Jump to content


Photo

JPEG frames


  • This topic is locked This topic is locked
36 replies to this topic

#1 Phil Rhodes

Phil Rhodes
  • Sustaining Members
  • 11937 posts
  • Other

Posted 20 May 2007 - 09:39 PM

I've just been looking at a couple of JPEGs on the Red site.

It's very creditable stuff. Highlight handling isn't great, but then, it's fundamentally a video camera. OK. So far, so good. You could shoot a movie with it, as much as you could shoot a movie with a Viper.

They're JPEGs, so it's hard to make any very critical judgments (which in itself makes me suspicious), but what strikes me is this: it looks like it's all been slightly unsharp-masked. This is an artifact typical of DSLR owners with feeble cameras trying to make very large prints look better in Photoshop, and it is somewhat effective.

However.

It is also an artifact typical of overoptimistic de-mosaic work, to the point that here I think it may have been processed so much to get every last iota of information out of it, that it's actually been damaged by the effort. The dark suspicion I can't avoid here - and honestly, only because the Red people have been so vocal about it - is that it has exactly the problem that I (and anyone who can count) said it would have, in that you can't make a very convincing 4K picture out of a 4K bayer array.

But I want to see TIFFs or something more concrete.

The one that particularly alarmed me is here:

http://www.red.com/s...till/009000.jpg

Look at the crosspoints on the mesh on his helmet.

Phil
  • 0

#2 Jim Jannard

Jim Jannard
  • Basic Members
  • PipPipPipPip
  • 361 posts
  • Cinematographer

Posted 20 May 2007 - 09:58 PM

Phil... the only thing you have forgotten to mention is that these were shot with Alpha prototypes (from two months ago) and we have been quick to point out that we had a lot of work still in front of us. We choose to be open about the process, including showing work from unfinished prototypes. You criticque is valid, but only if you consider what I have just said. We have come a long way since then. Your skeptcism remains valid until we post newer.

Jim

Edited by Jim Jannard, 20 May 2007 - 09:59 PM.

  • 0

#3 Jim Jannard

Jim Jannard
  • Basic Members
  • PipPipPipPip
  • 361 posts
  • Cinematographer

Posted 20 May 2007 - 10:31 PM

You could shoot a movie with it, as much as you could shoot a movie with a Viper.
Many would disagree with you here, even discounting the full frame 35mm vs. 2/3" sensor size.

They're JPEGs, so it's hard to make any very critical judgments (which in itself makes me suspicious), but what strikes me is this: it looks like it's all been slightly unsharp-masked. This is an artifact typical of DSLR owners with feeble cameras trying to make very large prints look better in Photoshop, and it is somewhat effective.
Agree that critical judgment is hard to make on a jpeg. No sharpening done in Photoshop (or anywhere in the process) as you would suggest. Demosaic has changed several times since this posting.

However.

It is also an artifact typical of overoptimistic de-mosaic work, to the point that here I think it may have been processed so much to get every last iota of information out of it, that it's actually been damaged by the effort. The dark suspicion I can't avoid here - and honestly, only because the Red people have been so vocal about it - is that it has exactly the problem that I (and anyone who can count) said it would have, in that you can't make a very convincing 4K picture out of a 4K bayer array.
You won't get any support on this from anyone who has seen the footage.

But I want to see TIFFs or something more concrete.
Maybe you would have a different point of view.

The one that particularly alarmed me is here:

http://www.red.com/s...till/009000.jpg

Look at the crosspoints on the mesh on his helmet.
What exactly?

Phil


Again... Alpha prototypes that produced an image that all said looked fantastic at NAB on a 4K projector. But we are already much further along since then.

Jim

Edited by Jim Jannard, 20 May 2007 - 10:34 PM.

  • 0

#4 Alessandro Machi

Alessandro Machi
  • Sustaining Members
  • 3318 posts
  • Other
  • California

Posted 20 May 2007 - 10:57 PM

The hot spot on the end of the rifle and one other one I saw, those would have had not completly blown out if it had been shot with film, no?
  • 0

#5 Jim Jannard

Jim Jannard
  • Basic Members
  • PipPipPipPip
  • 361 posts
  • Cinematographer

Posted 20 May 2007 - 11:24 PM

They are not blown out on the 4k... but if your point is that film has more dynamic range, that is correct.

Jim

Edited by Jim Jannard, 20 May 2007 - 11:26 PM.

  • 0

#6 David Mullen ASC

David Mullen ASC
  • Sustaining Members
  • 19759 posts
  • Cinematographer
  • Los Angeles

Posted 20 May 2007 - 11:48 PM

To be precise, color negative film has more dynamic range (not reversal, for example.)

But what I saw in the Peter Jackson film was certain workable, contrast-wise, considering the short was made under pretty harsh daylight conditions in NZ (sort of a worst-case scenario in some ways), and the Red team says they gotten a few more stops of dynamic range out of the camera since the demo was shot.

The ironic thing is that a certain percentage of people who have tried to color-correct the short film themselves invariably add more contrast and throw away some dynamic range to make the images conform to their tastes regarding "war photography" -- so while more dynamic range is always a good thing because it allows highlights to blow-out more gracefully, and shadows to roll-off into black more naturally, half the time we then take that footage and add more contrast for dramatic reasons. Better to have the info and toss some of it away than to not have it when we need it...

Also, some artifacts are more apparent in still frames than in motion (of course, others are more apparent in motion than in a still frame.)

My impression is that Graeme, Jim, and everyone else at Red have been nitpicking these images more than anyone else and know the artifacts quite intimately, hence the announcement of further improvements and tweaks on the way, both at the camera end and in the processing end. This is all very hopeful. I think it's going to be a great camera when it's finally released.

Due to color negative's non-linear response to light, you'd almost need to design a digital camera that exceeded the dynamic range of film in order to then apply gamma curves to compress the highlights and shadows and end up with a dynamic range that is the same as film. So far, I haven't seen any digital camera really match film's dynamic range, but it's also clear that the technology keeps improving so never say never.
  • 0

#7 Max Jacoby

Max Jacoby
  • Sustaining Members
  • 2955 posts
  • Other

Posted 21 May 2007 - 01:55 AM

Due to color negative's non-linear response to light, you'd almost need to design a digital camera that exceeded the dynamic range of film in order to then apply gamma curves to compress the highlights and shadows and end up with a dynamic range that is the same as film. So far, I haven't seen any digital camera really match film's dynamic range, but it's also clear that the technology keeps improving so never say never.

That has been my thinking as well. The problem with current digital cameras is not the latitude, but that as soon as it clips, it looks very ugly. Harris Savides says as much about the Viper also. The only way to remedy that would be a chip that never clips, similar to HDR images, so that one could blow out the highlight s gracefully in colorcorrection if one so choses.
  • 0

#8 Werner Klipsch

Werner Klipsch
  • Basic Members
  • PipPip
  • 86 posts
  • Other

Posted 21 May 2007 - 08:08 AM

So far, I haven't seen any digital camera really match film's dynamic range, but it's also clear that the technology keeps improving so never say never.

With due respect Mr Mullen, unless some person finds a way to make image sensors out of some brand new alternative to silicon, I am quite confident to say "never". Silicon Carbide might give 4-5 stops more, but we are nowhere even vaguely near to that achievement.

It may, (well, almost certainly will) get higher resolution with more pixels, but you will never get much more dynamic range than there is now. The dynamic range of even standard definition CCD has not improved all that drastically in even more than 15 years.

Every stop improvement needs a 6dB downwards shift in the noise floor. But the noise floor is already at rock bottom. Even the ultra-wealthy owner of Oakley simply cannot order silicon atoms to lie still :lol:

Cameras today seem to have lower noise due to clever signal processing, not so much that they generate inherently better signals.

I am quite happy to allow that the RED could be a great benfit to videographers, but cinematographers, no.

So please don't hold your breath. You will turn RED, blue, black, and various shades of decomposition before a video camera is the equal of movie film!
  • 0

#9 Eden Lagaly-Faynot

Eden Lagaly-Faynot
  • Basic Members
  • PipPip
  • 21 posts
  • Student
  • France

Posted 21 May 2007 - 09:27 AM

So far, I haven't seen any digital camera really match film's dynamic range, but it's also clear that the technology keeps improving so never say never.


Hello,

May I ask you your opinion regarding the Dalsa? I was given a demo at a meeting and the tech guy show how much details he could get from a blown-out sky, and that's really impress me, seemed just like an HDR done with my DSLR. The output is 16 bits though.

Thanks,
Eden.
  • 0

#10 David Mullen ASC

David Mullen ASC
  • Sustaining Members
  • 19759 posts
  • Cinematographer
  • Los Angeles

Posted 21 May 2007 - 10:46 AM

May I ask you your opinion regarding the Dalsa?


Just from the general shots that people like Dave Stump shot, I'd probably say that the Dalsa is the best I've seen so far for dynamic range in a digital cine camera, but that's not based on real testing.

All the high-end digital cine cameras are not too far apart though in effective dynamic range though.
  • 0

#11 Jim Jannard

Jim Jannard
  • Basic Members
  • PipPipPipPip
  • 361 posts
  • Cinematographer

Posted 21 May 2007 - 11:34 AM

I am quite happy to allow that the RED could be a great benfit to videographers, but cinematographers, no.


I guess you will be quite surprised at what films will be shot with RED this year...

Jim
  • 0

#12 Max Jacoby

Max Jacoby
  • Sustaining Members
  • 2955 posts
  • Other

Posted 21 May 2007 - 12:34 PM

I fully expect even big-budget Hollywood films to be shot on the Red, after all they have shot on the F900, Viper and Genesis as well and are looking at the Dalsa too.

But I'm very curious if those films actually go uncompressed or shoot Redcode.
  • 0

#13 Sam Wells

Sam Wells
  • Basic Members
  • PipPipPipPip
  • 1751 posts
  • Cinematographer

Posted 21 May 2007 - 01:11 PM

The 4K demo (projected with the Sony 4K) Dalsa footage I saw was quite good re response in the highlights - candles on a dinner table did not bloom etc and detail seemed to hold in the flames.... No hot sky through window type torture tests in that footage tho....

-Sam
  • 0

#14 Stephen Williams

Stephen Williams
  • Sustaining Members
  • 4708 posts
  • Cinematographer
  • Europe

Posted 21 May 2007 - 01:24 PM

But I want to see TIFFs or something more concrete.

The one that particularly alarmed me is here:

http://www.red.com/s...till/009000.jpg

Look at the crosspoints on the mesh on his helmet.

Phil


Hi Phil,

There is also some 'edge' around the gun & arm. What would that be if there is no detail used in the Red camera?

Stephen
  • 0

#15 David Mullen ASC

David Mullen ASC
  • Sustaining Members
  • 19759 posts
  • Cinematographer
  • Los Angeles

Posted 21 May 2007 - 01:38 PM

Could it be a JPEG artifact? I don't recall anything like that in the 4K demo of the footage, but again, things are moving. The Peter Jackson short film did not look "edgy" the way that many HD productions do.
  • 0

#16 Brad Grimmett

Brad Grimmett
  • Sustaining Members
  • 2660 posts
  • Steadicam Operator
  • Los Angeles

Posted 21 May 2007 - 05:25 PM

I guess you will be quite surprised at what films will be shot with RED this year...

Jim

Is this a certainty, or just a possibility?
  • 0

#17 Jim Jannard

Jim Jannard
  • Basic Members
  • PipPipPipPip
  • 361 posts
  • Cinematographer

Posted 21 May 2007 - 05:27 PM

I fully expect even big-budget Hollywood films to be shot on the Red, after all they have shot on the F900, Viper and Genesis as well and are looking at the Dalsa too.

But I'm very curious if those films actually go uncompressed or shoot Redcode.


Every single one (so far) is planning on shooting REDCODE RAW. I think "Crossing the Line" changed a lot of minds on what compressed raw can look like. There is no reason to carry huge drives around unless there is a meaningful visual advantage.

Before we will accept an order for a RAW Port, we will make sure that the customer has seen a side by side comparison and just isn't assuming that uncompressed is better. REDCODE RAW looks as good, shoots on portable media, is much easier to clone for backup, and has several very convenient workflow options.

Jim
  • 0

#18 Phil Rhodes

Phil Rhodes
  • Sustaining Members
  • 11937 posts
  • Other

Posted 21 May 2007 - 05:54 PM

It's not what it looks like straight off the camera, though, is it?

It's what it looks like after a whole post chain.

You seem to have convinced CML that compression is OK, though. I'm impressed.

Phil
  • 0

#19 Alessandro Machi

Alessandro Machi
  • Sustaining Members
  • 3318 posts
  • Other
  • California

Posted 21 May 2007 - 10:55 PM

Due to color negative's non-linear response to light, you'd almost need to design a digital camera that exceeded the dynamic range of film in order to then apply gamma curves to compress the highlights and shadows and end up with a dynamic range that is the same as film. So far, I haven't seen any digital camera really match film's dynamic range, but it's also clear that the technology keeps improving so never say never.


Great summation.
  • 0

#20 Carl Brighton

Carl Brighton
  • Basic Members
  • PipPipPip
  • 222 posts
  • 1st Assistant Camera

Posted 22 May 2007 - 05:34 AM

I guess you will be quite surprised at what films will be shot with RED this year...

Jim



Look mister, I for one have been absolutely AMAZED "at what films" have already been shot with absolute rubbish "cinematography" cameras, like, Star Wars for example? I still can't understand how Warner Bros allowed a flagship franchise like Superman to be shot on crap equipment; the picture quality of the 30 year old original was miles better.

Not that I'm saying your cameras are crap; I'm just making the point that I don't think that's a terribly good argument to use.

Why don't you make a new Oakley ad with the RED and arrange to get it shown in cinemas? Not only will you able to write it off as a business expense, but everybody will be able to see some RED footage without having to travel to Las Vegas or wherever.
  • 0


CineTape

FJS International, LLC

Ritter Battery

Glidecam

Abel Cine

CineLab

Wooden Camera

Tai Audio

The Slider

Paralinx LLC

Broadcast Solutions Inc

Rig Wheels Passport

Technodolly

Willys Widgets

Gamma Ray Digital Inc

rebotnix Technologies

Media Blackout - Custom Cables and AKS

Opal

Metropolis Post

Aerial Filmworks

Visual Products

Rig Wheels Passport

Wooden Camera

CineLab

Technodolly

Gamma Ray Digital Inc

CineTape

Aerial Filmworks

Willys Widgets

Broadcast Solutions Inc

FJS International, LLC

Metropolis Post

Ritter Battery

Abel Cine

Media Blackout - Custom Cables and AKS

Tai Audio

The Slider

Visual Products

Opal

Glidecam

Paralinx LLC

rebotnix Technologies