Coloring a B&W photo

[Chris333]

I had seen a few people using AI to colorize old photos. It somehow takes known parts of a photo to figure out the colors. I think it works like this. If there are trees in the photo it knows those are green and from there can figure out the rest of the photo.

So I tried it on a photo I knew what the colors should be. In this pic the loco should be yellow and the cars red:







The program is called DeOldify:
https://github.com/jantic/DeOldify
https://deepai.org/machine-learning-model/colorizer

[peteski]

That's pretty slick!

[Chris333]

Now for the smarter people here... how do I install this on my PC?

[CRL]

Nice find.

[DKS]

Don't let Ted Turner know about this. He'll want to colorize the black and white parts of Wizard of Oz...

[wazzou]

Don't let Ted Turner know about this. He'll want to colorize the black and white parts of Wizard of Oz...

Already been done unless of course that was the sarcasm font.   

[Chris333]

Well part of that movie is supposed to be in B+W. If you search YouTube for DeOldify you can see a few war movies that have been converted.

[learmoia]

I ran a few photos through it.. and unless it has wide range of 'known' colors to assume, it has a hard time rendering colors.

It would be nice to have an eye dropper so we could tell it known colors.
For example.. It doesn't know Brown and Orange..
[ Guests cannot view attachments ]
(from Railpictures.net)
https://www.railpictures.net/photo/652974/

[Chris333]

Yeah I tried a photo with a orange and green loco, it didnt figure it out.

[Chris333]

Seriously does anyone understand any of this ?

How to setup DeOldify
For those looking to try this themselves, here is a short tutorial on how to get things setup.
First, cd into the directory you would like to place DeOldify, then clone it:
git clone https://github.com/jantic/DeOldify.git DeOldify
Now cd into the newly created DeOldify directory and create a new “models” directory for your pre-trained weights:
cd DeOldify
mkdir models
You can now download the pre-trained weights through the command line:
wget https://www.dropbox.com/s/zkehq1uwahhbc2o/ColorizeArtistic_gen.pth?dl=0 -O ./models/ColorizeArtistic_gen.pth
If you don’t have wget, you can find links to manually download these pre-trained weights in the DeOldify README.
We also need to install dependencies:
pip install -r requirements.txt
Once the dependencies are installed, open the project in a Jupyter Notebook:
jupyter notebook
At this point, you should be able to open the ImageColorizer.ipynb file and continue from there. I hope this helped! Happy colorizing!

[Chris333]

This one came out pretty good:

[Simon D.]

Oh dear, someone left them out in the sun too long...

[MK]

Seriously does anyone understand any of this ?

How to setup DeOldify
For those looking to try this themselves, here is a short tutorial on how to get things setup.
First, cd into the directory you would like to place DeOldify, then clone it:
git clone https://github.com/jantic/DeOldify.git DeOldify
Now cd into the newly created DeOldify directory and create a new “models” directory for your pre-trained weights:
cd DeOldify
mkdir models
You can now download the pre-trained weights through the command line:
wget https://www.dropbox.com/s/zkehq1uwahhbc2o/ColorizeArtistic_gen.pth?dl=0 -O ./models/ColorizeArtistic_gen.pth
If you don’t have wget, you can find links to manually download these pre-trained weights in the DeOldify README.
We also need to install dependencies:
pip install -r requirements.txt
Once the dependencies are installed, open the project in a Jupyter Notebook:
jupyter notebook
At this point, you should be able to open the ImageColorizer.ipynb file and continue from there. I hope this helped! Happy colorizing!

Those look like a bunch or Unix of CP/M commands.  I'm sure one of our many resident computer geeks can figure it out.  I'm a bit too rusty at this time.

[peteski]

Those look like a bunch or Unix of CP/M commands.  I'm sure one of our many resident computer geeks can figure it out.  I'm a bit too rusty at this time.

Looks like Unix (Linux)  shell commands to me.

[Doug G.]

I want to know Toto's real color.

Doug