Photography Stack Exchange is a question and answer site for professional, enthusiast and amateur photographers. It only takes a minute to sign up.
Sign up to join this community
I have several photos taken in a studio, but the one photo I like is in black and white, and I lost the raw image file. Those photos I don't like, which have colors, are really similar to the one I like: same background, same guy, same shirt, although the camera positions are a little bit different. So I'm wondering if there's an easy way to use the colors from the other photos that I don't like to colorize this black and white one.
Thanks.
This is not what you want to hear, but the answer is no, there is no easy way to do it. I am not aware of any working algorithms which can automatically colorize a photo. given a colored template or not. I would rather try to find or recover (recuva, or some tools that can un-delte come free together with memory cards) the RAW which you lost. You don't write how you lost the raw; if you just accidentally deleted it you might be able to recover it.
If you're willing to use GIMP instead of Photoshop, there's a plugin named colorize-gimp just for that reason.
If you don't know or want to compile, apparently there's a compiled Windows version available.
Yes, it's possible, but a consumer software is probably not yet available.
One of the recent examples:
Let there be Color!: Joint End-to-end Learning of Global and Local Image Priors for Automatic Image Colorization with Simultaneous Classification by Satoshi Iizuka, Edgar Simo-Serra and Hiroshi Ishikawa from Waseda University
Abstract
We present a novel technique to automatically colorize grayscale images that combines both global priors and local image features. Based on Convolutional Neural Networks, our deep network fea- tures a fusion layer that allows us to elegantly merge local informa- tion dependent on small image patches with global priors computed using the entire image. The entire framework, including the global and local priors as well as the colorization model, is trained in an end-to-end fashion. Furthermore, our architecture can process im- ages of any resolution, unlike most existing approaches based on CNN. We leverage an existing large-scale scene classification data- base to train our model, exploiting the class labels of the dataset to more efficiently and discriminatively learn the global priors. We validate our approach with a user study and compare against the state of the art, where we show significant improvements. Further- more, we demonstrate our method extensively on many different types of images, including black-and-white photography from over a hundred years ago, and show realistic colorizations.
You can also use channels in Photoshop to help bring it back. A lot of photographs are restored by using this method. Here is a link to a tutorial. I have no affiliation or gain from linking to this website. Hope it helps! https://www.youtube.com/watch?v=C2Srw245R7U
Similar to the logic used by the GIMP plugin mentioned in Roflo's answer, you can also use the color picture to colorize the gray scale picture. You then first approximately align the two pictures (using e.g. Hugin). Then take the color image, transform to XYZ colorspace and attempt to correlate the gray values in the X and Z channels to the gray values in the Y channel. This requires partitioning the image into regions of similar color, make masks for each region and just divide the X and Z channels by the Y channel to do this in a crude way. With more effort you can do a better job by splitting the image for each region up according to ranges in Y and then making different fits for X and Z.
The next step is to use the fits to colorize the black and white image. You first transform this to linear colorspace, divide it up in the same regions as the color image (you may need to tweak the alignment if it is too far off). Then you normalize the average gray value of each region to match the average Y value of the color image, and then you apply the mapping to calculate X and Z. and add these channels t the picture. Finally, you change the normalization back to what is was (X, Y and Z are then all scaled by the same factor). When you're done with all regions, you put together all the parts and then transform the image from XYZ to sRGB.
You can try using http://www.colorizephoto.com to do this. You basically load the black and white image and the color image and then select colors from the color image to paint onto the black and white image. I hope this helps you!
