverse-Pulldown and interlace [ ] Advanced [ ]

Resolution:
[ ] SD
[ ] SD (Anamorphic)
[X] 720 HD
[ ] 1080 HD

Codec:
[ ] DV
[ ] DVCPro
[X] DVCPro HD
[ ] Apple ProRes
[ ] DNxHD

------------------
You have chosen to encode the source material (4k RAW @ 24fps) to (24fps 720p DVCPro HD)

[Process]




Excellent point and comment. I'm one of many that are "brain dead" and maxed out with keeping up with my software....and prefer to be behind the camera rather than behind the computer---unfortunately time behind the computer is necesarry. So a simple conversion tool makes a lot of sense to a simple minded type like myself: 4k or 2k IN; 1080p, 720p, 2k or 4k raw OUT using codec of mine or my clients choice.

Simple and sweet!

Andrew Young
08-03-2007, 06:58 AM
You use REDQUICK to put a quicktime wrapper on the file. Your computer sees this media as it does any other QT media. Now you can choose to tell that file to play back at a frame size of your choice. 4K, 2K, 1, K or .5K


I have heard from Red that right now 2K is the max playback of the 4k Quicktime files, which may be why they use 4K DPX files played from a server at their screenings. If I am wrong about this, someone from the Red Team please correct.

Nathan Troutman
08-03-2007, 07:01 AM
RED's biggest push should be for affordable, reliable, inexpensive compact flash media that can be used as the archive. They could really revolutionize post by creating a large demand for fast, inexpensive solid state media that is affordable enough to keep on the shelf.

Shoot ONCE
transfer ONCE
archive ONCE

If the CF media becomes the archive, like tape or film, then we have solved MANY interm problems. David

Anybody know the long-term reliability of CF flash as an archival storage medium?

If your footage isn't going to magically disappear in 5 years, CF could work and be another option to DLT tape based solutions - also provided that capacities grow, and price shrinks.

Jim Arthurs
08-03-2007, 07:25 AM
When you capture from an HD CAM tape you have the choice of capturing that to an uncompressed 10 BIT HD signal which is about 165MB/sec. This means that there is actually about six time more number crunching when you are working in a HDCAM uncompressed workflow. If you captured the HDCAM to DVCPRO HD it might be about 14MB/sec.David

For fun, I do have to point out that the actual RECORDED data rate of HDCAM to tape is about 22MB/sec... or about the same as 8bit uncompressed SD.

I know this, because I own an NLE system that used the raw Sony HDCAM codec data. This is the signal that comes form the HDCAM decks optional SDTI card. The system was a modified DPS Reality boardset that could ingest via the boards SDI inputs and output the same. Think "firewire style" editing elevated to 22MB/sec using the HDCAM codec instead of the DV codec. Problem was, Sony charged $4000 just for the codec license, and you needed an HDCAM deck with the SDTI card just to monitor the footage. Limited use, but a mighty fine way to make lossless dubs with edits though...

Steve Freebairn
08-03-2007, 07:29 AM
I think the big question might be:

How exactly does RED footage work in FInal Cut Pro?

We know that it REDCODE will be supported in a future release so this has me assuming that ANY red footage with a QT wrapper file can be used as any other CODEC in FCP.

What I am curious about is if FCP can edit the proxy files or whether those files need to be transcoded before being able to edit them?

This would obviously be a huge time saver.

Stuart?

Could you lay out a brief FCP workflow if I shot 4K REDCODE.

David


Stuart, please do so for FCP and Premiere.

battistella, didn't you mention in another thread here that you worked on the recent Superman? Didn't you guys use Premiere?

Even if Red is pro Apple, who here doesn't have a client or work with people who use Adobe products like Photoshop (which hopefully will be able to access the 4k files), After Effects, and heaven forbid Premiere Pro?

Someone please tell us the workflow and compatibility with Adobe products.:help:

David Battistella
08-03-2007, 07:55 AM
battistella, didn't you mention in another thread here that you worked on the recent Superman? Didn't you guys use Premiere?



No. I didnt work on Superman. We use FCP now, but I have experience with every NLE system dating back to the laser disk based EDIT DROID.

Now if that is not dating me I do not know what is, sheesh!

David

Steve Freebairn
08-03-2007, 08:37 AM
No. I didnt work on Superman.

David


Sorry, confused you with someone else on the board.

Still need to know from a member of the red team how Adobe works with Red.

Joe Carney
08-03-2007, 09:22 AM
I thought Rob Lohman was working on the DirectShow version. haven't heard anything about that in awhile.
Still with Adobe now being cross platform, Quicktime is still needed on Wintel boxes.
Adobe (Premier) has been very silent about all this, their Mac version seems to be released in stealth mode or something. hehehe.

M Most
08-03-2007, 01:34 PM
Workflow to get footage into an AVID or non REDCode supporting pipeline/system
........

So what's the problem with RAW again?

The problem is twofold. First, none of us have any idea how long those transcoding steps are going to take. Second, almost everyone here seems to make the assumption that playing back a proxy in FCP means you now have a deliverable product. As far as I can tell, this is not the case if you want the best quality demosaic, which I assume one would on a deliverable product. That requires a full render/demosaic from the original source files, i.e., an online conform/render. I don't know that I'm right about this, but I also don't know that I'm wrong, because I don't think we've really gotten a definitive answer.

By the way, this is exactly how it's handled in Cineform (real time demosaic is proxy quality, the final product requires a full render).

Jeff Kilgroe
08-03-2007, 01:41 PM
I don't think most everyone is assuming that realtime proxy playback is good enough for deliverables. But I do think that's a good indication of what down-res times will be like. In other words, I don't think a full demosaic and scale of the RAW footage is going to be a big deal. I would be more concerned about that crappy old DVCPROHD or HDCAM codec you're using and how efficeintly it's going to be able to render out the data to your target format. REDCODE and REDCINE are supposed to be fully multithreaded, if I recall Stuart's and Rob's statements several months ago, around NAB or a bit before. For some reason the transcoding aspect of all this doesn't concern me that much.

We also don't know what quality level we're looking at for the proxy generation and if there's any room to configure or set levels, even up to a full quality render. Fact is, right now we just don't know and no matter what possible theories we come up with, it's all just speculation. The RED team is fully aware of our concerns and yet they seem to have little outward concern for the matter. Perhaps I'm a looking at this through a glass that's half full, but it says to me that we don't have much to worry about.

Gavin Greenwalt
08-03-2007, 01:44 PM
Scaling can be done at faster than real time speeds. I can almost guarantee it. That's the sort of thing GPUs just love to do.

De-mosaic will probably be pretty snappy I would imagine. Especially in REDCine where the code can be multi-threaded and GPU accelerated.

I'm with Jeff on this one. The slowest step will probably be DVCAM/DVCProHD compression.