sony

[Adam S]

So where can I shoot that 4k rez that outperforms RED currently?

Let's just wait and see.

The comparative studies will come as soon as they are possible to do, don't you think.

Nobody outperform Red is resolution currently, which is what I was saying. But if Sony's new camera system can then Red will have to say they are the best at something else.

I am also saying we need to wait and see our own studies of these proposed design features BEFORE we say they are pointless and won't have any benefits and might actually have drawbacks.

I don't think it is productive for one of Red's top dudes to come on here and start bashing a sensor design of a camera that isn't even out yet, or even officially announced! Graeme is trying to bury Sony before they even come to market and create negative feelings towards their competing product. That is not a practice I want to see any equipment company in any industry do.

It's that kind of stuff that has caused backlash against Red. You can't engage others with that kind of attitude and make those kinds of claims and expect people to still react kindly to your product.

A more professional response by Red would be to either not respond or have a let's wait and see if there methods are better than ours. Either way it would project that they have confidence in their products, instead of being defensive. As if they know they have an inferior product and they want to hide that fact. Sony is doing that with the F35/SRW9000 vs R1MX/Alexa right now and it is very unattractive.

 

[Andrew M.]

My favorite line is "A Bayer pattern sensor can only resolve 30-38% of it's pixels" I think that's what he said.

I was listening carefully.
He stopped half the way there and corrected himself.
He said that Bayer pixel versus RGB pixel contains 33% of the pixels image information.
RGB pixel consist of 3 pixels and one Bayer pixel is only one.
(Bayer group could be arranged in 3 pixel pattern logically in software)

Basically he said that 1 out of 3 is 33% of 3

But at the end what counts is how many pixels do you have in any pattern and what you can do with it. What is the resolving power.
Just shoot the chart and you will get results very fast.

 

[Graeme Nattress]

I don't think Sony Engineers would do a "pointless 45 degree rotation." It must have some benefits or why would they go through "the hassle of converting all algorithms to work on rotated data." There has to be something to that design that is worthwhile to pursue. Engineers, regardless of employer, don't design things for not reason. They do it to solve certain problems. Sure there are trade offs for a lot of the design aspects of cameras and it's finding a balance between them all and figuring out ways to try to overcome some of those trade offs. A 45 degree rotation sounds like it gives you more color information.

Stop poo pooing everything others are doing so quickly, until you actually see the results. This is the mantra most of us try to live by and have to repeat again and again on these forums when it comes to new cameras. So I guess it's my turn to say it, unfortunately it's to one of the lead engineers at a camera company.

If Sony outperforms Red in resolution then Red has to come up with a way to outperform Sony. I think with Epic the form factor is a great design aspect that trumps everyone else when it comes to 3D shoots. So use that as a marketing point. So what if you can't use resolution as your big selling point anymore. I'm not concerned with that as much as I am with form factor/functionality, reliability, DR, and sensitivity anyways and I know a lot of others are the same way.

Well, the whole funny thing is that they don't convert algorithms to work on rotated data. They don't interpolate on the greens and hence get a much lower resolution out than if they did. Basically they've traded a simpler algorithm (just chroma interpolation) for needing twice as many pixels.

The reasoning behind this is probably because they spent a lot of time and effort trashing Bayer pattern colour filter array sensors, then realizing how much sense they make for imaging for high resolution motion picture cameras they had to think what to do.

By using a proper Bayer CFA and good demosaicing we'll get a higher measured resolution with lower aliasing than the same scheme using Q67. Based on measured numbers from my experience with RED One, and Epic, I conservatively estimated it would come out as Q71, a good 4 points superior in their numbering scheme, which incidentally, is nothing more than a pixel count.

Graeme

 

[Andrew M.]

they spent a lot of time and effort trashing Bayer pattern colour filter array sensors, then realizing how much sense they make for imaging for high resolution motion picture cameras they had to think what to do.
Graeme

They are trying very hard to differentiate themselves so they do not look like “me to” in 4K game.

Problem is that the best place was already picked by RED and now the 45deg CMOS with 33% of green pixels will be very difficult to balance. And this 45deg thing, I wouldn’t like to program for that.
Looks like accuracy of color reproduction is in trouble here.

BTW did they mention if their 4K sensor will be CMOS or CCD?

 

[Graeme Nattress]

I love their slight-of-hand where they double the number of pixels in their so called "Q67" compared to their traditional Bayer pattern. All the benefits they claim are only down to that doubling! They're only saying a 17mp sensor produces more detail than a 8mp one! Talk about stating the obvious.

What they don't tell you though, is if you'd compare a Q67 and Bayer with the very same number of pixels in both, you'd get a higher measured resolution and lower aliasing from the Bayer implementation.

Graeme

 

[Robert Frank]

When you speek with some company reps from other companies, it basically sounds like red never happened...

Trust me on this one, other companies will take credit for 4K acquisition and delivery and go so far as to rebrand it as their own idea.

Same goes for many other products and ideas that RED, IMO, is entirely responsible for making commercially viable.

 

[Taylor Morrison]

If Graeme and Hugo were car salesman...I would buy the Red one....Something off with Hugo and his powerpoints.

 

[Rob Ruffo]

When you speek with some company reps from other companies, it basically sounds like red never happened...

It's not like it's "bad" or "uninteresting", it just never happened.

RED only becomes insufficient, bad or uninteresting, if you really insist that you have tried one, and that it worked...

I once worked on an ad campaign for McDonald's. Marketing executives at that company live in a weird world where Burger King doesn't really exist, and if it does exist, it's not a competitor. I kid you not.

 

[Dan Kanes]

Ha!

It's like bringing a Hot Dog to a Taco Fight!

or Some other Mixed Metaphor...

 

[Bing Bailey]

this is their response to red , god could it be anymore limp. they are still obsessed with selling people crazy expensive decks they don't need

 

[Bing Bailey]

the world has moved on. EPIC's size and quality is light years ahead of this pastware

 

[Stephen Pruitt]

All I can say is Graeme Nattress is one incredibly smart dude, and that I'm awfully darn glad that he's on our team. The RED team.

Stephen

 

[Craig W. Bickerstaff]

I thought there was some interesting stuff in there but it had more to do with their roadmap for a 4k workflow than anything.
It will be interesting to see how Sony ultimately tackle a 4k camera but this video seems to show more of a hypothetical approach than an actual "This is what were doing, we are the best and everyone else sucks" so who knows maybe they come out with a completely different method.
On a side note, Road maps, this is a video of Sony talking about their road map for 4k, when I said Sony do roadmaps no one believed me but here it is Sony and their roadmaps.

 

[Bruce Allen]

Well, the whole funny thing is that they don't convert algorithms to work on rotated data. They don't interpolate on the greens and hence get a much lower resolution out than if they did. Basically they've traded a simpler algorithm (just chroma interpolation) for needing twice as many pixels.

Yup. But if you wanted to make cameras that can output a realtime 4K signal, doesn't that sound like a decent tradeoff?

Especially if you had were Sony and very extensive sensor manufacturing ability - but were a little behind in things like ASICs that do a realtime full debayer of an 18mp image?

The reasoning behind this is probably because they spent a lot of time and effort trashing Bayer pattern colour filter array sensors, then realizing how much sense they make for imaging for high resolution motion picture cameras they had to think what to do.

Naah, I think it's so that they can broadcast 4K sports today from a camera with today's level of simple image processing circuitry.

Even an Epic can't output 4K in realtime, can it... well, can it?

Personally I'd love it if you guys keep pushing the ASICs - I wouldn't be surprised at all if by the time Sony has their 4K Q67 cameras ready, you guys have an Epic Goliath Rocket Edition with 4K output.

Bruce Allen
www.boacinema.com

 

[paulherrin]

...tape

i'm pretty sure i heard someone start laughing when he talked about the blurry line between tape and drives. but that might have just been me.

really excited about sony's projectors going public. not so excited about their promotional techniques, it's pretty low. how bout those computer generated test plates? what's that? what sensor was the bayer example shot on? you won't say? how kind. oh yeah, you used a single image and put it in photoshop? okay, that's what i thought.

grasping for the wind...

 

[Jeff Coatney]

Clearly Sony must take an evolutionary approach to delivery of a 4K system. This may be the only way they can begin to work the problem. My fear is that from this approach, their ability to innovate a solution as ground-breaking as Red's will be limited. While their math is accurate, the paradigm they're locked into precludes dealing with data management in bold, unique ways. I really got a picture of where the rest of the industry is beginning to conceptualize their solutions to 4K. They have long established technologies that have calcified their thinking- because they want to save development money. They are pushing their legacy tech to deliver better performance. Often the path forward is only illuminated one brick at a time. I wonder what solutions they would arrive at if they were able to throw everything out the window and start with a clean sheet of paper?

 

[Joseph Hutson]

There's too much money in Tape decks to NOT go tapeless...

P.S. I guess this is valid for REDuser and can't be discounted as just another "Other Cameras" thread, as they bring RED into their little chat.

 

[Jannard]

Sony has just validated 4K as the future. We like that.

Jim

 

[Blair S. Paulsen]

Sony has just validated 4K as the future. We like that.
Jim

More industry validated standards for a 4K infrastructure is very welcome. With the terrific quality/size ratio of RedCode at acquisition and the promise of under 20mb/s 4K delivery via RedRay there is still one hill left to climb - Post. If Sony, with their large installed base can provide a mezzanine codec for 4K post that satisfies the requirements of both image fidelity from original elements and VFX material it sounds good to me.

I would rather have Graeme's version of a 4K mezzanine codec (preferably designed for an OpenEXR wrapper) for post, but ultimately widespread acceptance is the key.

Included in the Sony preso was a description of a "black box" called the ELLCAMI which seems very interesting for dealing with an array of source material all destined for a 4K DI path. These are the kind of tools that can drive acceptance of 4K as a reasonable trade off between cost (can be read as time just as easily) and asset value enhancement.

All ahead 4K.

Oh, and BTW, the 5K oversampled Bayer image from the Epic scaled down to 4K for post should prove substantially closer to 4K in measured resolution than the current RedOne's 4K nominal sampling rate. It is my belief that smart algorithms and oversampling can easily make up any theoretical mathematical advantage of the Q67 tech vs Bayer CFA. Of course there are a number of other variables in terms of sensor performance, OLPF, etc that could mitigate the theoretical aspects tremendously - but that is a whole 'nother thread.

Cheers - #19

 

[Jannard]

We don't even see theoretical advantages of Q67 over Bayer. See Graeme vs. Q67.

Jim