ACES...

[Eric Cameron]

Thanks Mike Most for saying what needed saying. While there is certainly still road to cover in getting everything perfectly in place for ACES to work as designed, this idea that it's hard and costly needs to die. There is nothing complicated in implementing it (designing is another story). I think that's evidenced by many of the people here talking about their path from camera to screen, many of which look an awful lot like brew-you-own version of ACES.

Half-float EXR is NOT necessary to derive most of the benefit of ACES. Using it is just icing on the cake. It really is a simple concept. Any image acquisition device -> IDT -> RRT/ODT. This is essentially how many, many people already work, just without standardized transforms. It'll get there, probably sooner than later. But many people are scared, I think partially because of all the talk about EXR containers. Fact is, you don't need them. Working in EXR would and will be great, but if the space and processing requirements are too much, don't sweat it. The core concepts of ACES work perfectly fine without it.

 

[Jannard]

Watch for RACES... the real solution.

Jim

 

[Gunleik Groven]

Watch for RACES... the real solution.

Jim

LOL

When the current reality doesn't suit you, create a new one!


(Suggestion for a new RED slogan. And I really like the idea!)

 

[Bruce Allen]

Watch for RACES... the real solution.

Jim

Hahaha! Awesome! RED - still creating greatness... still with a sense of humor about itself. All is good in the world

Seeing you guys cheerful and at ease is really great - a sign that good things are in the pipeline for us filmmakers

Thanks, Jim. I rest easy tonight.

Bruce Allen
www.boacinema.com

 

[Martin Stevens]

If ACES is... Academy Color Encoding Specification

Then might one day RACES be.... Red Accurate Color Encoding Specification

 

[David Mullen ASC]

The complex world of ACES will get worse before it gets better. ACES will cost more. ACES will take longer. ACES will eat up more space and require more support, more assessments and more adjustments. But in the long run, ACES will likely prevail and its benefits will strongly outweigh its limitations.

Gee, that sounds similar to the arguments surrounding 4K a few years ago...

 

[Blair S. Paulsen]

Gee, that sounds similar to the arguments surrounding 4K a few years ago...

Indeed.

After hearing so many "takes" on 4K at NAB last week I can report that the number of vendors bullish on 4K has grown substantially. Just wish I had recordings of many of those same companies (and in some cases the same people!) trashing the very notion of 4K in previous years who, now that they have 4K products, are extolling its benefits. Perhaps they're just preparing for a career in politics ,-)

Cheers - #19

 

[Tom Hamilton]

Occam’s Razor; parsimony as preference

Occam’s Razor; parsimony as preference

ACES appears to be attempting a single, global method to a very complex and diverse process, and that has significant implications, noted by some posters' concerns above. I very much agree with Michael on one point; the necessity of an extended gestation period in which all stakeholders should observe, participate, and contribute. Concerns notwithstanding, I am confident the birth of this baby will be an unusual and interesting one.

 

[Michael Cioni]

Well, if we're pointing out FUD, allow me to quote Cioni:

ACES is constantly getting better because its creators understand that there are problems at this point, and also because they understand color science better than just about anyone here or elsewhere on this planet and therefore have the experience and knowledge necessary to eventually work out those problems. ACES will cost more??? Really? The fact is that ACES is an industry initiative being done by the Academy of Motion Picture Arts and Sciences Scientific and Technical Committee, and whose work is free to all who want to use it, and will remain so. ACES will take longer?? ACES will eat up more space and require more support? Once again, the fact is that ACES can easily be incorporated into existing workflows. It basically replaces an input transform with an IDT, and an output LUT with an output transform, consisting of the RRT/ODT combination. You can use those transforms as part of a "live" working environment. Using an ACES workflow does not require creation of new files and does not require use of the half float EXR container. Those things are there for storage and image exchange.

I have already gone on record in this thread as portraying the ACES system as a work in progress, particularly the RRT. I have said that more work needs to be done to refine it, and that in its current state, there are issues that need to be addressed. What irks me is that I'm quite sure Michael is aware of all that, yet statements are made here that are patently false. I too believe in facts, not FUD. The facts are that it is a work in progress, is currently imperfect, but has a lot of potential for a lot of reasons, and will evolve over time. FUD is that it costs money, somehow adds time to the post process, and takes up more space.

Mike-
Please be respectful enough to avoid dramatizing my words as "patently false." I would never write or publish something I can't back up, so I'm happy to clarify for you and more importantly the readers of this thread, including the Academy, what I mean. In other words, I typically do not respond to your attacks, but in this case, I think the issues you raised would be helpful to other readers. Thanks for always being willing to dig deeper.
best,
m


+ I think the statement that the Academy knows more about color science than anyone on the planet is quite a claim. Let's not deify them just yet;-)

+ ACES is not supported by every color system. People (including myself) that want to use ACES are forced to invest to some degree in upgrades in order to offer it to clients. The result are investments in new on-set and in-post ACES compliant systems.
= COSTS MORE

+ I didn't ask for help in the area of image manipulation, but it sure feels like it's being pushed on me. But because there is demand from some groups for ACES, we answered the call and made the investment in the tools. But that investment is also manifested as research, development, testing, publishing and adjusting the system which takes a measurable tool on our staff.
= COSTS MORE

+ ACES is on-record as saying it is still "in development." Yet there are productions shooting now on shifting ground in which poor results are being re-worked, re-tried, and re-rendered. In other words, there are productions today acting as testing beds for ACES, which will benefit us all, but add significant time and expense to said productions.
= COSTS MORE & TAKES LONGER

+ I pointed out that ACES is under development and will require a lot of testing, evaluation, debugging, and the like. There is a lot of very detail-orientated research and testing that needs to be done with the plethora of cameras out there. This will take time on the administrative levels of every company.
= COSTS MORE & TAKES LONGER

+ Today, my company does 2 or 3 feature camera tests per week. Through these tests, we spend ample time explaining progressive options to clients and addressing their digital concerns. Adding ACES to the mix and the notion of ACES vs. Non-ACES or Log vs. Lin or DPX vs. EXR will require more literature, more presentation time, more testing, more evaluation time and solid answers to lots and lots of questions.
= COSTS MORE & TAKES LONGER

+ Full implementation of ACES requires a lot of very technically precise steps and this will add strain onto GPU and CPU architectures. While every company will do it slightly differently, the load on the system will require more GPUs in order to satisfy the arithmetic required by ACES, which is upstream and downstream of the actual color correction. For some systems, this will affect overall system performance and render times and (in some configurations) will require renders with and without some of the transforms, depending on the desired output/clients requests.
= COSTS MORE & TAKES LONGER & IS LARGER

+ While not the only way to use ACES, many people will transcode source media into a mezzanine format prior to manipulation. -This shouldn't be a surprise, since that's how many people do things with log today. But for those who prefer a more native approach to files, mezzanine files will increase data footprints and in many cases, render times and archive sizes.
= COSTS MORE & TAKES LONGER & IS LARGER


I like freedom of choice. I respect the lead set forth by the client. If a client prefers the look and path of ACES, then that is a perfect fit for them. If a client prefers the simplicity of gamma-corrected 709, or Log P3, then it is a perfect fit for them. If a client prefers to take an R3D in After Effects and use Color Finesse to do the image manipulation, then it is a perfect fit for them. -This is what I mean by being a pluralist. If everything is meant to go through an ACES pipeline, then ACES, by its own design, is subtractive. I prefer a world in which different options and pipelines and flows are uniquely chosen by the creatives - a world that's additive. I appreciate the fresh look on images that ACES offers, which is why Light Iron is enabling an ACES pipeline for clients who prefer it or (in some cases) are mandated to use it. But I believe no workflow is ideal by every standard of measurement, therefore I'm comfortable with creatives who continue to inspire us to sculpt solutions that make them comfortable...whatever that may be.

| m |

 

[Gavin Greenwalt]

Indeed.

After hearing so many "takes" on 4K at NAB last week I can report that the number of vendors bullish on 4K has grown substantially. Just wish I had recordings of many of those same companies (and in some cases the same people!) trashing the very notion of 4K in previous years who, now that they have 4K products, are extolling its benefits. Perhaps they're just preparing for a career in politics ,-)

Cheers - #19

My favorite NAB evolution of perspective was in a quote in the REDUser Event presentation about how those who were fighting against a lossless workflow were fighting progress (in light of the years of insistence that there was no use for mathematically lossless RAW).

 

[Bill Anderson]

"RACES... the real solution" to what: the Red end of the spectrum?

 

[Martin Stevens]

Thanks again Michael. Great insight. You are quite heavy with the light iron.

Heavy in the 60's groovy sense of the word.

I dig it.

 

[Gavin Greenwalt]

If everything is meant to go through an ACES pipeline, then ACES, by its own design, is subtractive. I prefer a world in which different options and pipelines and flows are uniquely chosen by the creatives - a world that's additive. I appreciate the fresh look on images that ACES offers, which is why Light Iron is enabling an ACES pipeline for clients who prefer it or (in some cases) are mandated to use it. But I believe no workflow is ideal by every standard of measurement, therefore I'm comfortable with creatives who continue to inspire us to sculpt solutions that make them comfortable...whatever that may be.

| m |

I disagree with this point. There is a reason that almost 100% of all 3D renders go through linear EXR now. It's because:

- Predictability and standardization means less errors, less mistakes and less miscommunications. No "What Gamma is this file in? What LUT needs to be applied." etc etc etc. You just open an EXR and you save out an EXR. No loss of data, no incorrect loading or saving.
- If you work with a standard then it doesn't matter what the client prefers... once the file gets into the software you can work in whatever color space you want. That's what's so nice about a predictable standard. This reminds me of the old FUD about LOG vs Lin that if you graded log DPXs you would get better results than grading linear EXRs (even though you could just apply a Log->Lin or Lin-> Log operation on ingestion). If your client wants to work in gamma corrected 709 then you can just import it and apply a gamma curve as step #1. But there is no ambiguity or uncertainty that you are viewing it wrong. If you do something to the image like grade in LOG gamma then it's because you intentionally moved the image into that color space.
- It was just better. Yes there is a plurality of options (rendering to AVI!) but most of those are just terrible. People rendering images to BMP for instance is an option but it is by every single metric a god awful and demonstrably bad option. Sometimes bad workflows just need to be killed.
- Everything becomes predictable. There are less surprises or curve balls.
- The single process is a little more complicated than any other process but you only have to understand the one and you always know what process you are working with.
- This ar tform is hard enough without something as simple as a file format being a source of worry.

I think it's also a little unfair to call out what's harder on ACES without also acknowledging what's easier. There were also a lot of things that made working with Linear EXRs 'harder'. But those negatives didn't slow down production or bog down studios since they were overwhelmingly offset by the positives. If you are going to say something is slower and more expensive then you have to take into account the aspects which make it faster and cheaper.

This was a problem when Digital was coming into its own and film people were comparing film vs digital. I saw a lot of comparisons mentioning things like "Transcode time". But they were completely ignoring the fact that film didn't get telecined by magic and also had its own time/cost analogous to transcoding.

Didn't you just give a whole talk on fighting the Status Quo.

 

[Nick Shaw]

What LUT do you use? If you give stuff to a VFX vendor do you give them a preview LUT for your RedLogFilm conversion to 709 / sRGB / P3?

Bruce, if it's any use to you I have REDlogFilm to REDgamma (1, 2 & 3) LUTs available from my website. The ones for sale in the store there are in Color and Resolve formats, but I can produce them in other formats too. I also have an unclamped float linear to REDgamma3 gizmo which I use myself in Nuke as a VLUT.

 

[Francis Kenny]

I can tell you from personal experience to avoid ACES at all cost regarding red footage. I have seen what it's done to my show Justified. Perhaps one day soon a generic development process like ACES will work with all cameras. It's a wonderful concept but currently it does not work with all cameras. It's use is limited. I've seen the difference up close and personal. Avoid ACES using Red. It doesn't work. I've worked very closely with ACES from it's beginning. I know where it works and where it will destroy beautiful images. I hope you take my advice. Thanks, Francis Kenny, ASC

 

[Marc Wielage]

Michael's post is hard to argue with. I just wish there was an easier way to reap the core benefit of a benchmark color space. Behind all the acronyms there is a very real upside to being able to take a wide variety of sources and put them in a reference container with fixed values for post/DI.

I think the ultimate goal (and concern) of the ASC is, they want to be able to stop in the middle of a DaVinci session, yank out all the files, run over to another facility with -- say -- Lustre, and continue the session, and have all the color corrections pop up in exactly the same way, with no room for error. And that's including layers, windows, secondaries, the works. And towards the end of the project, they can still yank all the files out, move over to a Pablo facility, and again have everything look identical to the other two places. Maybe even work in two facilities simultaneously, using different pictures, then merge the entire movie together and everything get put together and look hunky-dory.

Assuming identical monitors and set ups... it's theoretically possible. But getting there is going to be fraught with problems. I see Michael's point at the whole issue with 16-bit EXR files is that they're bloody enormous, and it's gonna slow down the pipeline like molasses in January. I've always agreed with his consensus that there are many ways to skin a cat, and not necessarily one perfect workflow that will work well for everybody -- especially in terms of time, budget, and available manpower.

ProRes 444 is viable. So is Avid DNxHD 220. Both are used every day for HD delivery for network and cable TV. Uncompressed DPX 10-bit log files have been around for over 15 years, and still get used every day, at a variety of aspect ratios. And there are people routinely using variations of HDCam-SR (tape and data), and it's workable for many projects. So is 4K, provided you have a fat pipe and a dedicated infrastructure.

The problems with colorspaces and LUTs is: there's so many to choose from. Every new camera that comes out seems to have a new one. This is getting worse, not better. SMPTE is trying to sort all these out, but the target is moving so quickly, every time they start laying the groundwork for standardization, three new cameras come out and the battlefield is changed all over again. It may never get stable enough for a standard to actually take hold. Until then, all we have is ad hoc standards for individual facilities. As long as you control the workflow end to end, this can work very successfully.

 

[M Most]

I appreciate the fresh look on images that ACES offers, which is why Light Iron is enabling an ACES pipeline for clients who prefer it or (in some cases) are mandated to use it. ....

I've read your post at least 3 times now, and with all due respect, I'm not buying it. If you substitute the word "RED" for "ACES" in every single one of your points, it is just as true. It is also just as true if you substitute the words "F65," "Arri Raw," "Cineform," "GoPro," "C300", or just about anything else. We work in a business that is at least partially driven by technology. New technologies come along all the time, and all of us are forced to spend internal time and money R&D'ing the best ways to deploy and support them. The primary differences I see with ACES are that it is open - and thus free - and it solves a problem none of the other things I just mentioned do, which is to create a common color space regardless of the original capturing device. The **potential** of that answers almost all of the issues we now deal with regarding new cameras and formats that we see almost weekly. In a world in which ACES is fully evolved (NOT the world we're currently in, and I've already said that...), it doesn't matter what formats different manufacturers want to invent as long as they also supply an ACES IDT for it. And the arguments about file size are moot because as I already mentioned, you don't need to use the half float EXR container unless you're exchanging files with another party or creating archival elements. You can use native file formats as long as the system you're on can play them.

You speak and write with eloquence, but words alone are not particularly convincing if the point they're trying to make is not well founded. If you, as LightIron, don't see the value in ACES and don't particularly want to support it, that's fine. But to come up with reasons that are basically just as true about any other new technology is not something I can agree with. Supporting Red was far more involved (still is), far more costly (still is, especially if you require a Red Rocket card in every device that needs to play it back in real time or render it), and far more "nonstandard" than a color system that is really, at its heart, just substituting an input transform for an input LUT and an output transform for an output LUT. Yes, it has a long way to go. Yes, it has specific issues at this point that need to be remedied. Yes, it has a "signature look" to it. All of those things are likely to be addressed over time, especially if the industry demonstrates support for the ultimate goal, which is to provide a theoretically unlimited color space that can accommodate all of the data from all capture devices known and unknown, and preserve that for archival integrity, along with providing a common mezzanine space for all of the varied formats we are constantly having to spend "time and money" on now. If you don't see the value in that, that's fine. But the points you made really don't stand up because they're true of just about everything we do. That's simply life in a technically driven industry.

 

[M Most]

The problems with colorspaces and LUTs is: there's so many to choose from. Every new camera that comes out seems to have a new one. This is getting worse, not better. SMPTE is trying to sort all these out, but the target is moving so quickly, every time they start laying the groundwork for standardization, three new cameras come out and the battlefield is changed all over again. It may never get stable enough for a standard to actually take hold. Until then, all we have is ad hoc standards for individual facilities. As long as you control the workflow end to end, this can work very successfully.

Sounds like you're making a good argument for ACES, which potentially solves all of those issues once its own issues are ironed out.

 

[Michael Cioni]

I've read your post at least 3 times now, and with all due respect, I'm not buying it. If you substitute the word "RED" for "ACES" in every single one of your points, it is just as true. It is also just as true if you substitute the words "F65," "Arri Raw," "Cineform," "GoPro," "C300", or just about anything else. We work in a business that is at least partially driven by technology. New technologies come along all the time, and all of us are forced to spend internal time and money R&D'ing the best ways to deploy and support them. The primary differences I see with ACES are that it is open - and thus free - and it solves a problem none of the other things I just mentioned do, which is to create a common color space regardless of the original capturing device. The **potential** of that answers almost all of the issues we now deal with regarding new cameras and formats that we see almost weekly. In a world in which ACES is fully evolved (NOT the world we're currently in, and I've already said that...), it doesn't matter what formats different manufacturers want to invent as long as they also supply an ACES IDT for it. And the arguments about file size are moot because as I already mentioned, you don't need to use the half float EXR container unless you're exchanging files with another party or creating archival elements. You can use native file formats as long as the system you're on can play them.

You speak and write with eloquence, but words alone are not particularly convincing if the point they're trying to make is not well founded. If you, as LightIron, don't see the value in ACES and don't particularly want to support it, that's fine. But to come up with reasons that are basically just as true about any other new technology is not something I can agree with. Supporting Red was far more involved (still is), far more costly (still is, especially if you require a Red Rocket card in every device that needs to play it back in real time or render it), and far more "nonstandard" than a color system that is really, at its heart, just substituting an input transform for an input LUT and an output transform for an output LUT. Yes, it has a long way to go. Yes, it has specific issues at this point that need to be remedied. Yes, it has a "signature look" to it. All of those things are likely to be addressed over time, especially if the industry demonstrates support for the ultimate goal, which is to provide a theoretically unlimited color space that can accommodate all of the data from all capture devices known and unknown, and preserve that for archival integrity, along with providing a common mezzanine space for all of the varied formats we are constantly having to spend "time and money" on now. If you don't see the value in that, that's fine. But the points you made really don't stand up because they're true of just about everything we do. That's simply life in a technically driven industry.

I can appreciate that, Mike. And I don't mind being called out or challenged in any way;-)

The bottom line for me is this:
My RU posts and all my previous statements about ACES demonstrate recognition of value to the system and for future proofing. For Light Iron, we think being cautiously optimistic and prepared to implement ACES is a better starting point than shooting in ACES and changing IDTs during the shoot. -Call me crazy, but really what I'm talking about is studying for the test so I can ACE it before it's given to me. There are movies out there right now that are doing the opposite and the images are paying the price. [See Francis Kenny's note above and ask around...]

While all of the camera formats you mentioned could easily apply into my post, here is where the math doesn't work: I'm not forced by the camera manufacturer to retrofit my facility to support their camera. It's a choice! (again, pluralism). Light Iron chose not to support film, which was a business decision. Though we knowingly gave up the opportunity for telecine clientele, we were never pressured or threatened to further film's prevalence in order to do any work. In contrast, I have been told by studio executive clients that, "If you don't do ACES, you don't work with us." The other key differentiator is if you insert RED into my previous post, you're ignoring the source of the request. In other words, Light Iron is clearly a known advocate for working with new and even beta technologies on a regular basis. The difference here is that creatives are typically the ones making requests for implementing a beta technology (such as F65 or RED) and they accept responsibility for its challenges, physically, financially, and visually.

Ask yourself this:
Hasn't history shown that some of the best technical implementation in motion pictures been motivated by creatives? From the Wizard of Oz to Terminator 2, to The Hobbit, I find when creatively driven, technological advancements can remain truer to the art and enhance the filmmaker's visions. But when the phone rings and I'm being told to use ACES on an upcoming project, even though it is admittedly incomplete and still in testing, it's never been a creative asking me to do this on the other end of the line.
That concerns me, Mike.
ACES has two faces. I am excited about ACES for the right reasons. I am troubled when ACES is being pushed on filmmakers by agencies instead of the other way around.
If ACES is purely an option, then it's a benign and powerful creative edge that should be leveraged when appropriate. But in my experience and the experience of many creatives around me, ACES is quickly snowballing into a mandate and not an option. That is starkly different from choosing to implement [substitute camera name here.]

m


"But I don't see any clothes!"
-little boy

 

[Blair S. Paulsen]

Kudos to the Mikes for the tenor of their debate - substance and respect, not rancor and chest pounding. Thank you.

It's too bad that at the point in time when the suits finally got hip to the upside of ACES, it wasn't ready for its closeup. I urge the academy and other stakeholders to seize the moment and put serious resources into making IIF/ACES viable before the opportunity is squandered. Props to RED for their commitment to working with the academy to optimize their IDT.

Shout out to Francis Kenny for his technical fortitude in (AFAIK) the first use of ACES for a premium cable network series (Justified, season 1 on F35) and going full Epic on Justified season 2.

Cheers - #19