Replies

Your 7:25 image is of noticeably lower quality than Kos's. The text is less crisp. The JPEG artifacts are more prominent. There is a green fringe above some of the text. The crease near the top is harder to pick out.

Kos's image is 2427px × 2369px, 519,907 bytes. Yours is 2426px × 2369px, 478,223 bytes.

It is important to minimize the number of JPEG generations. Each successive decompress / recompress cycle further degrades image quality. Graphic artists generally keep store their images in lossless form (TIFF) until they are done working on them.

I know that, which is why I made the first conversion from JPG to an uncompressed BMP and then back to JPG. If I had access to the original scanner bitmap, the original image would have been a bitmap, and not a JPG. Then I could have rendered all of it in one single pass.

I had to start with something, and I chose the highest quality image available to me, which was Michele's COLB, and made a bitmap of it.

I then edited it, and saved it at a JPG comression of 45 percent -- which is what I calculated to be the compression ratio of the KOs image, assuming that the size reduction, cropping, and JPG reduction were all done in one pass.

Like I said, it was not a pixel-by-pixel clone. There's no way to get it to be exactly the same, given that the original was a JPG.

One thing I did not do was to save it as a progressive JPG. Had I done that, the letters would have been crisper, and the artifacts less noticeable.

I might try going that route next time.