Why commercials are shot only on REDs and Alexas

[Björn Benckert]

Even us native English speakers confuse things. Particularly when we're talking about very specific things.

Although Jake seems to think the term "transform" directly implies a mathematical approach, that is not the case. Any method of converting one colorspace to another, or one set of values to another, is a "transform." A matrix is one way of doing that. A LUT is another way of doing it. A series of concatenated mathematical calculations is another. If one wants to be specific, your use of the term "mathematical transform" would be correct, as a LUT is not doing any math (it contains pre-calculated values). I work with one of the leading color scientists in our industry and he refers to all of these things - correctly so - as "transforms" because they take one set of input values and yield a different set of output values.

So to be "technically correct" I think it's fair to refer to "mathematical transforms" as opposed to "luts", but it's not really correct to associate the term "transform" only with the former.

Agree. To me a color transform is the same as a color change.
But again all that is just terminology.

All Im saying is c mos sensors fill up very simular / linear to clip. Some has more DR than others and the DRs are layed out differently etc. But in raw the recording is very much linear. The alexa luts have dynamic roundning just as the IPP2 luts
has it. But its simply wrong to say that one camera has it more than the other as its very much a post / user selectable / adjustable thing.

 

[Phil Holland]

So to be "technically correct" I think it's fair to refer to "mathematical transforms" as opposed to "luts", but it's not really correct to associate the term "transform" only with the former.

100% agree..... As does.... The industry.

 

[Neil Abeynayake]

Whoa..... Just how did this thread go from a guy asking a simple question "Why commercials are shot only on REDs and Alexas?" to all this colorist and technical mambo jumbos?

With all due respect, are any of you guys Electronics Engineers and/or software Engineers with firmware expertise, specialized in sensor technology?

 

[Phil Holland]

Whoa..... Just how did this thread go from a guy asking a simple question "Why commercials are shot only on REDs and Alexas?" to all this colorist and technical mambo jumbos?

With all due respect, are any of you guys Electronics Engineers and/or software Engineers with firmware expertise, specialized in sensor technology?

I'm sorry it's technical mumbo jumbo, but do be aware that Color Transforms are now apart of the RED Workflow.


As to answer the original question. Lots of cameras are used to shoot commercials. Some markets are different and certainly reveal trends. Some cameras certainly adhere themselves to more of a cinematic "feel". Some prefer specific cameras for their aesthetic or workflow.

Even shorter answer. Good cameras go to work. Especially if they are reliable and provide an image quality that everybody is happy with.


No ego stroking for me today, but I'll back Mr. Most here because.... Technicolor reasons. This is a particular point that has been beaten to into a dead horse however.

On a total side note, bare in mind REDuser is visited by a lot of industry professionals and veterans. Members of the Academy, ASC, etc..... Some of whom have been around for a long while during the time when our current standards were forming, heck, they might have even played a role in a few of those things. Some even contribute and that is one of the things that makes this forum often a better place than others.

 

[Gavin Greenwalt]

Accurately or inaccurately the Sony, Panasonic and Canon cameras are viewed as "video cameras". A lot of this bias though is IMO very well deserved. A lot of the design decisions were filtered through "what would a news cameraman want?" Example: look at how long it took to get progressive scan vs interlaced and 24 fps vs 30/60 fps as options. RED has earned a biased (and well deserved) reputation for just kind of blowing off well thought out standards as well but they could at worst be accused of going off on some random tangent from the great TV Video <-> High-end Film continuum religious wars. I think historically we will see that the religion-of-film-production side was "right". RED has tempered their rebellious tendencies and accepted a lot of film best practices and they have always at least ostensibly aspired to be a "film" camera. And Sony, Panasonic and Canon have to some degree in recent years paid lip service to swearing off their heretical "video" ways of the past. And yet, I just feel like every time I want to like a camera from one of the big 3 video camera makers it always ends up just feeling like a video camera in every choice that's made.

It's kind of like the difference between using Nuke vs using After Effects. The Foundry's customer base is mostly high-end film and commercial artists so its products integrate well with large pipelines and has lots of scriptability and customization. Then you have AE from Adobe which at every turn makes such idiotic, bone-headed choices it's obvious that no developer has ever actually worked on a big budget film or commercial. RED made a lot of these sorts of decisions early on... and still does from time to time. The Alexa is loved by camera assistants and directors because it's very obviously designed by people who regularly work on a film set. Then you pick up a Canon C300 and... well... it's obvious they aren't even pretending to listen to those users.

a superior result using actual linear light camera RAW recordings and for that he/she works using transforms, not LUTS

Pedantic and wrong. LUTs/Mathematical Transforms are no better or worse than one another in theory or usually in practice. LUT transforms are just cached\baked mathematical transforms. Let's take the most basic LUT: a 1D gamma lut. You can use a mathematical formula [RGB]^(1/2.2). Or you can use a 16bit LUT with the values baked in and use linear/bi-linear interpolation for everything in between. I guarantee you in a billion years you'll never find an instance where the end result is distinguishable. Saying "Good colorists use mathematical transforms, bad colorists use look up table transforms" is ridiculous. They're both transforms, they both do the same thing and in practice both with a sufficiently well made LUT should provide identical outcomes.

Bjorn's point that rolloff is a product of the gamma curve, not the camera is valid. The highlight rolloff of Alexa is just a product of its gamma curve. You can copy/paste that exact gamma curve onto RED footage or take the gamma curve from a canon and apply it to Alexa footage.

 

[Marc Wielage]

How is your monitor calibrated ?

Very well. I started a new feature in late February and had my guy come by with a Klein 10A colorimeter and we spent about 90 minutes getting it checked and tweaked. It's A-OK: I subsequently took the project into a DI theater two weeks ago and I think the results were just about identical to what we saw there. The pictures held up very well, which was a relief.

My colorist doesn't use luts. Yes, he saves a grade between similar shots on the same project, but when he gets to a problem image, he deals with that at the raw level.

That's what I generally do as well. I did provide the VFX department with viewing LUTs on my last project so the 120 shots they needed to be done would have a general look that would work on their displays; we still graded from Log space.

For the record, I think the high-end cameras from Arri, Panasonic, Red, and Sony all can make great pictures in the right hands. Commercials are more wide open in terms of equipment than TV series and features, and I've seen a little of everything used in LA, including film. I have not seen Canon being used for anything but shorts, but maybe the 700 will make some inroads for them.

 

[M Most]

I have not seen Canon being used for anything but shorts, but maybe the 700 will make some inroads for them.

The Canon C300 Mk. II has been used for the Netflix show "Grace and Frankie" for the last 2 seasons (a show that shot its first season on Red).

 

[Diego Gagliardi]

Agreed Neil. It turned a tangent into mumbo jumbo technical stuff. Let's get back on subject.

 

[Brendan H. Banks]

Accurately or inaccurately the Sony, Panasonic and Canon cameras are viewed as "video cameras".

I actually disagree with this a bit (for Sony + Panasonic, not Canon). I think that we are seeing the FS7's popularity and Panasonic's name being known now for the Lumix series....but on the high-end of the spectrum the Sony F65 and Panasonic Varicam are favorites for certain DPs and productions. We just don't really hear about them as much as we do others. I was going to comment earlier that cameras like the F55, F65, and Varicam are actually probably more prevalent than a lot of people think. Even moreso before the Alexa Mini came out. I think that certain people view those brands as being more "video", but also don't include the highest end cameras in that line of thinking. The Sony CineAlta name has been a big one in digital cinema for a while and I think will kind of quietly remain that way for a good bit longer.

That all being said, I know what you wrote was your personal opinion so I'm not arguing at all

 

[Hrvoje Simic]

Very well. I started a new feature in late February and had my guy come by with a Klein 10A colorimeter and we spent about 90 minutes getting it checked and tweaked. It's A-OK: I subsequently took the project into a DI theater two weeks ago and I think the results were just about identical to what we saw there. The pictures held up very well, which was a relief.

Knowing your experience I'd be surprised if that wasn't the case. So...which tool was applied to iron monitor's panel + processing originated very particular signal inaccuracies, manifesting throughout the whole signal range in variable effect and scope, measured and modified in three axis ?

Mathematical transform ? CDL ?

Bjorn's point that rolloff is a product of the gamma curve, not the camera is valid. The highlight rolloff of Alexa is just a product of its gamma curve. You can copy/paste that exact gamma curve onto RED footage or take the gamma curve from a canon and apply it to Alexa footage.

You can but you won't get the same results. Neither in overall colorimetry nor in particular signal behavior in extremes.
Because significant part of image performance comes from sensor and processing and predetermines what the mapping will be able to achieve.

Appropriate gamma curve alone is insufficient to get the best possible rolloff to overexposure from a digital camera's base signal.

 

[Neil Abeynayake]

Agreed Neil. It turned a tangent into mumbo jumbo technical stuff. Let's get back on subject.

Thank you Diego. I'll try to answer questions to your first post. As you know the purpose of a "commercial" is to influence a viewer one or more of following reasons and then some.

1). Introduce a brand or a product to the masses.

2). Sell that product to many consumers.

3). Build a consistent buying trend of your product.

4). Establish, promote advertiser's name and brand retention on viewers.

5). *** "Add your own reasons here............"

On average an urban resident of a mid-size city may see and hear somewhere to the tune of 2500 to 5000 commercial messages every day. TV, radio, internet, billboards, newspaper & magazine ads plus other areas not covered here. An advertiser needs to go to a specialist who can cover large area of this spectrum. A specialist who had done their research on behavior, habits and buying trends of consumers. Thus an "Ad Agency". They develop, implement and write a story-line of the product to capture your intended audience for 30 seconds (or more) with a commercial. So, how can you get people to watch your commercial? Make them attractive, better, cleaner, eye catching and entertaining to the masses. They have to differentiate your commercial from your competition. Change or modify average TV viewing experience from routine to add "Wow" factor momentarily. This is where using a digital cinema camera might be one of the best solutions in the mix.

RED camera captures digital stills at whatever the designated frame rate. I am not going into details about this technology but you could visit this webpage to learn more: http://www.red.com/learn/red-101/redcode-file-format

Bottom-line is you get a cleaner motion file than most camera brands would give you. This is what separate high-end commercials vs. "other" commercials. Yes, any professional camera (including film) with good glass in front of it in the right hands could make a great commercial. In our case you could argue people can see the difference, clarity and cleaner resolution of RED's wavelet based motion codec within the "digital cinema" arena. That and implementation of a good script in turn gets viewer's attention and helps advertiser to sell their product/s.

Also as other REDUser members have stated before, people tend to use whatever the tool they are familiar or comfortable with. There is a Native American saying: "In the end, people would protect only what they love. They'll love what they know about. They'll know of things that are taught to them by others". They'll try to protect their investment. Whatever the camera, workflow they are familiar with.

We can go into details about science, technicalities and other details later. I'm trying to keep this post as brief as possible to keep you and others getting bored. Hope my post answered at least some of your concerns.

 

[David Kudell]

Having shot quite a few spots on Red Epic and Dragon, I shot my last set of spots on a Sony FS7 as we were trying to hit a budget, and lighting requirements are much less with FS7's ISO of 2000 compared to 320 to 800 for the Red. The FS7 shooting Slog and graded well looks very much like an Alexa or a Red. I bet most wouldn't tell the difference. We lit well and used Zeiss Otus lenses. With a larger budget, I would have gone Red or Alexa, but sometimes it's a matter of finding the right tool for the job.

 

[Marc Wielage]

The Canon C300 Mk. II has been used for the Netflix show "Grace and Frankie" for the last 2 seasons (a show that shot its first season on Red).

I'm surprised Canon doesn't publicize this more. I think the camera clips early, but no question, in the hands of a great DP it can pretty much do anything they need it to do. I wonder why they didn't use the C500? I don't think the rental cost is that much higher, and the cameras are about the same size. Sharpness is clearly not what they're going for on Grace & Frankie.

The underdog to me is the Panasonic Varicam (and Varicam LT), which I think can make really good pictures. I don't see a lot of those used often, at least in LA -- and it's a shame, because it's looked stellar at past demos. I'd put the Varicam miles ahead of what I've seen from Canon.

 

[Hrvoje Simic]

Back to main topic...generalization is a common path to a distorted perception of reality.

Commercials are shot on many different cameras. Where Red and Alexa dominate in higher end and Alexa takes the most jobs (in my experience). Most cameras are not shooting highest quality codecs on board, which is a prerequisite for full quality post, meaning 12 bit codecs or raw, and aren't designed for typical film workflows but single camera operator, and compensating hassle is not worth it for many as it significantly affects on-set workflow relying on practicality and efficiency.

I worked on TVCs shot on cameras ranging from four to six figures. Few days from now we are shooting on one of Canon Cinema EOS cams, last one was on Alexa Mini, before that it was Ursa Mini Pro. Alexa Mini was a 2 day, 5 location, 40 people shoot for a known smartphone brand and Ursa Mini Pro a one day, one location, 20 people shoot for a barbecue. So not really the same expectations, not even close production budget and likely an exponential difference in target market size and media buy.

If it's a shoot with a larger budget and media buy, higher expectations and responsibility, no one with a healthy mental activity in charge for six to seven or even eight figure project (media buy included) chooses to save a few hundred bucks per day for rental difference of the most important tool on the set.

 

[Matt W.]

I'm surprised Canon doesn't publicize this more. I think the camera clips early, but no question, in the hands of a great DP it can pretty much do anything they need it to do. I wonder why they didn't use the C500? I don't think the rental cost is that much higher, and the cameras are about the same size. Sharpness is clearly not what they're going for on Grace & Frankie.

The underdog to me is the Panasonic Varicam (and Varicam LT), which I think can make really good pictures. I don't see a lot of those used often, at least in LA -- and it's a shame, because it's looked stellar at past demos. I'd put the Varicam miles ahead of what I've seen from Canon.

Pretty sure the C300 Mk II has a far better image than the C500, which has never really impressed me although it is ridiculously sharp.

Varicam footage looks good to me, I agree. Better than C500, but I'm not sure it's much better than the C700 or something.... Seems a bit better than the F55 maybe. All pretty solid.

 

[jake blackstone]

I understand what you're saying but there are a number of things you're incorrect on here. First, IDTs don't impart a gamma curve or any falloff characteristic. In fact, they do the opposite - they put the camera's information into a linear light, scene referred colorspace. Second, everything you're talking about here in an ACES pipeline happens in the output transform - usually the RRT, although as you mention, there are other rendering transforms available in certain software (like the Arri Photometric transform in Filmlight's products). But either way, it's the output transform you're referring to, not the IDT (which is the input transform). Finally, there are an awful lot of more than "decent" colorists who do not work in ACES. Many work in an "ACES like" pipeline in which all input data is converted to one common colorspace (Arri LogC Wide Gamut is very popular for this), but not ACES itself. This is due to many reasons, among them familiarity, but also (and perhaps more significantly) issues with the RRT itself that often manifest when dealing with very saturated blues and magentas. This is also the reason that a number of studio pictures that were slated to go through an ACES path have chosen not to, as these issues have not yet been satisfactorily addressed. If you don't have any material with that issue (often caused by photographing neon lights, car rear lights, or anything else with high brightness and high saturation), ACES is a good approach. If you do, it might not be.

Anyone working with modern cameras and multiple deliverables knows, that ACES workflow is the way moving forward. FilmLight correctly predicted, that having LUTs for every workflow situation is just not practical. You literally then must have hundreds of LUTs necessary for working with every imaginable camera and deliverable. That is why FilmLight designed the Truelight Color Spaces (TCS), that helps to navigate this video jungle. Of coarse, some still choose to stick with LUTs, because that is what they are familiar with. But with the proliferation of new HDR deliverables, LUT workflow is just no longer viable. Keeping track of all LUTs used is becoming almost impossible. Yes, RRTs are the foundation of the ACES process and, because of the proliferation of multiple deliverables FilmLight, again, pressing forward with the new and improved Truelight workflow by now introducing "families" of RRTs, which can assist with easier management of multiple deliverables. Otherwise, again, even with the standard transforms this process is in danger of becoming unmanageable- P3, Rec-709, Rec2020, 1000, 4000 nits, D55, D65 etc. The choices and their combinations are endless.
RRTs are the heart of the ACES process and are the artistic tools, that can impart a camera look signature. I know some here are upset, that the talk about cameras used for commercial shoots turned technical, but closing ones eyes on the problems facing modern workflow will not help, pedantic or not. It doesn't matter, that some lump all transformations, transforms and LUTs into transforms. You of all people know, that proper nomenclature matters. Transforms are precise mathematical formulas, that are performed on the fly, yet, in case of RRTs, those actually are "creative" transforms, created by color scientists for establishing a "pleasing" look. If you ever converse with FilmLight color scientists, that is the constant underlying theme of their work- to create the most pleasing image and not necessarily creating the "correct" image. Every camera has an underlying "look" and, I believe, it should be celebrated and cherished, like film shock used to have. And that, in a nutshell, the reason why going forward, there will be ever diminishing reasons to stick with LUTs and move toward a more comprehensive and artitic solution, like ACES and Truelight Color Spaces.
And finally, ACES is in it's infancy and as such, there are still number of teething problems. That is precisely why ACEScct quickly replaced previous version ACEScc. But even though ACES technology is still in it's infancy, despite being in development for many years, ACES already shows great promise in simplifying and qualitatively improving the modern workflow, which at times can be very daunting...

 

[Hrvoje Simic]

You literally then must have hundreds of LUTs necessary for working with every imaginable camera and deliverable.

That pretty much depends on what you use them for.
For some purposes you can use one per camera. I use 4 per each camera.

Of coarse, some still choose to stick with LUTs, because that is what they are familiar with.

That may be a reason to some but some may choose to incorporate them in their workflow despite not using them before and adapt the methodology to them due to their independent access to image particulars, which enables achieving very specific things, while reducing fiddling...and the fact that they can be applied everywhere.

Every camera has an underlying "look" and, I believe, it should be celebrated and cherished, like film shock used to have.

In every case ?

You celebrate and cherish human skin turning yellow, or magenta ? I rather avoid that and get a referent image in the start of grading. LUT form makes that possible.

Film stocks don't deviate in colorimetry as digital.

 

[Hans von Sonntag]

Anyone working with modern cameras and multiple deliverables knows, that ACES workflow is the way moving forward.

Not in Germany and not in commercial land in Europe. My Flame has a tricked-out colour science. The Flames of my colleagues as well, of course. Hardly anyone is using the new colour stuff, most leave it to "legacy"and virtually no one is using ACES. We all use log camera material and work in a Rec709 monitor environment. Actually many of the TVC finishing guys are working with graded Rac709 footage, yes this is true. Hegel calls it the normative force of the factual. ACES is something only 1% of the TVC industry professionals understand, not to mention IDTs, colour transforms and whatnot.

When we load the camera material in the the box we love to have it log and off we go in rec709. Comping is often done in linear or log. There are many reason to do so. One, of course, is ignorance and no will to understand a new concept. The other is the experience you'll have when setting your box to ACES. The colour transforms to Rec709 works well with cameras footage, albeit colour correction feels differently, but we all have to deal with graphics and the likes and these don't fit into ACES at all, at least I don't know how to comp those with a ACES workflow without funny experiences and unsolved issues.

I'm sure this all will be figured to at some point but I'm sure the vast majority of Redusers need today ACES as much as I, 0%.

Hans

 

[jake blackstone]

That pretty much depends on what you use them for.
For some purposes you can use one per camera. I use 4 per each camera.

That may be a reason to some but some may choose to incorporate them in their workflow despite not using them before and adapt the methodology to them due to their independent access to image particulars, which enables achieving very specific things, while reducing fiddling...and the fact that they can be applied everywhere.

In every case ?

You celebrate and cherish human skin turning yellow, or magenta ? I rather avoid that and get a referent image in the start of grading. LUT form makes that possible.

Film stocks don't deviate in colorimetry as digital.

You are operating from a very narrow point of view. I understand Hrvoje, that you have to promote your product and I generally stayed away from the subject, but because you addressed me directly, I have to admit, that frankly, I'm not a fan, sorry. I just never understood this business of buying and selling pre-made LUTs. But if someone feels like they need to use them, what do i care? Anyway, I'll leave it at that.
As I had mentioned, various modern cameras and a myriad of deliverables makes LUTs approach moving forward unmanageable, plain and simple. That is why ACES was originally introduced and further improved by FilmLight and now also adapted by BMD.
BTW, it's a well known fact, that at least Kodak film was always optimized for Caucasian skin reproduction and it was less so for any other types of skin. For obvious reasons, Fuji was much better at asian skin reproduction. And Agfa always was a reflection of European sensibilities in low film saturation and contrast in reproduction of pale European skin...

 

[jake blackstone]

Not in Germany and not in commercial land in Europe. My Flame has a tricked-out colour science. The Flames of my colleagues as well, of course. Hardly anyone is using the new colour stuff, most leave it to "legacy"and virtually no one is using ACES. We all use log camera material and work in a Rec709 monitor environment. Actually many of the TVC finishing guys are working with graded Rac709 footage, yes this is true. Hegel calls it the normative force of the factual. ACES is something only 1% of the TVC industry professionals understand, not to mention IDTs, colour transforms and whatnot.

When we load the camera material in the the box we love to have it log and off we go in rec709. Comping is often done in linear or log. There are many reason to do so. One, of course, is ignorance and no will to understand a new concept. The other is the experience you'll have when setting your box to ACES. The colour transforms to Rec709 works well with cameras footage, albeit colour correction feels differently, but we all have to deal with graphics and the likes and these don't fit into ACES at all, at least I don't know how to comp those with a ACES workflow without funny experiences and unsolved issues.

I'm sure this all will be figured to at some point but I'm sure the vast majority of Redusers need today ACES as much as I, 0%.

Hans

Funny, that you decided to speak for the majority of Red users, all German users as well as all Flame users. Just because you don't want to be bothered with ACES, doesn't mean, that you are qualified to speak for the rest of the users. On a TVC Flame is typically used on a few select shots and you definitely will not use it to grade a feature, like you'd use Lustre, right? Let me give you a hypothetical situation- you're grading a feature, where Arri is a main camera with Red camera used for high resolution VFX work and for a good measure some drone footage and some stock shots were also used. How about some high speed Fantom footage, that is thrown in the mix for a good measure? Using LUTs in this type of situation would have been a nightmare. With ACES it's a cinch- select your working color space- Red or ARRI or Sony or Canon or whatever you like and Baselight will use the material metadata to AUTOMATICALLY tag most of the material. All in all, it just takes a couple of minutes as opposed to many hours to manually apply various LUTs. The same goes for the deliverables- select your ODT family and FilmLight will do most of the work. Do you need to change the deliverable's white point? Just select it. Do you need to change the peak brightness? Again, just make a selection. With ACES you'd need to do very minor trim passes, unlike when used with LUTs. And then the biggest PIA all the material acts differently, depending on what it is. The new BaseGrade operator simply turns all the material regardless of it's origins into a linear light type of file. Yes, that includes even Rec-709 material. And now all of a sudden, not only all material acts the same predictable way, but now you have all those tools, that are normally associated with RAW controls- exposure, color temperature etc available for all of the material, regardless if it's RAW or not. It's just plain better, than the old way of grading. And all this wonderful new technology became possible thanks to ACES foundation and only ACES can provide a clean and simple workflow in this kind of situation. Can the work still be done with LUTs with lots of trim passes? Sure, it's still done this way all the time, but moving forward, ACES is a much more viable, precise and less time consuming solution, period...