*Physical: What we actually see
What the camera sees
*Psychological: What we think we see
*Cultural: What we are taught to see (Skin tone, etc)
Cameras see and capture more of the electromagnetic spectrum than we do , and they capture it literally.
The human eye brain interprets the information from the eye, and the colorist interprets the information from the camera.
1666 Sir Isaac Newton identifies the components of white light.
Richard of York Gained Battles in Vain.
Nobody really knows why he catalogued Indigo.
But it helped his analogy of colors represented by notes in a musical scale.
So What Color is it?
Depends on color of the light source
Depends on surface reflection and absorption of the object
Depends on the sensitivity of the eye
Colors do not have clear definitions
The lens of the eye restricts low frequencies (Blue) light and the atmosphere scatters blue light
Magenta is not a single frequency.
It is a compound of red and blue light.
Yet we see it as a single color.
Don't discuss with the client about the physical color. Talk about the feeling and emotion.
The brain is not allowing us the see physical color. We need to manipulate how people interpret color.
How Do we see Color?
We need light to see.
Objects can obstruct, absorb or reflect light.
The eye deciphers the light as colors.
Reflect red See Red
Absorb blue and green
Does light change color?
Not all light is white.
If the source light is filtered, the reflected light is also filterd.
Color temperature of Light:
The "whiteness" of light is defined by its color temperature and measured in degree Kelvin (K)
In theory a "black body" emits no light.
Heating this black body, adds energy, and creates light.
The heat in Kelvin defines the color of the black body.
Heating an iron bar makes it red hot…then yellow…then blue.
Color Temperature is a simple way to define the color white daylight
5500k----- Sun light, Cinema Projection standard (D55)
6500k------Daylight, TV standard. (D65)
6500k------Skylight. Hazy sky.
Color Temperature of Light
The D55 and D65 curves are useful for us because they display a flatter responser than other ranges.
Remember the sensitivity of the eye is different to that of a camera
changing the camera white balance alters the expected color temperature. This will have the reverse effect on the image!
If a scene has a color temperature of 5500 and the camera is set to 6500k, the image will appear warmer as if it were shot nearer sunset.
What color tools could achieve this?
Fluorescent lamps have irregular spectral distribution
The illumination has frequency spikes.
What color tools would correct these?
Impression - Visual Response
The Retina consists of Rods (Black and white) and Cones (Red Green Blue)
About 8% of men, but less than 1% of women have genet color blindness
Red and Green cones overlap more than blue. This minimizes chromatic aberration and increases acuity.
Red-blind persons are unable to distinguish between red and green,
Blue-blind persons cannot distinguish between blue and yellow,
Green-blind persons are unable to see the green part of the spectrum.
We see less red at night. Blue will fall out less.
Our eyes are attracted by motion. Then the contrast and color.
Our eyes are attracted by Red>Yellow…..
The dark corner is more for television, because it helps to focus on the center.
But it is not necessary for cinema. Because it is much bigger screen and you don't need to help your audience to focus.
CIE Chromacity Chart
Plotting the visible spectrum graphically shows all colors visible to humans.
RGB systems such as monitors cannot reproduce all of the possible colors.
Other color Models
CMYK used for printing
YUV used for television
XYZ is a theoretical "Maximum" space
Lab color gamut > RGB color gamut > CMYK color gamut
Red Green and blue is our first color space model
RGB systems include the eye, film, lighting, monitors, and telecines
RGB systems use additive color theory (R+G+B=white light)
Colorists refer to RGB controls as primaXXXX
Lab Color model
Defines visible colors as Hue, Saturation and Luminance
Colorists call HSL tools "Secondary"
There are no blue - yellows or red- greens
Not all hues has equal in saturations.
Munsell Color Model
The perfect color sphere is untrue
Hue is related to luminance
There are no dark yellows!
Orange and teal is the most signifiant contract.
Different colors have different characteristic. But the color correcting tools treated all the color equally.
Color are related to luminance
Actual color do not fit a regular sphere- Munsell described them as a tree.
There are no dark yellows.
(Munsell color tree, Munsell color system)
(Miholta 3D system)
The Red and canon is different because the red capture more amount of information of the colors.
At low light levels blue and green objects appear brighter (Purkinje Shift)
Rods are dominant so saturation is down too
(Useful for "Day for night")
Chromatic aberration causes different wavelengths to be focused at different planes within the eye
That means blues recede and red advance
Blue is good for backgrounds
The eye becomes saturated and sees an after image
The eye compensates to find a neutral color balance
Colorist needs to do job fast, or the eyes will adjust itself and make thins go wrong.
You need to use only one preference through out the work.
Hue, Saturation and luminance are all interactive with each other.
Hue, Saturation and luminance are relative to content and lighting
Color perception depends on context
Image composition affects perception
adjacent colors affect perception (color contrast)
There are 7 types of color contrast to consider
contrast of hue
Differentiation of Hue
At least 3 colors
Black heightens hues
Contrast of light and Dark
Light advances from dark
Fewer hues enhance the effect
Shadows take longer to read
Contrast of Cold and Warm
Warm advanced from cold
skin tones stand out
red stimulates adrenalin
blue haze looks far away
The color of warm is also relative
Orange is the warmest color
it is most luminant
it is most saturated
Blue is the coolest color
it is dark
it has low saturation
Colors are most vibrant against their complimentary hue
complimentaries are opposite sides of a vectorscope
adds balance to an image
will also be contrast of cold and warm
If the complimentary is missing the eye compensates
we see colors that are not there!
clouds appear yellow against blue
blacks appear blue against sepia
The ambient light conditions change our perception
Contrast of Saturation
Saturation stands out against saturated areas
reducing chroma in everything else is as effective
Contrast of extension
ratio of colors to create harmony or balance
Yellow=3 (Maximum purity)
What we think we see.
*Pixels Picture detail
*Bit Depth Color depth
*Dynamic Range realism
Film has bigger dynamic range than the digital videos.
Why size matters?
2K is a quarter of 4K,
The difference between 2K and HD is not so great.
Horizontally, but not vertically.
And the SD is very small.
The more resolution, the file is going to be bigger, and it contains more infomations.
What 's resolution?
*pixel = an abbreviation of "picture elements"
*Resolution = pixel in a single horizotal scan line
*Definition/Format = lines in a video frame = vertical pixels in a data
Benefits of Higher Resolution
Higher resolution, you gets much more details and infomation.
And higher resolution, the better compress result you will get.
Colorist always want to get a full resolution of the project.
Compress is different kind. one is lost, the other is lostless.
For the lostless, when you compress it, the size is becoming smaller, but you don't lose the information.
You need to know where is it come from.
2K image scan from a 4K head is more fantastic than 2K scan with 2K head.
2K image scan from a 4K head is almost the same as 4K from 4K head.
What is Bit Depth:
Bit = "Binary DigIT". A single element (1 or 0) of digital information
Bit depth = Color resolution. The number of available colors
One bit = either black (0) or white (1)
Two Bits =(00 to 11) displays 4 colors 2*2
8 bits = 2*2*2*2*2*2*2*2m seeb as 256 digital steps
How many colors?
10 bits= 4 times 8 bit, 1024 grays or 1 Billion colors
14 bits= 16 times 10 bit, 16384 grays or 4 TRILLION colors
Video is 8 or 10 bit.
Film is analog, but estimated as 12~14 bit
color correctors should be 16 bit or more
Are all bits equal?
No...interger processing is destructive
...floating point is not
Original Scene Negative Print density projection Television
8000:1 30:1 2000:1 200:1 20:1
Bit Depth represents the number of digital steps available
Dynamic Range represents the distance between the first and last step
Data has a greater dynamic range than video
Film - 10 stops usable
Video - 7 stops usable
Effect of Dynamic range
At normal exposure all cameras perform well
At +2 stops exposure the 10 bit linear YUV HD camera cannot capture highlights
At +4 stops exposure the 10 bit log RGB electronic camera suffers from poor saturation and the HD camera fails to record lighter colors accurately.
Linear Bit Depth for Video
Linear = equidistant steps
- If dynamic range is fixed (as in Video)
- more bits = smaller (smoother) steps
Not to Scale: Remember, 10 bit images have 4 times as many steps as 8 bit images.
Liner Bit Depth for data
The dynamic range of video is less than that of digital film
Colorists are skilled at grading the film range to suit the smaller video range
When the video range is recorded within the wider data range it appears flat on a data monitor, but correct on a video monitor. This is sometimes referred to as "TV Level".
To capture the full film range we use larger steps
- the risk banding artifackts
or use more steps of the same size
- which needs more storage and higher transfer rates
Not to scale: Remember 14 bit images are 16 times larger than 10 bit.
10 Bit Log
Log = steps that increase in size with luminance
Transfers and storage are optimized for 10 bit (Lin or log)
- Log captures the same range with fewer bits, but uses smaller steps in areas where film and the human eye most sensitive
- The relationship between steps is a mathematical function so it is possible to convert linear to log and interpolate log to linear
- However, since up to 15000 steps are "lost" in log encoding, it is essential to know the precise dynamic range and sampling method before interpolating a log encoded image back to linear
- For real images transparency of lin to log to lin cannot be guaranteed
- Many facilities therefore prefer to leav files log encoded
If we need to deliver a 4K image. We better start with 8K.
Look up table.
| | Kodak Strip Calibration
Scanner ------ DI suite ------- Film Recorder ----------- Film Lab
| Calibration lut |
Probe creates negtive
monitor profile | Recorded Digital Reference
After that, compare the two Digital Reference and really do the calibration.
DI: DI stands for Digital Intermediate and was originally intended as a replacement for film intermediates used to creat cinema release prints.
Digital Intermediate: "The process of captureing digital images, either with a film scanner or a digital camera, and manipulating the color for theatrical release."
*Feature film want to benefit from the same tools that television enjoys.
Data Inter Mediates
DI: Digital Intermediate is a term originally intended for film finishing.
The new DI is a "Data intermediate" for all post production
Data Intermediate: "The process of capturing digital images, either with a film scanner or a digital camera as data and producing a color graded, composited and conformed high resolution universal master, from which all film video and electronic formats can be produced"
1 Please the Cilent
Who's the cilent? Director? DP? Editor? Producer? Agency?
2 Optimize the source
3 Prepare for Post and Deliverables
Color Grading in Practice
Search: Find a reference, stars with the wide if you can
Primary:Balance black and whites
Control dynamic range and contrast
Layers: Add look
Secondary: Control Saturation
Correct isolated areas
Format for delivery
Start with the contrast rather than the color.
Start with the black and goes to the high light.
A color grading strategy
Optimize: Analyze source material
Balance as shot
Please: Match shot to shot
Enhance isolated areas
Prepare: Fix problems
Format for delivery
You take a film log or a film print, and put on a well calibrated look up table, you are actually looking at a print.
As proper colorist, you need the ignore the name the manufactory gives the tools, but figure out what the hell is the tool doing.
What's the biggest brush?
Black control in some of the system is the biggest tool. For example, Apple color.
When you do the lift, it lift all luminance in the same value. It could be called lift, it could be called offset, it could be call printer light.
But not in all the case, lift works this way. But Printer light works exactly this way.
Printer light lift all the pixels by the same amount. So it won't stretch the image.
The other type of lift is only effects on the black, but not white.
Use the first lift, you can see if theres some details in shadow and details.
FInd the part of the picture is the most important, and use the printer light to where the most important port you want it to be, and adjust with the other tool.
If you use the wrong tools to correct things, sometimes you will get yellow.
Deal with the Log image:
Look up table,
Log to linear converter.
if you are going to delivery the film print, you are going to put on a look up table, and grade on it, and take out the look up table and delivery.
In apple color
If you use a lift and a gain to gain the contrast, it will end up stretching the info in the mid tone, and lose the details. Better way to do it is using the RGB curve and change some saturation.
In film master, you use the contrast tool.
this is the best way to get the contrasted picture.
FOr the saturation, if you want to get a better quality, you can also remove the color, instead of reducing the saturation. Just a an option.
Colorist doesn't need to do things at an exact way.
For the contrast, you don't really need to adjust the black and white, you can use the curve.
For the saturation, you don't really need to pull the saturation, you can also do by add or reduce the color.
White Cliff is also a curve control. Is more use for very very high end, in the other word very high light. It will soften the image a bit. It will round very edge of it. It will chop some info in high light.
So, it is just effects on the small part of it.
So as Black Cliff. Some times you can use the Black Cliff to get rid of some noise in shadow.
For the blue sky, you can cliff out some of the red and green, it will leaves some blue in the sky.
Sometime Cliff can also solve some cross color problem.
It is a much more smaller tool. So, use the Cliff to more the end.
Before doing the secondary, start with bigger area, using bigger tools. And go to the smaller ones. And also, match the primary color correction before go into the secondary.
When using the track ball, go too far and come back as pulling focus. and Then go 90 degree to find the best place.
Saturation, Hue, and luminance is more for isolation control.
Generally speaking, any type of curve tool is isolation tool.
All system will have some type of keyer.
There's 3 type's of keyers: RGB, HLS, 3D.
It is awesome to have a matte imported into the color grading system to isolate a compositing shot.
Color key and all type of isolation:
Select first, do you color correction and adjust the key.
Because only after you do the color correction, you know how much you need to adjust.
After the isolation, use the biggest tool first, like printer light. Because it is isolated, So you don't really need to use like Lite gamma gain.
A trick to use a HSL tools is there's always one of them is dominance. Make it too big instead of too small.
RGB keyer can generate a pretty clean key, but it is hard to control.
3D keyer is actually doing the color correction in a color space. It is working in RGB space, so it is cleaner.
You can pick up different colors at once but not effect the color space in between.
You can use the isolation tool to expand the tool set. For example, you can use the Keyer to isolate the shadow or midterm or highlight and adjust them.
You can finish a look, then apply the look to the rest of the image, and start with the look. It will save you some time.
When you are doing the secondary, you don't really need to get a perfect key. Cause it is not really important.
Scan the film to log footage and make the look up table for the log shot and the editing suite.
Red, compress the file and edit it, and relink into raw file or log file to color grade it.
A look up table is that you put on the simulate the log footage on the theatrical screen. You grade on it. And export the log file again.
The log file look will be appear on the negative print, and when you make a release print, you can see how's that look. If there some sort of problem, you can just make another print. That is the advantage of working with log file.
And work with the log file, the benefit will be you get more dynamic range to work with and keeps more information.
It is better to darken or add contrast picture, instead of lift up or de-contrast the picture.
So log is better for colorist to work on.
First do the color correction. Second, do the color enhancement.
What's the role of the colorist?
Please the Client
Optimize the source
Prepare for Post and Deliverables: You need to know where are the final product is going.
Kevin's Tool Theories
Big Brush Theory
1 Use the tool with the largest and most uniform effect
1 Offset is less aggressive than Gain
2 RGB is less aggressive than HSL
3 Darker is less aggression than brighter
4 Flat is less aggression than contrasty
A color grading Strategy
Analyze source material
Balance as shot
Match shot to shot
Enhance isolated areas
Format for delivery
Traditional Film Workflow
Acquisition=Processing=Rush Prints=Edit offline=Edit Neg cut=Intermediates (lab grade)=Answer Print=Release Prints
(OCN) | |
| | |
---------------------------------- Telecine---------------------------------------------Video (Digital Grade)
Simple DI Workflow
Acquisition=Backup============Edit (Offline)==Edit (Online)==DI==Film Recorder (Intermediates)==Release Prints
Full DI workflow
Acquisition=Backup============Edit(Offline)==DI Conform==Data Master==Film Recorder=Release Prints
Digital Grade | (intermediates)
Telecine Shoot---------Shared Storage--------Edit
|------------Conform & Color Enhance
Tests Colorist DIT Colorist Dailies Colorist Pre-Grade Colorist DI Colorist Mastering Colorist
Design Acquisition Back-up Edit Offline DI Conform Data Master Film Recorder (inter neg) Release Tape/ Data Digital Grade |
VFX -------------------------------------------------------------Video Formats |___________________________________Digital Formats
Pre-Production Production Post-Production Distribution