How the hell does color correction work?? Mega reverse tutorial

lol, I just watched a man solve a problem I don't have... for an hour.

taran: is this balanced??? what does that mean vsauce: vsauce, michael here. what IS balance?

I can't wait for the day you figure it all out and make a mega 10 hour tutorial

Nothing in this video is relevant to my life at all yet I am still going to watch it.

Isn't there specialist about this out there somewhere? Or am I watching someone becoming one.

Hey Taran! I'm a full-time finishing colorist who works on TV, film, and web content (I also produce tutorials on my channel). To answer your questions in order: 0 - I know how frustrating this can be. It's important to understand two things about coloring. Firstly, at the end of the day, our job is to make the shot look good. When it comes to the "why" of many adjustments, you'll generally find a subjective answer (in most cases, not all). Secondly, every single project you grade is different. For this reason, understanding HOW the tools work makes the WHEN and WHY self-evident. That's the hope anyway. Grading is just visual problem solving, and so the tools you reach for are directly related to what problems need to be solved, and how you think they can be solved most effectively. Everyone solves problems differently, and it's why experienced colorists get paid big bucks to work their magic. There's no black and white, "one size fits all" approach. 1- There are few objective standards for what qualifies a balanced shot. In general, white balance should be neutral, and contrast/exposure should be what compliments the subject best. IMPORTANT: Consistency is much more critical than objective values. Whatever values you decide are appropriate for a scene should be used consistently throughout the scene. 2- You're right here. Clipped highlights in general should be kept at 100 IRE for the reasons you stated. You could bring down your midtones a little while keeping highlights at 100, but there's a limit to how far you can push the signal. This highlights the importance of good lighting in video, and making sure that levels/contrast ratios are where you want them to be in camera. 3- When you're looking for which control to grab here, you need to look at where the color cast is present in the signal. Is the color cast just in the highlights? Is it more in the mids? Shadows? In this case, the color cast looks like it's present throughout the entire signal (notice how the entire blue channel is lifted up), so I would tend to use Offset here. Robbie probably chose to use Gain because he thought the color cast was present more in the highlights than in the mids or shadows. 4- The behavior of the LGG wheels in Resolve is documented in the manual, but I'm not sure about Lumetri's. In general, each program has its own slight version of the wheels and curves, but they all behave similarly. 5- You need to use what's called a secondary correction here. Robbie is pushing the overall shot towards blue, then he's keying his warm tones and shifting them back towards orange. You won't be able to fully recreate this grade just using global adjustments. I have a tutorial on my channel that details how this is done in Resolve (it can be done in Premiere as well). 6- Few things here. First, you need to use secondary corrections to get the look as mentioned above. Second, the T/O look is lighting dependent, and it usually looks best with darker, higher contrast shots. The shot you're using here isn't an ideal candidate (in my opinion); it's too flat and evenly lit. 7- GIGO (Garbage In, Garbage Out). Consumer grade cameras don't capture nearly as much information as high end cinema cameras, so trying to reproduce the look of a high-end production with a "compromised" signal is an excercise in frustration. As colorists, we can only work with the information we're given; the integrity of the original video signal lays the foundation for everything we do. A GoPro will never deliver the same image as an Arri Alexa, no matter how hard you try. 8- Footage that is lit, exposed, lensed, and composed exactly the way the DP wanted it to look. The less you have to grade, the better the final product will be as a general rule. Whatever is in front of the camera is WAY more important than the technical specs of the camera itself. 9- See above. 10- Short answer: no. To perfectly match the two cameras, you would need to have the same information in both signals. Because the GoPro is capturing significantly less info than the Alexa, there's no LUT or magic correction that will bring that info back: it was simply never recorded in the first place. You may be able to get certain shots to look perceptually close through careful grading, but the success of your efforts would be highly dependent on the scene (it wouldn't work in every case). 11- I'm sure others have answered this one. 12- Bias lights are designed to control the viewing environment inside your coloring suite. For this reason, the color accuracy of a bias light is critical, as an imbalanced light will perceptually skew the colors on your monitor. For example, a low-quality bias light with a green color cast will make your reference display look magenta by comparison. A high-CRI rating ensures not only a fuller spectrum, but a more accurate match to its rated white balance. 13- Yes. CRI both refers to the completeness of the light spectrum, and how balanced different wavelengths are compared to one another. A higher CRI light ensures better color reproduction, and fewer unexpected color casts in the final shot. 14- Even though it seems counterintuituve, it matters. The RGB photosites on an image sensor are stimulated by many wavelengths of light, not just red, green, and blue. For this reason, colors in between RGB will be expressed more vibrantly under a high-CRI light than a low-CRI one. In general, you will see fuller and more accurate color reproduction under a high-CRI light, even when using an RGB sensor. Continued in my reply...

Thank you, now I have more questions than I started when watching this

I appreciate this video for the fact that you're being very up front and honest about what you do and don't know. I think some of the creators of many of the more amateurish tutorials probably don't have definitive answers themselves but instead of saying as much they just kinda go "the contrast and the dynamics...balance mumble mumble" and we're all just supposed to know exactly what it all means. Ignorance perpetuating ignorance.

I love how Taran not only asks all the questions I'm too afraid to ask, but also, a) asks a ton more than I'd ever think of asking and, b) teaches me a lot just by asking those questions.

I am film/video editor with an MFA in Film, I have been working for years and still haven't learned everything. They don't even teach it at Film School. Love your stuff by the way.

I'm a biologist and I think you could be an outstanding scientist. I don't know anything about video editing yet enjoy watching your stuff just because of how thorough, organized and clear your are. I wish you the best, now and always.

My answer to 24: The black point on HDR and SDR is the same. The difference is that more bits are allocated to the darker part of the image in HDR, allowing for more near black detail. I HIGHLY recommend you contact Vincent Teoh from HDTVtest if you have any more questions about HDR, EOTF curves, or display calibration.

This channel feels like math major trys art

UPDATE: There's a reddit thread, now: www.reddit.com/r/TaranVH/comments/bbduq2/how_the_h… The response to this has been significantly larger than I expected! It's taking a long time just to read and watch and discuss everything that everyone has contributed. But, I do intend to get to all of it. Thanks, everyone, for your help! Next time, I'll ask a lot fewer questions all at once!

I have no experience with colour correction. But from what you said, it seems like colour correction is like cooking, whereas you want it to be like chemistry. i.e. there are no absolute right answers, it's an art form, everyone has their own style which is refined through practice, you can make something new/different that some people will like and some will not

I can't comment the doc. Here is what I know about CRI: "Why do bias lights need to have a high CRI? (I’m not color correcting the wall!)" The light is a reference point for you. Let's say you are not just correcting, but you are creating a look.  When you make your image warmer to fit your narrative, after a while your eyes will adapt to the color shift, and your picture appears more neutral, than it actually is. Thats also why Editing software is usually in a neutral grey color. Professional colorists usually work with a big reference monitor, not within an editing program like Premiere, thats why they defiantly need a neutral grey wall behind there monitor, lit by a very accurate light source. Also the brightness of this light is very important, since for movies you often don't use the full range of brightness. Instead you sometimes have some very dark scenes to fit your narrative.  I would argue, for your kind of content, you should use your full Rec709 range, and keep caucasian skin tones at a brightness of around 70% (which is broadcast standard). "Is a high CRI important for the lights used on a set? Is higher always better?" Yes, generally the higher the CRI Index, the better. A CRI of 100 ist a perfect light source (for 5500K the sun is the reference point of 100, for 3200K a tungsten bulb is the reference point for 100 CRI.) CRI is measured, by comparing an ideal light source, with the light source you're testing. To do so, you use the light source on a test chart with 15 Color fields. https://cdn2.bigcommerce.com/server2300/43185/product_images/uploaded_images/high-cri-test-data-r-values.png?t=1455279594 Compared to an ideal light source, some color fields will no be as vibrant. For LED lights the R9 color field is usually quite low. While CRI can give you an indication, how good your light source is, it's important to understand: 1. CRI focuses on how a human eye sees this test chart, not how a bayer sensor sees it 2. these 15 color fields are not enough to make sure, your light source will look good in a camera. 3. Since every CMOS Chip manufacturer uses different filters for there red, green and blue sensels, each camera reacts different to a non-continuos spectrum of light. Since CRI is an important factor in lights for your living area, it saddens me, that LTT does not give a fuck about it, when reviewing light products (such as your Xiaomi Smart Bulb review) To judge how good a light is for filming, there are other ways. (as you pointed out TLCI) " If it’s important for a light source to have a continuous, rather than a discrete spectrum, but a camera only records the frequencies of R, G, and B anyway, then why does it even matter?" You do not necessarily have to have a continuous spectrum for a good image. If a led light is made developed to look god with your camera (most manufacturers make sure, their light looks good when filmed on Arri Alexa), it can look as good as an ideal light source. However: you cannot just use 3 coherent light sources in Red green and blue to light your scene. if an object reacts to a wavelength, that is not present in your light source, this object will be darker, or the colors will be shifted towards another color, since other wavelength are present in your light source, and will reflect of your object. Yes, your camera does capture the image through R G B, but a person, that is lit with a low CRI Led light, will look bad, since certain light-wavelength are missing, so the skin does not reflect light in a way in would with an ideal light source. In other words: Skin tons will look bad. When you illuminate a white surface, a low CRI light can look as good as an ideal light source, as long as Red greed and blue within its spectrum are balanced properly. But as soon as you light objects with colors, a low CRI light, will make everything look wrong, not just skin tones. But you do notice it mostly on skin tones, since humans notice much easier, if skin looks unnatural. In case this was confusing, let's say, you have an Led light source, that has a big spike in blue at around 450nm, a broader spike at 540-660nm: https://www.hydrohobby.co.uk/media/wysiwyg/porto/products/information/specturm-king-602-spectrum-hydrohobby.jpg lighting a white wall with such a light can look great an balanced. Bit since not all colors are luminated with the same intensity, other objects or people can look wrong. And this can usually NOT be fixed in post. "If a light source with a very low CRI is desirable for a specific shot, due to the monochromatic look that it would create,, wouldn’t it give you more control to just go with a higher CRI when filming, and then make the shot more monochromatic during color correction/grading?" This is hard to answer. Generally I believe, if you know what look you are after, try to achieve it while filming, if possible. If you are unsure, or want to be as flexible with your material as possibly, light your scene normally and do the effect in post. Footage shot with a Sodium lamp, can't be recovered. To make normal footage look similar to a Sodium pressure lamp is possible, but not perfectly, since a Sodium lamp will excite different colors in your subject. Desaturating and tinting the picture yellow will NOT look exactly as getting the look "in camera" I hope my text is comprehensible, If you don't understand something please ask!

After taking me down the Macro rabbithole and killing 1 year of my life, I am afraid to watch this video and get involved. Deja vu.

I'm about to watch this whole video. Not because it's knowledge I'll ever use, but something about the way Taran talks and explains things is so captivating.

Colour correction and grading are both extremely subjective. I tend correct, not grade my projects. I like to make humans and subject matter to look warmly lit and well exposed. These are all very loaded questions, so hopefully my super brief answers don't just raise even more frustrating questions. Answers to questions: 1. An exceptionally tricky shot. If you just try to "correct" the shirt (try to darken the shadows to make the shirt closer to black), his face ends up getting too dark. The fact that this is mostly backlit and lacks a fill light really makes this one tricky. 2. I'd attempt to remove the color cast from the bad white balance. It's still going to have a bit of a tungsten tint overall because of the lighting used in the factory (as getting the greys to look true grey just makes everything look really unnatural to my eyes). I avoid the eyedropper and Temperature sliders in Lumetri because they actually change the tint of pure white when working in 8bpc (which I usually do). I almost never want my blown out whites to not stay pure white. I personally lean towards using curves to remove color cast, or adjust the Midtone of the three-way colour corrector (not highlights, which also seem to tint pure white). The footage still felt dark so I bumped up the gamma/midtones. You can choose to use curves or sliders. There are quirks to each method and a lot of it comes down to personal preference. How do you feel most in control? 5. I feel like you honestly got really close to matching. 6. Teal and orange: Your original footage is very saturated, and has very saturated primary colours, and I do believe that's the biggest issue you're running into. Teal and orange is a stylistic look, and I do not believe that the amount of teal/orange in the original shot matters at all, and white balance should not be much of an issue in most cases. 7. I'm not Robbie, but working with GoPro footage myself, it's a combination of compression and 8bpc. You have much less to work with before you get artifacts, and Skyfall is incredibly punchy and saturated. GoPros also have almost nothing in terms of traditional manual camera settings, which could be a factor. 10. GoPro LUTs? Depending on the footage for a scene, yes. If most of the scene is shot in the same lighting conditions and is exposed similarly, a LUT will work. Different scenes have different looks, and different takes will need differnet approaches. 23. You are correct that you would lose information shooting in LOG for 8-bit. That being said, the control you gain when shooting flatter may outweigh the loss of information, depending on what you plan to do. GoPro has a flat mode that is nowhere near as flat as the example you showed, but it does give you a bit more control without much notable loss (in my experience). 26. Converting to B&W is 100% subjective. There is no "wrong" way. What gives you the best results just depends on what you're trying to accomplish. Using the newish Hue vs Luma in Lumetri to adjust the luma of specific colours before a desaturate should actually give you even more control than Photoshop's B&W because you can target any specific hue: not just Reds, Yellows, etc. 32. Not possible to use bezier handles in curves. I have a good theory as to why, but no room to type it here, so ask if you're curious 33. - Blown out highlights should stay pure white IMO. - the algorithms behind the scenes for sliders like "Blacks" and "Whites" are very tricky and I feel like you actually get more control using those specific sliders because they target such a small range of shades. Try to adjust just the top 10% of whites using curves while trying to keep the rest of the curve a straight line (aka unaffected). It's unreasonably difficult to do. Generally I find curves more powerful though overall, yes.

I can't decide if you should never, or definitely, take a photography class.