Format comparisons, a pair at a time: pick two from the same family. The sheet below lines up their specifications row by row, marks every difference and lists what each direction of conversion would lose. It also asks your browser whether it can open each one. For image pairs, it encodes two test pictures in both formats and measures the bytes each needs for the same quality.
Worked out from the rows above. If the source has a feature the destination can't store, it goes. If the destination only compresses lossily, every save loses a little more. Converting in the other direction doesn't bring a lost feature back.
Comparing file sizes at the same quality setting tells you nothing, because "quality 80" means something different in every encoder. Here each lossy format is searched by bisection for the smallest file that still decodes at 40 dB PSNR or better. That is about the point where differences stop being visible at normal size. Lossless formats are simply saved. The test uses two generated 360×240 pictures and your browser's own encoders.
Each cell counts what converting from the row format to the column format would lose, using the same rules as above. 0 means nothing is lost. Click a cell to load that pair.