Colorize Images is an automatic service based on machine learning to colorize photos in black and white, grayscale or night vision.
Note: If you have any questions or issues, send me a debug message using the Report a Bug feature in the app. The message includes information to help resolve the issues. Thank you!
Black and white images contain no clues to correct colors. The colorization of the photos is determined by training the machine learning model, but the model doesn’t know the context of the photos. The colors selected are not historically accurate and not necessarily correct.
Black and white photos for colorization can be selected from the gallery, or they can be shared from other apps such as Facebook, Twitter, Chrome or Google Photos. Colorized images of the result can be shared with other applications or saved in the gallery.
The app can be used for free by watching ads. The subscription to the application allows unlimited number of colorizations without interruption during the subscription period.
The service uses GPU servers to provide colorization functionality. This means that the source images are sent to a remote location on the network, colorized there, and the colorized images are then sent back to the device.
For best results, it is recommended that the source image have:
• good quality
• good contrast
• good concentration
• and correct orientation
Images that don’t have all of these properties may work fine, but some pre-processing (like contrast correction or sharpening) usually improves the result.
Based on the DeOldify project.
