Scan Tailor Advanced input format

[nightshift]

I'll using Scan Tailor Advanced (the fork that is still being worked on, as of several months ago) to process my images, but I'm curious about input formats. I'm "scanning" in DNG/Raw because I can't get ideal lighting, and processing the raw files lets me clean that up without experiencing data loss from re-compressing an already compressed jpg. But, when I save, I have the option of several image formats - is it worth saving as 16-bit (or even 8-bit) tiff for feeding to scan tailor, or is saving to low compression jpg fine? (some of my scans or just text with line drawings and a few black and white photos, but others have text plus full color photos or colored line drawings, if it matters)

[cday]

I think in general people worry far too much about possible quality loss when resaving a few times to JPEG format, and it should certainly not be an issue when book scanning. I did some tests once, and even after resaving an image with fine detail not best suited to the format many times, it was necessary to look hard at the original and final images to see any significant loss. This long thread refers, and this post is the start of my test.

The only caveat might be that camera-based book scan images are generally fairly low resolution images, but if you make s few test resaves I doubt if you will be able to see any differences between the images even in zoomed images, most easily viewed in a tabbed viewer so that successive images can be viewed almost instantly.