A Response To Jim Jannard and 1080P

[Tico Llaurador]

Hmmm... interesting discussion... now, if higher and higher spatial resolution will pave the way to the future, why are we still shooting at 24 fps? So we can make out the visual defects that come with 24 fps acquisition even better yet?

Seems to me that increased spatial resolution without an adequate adjustment in temporal resolution may only serve to exacerbate the visual defects inherent in 24 fps acquisition.

So, my question is this... will all this gain in spatial resolution finally make it almost necessary to bump acquisition up to 48 fps to better match it?

I mean, that'd be even more data to process and store.

Sometimes I wonder...

 

[paulherrin]

I recommend 16-bit LIN files - 16-bit DPX or OpenEXR....most high-end apps support EXR as well - and if we are talking mastering - I think the advantages of the flexibility of EXR far outweigh the fact that most apps can not handle EXR in realtime for playback.

I also want to confirm that I'll be able to encode to REDRAY directly from EXR.

I've got some more tests to run - and I'd love to hear other people thoughts on 16-bit DPX vs. EXR - but I guess that should be another thread.

But ... go 16-bit - and go LIN. Just say no to 10-bit LOG DPX.

Mark, the visual effects industry seems to be moving towards EXR as well, which means we should expect better, continuing support. There are a lot of advantages to the OpenEXR format, and they're beginning to show themselves.

 

[M Most]

I recommend 16-bit LIN files - 16-bit DPX or OpenEXR.

I'm currently weighing the pros and cons of each for OFFHOLLYWOOD's 4K Finishing - but - I'm leaning towards OpenEXR as it offers more flexibility as you can store different views in alt. channels - so ... for example - instead of mastering addition textless masters - on the frames where you have text over picture - you could/would have one channel with the "comp" (text over picture), one channel with the text elements, one channel with the alpha of text elements, and one channel with clean bg (textless).

More and more apps are supporting 16-bit DPX - Assimilate's Scratch supports it, RCX supports it - but most high-end apps support EXR as well - and if we are talking mastering - I think the advantages of the flexibility of EXR far outweigh the fact that most apps can not handle EXR in realtime for playback.

As IIF/ACES progresses (and it IS progressing, albeit a bit slower than I and a lot of other people would like), the "playable" variant of Open EXR will see a lot more use and become a lot better known and supported. You'll likely lose some of the multiple layering capabilities, as well as some of the currently acceptable compression codecs, but it is likely that any software capable of reading the half float, real time playable version will also be capable of decoding the "full" version. So you could potentially post with the ACES blessed version, allowing real time operations during those steps, and ultimately archive using the "full" version you're talking about.

BTW, putting overlay titles and possibly subtitles on an alternate layer in the EXR container is a good idea. I'm not sure in the end it's really better than a sidecar metadata file like we use now for DCP deliveries, and it would require the use of, as you say, somewhat bloated image files that likely won't be real time capable. But it would self contain these things for archival purposes, and since Open EXR is, well, Open, such things would be properly documented for later recovery and not in a proprietary format. So I guess I sort of think that maybe I might somewhat partially agree that it might perhaps possibly be a kind of semi-good answer for some things. How's THAT for a ringing, positive endorsement? :thumbup1:


Enjoy that one day delayed white Xmas we're hearing about....

 

[David Mullen ASC]

OK. One word: inevitability.

There are limits to how far ahead of the curve you want or need to be... widescreen color was "inevitable" from the long view of history, but does that mean that "Casablanca" should not have been made in 4x3 b&w, particularly since widescreen was not really available for another decade and the studio may have not made the movie at all if they had to pay for 3-strip Technicolor in addition, especially during WW2.

We make movies for the here and now, there are limits to how much we can future-proof them for all possible contingencies.

I read plenty of anti-color arguments in magazines from the 1930's and it's easy to dismiss them as short-sighted, but on the other hand, b&w did not disappear as a common studio option until the mid 1960's, 30 years later, which is an awful lot of b&w movies made long after people were predicting the color would take over. So were all those movies (the majority) shot in b&w in the 1930's and 40's "wrong" to be made that way because 30 years later, most movies would be shot in color? It starts to be like saying that paintings should have been done in oils decades before oil paints were available, or that Ansel Adams should have had the foresight to shoot his large-format landscape photos with a digital color camera.

Yes, I realize that 4K digital origination exists here and now, but it's relatively recent, and as for finishing all post work in 4K, including efx, that's a bit much to expect from TV shows made in the past decade, let alone all features.

There are simply practical and financial limits for every production on this planet to future-proof themselves for a 4K market that does not currently exist, as nice and far-sighted as that would be. I would cut some of these people in the past some slack.

Besides, it's the 1080P pioneers like Soderburgh, Rodriquez, Lucas, Fincher, Singer who have helped ease the way for the industry to move towards this 4K digital future. People here praise these directors for choosing Red while simultaneously implying that they were idiots for using 1080P technology just a few years ago. You can't have it both ways, praise their innovative spirit while saying that they were wrong to use anything less than 4K in the past.

 

[Mark L. Pederson]

BTW, putting overlay titles and possibly subtitles on an alternate layer in the EXR container is a good idea. I'm not sure in the end it's really better than a sidecar metadata file like we use now for DCP deliveries, and it would require the use of, as you say, somewhat bloated image files that likely won't be real time capable. But it would self contain these things for archival purposes, and since Open EXR is, well, Open, such things would be properly documented for later recovery and not in a proprietary format. So I guess I sort of think that maybe I might somewhat partially agree that it might perhaps possibly be a kind of semi-good answer for some things. How's THAT for a ringing, positive endorsement? :thumbup1:

Yeah - I've pondered the whole larger size EXR files vs. side-car files + XML thing a lot - and I just feel like if we are talking master archival - realtime playback is LESS important that clean, streamlined file sequences.

We do a lot of Supplemental DCP authoring - (changes, changes, changes) and although it works amazingly well - it just makes me nervous when the asset map is just full of side-cars - it can totally work - but I think there are a lot of advantages to having a "archival FRAME" that can contain alt versions of that specific frame - right inside the frames container.

I love the fact that you can open an EXR file in Photoshop and see the channels.

 

[Billy Barber]

Originally posted by Tico Llaurador

Seems to me that increased spatial resolution without an adequate adjustment in temporal resolution may only serve to exacerbate the visual defects inherent in 24 fps acquisition.

The visual 'defects'....well chosen word although many people might take issue with it. I remember when I was a music producer back when disco was still around (though I never did much of it myself-swear!) musicians used to call 120 beats per minute "God's tempo". There was just something about it that made people get up and boogie to use the lingo of the day. It's not a perfect analogy but there is something inherently attractive, seductive even about 24fps. In part because it's the film tempo we've all grown up with but maybe there's something more to it than that, who knows?

 

[M Most]

Besides, it's the 1080P pioneers like Soderburgh, Rodriquez, Lucas, Fincher, Singer who have helped ease the way for the industry to move towards this 4K digital future. People here praise these directors for choosing Red while simultaneously implying that they were idiots for using 1080P technology just a few years ago. You can't have it both ways, praise their innovative spirit while saying that they were wrong to use anything less than 4K in the past.

My guess is that very few of the posters here even know that David Fincher made 2 major studio pictures on Viper cameras years before "The Social Network". Or that Bryan Singer made the first major studio picture to use a camera with a 35mm sensor (the Genesis). Or that Steven Soderbergh not only made at least one picture with an F900, but a few years before that made one with DV cameras. Or that Robert Rodriguez has probably produced more box office revenue for digitally captured material than anyone else in the industry ("Avatar" was primarily CG, not digitally captured live imagery).

It is all too common for people here to dismiss the past without actually being aware of it.

 

[Stuart English]

RED RAY PRO

RED RAY PRO

What I hope, is that RedRay won't be locked to 24/48 fps, like DCP is....

No, it's not locked to 24, we allow the options of 23.98, 24.00 or 25.00 fps for 4K cinematic playback...

 

[M Most]

Yeah - I've pondered the whole larger size EXR files vs. side-car files + XML thing a lot - and I just feel like if we are talking master archival - realtime playback is LESS important that clean, streamlined file sequences.

We do a lot of Supplemental DCP authoring - (changes, changes, changes) and although it works amazingly well - it just makes me nervous when the asset map is just full of side-cars - it can totally work - but I think there are a lot of advantages to having a "archival FRAME" that can contain alt versions of that specific frame - right inside the frames container.

One thing about studio archives that has become pretty clear is that they ultimately want archives on open formats. Saving proprietary formats like ProRes or R3d is nice, and in many cases represents the original data best, but in the archivist's world, proprietary is unacceptable and open is required. That's one of the primary advantages of DPX and EXR, but if you want to go down the multilayer EXR path, I would suggest talking to some of the studio archivists, or those who have been there (Garrett Smith would be a good start....) and look for some industry consensus. Right now that consensus is that DPX is the current accepted format, but many are looking for something a bit less limited. I brought up ACES for a reason, because I really do think that many of these archival issues will be greatly impacted by it over time. In a positive way.

 

[Mark L. Pederson]

I brought up ACES for a reason, because I really do think that many of these archival issues will be greatly impacted by it over time. In a positive way.

I'm a little more open-minded and hopeful now about ACES - but OpenEXR is here now.

 

[Stuart English]

1080p

1080p

Besides, it's the 1080P pioneers like Soderburgh, Rodriquez, Lucas, Fincher, Singer who have helped ease the way for the industry to move towards this 4K digital future. People here praise these directors for choosing Red while simultaneously implying that they were idiots for using 1080P technology just a few years ago. You can't have it both ways, praise their innovative spirit while saying that they were wrong to use anything less than 4K in the past.

and

My guess is that very few of the posters here even know that David Fincher made 2 major studio pictures on Viper cameras years before "The Social Network". Or that Bryan Singer made the first major studio picture to use a camera with a 35mm sensor (the Genesis). Or that Steven Soderbergh not only made at least one picture with an F900, but a few years before that made one with DV cameras. Or that Robert Rodriguez has probably produced more box office revenue for digitally captured material than anyone else in the industry ("Avatar" was primarily CG, not digitally captured live imagery).

It is all too common for people here to dismiss the past without actually being aware of it.

Agreed. At one point in time, not so long ago, 1080p "CineAlta" - either as a Camcorder or Camera plus separate Recorder - was really the only option in electronic cinema imaging, so it's inappropriate to criticize the choice of 1080p as a record format at that time. But Jim's original post was about now...

 

[M Most]

I'm a little more open-minded and hopeful now about ACES - but OpenEXR is here now.

Yes, it is, which is why I'm certainly not being dismissive of it. As a format for in-house archiving, or for indie clients, it's a very good idea on a lot of levels. At the studio level, well, they want what they want - not that they can always figure out exactly what that is. But ultimately, they usually adopt things that are outgrowths of industry wide efforts, in fact, they often fund those efforts, as they did with the DCI, the DVD consortium, and the Motion Picture Academy's SciTech Committee. Sometimes, if something attains almost universal popularity, it gets adopted even if it's basically proprietary (ProTools session files, for instance), but in most of those cases, an open format version of that same material is also archived. I think you're pretty safe with the current incarnation of Open EXR, but before "selling" it to clients that have studio deliverables, you might want to check into where some of this might be going.

Of course, you could also spearhead an effort to move that forward. Maybe we should talk about this stuff...

 

[Gunleik Groven]

My guess is that very few of the posters here even know that David Fincher made 2 major studio pictures on Viper cameras years before "The Social Network". Or that Bryan Singer made the first major studio picture to use a camera with a 35mm sensor (the Genesis). Or that Steven Soderbergh not only made at least one picture with an F900, but a few years before that made one with DV cameras. Or that Robert Rodriguez has probably produced more box office revenue for digitally captured material than anyone else in the industry ("Avatar" was primarily CG, not digitally captured live imagery).

It is all too common for people here to dismiss the past without actually being aware of it.

Just for the record:

1. I see your point very clearly

2. I was aware of all this..

As to Open EXR...

We started a transition in that direction in October, and like many other transitions it seems like it will gain traction to more uses than we initially imagined.

I like the idea Mark presented of several versions of the same frame in the same archival master a lot.

Have to admit NOT having tested this, but as I read it, one could possibly store a REC, a P3 and both with and without subtitles frame in the same file.

That'd be very usefull.

Now... audio... - LOL

 

[Mark L. Pederson]

Have to admit NOT having tested this, but as I read it, one could possibly store a REC, a P3 and both with and without subtitles frame in the same file.

That'd be very usefull.

Now... audio... - LOL

Not only that ... but you COULD put the actual R3D Redcode frame inside a channel - just using the channel for a container.

 

[Gunleik Groven]

Not only that ... but you COULD put the actual R3D Redcode frame inside a channel - just using the channel for a container.

Now, that's nice...

Then I'll just have to work out how.

Could you store corresponding audio with the frame, too?

(Asking too much again, I know...)

 

[Mark L. Pederson]

I think you're pretty safe with the current incarnation of Open EXR, but before "selling" it to clients that have studio deliverables, you might want to check into where some of this might be going.

Of course, you could also spearhead an effort to move that forward. Maybe we should talk about this stuff...

Well ... I've seen studio deliverables with VERY loosely spec'd DCDM (16-bt TIFF) as archive master - I've made those - and continue to make those as deliverables - but I'd rather have a BETTER 4K 16-bit master than one with no meta-data per frame.

No reason I can't sell clients a master archive that is BETTER for creating additional deliverables in the future. That's basically my plan.

Right now I see DCDM on deliverable docs - and you can make a DCDM from 4K OpenEXR in the Clipster - no problem.

For example - I'll convince my clients to master to 4K, 16-bit OpenEXR. I make a 2K DCDM or 4K DCDM per deliverable doc from distributor.

 

[M Most]

Just for the record:

1. I see your point very clearly

2. I was aware of all this..

It was a general statement, not aimed at anyone in particular.

Have to admit NOT having tested this, but as I read it, one could possibly store a REC, a P3 and both with and without subtitles frame in the same file.

That'd be very usefull.

Now... audio... - LOL

I assume by "REC" you mean Rec709 for HD deliverables. What you say is not really possible because the formats that you mentioned have different standard frame dimensions (DCDM is 2048x1080, and HD is 1920x1080). An image file can only be one size, regardless of how many layers it has. You could, of course, make the file the dimensions of the largest contained frame and pad the rest, but that would somewhat defeat the purpose because it would require reprocessing. As for audio, you can actually put audio chunks in DPX files now. There's even a portion of the header reserved for it. SpectSoft uses that capability in their Rave HD product, and other manufacturers also make use of it. OpenEXR could theoretically be used the same way, although I'm not sure that a specific header tag has been defined for it. And for archival purposes, a format with limited track capabilities would not be particularly useful anyway. And for sound archives on a motion picture, you need LOTS of tracks (7.1 stems, foreign dialogue, etc., etc.)

Nice thought, though.

 

[Gunleik Groven]

Thanks Mike.

Just thinkin' out loud.
for now the projects with 2k (P3) and 1080 (rec 709) masters have not been the same, so it is just a bit of pondering...

 

[Uli Plank]

Regarding 24 fps or more, i.e. bringing temporal resolution into the discussion as well: AFAIK there has not yet been any research of scientific validity about the difference in perception of fiction in 24 fps vs. say 50 fps.
Our brain has a CPU clock frequency too, it's frequencies are connected to certain states of activity, and flashing light can even induce epilepsy in rare cases. What the heck, the brain can even be kind of "overclocked" or "underclocked" with flashes of light (i.e. mind machines, which were quite popular years ago).
So, who says that we didn't run into a frequency range well adapted to a fictional attitude by pure coincidence? After all, the choice of 24 fps was not based on any serious research, just "good enough" (sic!) quality for sound vs. amount of footage needed.
When I first saw a really good projection of 2K in 50 fps it felt to me like looking at a theater play through a well-cleaned window…

But I guess I'm kind of hijacking this thread, sorry.

 

[M Most]

For example - I'll convince my clients to master to 4K, 16-bit OpenEXR. I make a 2K DCDM or 4K DCDM per deliverable doc from distributor.

I do think that's very forward thinking. I've actually advocated "house" mastering formats for quite some time, as long as the "house" format can be used to directly produce any deliverable that's called for. Your approach seems to have that covered (for now... :undecided:..).

Do you know of any method other than command line tools for actually creating multilayer OpenEXR files with the differing types of data - along with metadata - you're describing? Does Clipster actually do that? Not that I have anything against command line (it's actually my first choice in many, if not most, cases), but I'm trying to shortcut questions that others will likely ask...