ENG and Cinema...

[Casey Green]

Stuart,

We still haven't heard if it is technically possible to have the user change (upload) the firmware on the FPGA to decide which formats they would like to shoot in...

Can you please chime in on this? Wouldn't that give us the best of ALL worlds?

 

[Stuart English]

What producers want..

What producers want..

Producers don't need to come up to anything. But after my Red investment I'm going to need to make my money back. Yes, eventually Red will prove to everyone that its formats are superior, but until then there are just too many clients that want 720 and 1080. And web commercials and corporate are not just ENG with a B4 lens. Those clients want a high end cinematic look too and they are largely coming up in budget and scope. They want a 35mm look, but in 720 or 1080.

So lets take that 35mm analogy. Film isn't a video format, but its understood how to get it into the video domain at a variety of resolutions. The essence of this discussion is if we can RAW to video in a transparent manner, where then is the issue?

 

[Poi Boy]

shoot raw then convert to whatever the client wants,pretty simple. If you think that is not fast enough opt for a tape based camera.
Aloha
-A

 

[PaulClements]

Stuart is this more of an issue of removing RGB than it is of 720/1080p?

And if we do away with RGB completely is 96fps still the most we can expect at 2k or would there be more room to hit the 120 barrier?

 

[SF Geek]

Poi boy, that's the thing. I don't want to opt for a tape based camera. I'm trying to get rental on the Red.

 

[David Newman]

what about 1080p raw, would that not be possible?

One reason to support 1080p RAW is to avoid a 2048 to 1920 down-scale, this 6% downres is not ideal, as most scalers would soften the image as compared to a 1080p crop. If you going to crop in post, why not crop in camera. Of course shooting 2048 does offer some subtle re-framing options if you going to crop, but based on my experience with the SI-2K, I shoot 2K when I want that and 1080 otherwise -- it speeds up the post the post to avoid any re-framing or scaling. 1080p RAW is a fine option if that can happen.

 

[SF Geek]

Hey Stuart,
The issue is that I need to be able to get footage to the client the day of the shoot. If Redcode Raw and Redcine can do that without too much thought and manpower, then I'm ok with that. I'm just worried that my Red rental will lose out to a varicam or other camera because of convenience for the client.

 

[PaulClements]

One reason to support 1080p RAW is to avoid a 2048 to 1920 down-scale, this 6% downres is not ideal, as most scalers would soften the image as compared to a 1080p crop. If you going to crop in post, why not crop in camera. Of course shooting 2048 does offer some subtle re-framing options if you going to crop, but based on my experience with the SI-2K, I shoot 2K when I want that and 1080 otherwise -- it speeds up the post the post to avoid any re-framing or scaling. 1080p RAW is a fine option if that can happen.

Why not use 6% less of the sensor?

Record only 3840 x 2160 pixels of the entire sensor and then divide the image by 2 when using 35mm lenses and record only 1920 x 1080 of the sensor with S16mm lenses and output as RAW. Wouldn't this just be a simple piece of the math for the brains of the RedOne to remove certain pixels.

 

[I Bloom]

how long is it going to take?

This question keeps coming up. It seems to be the main piece of information that we need in this debate.

I would add another question to the mix. Can we compromise anything to save time?

IBloom

 

[Häakon]

shoot raw then convert to whatever the client wants,pretty simple. If you think that is not fast enough opt for a tape based camera.
Aloha
-A

This is essentially how I feel, too. For me, RED is going to be my alternative for 35mm film. No client gets instant film dailies five minutes after the shoot is over. They can get a hard drive - just like we've been doing with all of the other tapeless solutions - and they can watch the footage instantly with "Redquick." If they need LUTs added, etc., they can wait for a transcode of the RAW footage just like they'd wait for film to be developed and processed. Seems to me the client is covered this way; they have something to watch at the end of the day and they have the ability to see a treated version shortly thereafter if they need. Because of these reasons, I'm really not that concerned.

HOWEVER, as of this writing, all we know is that the only way we can get faster than 30fps readout of the entire sensor is through use of 1080p. I just can't support losing the format if it's the only way I'm going to be able to get higher speeds from the 35mm area. 2K and 1080 are NOT the same thing, people need to understand that. One is windowed and one is scaled; they will behave differently and more often than not need a different set of lenses to work with. Just because they're close in resolution doesn't mean we can just ditch 1080 because 2K may offer us up to 96fps.

Perhaps the RED team knows more about getting higher framerates out of 4K and they are just holding back until they can squeeze the most out of the camera that they can... maybe they're trying to wait until the last moment to give us one last amazing bullet point before the camera ships. Maybe they've done all they can and 30p is the most we can expect out of 4K onboard. I don't really know. But I cannot vote for canning 1080 until we have that information. If it was just a simple case of "the client needs 1080 immidiately!" then I would say there are several ways of shooting in 4K and giving them 1080 content. But 1080 offers something that neither 2K or 4K do - 60p from the full sensor - and I can't believe anyone would want to get rid of that.

 

[David Battistella]

We really just need the numbers Jim. RED needs to look at what its going to cost their average customer to deliver various kinds of footage. How much computing power do they need and how much time.

IBloom

Wait. This workflow is more efficient, smarter and economical than P2. THink about it. You can shoot all of this to CF cards.

There is no difference to what they are proposing from P2. The wrapper allows you to play the footage back at the rate you want. if QT plays it back in realtime then so will FCP, the work is being done in the codec.

If you are delivering 720P then you would edit in the 1K version of the 4K file and RED is doing the conversion on the fly.

Just take a step back and think about how amazing this is.


David

 

[Dexter Gregoire]

Wouldn't it be great to shoot 4k or 2K@ 96 fps then;

Remove the Drive or CF Card,
Plug it into your computer and;

Output to anything you want using the software that comes with the Camera. Apples or Oranges 1080p or 720p. Golden.

It doesn't get any easier than that!

Varicams need a Deck. Decks are expensive. The beauty of what the RED Company has developed makes it easy for Renters, ENGatherers, Junior film makers and Professionals alike to Shoot, Transcode, Edit in the software of your choice and Deliver excellent quality images all at an affordable price. I couldn't ask for anything more, except for 2K@96fps since Jim Jannard brought it up.

 

[David Cubbage]

This debate reminds me of the choices you got with the JVC HD100 camera when it first came out. You had the choice of shooting HDV720p or SD straight from the camera. Now although most people's delivery was only going to be SD at the end, did they set the camera to shoot SD? No, they shot in HDV720p, edited natively and down converted to SD for delivery. Why? Because you got a better end result. I know because that's the way I shot. You only have to read the boards on DVInfo.net to see that's how most HD100 owners worked. The beautiful thing is about this way of shooting is that if, at a later date you wanted an HDV720p version, you already had it. JVC called it "Future Proofing".

I am not comparing the HD100 to the Red but what I am saying is that obviously when JVC designed their camera they thought that they ought to give shooters the option of shooting in a lesser format in camera because their HDV 720p was a completely new format and users might want the choice of giving their customers SD if needs be. But, as I stated in the above paragraph, most people didn't use this option after all.

I think that Jim's suggestions are very sound; the opportunity to shoot in 4K and convert to any other format in in Redcine is truly amazing; why the need to shoot 1080 or 720 in camera when you have all these choices. I think that when our customers see the fantastic footage the Red camera can produce they will all be asking for higher formats than 1080 and 720. Of course it's too late if your master in 1080 or 720.

Why do you think they still use film for major TV series when, apart from the HD channels, you still have to accept broadcasts in SD? Because if your master is the very best quality this shows in your finished product right down the line. With the workflow Red is giving us I see no reason why I should shoot in anything less than 4K with 35mm glass and DOF and down convert to what ever format my future customers demand. And if they change their minds I have the original 4K masters on disc.

I have read every post since Jim's suggestion on this argument, except for the ones I've missed while typing mine, and, while I am sympathetic to some of the arguments against the motion; I still see Jim's suggestion as a positive step forward in the development of Red One.

 

[Michael Hastings]

1080 offers something that neither 2K or 4K do - 60p from the full sensor

1080 offers something that neither 2K or 4K do - 60p from the full sensor

I just can't support losing the format if it's the only way I'm going to be able to get higher speeds from the 35mm area. 2K and 1080 are NOT the same thing, people need to understand that. One is windowed and one is scaled; they will behave differently and more often than not need a different set of lenses to work with. Just because they're close in resolution doesn't mean we can just ditch 1080 because 2K may offer us up to 96fps. But 1080 offers something that neither 2K or 4K do - 60p from the full sensor - and I can't believe anyone would want to get rid of that.

This seems like a pretty big issue, people are paying thousands for 35mm format converters for their HVX200 because they want the DOF characteristics. If 2K can't do that it's a problem.

My question: why can the camera handle 1080p RGB realtime conversion and downsampling of the 35mm sensor but 2K raw can't be done from the full sensor? Doesn't make sense to me.

 

[Jeff Kilgroe]

My question: why can the camera handle 1080p RGB realtime conversion and downsampling of the 35mm sensor but 2K raw can't be done from the full sensor? Doesn't make sense to me.

It has to dow with the sensor data itself and how the bayer pattern and info is dealt with. According to Graeme / Stuart, this can't be scaled unless it is converted into RGB color space first. So this is why the 1080p scaled mode has to be RGB and not RAW.

Of the RGB modes, the only one I'm interested in keeping is the 1080p mode. And only because it would allow for capturing the full 35mm sensor area at rates from 30fps to 60fps. Jim has said that 4K @ 60fps is in the specs, just not for this first release. I'd love to know some more details about that... Are we looking at a minor revision and upgrade in the not too distant future or will this involve a significant upgrade at some much later date?

 

[Häakon]

My question: why can the camera handle 1080p RGB realtime conversion and downsampling of the 35mm sensor but 2K raw can't be done from the full sensor? Doesn't make sense to me.

Well the problem is you can't scale RAW data, you have to convert to RGB first. So whereas 2K is RAW (because it's just taking a crop of actual pixels), 1080p is scaled because it's taking all of the pixels and oversampling them into a smaller image. You lose RAW functionality when you do that, however.

The only way to get RAW data from the full sensor is to shoot with the full sensor (in other words, shoot 4K). Given that, 4K is really the only format I would be interested in. The caveat, however, is that (as of right now, anyway), 4K REDCODE RAW is limited to 30fps onboard. If you need more than that (which many do), the choice is to either oversample or just take a center portion of the sensor so that the data rates are smaller and the camera can handle it. Then the question becomes what's more important to you - the RAW functionality, or the 35mm image characteristics? Resolution is negated at this point because 1080p and 2K are so similar - and the 1080 image should theoretically look better because it's a 4X+ oversample.

Of course if we could have 4K at 60fps REDCODE RAW onboard none of this would be an issue, but I'm sure that if they could offer us this they would in a heartbeat.

 

[Antoine Baumann]

I don't really think going from Redcode raw to cineform raw is a super great solution. I'd like to hear from a member of the Red team, to confirm this (that premiere and after effects can handle redcode raw).

yeah sure they can, it have been stated many times, that any piece of software that support qt codec will hadle redcode raw.

I don't see the point to go from redcode raw to cineform raw...depending on your workflow it might be usefull, but don't see why

antoine.

 

[Häakon]

You beat me to it, Jeff.

As for the 4K/60p...

Jim has said that 4K @ 60fps is in the specs, just not for this first release. I'd love to know some more details about that... Are we looking at a minor revision and upgrade in the not too distant future or will this involve a significant upgrade at some much later date?

This *is* already in the spec, but only in the (uncompressed) RAW mode out the high-speed data port. And as far as I can recall, the HSDP isn't going to be ready right when the camera first starts shipping either. The sensor is capable of reading the entire 4K area 60 times a second, but there isn't enough juice to take that image, compress it to REDCODE RAW, and write it to the media that quickly. I thought this was more of a processing issue than a bandwidth issue, but some of Jarred's recent comments seem to indicate otherwise.

If they can't get 4K/60p onboard enabled by the time the camera ships, it certainly would be great to know that it's something a firmware update could address. Up until this point, however, I understood it to be a hardware limitation.

 

[M Most]

I repeat something that someone else asked about 5 pages ago, but it hasn't been answered and bears repeating:

How will these reference files compare to a full 4k decode and scale to 1080p in Redcine? (Or to a on board 1080p)

The notion that playing from a Quicktime file in Final Cut is a post solution is only valid if there is no difference between a real time demosaic operation, and a rendered demosaic. As far as I know this is not the case - in order to get full quality, you need to render. That would mean that regardless of the convenience for editing, playing the Quicktime file is not a delivery solution, and that at some point, a full render will always be required to create a deliverable product, regardless of whether it's done in Redcine or somewhere else. Am I wrong here?

 

[Jeff Kilgroe]

As for the 4K/60p...

This *is* already in the spec, but only in the (uncompressed) RAW mode out the high-speed data port.

Yeah, I'm fully aware of all that... But I can't justify having the RAW port factory installed (and eliminating my FLASH option) for a feature that I will only use every once in a while. Besides, it would also require a large external RAID and/or powerful system to process and compress the data stream to REDCODE RAW in real-time.

In Tom's recent thread from the last day or so, making one last plea for 4K @ 60fps on-board, Jim responded. He said that 4K @60 was in the spec, just not in the first release... So I take that as an encouraging sign, but I think what it means is just what it literally says. I think RED is planning for 4K at 60fps and beyond to be recordable on-camera. And that will come in a future release (as in version 2.0 of the camera or however you want to look at it) and it will require first release cameras to be sent in for an upgrade.

It's the on-board processors (as far as I've always understood it) that are the problem with recording beyond 30fps in 4K. They just can't cope with that amount of data to crunch.... This led to all that discussion about RAM buffers or other options that would allow for recording 4K @ 60fps on-board rather than being tethered to some box the size of an under-counter refrigerator.