Resolution, IBC and 4K +

[Johan Lindgren]

Ok thanks Graeme. I have seen the baked in "flare effect" too. Do you think it could be something in the DualGainArchitecture process? The high gain pipe clips and the low gain information comes on top of that as Arri explains it. Its a process starting in the analogue domain finishing off after the A/D. Im sure there must be some "smoothing" required in order for that to work decent.

Johan

 

[Graeme Nattress]

No, the dual gain shows up in other ways, notably increased horizontal flare even on un-clipped portions of the image.

Graeme

 

[Blair S. Paulsen]

I carry a Schneider #1 Digi-Con filter for my Epic which mimics some of the low-con "look" of the Alexa. It's certainly not identical and because it is in front of the lens it interacts with speculars differently. Even so, it's a useful tool in the right circumstances.

Cheers - #19

 

[Martin Stevens]

So Graeme, will we be seeing your RedUser signature go from 5K to 6K when the Dragon is unleashed?

 

[Nick Morrison]

I carry a Schneider #1 Digi-Con filter for my Epic which mimics some of the low-con "look" of the Alexa. It's certainly not identical and because it is in front of the lens it interacts with speculars differently. Even so, it's a useful tool in the right circumstances.

Cheers - #19

Another one for the list Blair, thanks for sharing!

 

[Ketch Rossi]

You do realize a lot of the "Alexa look" is down to the in-built low-contrast filter as part of the OLPF stack. At RED we don't think the choice of filtration should be ours, but that of the creatives on the project and the DP.

Graeme

I could not agree more with this statement Graeme, also one of the reason I much prefer the Look of lenses like the super Clean Master Primes on EPIC, as it brings a RAW Clean Neutral image to your Post to do as you please, NOTHING BAKED IN!!

 

[Roberto Lequeux]

I carry a Schneider #1 Digi-Con filter for my Epic which mimics some of the low-con "look" of the Alexa. It's certainly not identical and because it is in front of the lens it interacts with speculars differently. Even so, it's a useful tool in the right circumstances.

Cheers - #19

Makes you wish all lenses came with filter slots. I've thrown arround lots of wacky ideas, but a built in system to place filters directly in front of the OLPF really seems doable. As long as the design made the placing and removal safe and a pair of companies agreed to put out a set... Why mot?

 

[Graeme Nattress]

It throws off your back focus for starters.

Graeme

 

[Blair S. Paulsen]

At least now we have relatively simple back focus adjustment ;-)

Shimming is way less fun than it sounds like...

Cheers - #19

 

[L. Langer]

Does anyone know if the extension of the H.264 standard using EAVC4 Langer mentioned could co-exist with current H.264 implementations?

I can't quite figure out the answer to that question. The public information thus far - what little of it there is - simply states "h.264 compliant codec" and really only references encoding processes and capabilities so I imagine that there is some distinct differences in the decoding processes required but they have not stated that it changes anything on the receiving end of broadcast infrastructure or file-based decoding via GPU or whatever. There is a whitepaper available if you want to request it from the technology company and I have not read any of it or requested it, so I cannot really say anything more about the codec.

 

[Blair S. Paulsen]

Sometimes I wonder if we should "elect" a codec every four years and on the following Jan. 20th it becomes the default deliverable format ;-)

I am under the impression that with a decent bit rate H.264 can hold its own quality wise. Since its so widely supported perhaps its the best option around...

I am a big fan of JP2K (JPEG 2000) which, AFAIK, is the underlying codec for RedCode and the DCI (Digital Cinema Initiative) spec used for distributing feature films to the digital projector in your local theater. At one point there was talk of a version of RedCode that would carry an RGB image rather than the Bayer pattern data encased in the r3d files using current RedCode. I have blathered on about how much I would love to see such a codec available for use as a post enabling mezzanine codec in the 100-250MB/sec range for 4K. Still think it would be a great tool, BTW, but perhaps a lower bit rate version, 20-50MB/sec @4K, for mastering and transit to distributors would do more good.

IIRC the reason JP2K has not been utilized more is due to demanding encoding requirements that require beefy hardware and/or lengthy renders. Is that still a major hurdle in 2012? I know with RedCode the decoding is non-trivial as well so perhaps there are client side issues as well. If Mike M., Graeme or some other well informed person can explain why JP2K is a poor choice for applications beyond its current implementations then I'll shut up. Is there a more modern wavelet codec out there? Sure would be nice to have a codec that offered "extraction layers" so one could master at a good quality bit rate and from the same file you could extract a lower quality version suitable for distribution rather than having to author multiple versions.

Cheers - #19

 

[Bill Anderson]

Sometimes I wonder if we should "elect" a codec every four years and on the following Jan. 20th it becomes the default deliverable format ;-)

That sounds quite appropriate Blair, since every four years or so our elected throw most of their promise away, and perform only well enough to preserve a semblance of their ideal state. I would hope the codecs perform a little better though. ( :

 

[L. Langer]

The only problems I ever read regarding JPEG2000 were the questionable performance (this was many years ago when it was a new and untried tech as far as implementing it in software and hardware) in respect to JPEG at the time and the issues of patent trolling - even though all known stakeholders agreed to make implementation of JPEG2000 creation free - which have since popped up for JPEG. With current hardware-accelerated tech and software know-how, it would be a viable replacement for JPEG but it just seems like nobody wants to implement it in cameras.

 

[Les Dittert]

J2K takes too much processor for sure. Cineform is much lighter, I'd like a rehash of why that is shunned .

 

[Blair S. Paulsen]

As noted by Mike M. and others its hard for an "outsider" like Cineform to reach the necessary critical mass. It's even tougher to build a sizable user base when the authoring side requires a cash outlay roughly equivalent to buying a low end NLE. I have urged the Madman to buy CineForm and give it away to boost the 4K ecosystem but that ship seems to have already sailed...

Cheers - #19

 

[N.D. Paolo]

4 years to the date of the launch of the Canon 5D Mark II and DSLR video movement propelled by V. Laforet,
I LOVE this quote from Bruce Logan, ASC in Philip Bloom's website interview about the Panasonic GH3 launch and Digital Cameras today:

BLOOM: "honestly…do you miss film?"
LOGAN: "I’m over film. It’s in the past. Yes I hanker for a great black and white print where you can actually see the the crystals of silver halide dancing around on the screen, but modern color film emulsions have grain structures which look just like electronic noise to me. Of course in film the dancing grain actually help to define the image in greater detail, where as in digital the noise is just superimposed on the image, but all digital has to do is double it’s resolution (which it seems to do every five years) and the result is the same or better. I love file based workflows and desk top color-control technologies which put the control that lots of people had on my projects back into my hands. I’m not going back!"


Genesis
http://vimeo.com/49420579
Written and directed by: Bruce Logan
Producers: Elliot Lewis Rosenblatt and Bruce Logan
Director of Photography: Philip Bloom & Bruce Logan ASC


Good thing is Epic / Scarlet has that resolution today- no need to wait 5 years, Bruce!
I guess that relevant opinion didn't much make it into the Revenge/Zacuto shootout a couple of months ago,
but I'm glad he tells it like it is!

Bruce's talent was on display in that short, especially considering the minimalist approach in the BTS.

PS. Could Red entertain Logan's fancy with a B/W shoot using the Epic- Monochrome?

Would be cool to see...

 

[PatrickFaith]

Sometimes I wonder if we should "elect" a codec every four years and on the following Jan. 20th it becomes the default deliverable format ;-)

I am under the impression that with a decent bit rate H.264 can hold its own quality wise. Since its so widely supported perhaps its the best option around...

I am a big fan of JP2K (JPEG 2000) which, AFAIK, is the underlying codec for RedCode and the DCI (Digital Cinema Initiative) spec used for distributing feature films to the digital projector in your local theater. At one point there was talk of a version of RedCode that would carry an RGB image rather than the Bayer pattern data encased in the r3d files using current RedCode. I have blathered on about how much I would love to see such a codec available for use as a post enabling mezzanine codec in the 100-250MB/sec range for 4K. Still think it would be a great tool, BTW, but perhaps a lower bit rate version, 20-50MB/sec @4K, for mastering and transit to distributors would do more good.

IIRC the reason JP2K has not been utilized more is due to demanding encoding requirements that require beefy hardware and/or lengthy renders. Is that still a major hurdle in 2012? I know with RedCode the decoding is non-trivial as well so perhaps there are client side issues as well. If Mike M., Graeme or some other well informed person can explain why JP2K is a poor choice for applications beyond its current implementations then I'll shut up. Is there a more modern wavelet codec out there? Sure would be nice to have a codec that offered "extraction layers" so one could master at a good quality bit rate and from the same file you could extract a lower quality version suitable for distribution rather than having to author multiple versions.

Cheers - #19

I'd love to see a jp2k option in openexr, the PIZ wavelet option just doesn't drop the file size down enough for me. Especially in openexr 2.0, is super nice to have 5 to 9 layers per frame .... But this gets massive at 4k frames.

 

[Blair S. Paulsen]

I love OpenEXR, but in order to accelerate the adoption of 4K mastering as SOP I think we need something more manageable. Especially with the speed of single SSD solutions these days I think 200MB/sec @4K is a good balance of size to image integrity with an affordable point of entry. DPX 10 bit log @ 4K is over 1GB/sec and uncompressed OpenEXR at 4K 16bit, half float is at least 1.6GB/sec which really limits the field.

Cheers - #19

 

[Gavin Greenwalt]

I love OpenEXR, but in order to accelerate the adoption of 4K mastering as SOP I think we need something more manageable. Especially with the speed of single SSD solutions these days I think 200MB/sec @4K is a good balance of size to image integrity with an affordable point of entry. DPX 10 bit log @ 4K is over 1GB/sec and uncompressed OpenEXR at 4K 16bit, half float is at least 1.6GB/sec which really limits the field.

Cheers - #19

EXR has ZIPS which would drop 4k 16bit down to 6-800MBs. That's 2x SSDs. But yeah still quite a bit.

 

[Wayne Morellini]

Cineform is trying to make it a vc-5 etc standards under go pro. I would love to see a goprohd camera with it.