A compression algorithm and image file format that stores images that are highly compressed and have a very small file size.
The image is loaded into memory and 7 central pixels are chosen as samples. The algorithm then finds what sample pixel each pixel of the image is closest to. It then stores the pixel as a value to 1-7 referencing to what source pixel it is. These values are written to a binary file with the 7 source pixel's values and the image's dimenensions as the header of the file. This file is then 7zipped and stored as a .dan image. Viewing is same steps in reverse order.
Source image:
becomes this when its saved as a .dan file:
Let's look at the file sizes:
| Image | Size(kb) |
|---|---|
| tree.bmp | 2214.6 |
| tree.png | 960.6 |
| tree.jpg | 163.8 |
| tree.dan | 75.4 |
Wow! The .dan image is less than half the .jpg size, or 3.4% of the size of the .bmp image. That's a compact format, and it's pretty close* to the source image.
Here's a couple more demos:
Yes, I know this is a terrible image format and is not meant for any serious use. It was fun to make and it does work to produce a very small image file. You can probably get better results using a low colour depth .png file.
- SevenZip JBinding. Requires libs sevenzipjbinding.jar and sevenzipjbinding-PLATFORM.jar. Can be found in /libs directory of archives found here.
- Swing GUI
- Run Danpression.jar with an up-to-date JRE installation.
java -jar Danpression.jarTo quickly view an image, launch withview /path/to/image.danarguments. Builds can be found on the release tab