printf("Thoughts");
Adventures in C, Linux, and open source
Steg-Image
| Source - GitHub | License - MIT |
There are countless ways to obfuscate data, some specifically apply to embedding data in images. This method uses steganography, which involves storing data inside the color information (pixels) of images.
If you run steg-run -d -i demo_img_400_e.png for the above image, you'll get a text string reading "Hello World!"
The technique I'm using here is replacing the N least significant bits (where 1 ≤ N ≤ 8) per byte of each color channel with the data I want to embed. Essentially, if you have an image with four 8-bit color channels (RGBA), you can store (N/2) bytes of data per pixel. Using more bits increases the storage capacity of the image, at the cost of noticeably changing how the image looks visually.
More details about how it works →
While this is really cool, there are several disadvantages that make steganography impractical:
- Only possible with lossless formats (i.e. PNG)
- Any kind of image manipulation will corrupt embedded data
- Most image hosting platforms perform some image optimization, which also corrupts embedded data
Despite the fragility of the embedded data, there are a few reasons you might use steganography:
- Most image tools look at metadata and don't expect encoded data in the actual color information
- Theoretically, steganography doesn't change the file size of an image (in this case, I'm using an inefficient compression library, so the steganographied image will be larger)
- It's cool!