A "real" 4k from a 5k - 6k camera

[Joe G.]

Much has been said about how the 4k from a Red One is "effectively" a 3.2K or some even say 2.5K "effective" resolution. Is there a white paper that puts this into a coherent explanation?

So, at what point do these new cameras become an "effective" or a "real" 4K image?

What determines the actual resolution?

How can these 5 or 6K images be turned into real 4K quality images suitable for the big screen and a high end 4k post production workflow?

When will we learn more about this workflow, which at the moment seems non-existent?

[Andrew Walker]

Much has been said about how the 4k from a Red One is "effectively" a 3.2K or some even say 2.5K "effective" resolution. Is there a white paper that puts this into a coherent explanation?

So, at what point do these new cameras become an "effective" or a "real" 4K image?

What determines the actual resolution?

How can these 5 or 6K images be turned into real 4K quality images suitable for the big screen and a high end 4k post production workflow?

When will we learn more about this workflow, which at the moment seems non-existent?

The Red One is already suitable for the big screen. At least what I've seen on the big screen at Warner Bros. Both on film and digital.

[Jannard]

The highest measured "real resolution" of 35mm film is 3.2K. And that is slow film. ISO 500 film is under 3K. Just for your reference...

Jim

[Ryan Patch]

Effective resolution is measured by the resolution that can be "resolved" from details on screen, like a camera test chart with small details.

Your post seems to be putting 4k at the gold standard of what you think is suitable for the big screen. Remember that Star Wars was shot at 1080, and Lord of the Rings was scanned, DI'd and printed back at 2k.

When you say that you want a workflow for a "real" 4k image, you're saying that you want a workflow for 5k or 6k, right? The answer is... probably when a) cameras are released to the public that image at 5, 6, etc., and b) when you make it happen. If you want workflow that is set in stone and everyone knows how it's supported through and through, go shoot some film. If you're going to pick up a device that is brand new, is rewriting several rules, less than a year after the first models are released, you're going to have to accept some of the growing pains.

And, practically, the workflow looks like...

Edit with proxy files, conform and color with a 5, 6, 9k master, then render out 4k images.

[Ryan Patch]

Also, this should tell you some more about resolution, etc. Keep in mind it's written by some people with an agenda, though, just as we, here, have an agenda.

http://www.arri.com/entry/4kplus-systems.htm

Ryan

[Graeme Nattress]

Effective resolution is the measured resolution. That means pointing your camera at a test chart, and reading off the result. However, in reality it's more complex than that.

Many "HD" camera systems are so desperate to appear sharp and detailed that they do not adequately filter out the detail that lies above which the camera can record. At RED, we're a bit conservative with our filtering. We want to ensure that you don't get any false details, or aliases appearing in your images. There are a number of reasons why this is important, not least how these false details can increase the perception of motion judder, make it harder to compress for broadcast or distribution via a motion adaptive codec, or just plain look ugly.

Instead, with RED, you get a more natural level of detail, not forced, that looks to me more how the eye sees things.

Graeme

[Tico Llaurador]

Instead, with RED, you get a more natural level of detail, not forced, that looks to me more how the eye sees things.

Graeme

And *THAT* is exactly why I love the look of images produced by the RED One. They have a very "organic" feel to them.

[Graeme Nattress]

Correct answer Tico!

Graeme

[David Mullen ASC]

Trouble is that it is hard to quantify sharpness in relation to aliasing, because how much aliasing you want to accept in order to gain the impression of more sharpness is a judgement call on the part of the camera designer.

Simply telling you the number of lines resolved, or the MTF of the system, etc. doesn't give you the whole picture, so to speak, in regards to sharpness, detail, resolution, and the perception of those things (and perception is a valid thing to consider). Which is why opinions can vary so much, and why you should shoot your own tests and make your own judgments.

The real problem with this whole 4K resolution benchmark, while a theoretical ideal, is that there really isn't a real world example to compare against. You'd almost want to design a test image shot with a 4K monochrome sensor in three passes filtered red, green, blue as some sort of digital 4K RGB standard to measure all movie cameras against. But even then, you'd have to decide how much aliasing to accept in the system.

I'm not sure there really is a simple engineering figure or measurement that can sum up the whole picture regarding resolution.

I'm not sure studying single frames either is a completely accurate picture since detail on film lies on different grains on each frame, and our eye averages information of a couple of frames during projection. Some people use the term "temporal resolution" to describe some of the effects of shooting and projecting at higher frame rates, though I'm not sure if that's a real engineering concept.

I think many of us can agree that what we want is an image containing a lot of fine details without aliasing artifacts and without the impression of artificial sharpening tricks. One way to get that is by oversampling, hence why 70mm and IMAX photography looks so nice -- lot of fine detail, low grain, but not edgy-looking. So larger sensors with more photosites will help in this regards.

[Graeme Nattress]

David, with answers like that you're putting me out of a job!

Graeme

[Michael Lindsay]

Many "HD" camera systems are so desperate to appear sharp and detailed that they do not adequately filter out the detail that lies above which the camera can record. At RED, we're a bit conservative with our filtering. We want to ensure that you don't get any false details, or aliases appearing in your images. There are a number of reasons why this is important, not least how these false details can increase the perception of motion judder, make it harder to compress for broadcast or distribution via a motion adaptive codec, or just plain look ugly.

I think it would serve us all very well if Red commissioned an education orientated micro site full of tests and white papers... The above needs to be screamed loud somewhere you can point people to...

I've pointed loads of people to the Panavision/canon lectures despite the anti-red bias just to get people to understand that they don't have the understanding framework they think they have (the megapixel + latitude reductive thinking challenge)





I'm not sure studying single frames either is a completely accurate picture since detail on film lies on different grains on each frame, and our eye averages information of a couple of frames during projection. Some people use the term "temporal resolution" to describe some of the effects of shooting and projecting at higher frame rates, though I'm not sure if that's a real engineering concept.

I've been curious about this for a while and would love to know more...

The augmentation of resolution through time and tiny spatial differences in detail is analogous to 3d resolution augmentation. 3d rez augmentation feels about 30-50% ... (obviously this only happens with certain active display technologies)

regards

Michael L

[Graeme Nattress]

I wish we had time Michael, but we're way too busy making cutting edge cameras and innovative image processing and codecs. The best we manage is a quick post on this forum to help explain the issues.

Graeme

[Michael Lindsay]

I wish we had time Michael, but we're way too busy making cutting edge cameras and innovative image processing and codecs. The best we manage is a quick post on this forum to help explain the issues.

Graeme

Hey I'm a amazed you manage what you do...I've personally learned quite a few things... maybe two literate technology interns for 1 year (with full Red resources) and you oversee their work?

just an idea..

I think its important because you (personally) make the same points over and over again every few months to a different load of REDUSERS.... And David (near to being made a saint) Mullen would probably benefit too :shifty:

thanks again

Michael Lindsay

[Patrick Tresch]

The highest measured "real resolution" of 35mm film is 3.2K. And that is slow film. ISO 500 film is under 3K. Just for your reference...

Jim

Are you talking about the "native" 35mm negative resolution or the resulting resolution of a 35mm print passing the negative / interpositive / internegative / positive distribution process?

35mm neg /positive direct print process is beautifull.

Patrick

[David Mullen ASC]

I believe he's talking about what can be measured on a 35mm negative.

What gets confusing is that there is a difference from measurable resolution of a negative versus optimal scanning resolution of a negative, which tends to be higher than measurable resolution. There are some benefits, again, with oversampling the negative in a scan, which is turn may be oversampling if you end up in 2K or HD for a final master.

While I haven't shot my own tests, I can believe that 35mm color neg falls in the 3K to 4K range in terms of measurable line resolution, just from my experience doing 2K and 4K D.I.'s.

Of course, color neg doesn't have any aliasing artifacts (plus it has a wider dynamic range.)

Also, 3K or 4K describes horizontal pixel resolution -- some film formats like 4-perf 35mm anamorphic benefit from greater vertical resolution than 35mm formats that involve cropping to achieve 2.40, if that's your intended aspect ratio.

We're obviously getting into that zone where many modern digital cine cameras are in the ballpark of matching 35mm resolution, but that also means that now we're getting into the nitty-gritty details of the particular "feel" of how these cameras handle fine detail versus how film does, which is why I feel that we don't have adequate simple engineeering numbers to describe that difference.

I recall a discussion I had with Dale Launer about audio engineering, he being a real buff on the topic, and he mentioned that when digital sound came out, every engineer pointed out how it was measurably better than analog sound, yet musicians kept saying that certain aspects were not as good to their ears. The engineers would sort of dismiss those comments until later it turned out that they simply weren't measuring the sorts of things that the musicians' ears were hearing. So the problem is that we may not have adequate tools to quantify what we are seeing with our own eyes, at least, not in a way that gives us a simplistic numerical value. Hence we fall back into inaccurate or inadequate terms like "4K" to describe sharpness.

[David Mullen ASC]

35mm neg /positive direct print process is beautifull.

Patrick

It IS beautiful, but oddly enough, emulsion to emulsion contact printing actually is less sharp than step optical printing of the neg to print. But some people prefer the look of contact printing anyway since optical printing (assuming the same number of generations involved) tends to oversharpen the grains.

[Graeme Nattress]

Film should suffer from aliasing, but due to the random sampling used by the film grains, it's not particularly visible. You've essentially traded random noise from the grains for lack of aliasing.

When you scan film, that's a fixed sampling process, and you do need to be very careful about aliasing, especially if you scan at a high resolution then downsample. I've seem some pretty stupendously bad aliasing artifacts from film sourced material that has been downsampled from a high rez scan, and they're not pretty.

David, you're right that the "feel" of an image is often hard to quantify. I think we said earlier on that at RED we prefer the rich taste of a nice dark chocolate in our images as an attempt to describe the image feel we were going after. Words work better than numbers for this.

However, in the new year, I get the feeling that the rich bold delicious melt-in-the-mouth taste of RED is going to get a whole lot tastier, but still keep the same great taste we know, enjoy and love!

Graeme

[Jannard]

The ONLY reason to "go digital" over film is that film is problematic. While we ALL agree that it is BEAUTIFUL beyond measure, it is expensive, not very earth-friendly, takes up space (ever seen what a feature's worth of film looks like?), it cannot be quality-checked without processing, etc. For the "switch to flip" to all digital, we need to equal or beat film in every aspect. The 1st digital programs did not do it. But they got the ball rolling. Now we are close in many ways... and it is now easy to see that digital will surpass film in every way in the future.

The question is no longer if... it is when. The RED ONE went a long way to bridging the gap between "early digital" and a film replacement. We call it an acceptable "film alternative". The film replacements are on the way. Whether it is RED or someone else... it is coming. As DNR continues to improve and workflows follow, it will be hard to make a case to shoot film in the future. I did not say now.

A few facts... there are NO new film cameras in development that we know of. The film camera companies are looking for digital solutions, even though they know it puts pressure on their legacy.

It seems to me that the struggle between film and digital is counter-productive. The push, IMHO, should be to get the digital camera companies to make better digital equipment so there is no longer any compromise as a film alternative.

Jim

[Joseph Ward]

The ONLY reason to "go digital" over film is that film is problematic. While we ALL agree that it is BEAUTIFUL beyond measure, it is expensive, not very earth-friendly, takes up space (ever seen what a feature's worth of film looks like?), it cannot be quality-checked without processing, etc. For the "switch to flip" to all digital, we need to equal or beat film in every aspect. The 1st digital programs did not do it. But they got the ball rolling. Now we are close in many ways... and it is now easy to see that digital will surpass film in every way in the future.

The question is no longer if... it is when. The RED ONE went a long way to bridging the gap between "early digital" and a film replacement. We call it an acceptable "film alternative". The film replacements are on the way. Whether it is RED or someone else... it is coming. As DNR continues to improve and workflows follow, it will be hard to make a case to shoot film in the future. I did not say now.

A few facts... there are NO new film cameras in development that we know of. The film camera companies are looking for digital solutions, even though they know it puts pressure on their legacy.

It seems to me that the struggle between film and digital is counter-productive. The push, IMHO, should be to get the digital camera companies to make better digital equipment so there is no longer any compromise as a film alternative.

Jim

When I saw Che I knew it already happened. Thank you Jim.

[J Davis]

not very earth-friendly,

Jim

Best reason by far,
and the less plastic you use (in construction of the camera) the better.
Glad to hear RED is a little green.

EDIT: Merry Christmas everyone!

[Joe G.]

I take it that the real answer to my actual question is: 'we don't know until the things are farther along.'

It was a straightforward question about the measurable resolution of "5k" or "6K" sensors.

No sophistry was needed.

[Graeme Nattress]

Measurable resolution will be somewhere in the region of 78% of the linear pixel dimensions, given similar OLPF to RED and assuming the lens is not the limiting factor. If any of those conditions change, that % changes.

Graeme

[Tom Lowe]

To answer the OP, in order to achieve a pristine "True 4K" you're going to need 5- or 6K sensors. Better to err on the side of oversampling. That's why I'm so excited about FF35 @ 6K. It's going to result in mind-blowing motion pictures.

[paul gill]

with regard to the film/digital issue i think it's always interesting to look at the audio world. Most of the development of the visual side of things has been reflected already in the audio side in recent years. Most audio pro's agree that analogue audio added a certain warmth to the recording which digital could not replicate. But now it is very very much fully digital apart from analogue ecenentricity. There was for a time a big surge in recreating the analogue characteristics via digital methods, and while they are not 100% equiv, the good ones are pretty damn close, and with their ease of use, without all the problems of tape, good enough for the majority of the industry to abandon them. At some point Film will be the same, and Red are seemingly going to be part of the breaking point!

[Andrew Walker]

Well as I've heard the flip over to digital is coming rather fast. A 12 screen theater down near Huntington Beach is going to change over to all digital by the end of this coming year...or so its been said to me. They already have a couple 2K's up there now.

Jim had some really solid points, as Jim often does, film is bad for the environment. Its crazy how much film is just wasted. Plus once those film prints are done they usually get cut up into little pieces so no one can use them. Plus they are usually so trashed from being out in theaters you wouldn't want to watch them anyways.

Also 28K down-converted to 4K will look amazing. I've seen 65mm down converted to 4K and that was some of the best 4K material I've seen.

[Patrick Tresch]

film is bad for the environment. Its crazy how much film is just wasted.

Strange argument... we shoud also stop dragster races... and drive Hummer... :usd:

Hope we will be able to archive DATA for 50/100 years like we do now with film...

We are at a point where digital archiving will be problematic. A major that wants to erase a DCP or deny acces to the key will make the movie lost for ever. No hard copy that could be restored...

The manual about the DCP key is bigger than the DCP specifications itself.

There are still stuff do be worked out...

Patrick