"This tool allows you to add arbitrary-size caption to your photos without losing image quality."
Does this mean that, when the image is saved, the original JPEG data isn't recompressed? Similar to LLCrop on Android?
I thought that opencv.imread() would decompress the image into a Numpy array, and that a later call to imwrite() would need to (lossily) recompress the uncompressed data.