I’m attempting to create a very large volume of images with very little space, and the one bit or “true black and white”, as opposed to grayscale, seems appealing. I have gotten familiar with the styles that old newspapers and books used, as the printing presses of the time could only apply ink without any control of value. However when I try to create a new project in Krita with a bit depth of one, I can’t. 8 bit grayscale seems to be the lowest it will go. I guess I could create a brush with no aliasing and draw that way but I’m still creating an 8 bit grayscale image which I imagine takes up more space than a true one bit image. I did find a website that “crushed” images into a one bit image, and this produced an image that was 162 KB with a 32 bit depth (according to the image properties). Taking the same image before it was “crushed” through Krita, converting to an 8 bit depth grayscale produced an image that was 244 KB with 8 bit depth (also pulled from image properties). Shouldn’t the 32 bit depth image take up more space than the 8 bit one? Is there a way to directly, or with a plugin, make one bit images with Krita or another software?
The website that did the crushing: Ditherlicious 1 Bit Image Crusher - 29a.ch
I’ve found that Microsoft paint will let you save an image as a monochromatic bitmap, but something strange is happening. The 8 bit depth png image Krita produced is smaller than the Microsoft paint produced 1 bit bitmap. (180 KB vs 190 KB)
JPEG will easily drop the size, but also adds a bunch of noise to the image. Even with 100% quality, the image looks “fuzzy”.
That is not unusual, because PNG is compressed lossless.
If you use some dithering algorithm to reduce the bit depth, you essentially introduce noise, which compresses a lot worse than flat colors. So depending on the noisiness of the original image, it may become larger instead of smaller.
Also keep in mind applications may store various amounts of metadata, like ICC profiles, Exif data etc. that can add a few kB.
And JPEG is hardly state of the art anymore, AVIF and JPEG XL should produce a lot less blocky/noisy artifacts and generally compress a lot better. They are however pretty aggressive at removing subtle noise.
If it’s a color image, the fuzziness may come from chroma subsampling.
GIMP can use a 1-bit indexed colour image and convert another image to that using
Image → Mode → Indexed.
For 8-bit greyscale with no alpha channel stored as .png vs. 1-bit indexed .png no alpha channel, I see 41kB vs 13 kB file size.