Are the colors based purely on their assumptions/imagination, or do the greyscale images retain some sort of data that tells what color on what part?
In: 641
Different shades of grey align with different colors. So a really dark shade would likely be a dark blue/red/purple. Then it comes down to knowing the fashion trends and what colors were used at the time,in that region, and the economic class of the person. They can also consult written accounts for colors of things like military uniforms.
I often color old black-and-white photographs as a hobby and, most of the time, you just have to guess when you colorize an image.
Sometimes, for famous people, you can do historical research to find out the color of their hair and eyes or even their homes’ furnishings (for instance, Empress Alexandra Feodorovna had reddish gold hair and blue eyes and her mauve boudoir was all lavender with pistachio green carpets). You can also do research to get a general idea of the kinds of colors that were popular, say, in the Victorian or Edwardian eras… but that’s pretty much it.
It is based on assumption. The grayscale image only contain the brightness of an object but not its hue or saturation. The artist will therefore have to make up these by themselves. Often you can research a bit about the objects to find their hue. For example peolpe have the same hue skin color, militaries have the same hue uniforms, etc. The saturation on the other hand is more or less guesswork. And I have seen restaured photos with the completely wrong hue for objects.
>Are the colors based purely on their assumptions/imagination, or do the greyscale images retain some sort of data that tells what color on what part?
More of the former and less of the latter.
I colorize photos as a hobby and have posted a few on Reddit. When picking a color it’s a process. For me it’s all about context and light levels. Light levels are one of the easiest problems that will make a colorized photo look weird. A bright color being added to a part of the photo that was originally dark or vice versa will just make it look abnormal.
My process for picking colors tends to go like this
1. Research – find color photos or descriptions of the subject to use as a template. I’ve spent many hours looking up tiny details like military service medals.
2. Balance – If research doesn’t work, find colors that match the overall look of the photo
3. Guess – If neither 1 nor 2 work, just throw a color on there and hope for the best (make sure your light levels match up)
I had one photo of a building from the 1930s I worked on for about a month. My research told me it was a brick building from the 1800s to the current day. I kept trying to add red brick colors to it and it always looked weird and too bright. After a month or so, I noticed the department store that occupied the building from the 1920s to 1940s always painted their other buildings white. I had been trying to add a dark red brick color to a white building and it just looked weird. I ended up scrapping my work on that photo and moved on because all the buildings on the street were some form of white or grey.
Hobbyists go on assumption if there are no color representations to compare to. Higher end operations (that colorize films for example), use software that can analyze known colors in a shot (for example, the grass, bricks, even clothes or hair color that there is a color record of) and extrapolate.
I’ve done photo restoration for a bit over a decade now, and there’s a few tricks here. The biggest one for me is research; with a little research it’s not hard to find out what colors certain things were exactly (uniforms, cars, buildings, etc). Adding onto that, colors have popularity over time, and knowing which colors were popular when/where help a lot at guessing what a color is.
When I’m *fully* stumped on a picture, I’ll shoot it over to my grandmother and her and her friends usually have a good idea of what the colors were. If all else fails, I just assume what the colors are based on the lightness/clarity of the greyscale and the context of the image. Though I do prefer to have some kind of confidence in the colors I use in a restoration, so I try to guess as little as possible.
In the absence of independent evidence it is mostly supposition.
Another factor I have seen that can cause error in colorization is the difference between pancromatic and orthochromatic black and white film.
Orthochromatic film was used frequently in earlier times & while it gives striking images it is insensitive to red light while being quite sensitive to blue and violet.
Pancromatic film was less frequently used in earlier times and was more evenly sensitive to different colors
So, a colorist taking an orthochromatic print of an orthochromatic negative may color items in the scene that were mostly red far darker or assume a different color because they show as practically black. While coloring blue items in the scene as very light tone even though in reality they were a mid blue.
[This video](https://youtu.be/vubuBrcAwtY) focuses on one team that restores and colorizes photos for archival purposes. It can take lots of painstaking research to ensure accuracy if the team takes it seriously.
Now there is also the artistic side where artists can take liberties to place their own ideas of colors onto photos, which at that point just relies on understanding light and shadow and how that can affect colors on different surfaces or materials
They don’t.
Some things are well known because they still exist. They know coca-cola is probably red, military uniforms are olive green. Maybe that particular skirt wasn’t this pastel color but some of them were. You can look find them in museums or in books with color photographs from later.
But other than that they mostly just guess. Maybe that’s the wrong word, they choose. They get some info from the greyscale but at a certain point they just pick and go. Modern digital tools mean that they can even change colors at any point in the process and just make it look good.
Latest Answers