"Color Science" Makes Me Laugh
Vložit
- čas přidán 10. 09. 2024
- Good photographers and filmmakers understand what their equipment does, not what they're told it can do, or what they think it "should" do.
I talk about camera sensors, their color filters, the ultimate tradeoffs all manufacturers must make, and which can't be changed once the devices leave the factory.
It's a fun phrase, but I totally agree with you. Camera do have their "look" by all the factors you've listed, but adding the term "science" to it makes it sound like something unique and special when it isn't - it's all in the manufactures preferences. With proper LUTs and grading it really doesn't matter what you shoot with anymore.
That's my limited experience!
Excellent video. For those of us who started photography in the film days (and particularly for those who worked in photo labs), this was common knowledge with regard to film choices. Kodak typically emphasized reds and yellows, Fujifilm blues and greens, etc. The same people who loved the exaggerated greens of the grass in Fujifilm thought Kodak made landscapes look dull. Conversely, people who wanted Kodak's skin tones wondered why Fujifilm looked a little sickly. It's really the same thing - it all depended on the emulsion coating on the film, and even the color of the film base itself.
Cheers!
I couldn't afford to shoot much color ;)
There are some sensors used on cameras from multiple brands. I do not know whether the sensor filters are similarly shared. If so, I should think that the raw data should be invariant across the brands even if the file format is different (NEF for NIkon etc.). But I don't know if raw formats all give the exposure received by each individual photosite, or whether the conversion to the output resolution is done in-camera even for raws. If the sensor data is fully recorded, you'd expect the exact same image to result regardless of brand, but I think it actually wouldn't. Raw files contain not only sensor data, but a lot of other information including the sensor metadata's color profile. And then the program reading the file gets its chance to modify what you'll end up seeing too. :) I can remember lots of times I'd open a raw for a new camera in Lightroom and find unexpectedly low image quality, which improved with Lightroom updates. :)
As far as I know, each photosite's electrons are written to the RAW file exactly as they are "counted". CMOS works by writing them one row at a time, which is why you have "rolling shutter". I believe those in the industry who work on sensors know perfectly well how similar each is to the other, but it's a lot of math and things to know to understand the differences, if any. Why try to talk to trolls about it? haha.
Canon uses dual pixels (scattered pixels for focus), not sure if they are the same as normal photosites or physically different. I do know what the data enters RAW as is, because they turn up red or green in Magic Lantern. My guess is there is some demosaicing in Canon cameras with RAW recording.
Most of the work is in the color matrixes that camera manufacturer suggests the software use to deal with the image. For whatever reason, Adobe or others will use their own calculations. But I don't feel any of that changes the fact that the RAW data is straight from the sensor unadulterated.
Maybe someone who knows will comment.
Hey man, colors are colors to me. And some camera colors just suck, you won`t get them right in post.
Some camera colors are great and are easily adjusted in post, that`s how i see it. And that why i choose Sony.
Yes. For video, the reason many find it difficult to get the colors right in post is they throw them out using LOG shooting GAMMAS, a tradeoff for brightness detail at the extremes. Although I have cameras from all the brands if I had to pick one it would be Sony. Thanks for comment!
Interesting topic. I do feel that "color science" has become an acceptable universal term in digital photography. It combines the two words used in the overall process, one being subjective and other involving science / technology. At the end of the day we know what it means, and the RAW file results are consistent and generally easy to differentiate.
I screwed up by not saying I was talking mostly about video. Yes, you're right of course. Any photographer who works with RAW in a deep way can separate the science from art. The problem for those who shoot video in non-RAW is the camera makes the decision what "colors" to choose and the debayering method when writing to 8 or 10 bit. In other words, most of the science is hidden from them.
Excellent explanaition
Thanks for the kind words!
Actually, we are very bad at getting „absolute colours“. We can discern between two shades of colours (if they are far enough apart) but when I show a single sheet of „red“ paper, we all may see a slightly different shade of red (or maybe not, we do not know…). This makes it very hard for manufacturers to get „good colours“. We are probably seeing the preferences of the execs of the company in the JPEGs 😂
And the sun is a slightly different color in Japan and that mixes with their displays and gives them a different red than we get in the West. Still, we must blame the execs for sure ;)
I can't understand why in movies in bright light they say that you need to shoot on the second 2nd ISO, say 12800 with ND filters (to catch all the highlights), and in dark light on the first ISO 800. Why not the other way around? They said that the dynamic range somehow captures the largest shadows and damn, do you also agree with this statement?
I would think in a movie with a real budget they'd light for the effect they want and expose for the subject. Setting ISO is for us armchair filmmakers ;)
@@MaxoticsTV There are tests of the study of the volume of the well of the pixel of the Maritsa, and often it does not depend only on the size of the pixel but also on how this cell absorbs light or rather the volume of this light. That is why the color of Panasonic video cameras is so beautiful, unlike mirrorless cameras or Sony cameras, they simply force their pixels to collect more light per unit of time.
@@ggdfggdfgdffgfddg34 Interesting. Anywhere I can read/learn more about that?
Keep in mind that "color science" is different for different sorts of media -
You'd be mixing apples & oranges if you compare physical pigments, for example, with electronic media.
Yes, I should have made it clear that I'm talking about filmmakers who talk about "better color science." Thanks for comment!
😊
Replace color science with color reproduction, which is a combination of the censor and the processing, that’s why a censor of brand x might look different in a camera produced by brand y compared to the same censor used in camera produced by brand z. Color science is just another popular term like full frame, which we all understand. The Canon 5D has a certain look, the Nikon D700 has a certain look, the Fujifilm XPro1 has a certain look etc. I understand why you dislike the word science here, but you can argue about the meaning and interpretation of many words, while ignoring the fact we all understand what is meant. So I’ll stick to Canon vs Nikon vs Fujifilm color science - none of which are even fixed.
You're right, a better title would have been "Better" Color Science makes me laugh.
@@MaxoticsTV well looks like I mixed up sensors and censors, the latter I have been encountering too frequently on YT 😬
Beside preferences, our eyes also see depth of color very differently very similar to taste bud.
Yes, and females have more refined tastes than men--empirically.
Great video! The third piece of the puzzle (in addition to sensors & displays) is how our eyes work. We have three types of cones, each sensitive to a different wavelength range. This is why filtering the light entering a camera into red, green & blue allows us to produce images that are interpreted by our brains to have colors which are the same (or very similar to) the original objects. If our eyes were able to detect the wavelength of light (instead of which wavelength range it is in) it would be much harder to produce images that looked right to our brains.
Interesting! I had never thought about wavelength precision might actually be harder!
Yep. I tend to have all my cameras of differing makes as neutral as possible, so that I can dodge the in-camera jiggery-pokery that's going on before the raw is saved (and let no one try to tell you the raw is unadulterated).
Excellent! Most camera detectors are sensitive to the spectrum of visible light, filtered to red, green, blue primary colors. The RBG values are stored in 8-bit RGB values of 0-255 in jpg files. A 16-bit RAW or Tiff file stores values in 16 bits as 0-65,535 RGB values. There are 2 types of primary colors: additive primary colors as light with Red, Green, Blue the primary colors;
this is how color display devices such as TV, computer screen work. Subtractive primary color or pigments: Cyan, Magenta, Yellow which is how most color printers work (usually including black pigment also).
Yes, a lot of the technology was adapted to work with legacy TV (NTSC/PAL) and printing.
@@MaxoticsTV My background is remote sensing and much of this in satellite sensors is similar, except we work in spectral regions beyond the visible spectral region.
@@RetrieverTrainingAlone How do you infer what wavelength you're reading, or do you just assume a range?
@@MaxoticsTV Most humans have 3 type of photo receptors in the cones of the retinea sensitive to blue (peaking at around 450nm, green (peaking at around 550 nm) and red (peaking at around 650 nm). The rods in the retina are typically sensitive to a wider range of wavelengths (400-700nm) and therefore can sense in much dimmer light.
Color science is just the current gimmick they’re pushing now, it was megapixels, codecs and pixel size just a few years ago. Ironically a lot of older cameras had better “color science”, Canon C300 mk1, Sony F3/F35/F23, Red One, etc all had a distinctive film look and feel to their image.
I am relatively new to photography, but after a few years, I'm at the point of trying to educate myself on the many terms "erroneously" thrown around in the photography world. I see much the same in the dance world - and very few who attempt to be accurate. I started tackling this area about 2 weeks ago - so I understand where you are coming from (i.e. light is composed of spectra (not colors - the label we put on a certain spectra) and colors are subjective depending on what type of vision you have (i.e. if you are color blind or have an extra type of cone). I also understand that the color recorded is a byproduct of the Bayer filter dyes, infra-red filter (surprised you left that one out), and the silicone composition in the sensor - to to mention the post capture picture profiles and processing - if shooting JPG. I really appreciate your efforts here and details outlining this. Can I ask you a favor? In the last week, I have reviewed no less than 10 articles - and a few CZcams videos - attempting to explain "gamma" as it relates to photography. In my review, noone I can find has done a start to finish (us humans visualizing the final image) explanation that is consumer friendly. While I believe I finally understand this concept from beginning to end, I'd love to hear someone else tell it to confirm my understanding, and then outline it's relevance to post-processing. This is to say, I do get how are eyes see a dynamic range scene differently (not linear secondary to pupil constriction or dilatation) compared to the luminance seen by the sensor. I'd love for someone to take it from the beginning and also show how the bit-depth can allow a smoother luminance gradient (i.e. where in the process that is applied and how) to the final output. Just thought I would ask, even though I'm not sure how popular the video would be - That said, it would be one of the few of it's kind. Thanks in advance.
My mother was a professional ballerina, then a ballroom dancer later in life. She loved the science of it though few recognize such an aspect exists.
I did a video covering GAMMA six years ago. I'll try to redo it for you. But if you have 30 minutes to listen to "ums", it's here: czcams.com/video/ONfdarINvmw/video.htmlsi=8rn6UvCLnQebkzHE
Your comment has put me into deep thought. And yes, few viewers for videos like this.
@@MaxoticsTV Just some fun facts for you: 1. My father was a Jazz musician (arranged for Duke Ellington, Oscar Peterson, Count Basie, Sy Oliver (his teacher), trained with Billy Taylor - and many more. Worked at Decca Records in N.Y.C.. He was also into computers and introduced me to a guy by the name of Bill Gates in the late 1970s. He passed away last year. I danced briefly (swing) with Janie Parker (PB of Houston Ballet). Years later I met Li Cun Xi (her partner) - and the defector from China, because friends of mine had let me stay at their home when he defected and he returned (he now heads the ballet in Australia) to cook a birthday meal for them and recruit dancers from Canada. He was an absolute delight and the subject of the movie "Mao's Last Dancer". Thought you might enjoy some of that. ;-) Thanks for the link. I'll head that way.
What a hell are you talking about😂😂😂 remember take your head medicine😂😂
What really gets me is when someone compares different manufacturer's color "science" by opening the raw in Lightroom, since at that point they are not even looking at manufacturer colors, but Adobe colors. Even the camera standard profiles are (as far as I know) made by Adobe engineers trying to match the camera JPEG.
Yes, the camera manufacturer conversion of RAW to in-camera JPG is proprietary. Adobe would have to reverse engineer. Adobe or all end-user software hides a lot changes made on their end.
But still, regardless of "translation" by Adobe or any other software - we are getting consistent, measurable results from different manufacturers in the same software. Adobe and others work closely with camera manufacturers to get this right. A Nikon RAW in LrC is way different than a Sony RAW in LrC, one example. Science is absolutely involved in this process.
I agree "color science" is not an accurate term, it is simply a hardware encoded way to represent colors. I am interested in how skin tones are represented. In case of Nikon d850 would argue against coolness, it errs on yellow side...
Since, I believe, Sony manufactures Nikon sensors that makes sense ;) But I'm just repeating what I've heard. My general non-professional opinion is all cameras can be matched up so you couldn't tell the difference in a blind test. From RAW of course.
Though I believe sophisticated data tests could tell sensors apart. A project I'd like to do at some point. There is software, but it's very expensive.
@@MaxoticsTV Well, Sony might make the chips, but there are numerous hardware-dependent steps involved-pre-ADC, ADC, and post-ADC-that can be influenced by design choices. Nikon claims to do its own design work in these areas, so their influence on the final image could still be significant.
Actually, objectively measuring the color output would be an excellent idea, especially under artificial lighting. While most cameras perform well in broad daylight, handling skin tones and complex artificial lighting conditions is crucial. The way different cameras convert these lighting conditions could reveal significant differences that are not often compared.
@@astraeusone Yes, I once argued that RAW is not manipulated by the ISO dial but someone said Nikon has better DR at very high ISOs (amplification). I did some tests and they were right. Now, I don't believe anyone would shoot at such extreme ISOs or that the differences were noticeable...still, proved what you just said. A lot of steps.
Even worse. The colour of brand X is very little more than the default setup for the Raw data interpretation. So like you say, there is no such thing as "Canon colours" or "Leica colours" etc. If you shoot Raw, the "colour" you see is nothing more than an interpretation of the data. Yet, people who know this still push this nonsense on their reviews
110% What I find fascinating is people don't realize how they're brainwashed by those who push that nonsense. When you try to show them the truth, like those in Plato's Cave, they get angry with you. Brand X is the best. Go away!
Thank you for a great explanation. We also do perceive colors differently. We see with our eyes and brains and the interpretation of colours depends on our memories and experiences.
Then explain to me please how I can easily point out Nikon / Sony / Leica etc RAW files. This is what photographers are referring to with "< brand > colors". Each manufacturer has its own proprietary formula for a given sensor to achieve consistent results, and this is measurable. This is exactly what people are referring to.
As someone who works with an aerial imaging solutions company and with over a decade of experience as a commercial photographer, I could attest that you've done a great job of putting across the point. When working with RAW images, there are options for the editor(read interpreter) to balance these different colors to output the final image. Commercial photographers and editors anyway perform color transforms on the images to achieve the desirable color palette for images based on color theory. The RAW image is a great starting point as far as I can see. I believe what the average person confuses with color science is color theory and its perception.