Color Pseudoscience

[Guest]

What i actually find odd, since so many people shoot a good deal more footage than they grade, then why not enjoy or want a camera that looks good from the beginning?

Indon’t thinknthere is a single person that *relied to this this thread that replied on the camera to make good images.

 

[morgan_moore]

Baked.

The word makes me shake in my boots. Its my weakness.

The only thing Ive done that was more scary was 'slide on the train'. Where one, in stills, did jobs for magazines and couriered them to the client without processing. But with stills one could 'braket' - take the same still at three or four Fstops.

Mainly during my career Ive developed my own work, which was stills for the first 15 years. Starting with a B+W dark room, then C41 colour and scanning (in home dark room) then digital. Up to 16bit raw.
Even black and white (up to 1995) it was 'normal' to dodge and burn and one had the latitude of film (being film after all). Stills strobes also outlight an M18 no problem (http://www.sammorganmoore.com/escape/)

Basically I 'grew up' on resolution and latitude.

Then I bought and EX1 - what a shock - no lights no latitude

So I think getting it right in camera is hard, especially without HMI because its a 14 stop world.. and plopping that in 7 is hard/impossible

Additionally on location we tend to have terrible small monitors which dont have great colour - everything about baked works against one until one has a full movie set up. HMI, and a monitoring tent.

 

[Run&Gun]

Baked.

The word makes me shake in my boots. Its my weakness.

The only thing Ive done that was more scary was 'slide on the train'. Where one, in stills, did jobs for magazines and couriered them to the client without processing. But with stills one could 'braket' - take the same still at three or four Fstops.

Mainly during my career Ive developed my own work, which was stills for the first 15 years. Starting with a B+W dark room, then C41 colour and scanning (in home dark room) then digital. Up to 16bit raw.
Even black and white (up to 1995) it was 'normal' to dodge and burn and one had the latitude of film (being film after all). Stills strobes also outlight an M18 no problem (http://www.sammorganmoore.com/escape/)

Basically I 'grew up' on resolution and latitude.

Then I bought and EX1 - what a shock - no lights no latitude

So I think getting it right in camera is hard, especially without HMI because its a 14 stop world.. and plopping that in 7 is hard/impossible

Additionally on location we tend to have terrible small monitors which dont have great colour - everything about baked works against one until one has a full movie set up. HMI, and a monitoring tent.

And it’s the exact opposite, for me. I cut my teeth in an SD Betacam world, where “fix it in post” was said tongue-in-cheek. You HAD to get it right in-camera. That’s why I laugh at the people that complain about the DR of the cameras today and that they can’t shoot anything unless it’s LOG or RAW and the camera is 16 stops. Is it great that we have the tools that we do today? Absolutely. But people shouldn’t act like it can’t be done in-camera, in the field, at capture. The tools and tech we have today are so far beyond what we could have imagined 20+ years ago, but I see more blown-out, overexposed, poorly colored(white balanced) video than I ever saw, before we had all these tools at our disposal. And we were usually doing it with just a B&W CRT VF.



Any other old-timers get the hidden joke?

 

[morgan_moore]

Yep starting an black and white photography one had a wide wange of control from day one - and felt robbed of this when I got my EX1.

My photos could be as good as McCullin or Leibovitz or anyone - I had the same kit as them (a nikon and a 50 - for $200) and the only restriction was my subjects and skills. But kit was not stopping me from competing with the best.

Beta people were not there and could not really compete with the echelon shooting on 35 or vista. I respect the job those Beta people did.. but ultimately they were massively held back by kit. 35 and vista stand up today - beta does not - even though some amazing historical content was captured by very talented people

Of course it can be done in camera.. but that could involve gelling windows or finding a dark spot to shoot, or rolling in a truck of HMI or only shooting in cloud or at magic hour - things that are compostiionally restrictive or beyond a typical corporate budget for time or kit.

Off WBand poor exposure is indeed today inexusable.

As it happens a lot of cinematographers are clamping thier image to an on set LUT, which I feel forces them into using loads more kit than they need and is not responsible cinematography.

 

[Mitch Gross]

And it’s the exact opposite, for me. I cut my teeth in an SD Betacam world, where “fix it in post” was said tongue-in-cheek. You HAD to get it right in-camera. That’s why I laugh at the people that complain about the DR of the cameras today and that they can’t shoot anything unless it’s LOG or RAW and the camera is 16 stops. Is it great that we have the tools that we do today? Absolutely. But people shouldn’t act like it can’t be done in-camera, in the field, at capture. The tools and tech we have today are so far beyond what we could have imagined 20+ years ago, but I see more blown-out, overexposed, poorly colored(white balanced) video than I ever saw, before we had all these tools at our disposal. And we were usually doing it with just a B&W CRT VF.



Any other old-timers get the hidden joke?

Get off my lawn!!

 

[Mitch Gross]

The simple fact is that most people don’t know how to use their tools. They may understand in theory but in practice it’s a glancing knowledge. In the early Beta days a camera “menu” was perhaps three screens of basic choices. There were a couple of companies that made a business of opening up the side of cameras and tweaking pots to customize the image. Most people just shot to whatever this was. I used to work for a company that was known for its scene files. People would load those into their cameras and just white balance and expose to them. Even now people want a magic LUT that will simply fix everything for them.

Guess what — they’re not necessarily wrong. Light to the look. Just because something can be adjusted doesn’t mean you should. Always saving it for post in many if not most situation is wrong for the job at hand. So be it.

Back my day we had to walk uphill in the snow both ways to school.

 

[Guest]

...There were a couple of companies that made a business of opening up the side of cameras and tweaking pots to customize the image.

I can imagine Combatentropy getting in there and futzing about inside the cameras back then. Haha

... Back my day we had to walk uphill in the snow both ways to school.

Back in my day, “highspeed” internet was 30kbps both ways!

 

[Run&Gun]

Get off my lawn!!

I always like this one...

 

[Run&Gun]

The simple fact is that most people don’t know how to use their tools. They may understand in theory but in practice it’s a glancing knowledge. In the early Beta days a camera “menu” was perhaps three screens of basic choices. There were a couple of companies that made a business of opening up the side of cameras and tweaking pots to customize the image. Most people just shot to whatever this was. I used to work for a company that was known for its scene files. People would load those into their cameras and just white balance and expose to them. Even now people want a magic LUT that will simply fix everything for them.

Guess what — they’re not necessarily wrong. Light to the look. Just because something can be adjusted doesn’t mean you should. Always saving it for post in many if not most situation is wrong for the job at hand. So be it.

Back my day we had to walk uphill in the snow both ways to school.

Did you work with Macie, back in the day? In the Beta days, he was a genius. That set-up made so many of us look like hero's. Which is why I can't understand why they just couldn't figure out the VariCam's. I saw their set-ups on a client's 27H's when I bought mine and thought it was God-Awful. Never sent mine it. Against my better judgement, I sent one of my 2700's in to them and after getting it back, I went back to my personal set-up.

 

[Guest]

@combat,

The irony of all this is, i own a camera that doesn’t look good out of box. Well, it looks good, but isn’t the bench mark for “mojo”. Still designing a LUT that doesn f me over when i get to post. Designing a good LUT for each camera takes time and some trial and error.

Perhaps it is my dislike for constantly needed to tweak the basic things. I should have bought an Alexa Classic, but i do like 4K and up for texture and tonality, so, i am willing to go through the fire of LUT designing and post colouring.

I figured if i bought a cheap classic, i’d have to spend another $10K upgrading my support gear, and i’d have to worry about my spine on a whole new level when shooting handheld.

So, DR and Colour Science aren’t everything.

The problem is, just few cameras have the right ratio of features. Or... they are are all kind of good these days too.

 

[CharlesPapert]

I had my own secret sauce with analog then digi-beta that involved endlessly twiddling the outboard CCU unit plus filtration like pro-mists and Ultra-cons (I just sold my set of 4x4 Ultracons that dated back to somewhere in the early 90's, hadn't touched them in many years) and when possible I used to send stuff to Filmlook for the 24 frame conversion. . Everything was in the service of trying to get more latitude, softer rolloff and overall just less video looking. I can also remember the very first tape-to-tape session I was in that had a rudimentary version of color correction (cannot remember which desk they were using) and the first thing the colorist did was apply magenta to the image, they said "all Sonys shoot green". I was shocked to realize this. From that point on, I added an 1/8 and 1/4 plusgreen to my little book of CTO/CTB filters I'd carry around to cheat the white balance (remember those?). That alone was a huge difference, I remember clients saying "how do you get the skin tones to look so good?" and it was because they weren't slightly green! Amazing to think, baked-in color and few people doing any kind of color correction--such a different time.

 

[John Brawley]

Basically I 'grew up' on resolution and latitude.
...
So I think getting it right in camera is hard, especially without HMI because its a 14 stop world.. and plopping that in 7 is hard/impossible

.

I started off shooting reversal 16mm film and Betacam. You learned to work around the DR limitations, like one does in any creative endeavor.

ARRI Alev III based camera systems (including the latest 65, LF, mLF) are still the only camera systems that provide true, usable, full stated DR. The sensor is 16-bit linear where the result is mapped to 12-bit log.

Few things...

Alexa isn't the only 16 bit linear mapped to 12bit log and dual gain sensor. There is another company that does this. And what I find is that this leads to better "tracking". That is, as you slide up and down the exposure scale in post, the colours don't start skewing off as widely as you go to near clipping or in near blacks.

The colour correction "tracks" neatly through the exposure range.

Alexa also has "phat" pixels. The main reason their sensor's also haven't been 4K till the "new" sensor launches sometime next year. Their pixels are freaking huge. While companies like RED chased resolution by making smaller and smaller pixels for the same surface area, Alexa's big gigantic pixels are the opposite.

Here's another dirty little secret.

All manufacturer's DR numbers are theoretical because they are CALCULATED based on the sensor performance. No one is sitting in a lab measuring it to generate those marketing numbers. It's based generally off the signal to noise spec of the sensor.

There is NO STANDARD for measuring DR. It's an entirely subjective judgement. With the same visual information, two individual viewers will come up with two different numbers. Because....noise....one person's unacceptable noise is another's filmic grain. This is why you get so many meaningless phrases like useable or bankable DR. It's all word play to get around the fact that everyone has a different opinion about it.

That's why I like side by sides. Create a high dynamic range scene and compare one to another.

Add to that the number of cameras that are doing in-camera processing to reduce that noise and it starts getting even more tricky to measure. And something I rarely see mentioned when everyone posts their XYLA charts...

Motion.

We don't shoot test charts as a single frame to pixel peep. They're 24 frames per second, and you perceive LESS noise when it's in motion and it's different again when projected or seen on a monitor. Not only that but the type of noise signature alo changes how offensive the noise threshold is. I find the Sony noise to be the most egregious once it shows up. I really liked using Venice in some recent testing I did but it still falls into a hole once you try and lift it beyond it's comfort zone.

Alexa always seems more forgiving to me. The Alexa 65 had more noise but it was "nicer".

We're straying beyond the OP's question, I still think we do perceive differences in cameras because of the many choices that affect the image pipeline, some we control and some we have no idea about, and while there's plenty of overlap and many cameras can be "graded" to be the same visually, there are certainly very significant strengths and weaknesses that give each camera platform it's unique visual signature.

JB

 

[Guest]

@combatentropy,


Several years ago, after being a long time lurker, and unabashed camera basher, i decided to try and actually contribute something semi-useful to the dvxuser community, and started sharing my experiences with the Canon Picture Style editor. I sold my Nikon D7000 and started running with a 60D due to my preference for the 60D ergonomics and ability to adapt more lens variety.

On this page, i had begun to settle on a picture style i called R3. I forget what “R” meant, but 3 was the category. I started designing with R1. I had a system, where the letter denoted the look i was after, and the three following numbers represented changes in tone curve, saturation/colour, and other variations (like sharpness or hue etc). R303 was the one where i started prefering it to some of Canon’s default styles.

http://www.dvxuser.com/V6/showthrea...s-in-the-PSE&p=2461734&viewfull=1#post2461734

Eventually a few others added even better picture styles and advice.

Now that JCS has a new Canon picture style, i am thinking of updating that thread.

And when i worked news, i changed the camera settings on my designated camera. When i left the station, one of my colleagues told me he has to get my camera when i leave.

But after all the tweaking, i almost just like something similar to a good default lut, like Arri’s Rec709, or JCS’s picture style for Canon. Although, Canon also has done a good job on their high end cameras. I newrly can just run with their defaults and be pretty happy. Just wish more camera companies preserved the highlights in a super subtle roll off rather than truncated curves. Arri does that pretty well, and Canon wide dr or sony hyper gamma are all variations of that.

 

[morgan_moore]

I started off shooting reversal 16mm film and Betacam. You learned to work around the DR limitations, like one does in any creative endeavor.

Of course talented individuals such as you and Charles made the best of the tools available to you at the time. I still guess you would prefered then to have been shooting on 35 and now you would choose 35 or high end digital, with a decent grade today.

16 still seems to have something today from my observations (ive not used it)

 

[Guest]

People that talk about shooting even with limitations, still tend to choose better tools with less limitations when available. Curious that, no?

 

[John Brawley]

People that talk about shooting even with limitations, still tend to choose better tools with less limitations when available. Curious that, no?

Yes, you could type out a novel using an iPhone, but why would you if you didn't have to ?

JB

 

[jcs]

With my somewhat limited range cameras on hand, the 1DX II and EOS R (don't use the 80D much anymore), I have to shoot for an in camera look, or WYSIWYG. While the EOS R supports CLog, I prefer my custom setting for skin tones (CLog has a hair more DR in the highlights compared to my 'match as close as possible to CLog' profile, where the DR edge is slightly more color retention, not white clip).

Indeed I agree that XYLA DR charts aren't very useful for determining usable DR. What is useful, is as JB noted, side by side over/under shots and seeing what can be recovered and how exposure changes color. Good/great color science provides exposure invariant color, as the Alexa examples illustrate. Since I have to do camera, light, sound, directing, and editing (audio & video, + writing or helping with writing), my goal is simply accurate color at capture with no surprises in post. Now that I know that exposure invariant color science is possible (ARRI), that's the bar of quality I'm looking for in a new higher end camera. If color skews with exposure, that's simple sub-par color science. AI could be helpful for those still playing catch up to ARRI.

Since I need something small, light, and not-attention-grabbing, DSLR/mirrorless is currently the best match, and Canon in my experience provides the best skin tones once tweaked (better than my old C300 II). This is a somewhat challenging shot from the 1DX II + Filmic Skin shot in green tree shade:

While it's holding together OK, we can see that highlights and shadows have different color bias (perhaps a bit blue in highlights (hair) and yellow in shadows (neck)). White is white, so good enough. However in more challenging conditions when doing more radical grading or fixing a mistake in exposure, or couldn't control light etc., Canon color science becomes more challenging to fix (and it's worse for Sony & Panasonic, though they get a little better every year).

 

[Guest]

Yes, you could type out a novel using an iPhone, but why would you if you didn't have to ?

JB

My point exactly.

Just always see those suck-it-up post when someone notices a flaw in any given camera. Photography is far more competitive than camera manufacturing. By that fact, chances are photographers are outperforming cameras.

 

[Mitch Gross]

Did you work with Macie, back in the day? In the Beta days, he was a genius. That set-up made so many of us look like hero's. Which is why I can't understand why they just couldn't figure out the VariCam's. I saw their set-ups on a client's 27H's when I bought mine and thought it was God-Awful. Never sent mine it. Against my better judgement, I sent one of my 2700's in to them and after getting it back, I went back to my personal set-up.

Roger Macie was the king of tweaking those pots to "fix" the Sony betacam look, which basically meant taking it away from the preferred Japanese look to the preferred American look (the Japanese like things COLORFUL and SHARP). I worked at AbelCine, which did in fact conquer the VariCams with excellent scene files, including the famous "JR45" look, which was simply Jesse Rosen's 45% gamma settings in both FILM REC and VIDEO REC. You should have sent your 2700s to AbelCine -- I must have loaded those looks 100 times and helped translate the original JR45 Scene Files to a dozen differnt camera models, including the 2700s.

 

[Mitch Gross]

Here's another dirty little secret.

All manufacturer's DR numbers are theoretical because they are CALCULATED based on the sensor performance. No one is sitting in a lab measuring it to generate those marketing numbers. It's based generally off the signal to noise spec of the sensor.

There is NO STANDARD for measuring DR. It's an entirely subjective judgement. With the same visual information, two individual viewers will come up with two different numbers. Because....noise....one person's unacceptable noise is another's filmic grain. This is why you get so many meaningless phrases like useable or bankable DR. It's all word play to get around the fact that everyone has a different opinion about it.

That's why I like side by sides. Create a high dynamic range scene and compare one to another.

Add to that the number of cameras that are doing in-camera processing to reduce that noise and it starts getting even more tricky to measure. And something I rarely see mentioned when everyone posts their XYLA charts...

Hey John, I can't speak for all manufacturers but I can sure speak for Panasonic. We absolutely do test the finished cameras to determine Dynamic Range. In fact I personally got in a bit of trouble back when the EVA1 was announced but final specs not yet announced because I published a picture of a waveform and a DSC Labs XYLA 21 chart from the output of an EVA1 still being finished and tweaked. Panasonic had not yet announced a DR number because we're a pretty conservative company and don't publish such things until we know we can back it up (unlike others). The chart I showed indicated 14 stops of range, which is in fact what we deliver on the camera. But if tweaking had meant that the range would drop to 13 then my photos would have proven problematic. Because we really do test and anyone can easily repeat the testing we do so it has to be right.

RED got into a bit of trouble when they were first testing their Monstro sensor and stated that on a testbed they were able to delineate something like 17 stops. Of course that was on a testbed and they did clearly state it, but people took that to mean that a finished caera would actually deliver such range and no way was that ever going to happen. RED learned an embarrassing lesson, which was to wait until you know what you can actually deliver because otherwise people will be pissed.