What is vectorization, and how is it different from format conversion?
Ordinary format conversion (such as PNG → JPG) only changes the storage format; the pixels stay the same.
Vectorization lets an algorithm analyze the shapes and colors of the image and re-describe the pixels as a set of paths (curves and filled color blocks). It is a rebuild, and the result will not be exactly identical to the original.
Which images vectorize well?
Graphics with sharp edges and simple colors give the best results: logos, icons, text, line art, QR codes, stamps, and similar content.
After vectorization these graphics can be enlarged infinitely without blurring and can be edited like vector art. This is where vectorization is most valuable.
Why are complex photos not suitable?
Photos contain large amounts of gradients and fine textures. The vectorization algorithm can only approximate them, so the result is either heavily distorted or produces a huge number of paths.
The SVG produced from a photo can be much larger than the original, and when enlarged it still looks like patches of color blocks, so the trade-off is not worth it. Keep using bitmap formats for photos.
How to choose between black and white, grayscale, and color modes?
Black and white: only two colors, black and white. Suitable for text, stamps, and black-and-white icons, with the smallest files.
Grayscale: layers by light and dark. Suitable for line art and sketches with shading, with moderate file size.
Color: layers by number of colors. Suitable for color logos and icons. Fewer colors mean smaller files, but too few colors may cause more distortion.
What do colors, detail, and smoothing mean?
Colors: the number of colors allowed in color mode. More colors give higher fidelity and larger files.
Detail: controls how fine the paths are. Higher values stay closer to the original but produce more paths.
Smoothing: rounds the paths. Higher values look softer but may erase sharp edges. Start from the default value and fine-tune.