All the concepts that you explain in your videos; I have never understood them with such clarity before. Your narration and explanation technique is elegant. I was able to follow you. Your content is devoid of unnecessary flashy video effects, shouting, and overacting. So no distractions or bs. Please don't change your style. Massive respect. You are awesome!
this 16 min video took me one and half hour to complete. Never been into this so it took me time to really understand what you are teaching. My mind is blown brother. This is what I needed since don't know how many days.
If my screen is sRGB and it cannot show colors outside the triangle on CIE diagram then how can my screen show the full diagram? Shouldn't I see just a triangle?
The colors which are outside your display's gamut are shown as being the most saturated possible color of that hue. So the part of the diagram which corresponds to your display is accurate but the parts outside of it are not. The colors on the diagram are only used for visualization purposes
You are singlehandedly rendering it possible to understand comprehensibly subject matters that are core to an entire industry I intend to build a career in. THANK YOU ❤️
The human retina has two types of light sensing cells, rod and cones. The rods are only sensitive to a narrow range of wavelengths in the green region but significantly are nearly 3000x more sensitive to light intensity than the cones which sense red, green and blue, respectively. The rods cover the entire retina, but the cones are mostly concentrated around the optic nerve. The greater sensitivity of the rods and their location explain why at light we can see movement at the periphery of our vision better than in the center and why when our eyes fixate on objects we get “tunnel vision” which is the brain’s way of mentally tuning out the much stronger signal from the rods. The rod / cone physiology explains why the CIE*xyz graph used to represent color space is much larger in the green region and why staring at bright green colors becomes fatiguing. Red lighting is used at night on control panels because the rods don’t detect it making it easier for them to detect any objects in other areas viewed.
rods are sensitive of a broad range of wavelengths. The peak is near to cyan, and excludes reds, but the range is broad. Sensitivity or rods are not derived from a single cell, since a lot of rods are combined in a single ganglion. That means a collection of small sensitivities added together, since the small size of every rod and the huge population of them. Cones are concentrated in the fovea (not around the eccentricity of the optic nerve) and rods do not cover the fovea at all. Because of that you cannot see a star at night with the center of your vision. Movement definition of rods are related with the speed of the visual cycle of them, not from sensibility.
Correction: RGB is a non-perceptual color model, not a colorspace. HSL is a perceptual color model that takes into account human vision. Rec. 709 is a colorspace that defines primaries in locus points to derive relative coordinates to CIE 1935. You can convert to other colorspaces using relative or absolute colormetric intent with prioritizing white balance against saturation. (so even the transform conversion itself can be perceptual). Additionally the camera's RGB sensor array only exists in the world of a standard exposure (usually 18% grey) so it's data is not truly 3D in nature since camera's sensors give non-linear response to light. In essence, every camera has a custom color model. This is why it's hard to match cameras that have varying scene referred color over an exposure range. And adds to the reason why lookup tables do not work as well as anticipated.
HSL is "perceptual"? No, is just a projection model from an RGB model by using a very simple transform. Not a colorspace, just a different color "model". Camera sensor are possible because the photoelectric effect, which is "linear" in transduction by nature, that means all sensors are linear in response, but each can use a different encoding at the time of saving data only. Cameras has a custom "colorspace", not color model.
@@EdiWalger The HSL representation models the way different paints mix together to create colour in the real world, with the lightness dimension resembling the varying amounts of black or white paint in the mixture (e.g. to create "light red", a red pigment can be mixed with white paint; this white paint corresponds to a high "lightness" value in the HSL representation). Different cameras can have the same colorspace but different spectral response in the sensor. The spectral response over overexposure is what causes irregularities in matching cameras and that mapping process is non-linear despite transduction.
@@robertulrich3964 the QE of every sensor can be different by wavelength, and the colorspace is the result of the camera calibration that you need to reproduce real colors. Every wavelength stills with a linear (proportional to radiometric intensity) response, but changes for every wavelength. L (lightness) is an abstraction that must be defined mathematically, and never match with any "perceptual" concept, specially in the HSL projection: you can define a color with hue/sat but no lightness, which is impossible to realize. Because of that HSL is "derivative" from RGB that defines the real color mixing, and not a colorspace made from abstract attributes. Just an indirect manipulation for the RGB model which results in a non-perceptual match: color cannot be fully represented by euclidean dimensions.
I spent 3 hours on Wikipedia and a bit of RUclips trying to understand why if the 3 lights making up a pixel correspond to the 3 cones in the eye, my monitor can't reproduce the entirety of the visible spectrum. The CIE 1931 (I read that wikipedia page) and THEN the R G B color space put it together. I thought this was a really simple question when I started but wow, nope.
Well done. I've been a professional colorist for years... Now I teach color at a university in the film department... I was racking my brain for a good way to explain color spaces without overwhelming my students. This has helped exponentially.
I wish you could've shown more the difference between the colorspaces and any benefit of the gained range vs just a "it'd be more expensive". Great video otherwise.
11:53 The CIE1931 is not an "absolute colorspace", since Y means a percent of a "luminosity scene" estimated from a luminosity function (CIE1924). A real absolute colorspace uses cd/m2 for luminance, like ICtCp. But 1931 is an "absolute gamut", in terms of chromacity inside of an unknown dynamic range.
It's kinda cool that metametism isn't actually a phenomena of physical light, but a limitation of our 'three cone' based measurement tools :) While it makes sense that we use RGB sensors to account for this, I kinda wonder how useful it would be to use a different architecture to capture the light spectrum so the data can actually differentiate from a spectral 'yellow' yellow and a yellow created by red/green pollution. Things light scene light pollution could be targeted directly in colour finishing without effecting something which is actually supposed to be that colour.
Glad to see some objective science being applied to color theory. Note there is a huge difference between light color theory and pigment color theory. To start 6 Primary 3 Light >>> Red Green Blue 3 Pigment >>> Magenta Yellow Cyan Pigment Derived Red=Magenta×3+Yellow Green=Yellow+Cyan Blue=Cyan×3+Magenta 6 Secondary Pigment 3 Common >>> Violet=Cyan+Magenta Rose=Magenta+Red Orange=Red+Yellow >>> 3 Obscure >>> Chartreuse=Yellow+Green Aquamarine=Green+Cyan Cerulean=Cyan+Blue Other Purple = Blue+Red >>> >>> >>> >>> Rant ROY G BIV is a mistake and there are several issues. 01 Cyan is mislabeled as baby blue and Magenta gets left out most of the time 02 Indigo is not necessary, it's not even a secondary color. It's a dark dark violet that is more on the blue side. 03 Purple would make more sense than Indego, and fyi purple is not true violet. Purple = Blue + Red, Violet = Cyan + Magenta 04 Blue and Violet should have been separated correctly but instead indigo gets used 05 In light, Violet exists in two different ways, it has its own distinguishable frequency and wavelength, but it can also be created by a mixture of blue and red wavelengths (the human eye will except both versions and they are not distinguishable to us). This duality makes it a bit confusing in some situations. Note there is a huge difference between light color theory and pigment color theory. 06 Purple is a type of violet. Slightly darker with more red. Purple = Blue + Red, Violet = Cyan + Magenta 07 Indigo is a dark dark violet, closer to the blue side. 08 Violet is a wider spectrum (than the other colors) because it can be a mixture of blue and red light at different ratios. 09 Magenta has a similar issue as violet but magenta actually does not exist as a single wavelength/frequency in light. Magenta is the 1 to 1 ratio (1:1) of red and blue Light. Magenta does not exist on the rainbow but we can clearly see it's place in color theory. 10 Indigo and Purple are not necessary when considering primary and secondary colors (6 primary 6 secondary [derived]), they just fall under violet. 11 M ROY G CBV M, I distinguish 8 unique colors that we seem ti care about most, the 6 primary plus orange plus violet. 12 Aquamarine (green-cyan) is on the same level as orange and violet but in our culture we don't value it as unique. 13 If you only consider the rainbow then you would not get Magenta and the 8 would become 7 which is what you see in the rainbow ROY G CBV.
This is great content.❤❤ My questions are how do you identify color transformation / color space made by LUT ? What reference and tool should I learn if i Interested in LUT / look development? I'm learning color grading using davinci and intersted in look development and film LUT.
Amazing explanation, best i have seen.. what i dont understand, if our eye has three receptors why do we manage to get a parabola, and all technology with 3 colours have always a triangle. Sure with a triangle less mixing area is possible, but why is it even a triangle if biology already allows a parabolic contour form.
That video was really helpful for me! I'm glad you decided to create a better one! On another note, would you be able to make a video explaining monitor calibration types? Cheers!
Very understandable and useful. Good job! To further understand, why there are always two opposing frequencies that produce the same colour impression, one could add to 6:37: With the assymmetric overlapping of the susceptibility curves and the approx. proportional response of each cone to the combined weighted intensity of all frequencies, Grassmann's Laws hold as follows: Any vector of three visible spectral frequencies (f1, f2, f3) maps linearly to (S, M, L) spanning a convex subset of a 3d vector space. Linear algebra then shows, that due to linear independence in 3d, even after projecting out intensity (luminence) to a 2d subspace, you can counteract any shift of one spectral frequency with a shift of the other frequency to get the same S:M:L impression. Then you tweak the horseshoe shape to make the lengths all the line segments, that meet in the same color impression, to reflect their mixing ratios. And this, too, is possible because of linearity and smoothness of the cone response.
Hey, This video was very intuitive. I am actually doing a mathematics project on modelling a color space. Do you know a good source where you can find the corresponding spectral power distributions/metamers for given hues. Like a database that shows the spectral power distribution for many colors?
Could you explain the conversions that goes on in between a camera and computer, photoshop, monitor so I can understand when and how to deal with color profiles and conversions?
Those were in earlier drafts of this video, but I ended up cutting them because this video was pretty dense already. I'll most likely cover them sometime soon!
Great video. Really helps grasp the idea of a very mathematically implemented concept that is derived from a subjective experience. I would only mention that I believe the equation at 7:18 should swap the position of your transformation matrix and your LMS vector in order to make it a valid operation.
This is really great stuff. However there is an incorrect concept that is stated and referred to over and over that, as a Disney artist of 20 years, has me uncomfortable. The statement "these colors only exist inside our brains color isn't really a property of the physical world. It's just a subjective sensation that we experience in response to the light our eyes receive" is incorrect. The way this is presented in the video leads the user to believe that all color is fabricated in our minds and that it's all just a big guessing game where we hope other brains create the same color. Color is not truly subjective and is not just a sensation created only in/by our brains. Color is a real thing that does exist in the physical world in the form of different frequencies of light spectrum. Human eyes are capable of see a good amount of this light spectrum. If your human eyes are perfect then you are seeing the most true and accurate representation of these colors we know of. If you have imperfect eyes or are colorblind, for example, then your eyes are not seeing the true light spectrum that does exist in the world. Yes it is a "sensation that we experience in response to the light our eyes receive", and yes your exact observation only exists inside your specific brain because you're the only one with those particular eyes. But the light that is entering your eyes does so because it does exist in the physical world first. There is a true color in the world. Truth exists. Just because you can't see it or see it correctly doesn't mean it doesn't exist. So NO, colors DON'T only exist only inside our brains, and YES color IS really a property of the physical world. Truth exists outside your head, it's up to you to see it correctly. You don't make up your own truth. "Your truth" is a misnomer for "your interpretation", be it correct or not. These are fundamentals that can enormously effect your progress especially if you have a profession that involves creativity and color theory.
Thank you for your comment! My intention behind claiming that color only exists in the mind was to highlight the difference between the objective (light in the physical world) and the subjective (the sensation of color.) Yes, the subjective sensation is directly caused by physical light, but it's also true that different physical light can result in an identical sensation inside the brain. I think perhaps my choice of words wasn't the best. My goal was to get the viewer used to the idea that there may be a disconnect between physical light and subjective sensation, so that they'll understand why we need color spaces like CIE 1931 to perform a translation between the makeup of physical light and the color that light will create in a viewer's mind.
all of these are 2d colour spaces, you'll get poor gradients in these basic colour spaces. i want to know more practical application of 3d colour spaces like OKLAB
The 2D representations of CIE 1931 and it's derivative color spaces like Rec.709 are just simplifications of the 3D reality. Color is inherently three dimensional, so all the color spaces mention in the video are 3D as well
Great video, I'm on the hunt for good, concise, videos on this topic to share with some work colleagues. This knowledge is not taught in most software courses so is a bit of a black box abyss for most software engineers. I keep running into "two wrongs make a right" scenarios that appear out of nowhere when one colospace bug is fixed. Thank you for making this content. If there's not another video from you that I've not seen yet, I hope you cover gamma transfer functions as well, PQ, slog, sRGB, etc. Because that's arguably even more confusing than the colorspaces, especially when sRGB color primaries (Rec.709) and sRGB gamma transform are often used interchangeably online.
What I still haven't been able to find is if using YCBCR444 on my PC is noticeably inferior to RGB Full. I don't use YCBCR by choice, but more because using RGB has a weird bug that causes horrible corrupted uneven banding effects in gradients.
I hava a doubt. I know it's a dumb question. If we can only see the gamut of the colors in our monitors, how can we still see the whole CIE diagram. So they're not true colors what our monitors have been emitting.
Great explanation of a very complicated subject 🌟 love the graphics (!) really helpful with the visualisation process... 🌟 would you consider making a video focussing on the 'OKLAB' colour model >> I think that would be very interesting >> keep up the good work 👌💯👀🎯🖼😎🌠
So would it be accurate to say that the difference in color spaces is the size of the triangle (R, G, and B coordinates) overlaying the CIE 1931 blob graph?
Thanks for a great video 😃. It was educational, interesting and has cleared up many of the misunderstandings I had about color in my ThreeJS computer graphics application. You did a great job of explaining a complex topic!
Thanks a lot, Camon for explaining all the concepts in a crisp manner. Very much appreciated. You have explained these complex concepts in very simple maner with the diagrams.
So does this mean that TV output is just inherently nonsense to animals since the RGB system is so specifically tailored to human vision? Example: dogs have blue and yellow cones, but a RGB display won't show pure yellow, it will show red and green wavelengths together, both of which the dog has no cones for.
What your previous video does still contribute to this topic is the colormetric transformation needed between color spaces to ensure that what the camera saw is what you see on the monitor. Whether that transformation is absolute or relative is e.g. part of the answer to a question below on why someone still sees ‘all’ the colors of CIE diagram.
![NOSFERATU - Official Trailer [HD] - Only In Theaters December 25](http://i.ytimg.com/vi/nulvWqYUM8k/mqdefault.jpg)
Broooo you resume hundreds of hours of "masterclasses" in a single video. Thanks a lot!
Just found this channel. Gonna binge tf out of it lol
All the concepts that you explain in your videos; I have never understood them with such clarity before.
Your narration and explanation technique is elegant. I was able to follow you.
Your content is devoid of unnecessary flashy video effects, shouting, and overacting. So no distractions or bs.
Please don't change your style. Massive respect. You are awesome!
this 16 min video took me one and half hour to complete. Never been into this so it took me time to really understand what you are teaching. My mind is blown brother. This is what I needed since don't know how many days.
Holy shit I appreciate you so much for this. Incredible video, this is what I wish all of RUclips would be more like
this is the single video on the internet getting "what is color" almost perfectly correct. congratulations.
If my screen is sRGB and it cannot show colors outside the triangle on CIE diagram then how can my screen show the full diagram? Shouldn't I see just a triangle?
The colors which are outside your display's gamut are shown as being the most saturated possible color of that hue. So the part of the diagram which corresponds to your display is accurate but the parts outside of it are not. The colors on the diagram are only used for visualization purposes
Amazing summary - thank you very much!
7:17 the column matrix {L M S} should be at the right of the 3 * 3 matrix
You are singlehandedly rendering it possible to understand comprehensibly subject matters that are core to an entire industry I intend to build a career in.
THANK YOU ❤️
you'll soon find that virtually everyone misunderstands this subject and try to avoid it
great work thanks
The human retina has two types of light sensing cells, rod and cones. The rods are only sensitive to a narrow range of wavelengths in the green region but significantly are nearly 3000x more sensitive to light intensity than the cones which sense red, green and blue, respectively. The rods cover the entire retina, but the cones are mostly concentrated around the optic nerve. The greater sensitivity of the rods and their location explain why at light we can see movement at the periphery of our vision better than in the center and why when our eyes fixate on objects we get “tunnel vision” which is the brain’s way of mentally tuning out the much stronger signal from the rods.
The rod / cone physiology explains why the CIE*xyz graph used to represent color space is much larger in the green region and why staring at bright green colors becomes fatiguing. Red lighting is used at night on control panels because the rods don’t detect it making it easier for them to detect any objects in other areas viewed.
rods are sensitive of a broad range of wavelengths. The peak is near to cyan, and excludes reds, but the range is broad. Sensitivity or rods are not derived from a single cell, since a lot of rods are combined in a single ganglion. That means a collection of small sensitivities added together, since the small size of every rod and the huge population of them. Cones are concentrated in the fovea (not around the eccentricity of the optic nerve) and rods do not cover the fovea at all. Because of that you cannot see a star at night with the center of your vision. Movement definition of rods are related with the speed of the visual cycle of them, not from sensibility.
@@EdiWalger Thanks for the multi-spectral enlightenment 👍❤️
This is my new favorite channel!
This clears up so much. Such a great video from such a small channel
Goes to say people don't like it when it gets technical, but they miss out on good content.
I've been working on some videos in this area too, and man, you really nailed it! Nice job! Far and away the best content on YT on this topic
Correction: RGB is a non-perceptual color model, not a colorspace. HSL is a perceptual color model that takes into account human vision. Rec. 709 is a colorspace that defines primaries in locus points to derive relative coordinates to CIE 1935. You can convert to other colorspaces using relative or absolute colormetric intent with prioritizing white balance against saturation. (so even the transform conversion itself can be perceptual). Additionally the camera's RGB sensor array only exists in the world of a standard exposure (usually 18% grey) so it's data is not truly 3D in nature since camera's sensors give non-linear response to light. In essence, every camera has a custom color model. This is why it's hard to match cameras that have varying scene referred color over an exposure range. And adds to the reason why lookup tables do not work as well as anticipated.
HSL is "perceptual"? No, is just a projection model from an RGB model by using a very simple transform. Not a colorspace, just a different color "model". Camera sensor are possible because the photoelectric effect, which is "linear" in transduction by nature, that means all sensors are linear in response, but each can use a different encoding at the time of saving data only. Cameras has a custom "colorspace", not color model.
@@EdiWalger The HSL representation models the way different paints mix together to create colour in the real world, with the lightness dimension resembling the varying amounts of black or white paint in the mixture (e.g. to create "light red", a red pigment can be mixed with white paint; this white paint corresponds to a high "lightness" value in the HSL representation). Different cameras can have the same colorspace but different spectral response in the sensor. The spectral response over overexposure is what causes irregularities in matching cameras and that mapping process is non-linear despite transduction.
@@robertulrich3964 the QE of every sensor can be different by wavelength, and the colorspace is the result of the camera calibration that you need to reproduce real colors. Every wavelength stills with a linear (proportional to radiometric intensity) response, but changes for every wavelength. L (lightness) is an abstraction that must be defined mathematically, and never match with any "perceptual" concept, specially in the HSL projection: you can define a color with hue/sat but no lightness, which is impossible to realize. Because of that HSL is "derivative" from RGB that defines the real color mixing, and not a colorspace made from abstract attributes. Just an indirect manipulation for the RGB model which results in a non-perceptual match: color cannot be fully represented by euclidean dimensions.
Thanks for your useful comment.
You’re very didactic.thanks
youtube at its best!
Anybody know the name of the music piece at 0.40?
"In the hall of the mountain king"
I assure you this is not for beginners
I spent 3 hours on Wikipedia and a bit of RUclips trying to understand why if the 3 lights making up a pixel correspond to the 3 cones in the eye, my monitor can't reproduce the entirety of the visible spectrum. The CIE 1931 (I read that wikipedia page) and THEN the R G B color space put it together.
I thought this was a really simple question when I started but wow, nope.
Well done. I've been a professional colorist for years... Now I teach color at a university in the film department... I was racking my brain for a good way to explain color spaces without overwhelming my students. This has helped exponentially.
I wish you could've shown more the difference between the colorspaces and any benefit of the gained range vs just a "it'd be more expensive". Great video otherwise.
11:53 The CIE1931 is not an "absolute colorspace", since Y means a percent of a "luminosity scene" estimated from a luminosity function (CIE1924). A real absolute colorspace uses cd/m2 for luminance, like ICtCp. But 1931 is an "absolute gamut", in terms of chromacity inside of an unknown dynamic range.
It's kinda cool that metametism isn't actually a phenomena of physical light, but a limitation of our 'three cone' based measurement tools :) While it makes sense that we use RGB sensors to account for this, I kinda wonder how useful it would be to use a different architecture to capture the light spectrum so the data can actually differentiate from a spectral 'yellow' yellow and a yellow created by red/green pollution. Things light scene light pollution could be targeted directly in colour finishing without effecting something which is actually supposed to be that colour.
I am not even halfway through. This video is incredible!
Keep up the good work!
brilliant, thanks for helping me build knowledge on a strong base
Very good presentation! Keep it up
Glad to see some objective science being applied to color theory.
Note there is a huge difference between light color theory and pigment color theory.
To start
6 Primary
3 Light >>> Red Green Blue
3 Pigment >>> Magenta Yellow Cyan
Pigment Derived
Red=Magenta×3+Yellow
Green=Yellow+Cyan
Blue=Cyan×3+Magenta
6 Secondary Pigment
3 Common >>>
Violet=Cyan+Magenta
Rose=Magenta+Red
Orange=Red+Yellow
>>>
3 Obscure >>>
Chartreuse=Yellow+Green
Aquamarine=Green+Cyan
Cerulean=Cyan+Blue
Other
Purple = Blue+Red
>>> >>> >>> >>>
Rant
ROY G BIV is a mistake and there are several issues.
01 Cyan is mislabeled as baby blue and Magenta gets left out most of the time
02 Indigo is not necessary, it's not even a secondary color. It's a dark dark violet that is more on the blue side.
03 Purple would make more sense than Indego, and fyi purple is not true violet. Purple = Blue + Red, Violet = Cyan + Magenta
04 Blue and Violet should have been separated correctly but instead indigo gets used
05 In light, Violet exists in two different ways, it has its own distinguishable frequency and wavelength, but it can also be created by a mixture of blue and red wavelengths (the human eye will except both versions and they are not distinguishable to us). This duality makes it a bit confusing in some situations. Note there is a huge difference between light color theory and pigment color theory.
06 Purple is a type of violet. Slightly darker with more red. Purple = Blue + Red, Violet = Cyan + Magenta
07 Indigo is a dark dark violet, closer to the blue side.
08 Violet is a wider spectrum (than the other colors) because it can be a mixture of blue and red light at different ratios.
09 Magenta has a similar issue as violet but magenta actually does not exist as a single wavelength/frequency in light. Magenta is the 1 to 1 ratio (1:1) of red and blue Light. Magenta does not exist on the rainbow but we can clearly see it's place in color theory.
10 Indigo and Purple are not necessary when considering primary and secondary colors (6 primary 6 secondary [derived]), they just fall under violet.
11 M ROY G CBV M, I distinguish 8 unique colors that we seem ti care about most, the 6 primary plus orange plus violet.
12 Aquamarine (green-cyan) is on the same level as orange and violet but in our culture we don't value it as unique.
13 If you only consider the rainbow then you would not get Magenta and the 8 would become 7 which is what you see in the rainbow ROY G CBV.
I learned a ton binging your whole channel a few months back. So excited to see another explainer from you. Great content!
You deserve millions of views dude….. stay consistent 👍🏻
zero views*
This is great content.❤❤
My questions are how do you identify color transformation / color space made by LUT ? What reference and tool should I learn if i Interested in LUT / look development?
I'm learning color grading using davinci and intersted in look development and film LUT.
Amazing explanation, best i have seen.. what i dont understand, if our eye has three receptors why do we manage to get a parabola, and all technology with 3 colours have always a triangle. Sure with a triangle less mixing area is possible, but why is it even a triangle if biology already allows a parabolic contour form.
My camera shoots only natural, flat, sepia and etc. profiles in 8bir. There are no log or raw. So how do I convert my footage to proper rec.709?
That video was really helpful for me! I'm glad you decided to create a better one!
On another note, would you be able to make a video explaining monitor calibration types? Cheers!
Very understandable and useful. Good job!
To further understand, why there are always two opposing frequencies that produce the same colour impression, one could add to 6:37: With the assymmetric overlapping of the susceptibility curves and the approx. proportional response of each cone to the combined weighted intensity of all frequencies, Grassmann's Laws hold as follows: Any vector of three visible spectral frequencies (f1, f2, f3) maps linearly to (S, M, L) spanning a convex subset of a 3d vector space.
Linear algebra then shows, that due to linear independence in 3d, even after projecting out intensity (luminence) to a 2d subspace, you can counteract any shift of one spectral frequency with a shift of the other frequency to get the same S:M:L impression. Then you tweak the horseshoe shape to make the lengths all the line segments, that meet in the same color impression, to reflect their mixing ratios. And this, too, is possible because of linearity and smoothness of the cone response.
Hey, This video was very intuitive. I am actually doing a mathematics project on modelling a color space. Do you know a good source where you can find the corresponding spectral power distributions/metamers for given hues. Like a database that shows the spectral power distribution for many colors?
Thank you very much, Camon, for your excellent videos. You are a very cogent thinker and presenter.
Great video! what software did you use to make the cieLab visualization?
Thanks!
amazing video
Could you explain the conversions that goes on in between a camera and computer, photoshop, monitor so I can understand when and how to deal with color profiles and conversions?
Best of the best. Thank you .
Amazing video, it deserves more views. I actually didn't know what color spaces were and now it is super clear. Thanks for the explanation!
At 9:12 I have flash backs to calculus 2 calculating volume using cross section area... Good times
excellent!
thank u sir!! Thx for educational info!!
Amazing
Thank you so much for it. And, please add CIELAB and HCT also brother.
Those were in earlier drafts of this video, but I ended up cutting them because this video was pretty dense already. I'll most likely cover them sometime soon!
@@VideoTechExplained Lots of love from this side man. I really appreciate your hard work!
How to calculate color matching x bar, ybar ,z bar
Bro knows math I love it
Great video. Really helps grasp the idea of a very mathematically implemented concept that is derived from a subjective experience. I would only mention that I believe the equation at 7:18 should swap the position of your transformation matrix and your LMS vector in order to make it a valid operation.
Great video dude! I learned a lot! You might need to find someone to iron your shirt though lol.
I'm not the video is completely accurate but very good over all.
Thank yiou Camon, this is insightful information
I hereby declare you, King of Color Nerdism! Congrats and well done! Love your channel
Chapeaux, very good. Good video well done.
This is really great stuff. However there is an incorrect concept that is stated and referred to over and over that, as a Disney artist of 20 years, has me uncomfortable. The statement "these colors only exist inside our brains color isn't really a property of the physical world. It's just a subjective sensation that we experience in response to the light our eyes receive" is incorrect. The way this is presented in the video leads the user to believe that all color is fabricated in our minds and that it's all just a big guessing game where we hope other brains create the same color. Color is not truly subjective and is not just a sensation created only in/by our brains. Color is a real thing that does exist in the physical world in the form of different frequencies of light spectrum. Human eyes are capable of see a good amount of this light spectrum. If your human eyes are perfect then you are seeing the most true and accurate representation of these colors we know of. If you have imperfect eyes or are colorblind, for example, then your eyes are not seeing the true light spectrum that does exist in the world. Yes it is a "sensation that we experience in response to the light our eyes receive", and yes your exact observation only exists inside your specific brain because you're the only one with those particular eyes. But the light that is entering your eyes does so because it does exist in the physical world first. There is a true color in the world. Truth exists. Just because you can't see it or see it correctly doesn't mean it doesn't exist.
So NO, colors DON'T only exist only inside our brains, and YES color IS really a property of the physical world. Truth exists outside your head, it's up to you to see it correctly. You don't make up your own truth. "Your truth" is a misnomer for "your interpretation", be it correct or not. These are fundamentals that can enormously effect your progress especially if you have a profession that involves creativity and color theory.
Thank you for your comment!
My intention behind claiming that color only exists in the mind was to highlight the difference between the objective (light in the physical world) and the subjective (the sensation of color.) Yes, the subjective sensation is directly caused by physical light, but it's also true that different physical light can result in an identical sensation inside the brain.
I think perhaps my choice of words wasn't the best. My goal was to get the viewer used to the idea that there may be a disconnect between physical light and subjective sensation, so that they'll understand why we need color spaces like CIE 1931 to perform a translation between the makeup of physical light and the color that light will create in a viewer's mind.
Very nice video. Instant subscribe.
all of these are 2d colour spaces, you'll get poor gradients in these basic colour spaces. i want to know more practical application of 3d colour spaces like OKLAB
The 2D representations of CIE 1931 and it's derivative color spaces like Rec.709 are just simplifications of the 3D reality. Color is inherently three dimensional, so all the color spaces mention in the video are 3D as well
What is a color outside of the xyz color space?
how to convert the cy cx value to colour temperature?
Absolutely fantastic video.
Could you please recommend a really good text on color theory. I am not afraid of the maths. Thanks.
Wikipedia.
Linus Tech Tips vibes, good video I like
Perfect 👍👍
I'm wondering about the matrix transformation at 7:17. Any additional information on that? Is it based upon human vision?
Found it: en.wikipedia.org/wiki/CIE_1931_color_space#Meaning_of_X,_Y_and_Z
Man this is uni level quality of education
How did equation at 7:21 was derived?
Thanks
Thank you
This kid is a riot! Bravo!
Great video, I'm on the hunt for good, concise, videos on this topic to share with some work colleagues. This knowledge is not taught in most software courses so is a bit of a black box abyss for most software engineers. I keep running into "two wrongs make a right" scenarios that appear out of nowhere when one colospace bug is fixed.
Thank you for making this content. If there's not another video from you that I've not seen yet, I hope you cover gamma transfer functions as well, PQ, slog, sRGB, etc. Because that's arguably even more confusing than the colorspaces, especially when sRGB color primaries (Rec.709) and sRGB gamma transform are often used interchangeably online.
🔥🔥
Your content is such a hidden gem in RUclips ocean.
What I still haven't been able to find is if using YCBCR444 on my PC is noticeably inferior to RGB Full. I don't use YCBCR by choice, but more because using RGB has a weird bug that causes horrible corrupted uneven banding effects in gradients.
I hava a doubt. I know it's a dumb question.
If we can only see the gamut of the colors in our monitors, how can we still see the whole CIE diagram. So they're not true colors what our monitors have been emitting.
The colors outside the gamut in the CIE diagram are not correct. They are just approximations, the best that can be shown on your monitor.
Thank you for the confirmation.
brilliant
Great explanation of a very complicated subject 🌟 love the graphics (!) really helpful with the visualisation process... 🌟 would you consider making a video focussing on the 'OKLAB' colour model >> I think that would be very interesting >> keep up the good work 👌💯👀🎯🖼😎🌠
So would it be accurate to say that the difference in color spaces is the size of the triangle (R, G, and B coordinates) overlaying the CIE 1931 blob graph?
Thank you! I wasted a hour of my life trying to decypher the Wikipedia page and still don't understand it. This video explains everything immediately!
Excellent, simply excellent!!!
I expected a British accent when I saw the thumbnail.
Not for me since he looked like Scott the woz.
Holy shit this was so thorough
What a great video!!!!!!!!!!!!!!!
great content. I need a summary of that topic to remember school ^^' And your video is accurate and right ! :)
I finally understand where the CIE 1931 color space got its shape from. Thanks!
Thanks for a great video 😃. It was educational, interesting and has cleared up many of the misunderstandings I had about color in my ThreeJS computer graphics application. You did a great job of explaining a complex topic!
Insane video
Best explanation!
Congrats and thank you for the explication. Really well done
Thanks a lot, Camon for explaining all the concepts in a crisp manner. Very much appreciated. You have explained these complex concepts in very simple maner with the diagrams.
Great explanation.
Thank you. I will be returning to this video for many reviews until I firmly understand and always remember.
This is great!
absolutely loved it's one of the most helpful video i've watched in 2023 thanks for creating this mate
So does this mean that TV output is just inherently nonsense to animals since the RGB system is so specifically tailored to human vision?
Example: dogs have blue and yellow cones, but a RGB display won't show pure yellow, it will show red and green wavelengths together, both of which the dog has no cones for.
I'm not an expert in animal vision, but I'd expect so, yes!
Excellent video, really helpful and insightful. Thank you!
What your previous video does still contribute to this topic is the colormetric transformation needed between color spaces to ensure that what the camera saw is what you see on the monitor. Whether that transformation is absolute or relative is e.g. part of the answer to a question below on why someone still sees ‘all’ the colors of CIE diagram.
12:28 Prisoner 709