Copy link to clipboard
Copied
I've only recently become interested in printing my images. I was surprised to find that photo labs with a reputation for high quality fine art printing (e.g., Bay Photo or Whitewall) request 8-bit images. I've seen an obvious advantage in 16-bit rather than 8-bit editing, especially with respect to very gradual color transitions (e.g. gradients or mist over a field) where there is obvious banding in the 8-bit file. As an experiment, I made a copy of a 16-bit file, converted it to 8-bit, then compared the 8-bit side-by-side to the 16-bit original at print view (in this case, 22"x40"). I found no banding in areas that included gradual color transition and a barely perceptable difference in color limited to highlights. It seems like it makes sense to edit in 16 bits even if the file needs to be converted to 8 bits for printing. But I have no clue why 8-bit editing is problematic but conversion from a 16-bit to an 8-bit file does not seem to result in a degradation of the image. Can anyone explain that or point me in the direction of an article that does explain it?
It's because 8 bit editing is cumulative. You end up with the 256 values all over the place, with gaps and "combing". Everything gets exaggerated as you go.
When you work in 16 bits and then convert to 8 at the very end, the 256 values are perfectly evenly distributed. You might still see those 256 values in very smooth gradients, but you won't get the multi-value jumps.
It’s the difference between the requirements for editing, and the much simpler requirements for final delivery.
In editing, you need maximum flexibility, overhead, resolution, and margins to make changes, whether that’s redistributing the tones, preserving highly saturated colors, or making local adjustments. 16 bits per channel (and similarly, the full original color gamut), gives you the resolution and room to maintain quality as you make change after change. The tonal resolution of 16bpc he
...Copy link to clipboard
Copied
It's because 8 bit editing is cumulative. You end up with the 256 values all over the place, with gaps and "combing". Everything gets exaggerated as you go.
When you work in 16 bits and then convert to 8 at the very end, the 256 values are perfectly evenly distributed. You might still see those 256 values in very smooth gradients, but you won't get the multi-value jumps.
Copy link to clipboard
Copied
It’s the difference between the requirements for editing, and the much simpler requirements for final delivery.
In editing, you need maximum flexibility, overhead, resolution, and margins to make changes, whether that’s redistributing the tones, preserving highly saturated colors, or making local adjustments. 16 bits per channel (and similarly, the full original color gamut), gives you the resolution and room to maintain quality as you make change after change. The tonal resolution of 16bpc helps prevent the banding that large changes might create.
But when you are done with changes, it is much less necessary to preserve all of the overhead. You can just keep the final results, and drop much of the rest. The copy sent to a print shop can be flattened, 8bpc, and even JPEG (if high quality), and the final print will look good, because the penalties for stepping down the specs don’t happen if the print shop makes no further major changes to that file.
Another reason is that a print can only reproduce a smaller dynamic range and color gamut compared to the original file and on-screen editing, so print can’t reproduce it all anyway. What can be reproduced in print is typically reproducible at 8bpc, again if no additional major edits will be made.
This works the same way with the digital music and movies you enjoy. The digital files that you listen to or watch sound and look great. But the rendered digital files delivered at the movie theater or on your home AV system do not match the file size, bit depth, and quality level of the original media files (stacks of tracks and clips) that were put together to create the final movie or song.
We even know this with food. What we eat is a plate of food, and that’s exactly what we expect. But the meal could not have been prepared only with the plate and the space on it. Preparing the meal properly actually requires a much larger work surface, more ingredients (like it needed half an onion, but the store will only sell you a whole one), and a lot of water, energy, and a range of tools. But delivering the meal no longer requires all that overhead, all we need at the table is that one plate with the finished dish on it, that will not be altered any further (except maybe with a little salt and pepper).
Copy link to clipboard
Copied
That was really interesting and helpful. Thanks very much. This forum is a treasure trove.
Copy link to clipboard
Copied
@Doc_Pit good of you to come back with thanks, yeah, I agree, work in 16 bit for the seasons that @D Fosse gave - then convert a flattened copy to 8 bit (archive your layered 16 bit original inn case of adjustments down the line) - there's no point printing 16 bit
I hope this helps
neil barstow, colourmanagement net - adobe forum volunteer - co-author: 'getting colour right'
google me "neil barstow colourmanagement" for lots of free articles on colour management