Time to be with your loved ones

Time to be with loved ones

by sat

submit your photo

Hall of Fame
View past winners from this year

Please participate in Meta
and help us grow.

Take the 2-minute tour ×
Photography Stack Exchange is a question and answer site for professional, enthusiast and amateur photographers. It's 100% free, no registration required.

When working on JPEG files, should I start by converting the images to 16 bit depth? Supposedly that would make for more room for the adjustments I'm going to make on the photos. Two occasions that recently popped up while I was reading around PhotoSE questions mention this conversion:

  • Philw in his answer to "What is the optimal order of post-processing steps?"

    Back in the day people would worry about doing most work in the maximum bit-depth and then converting down for output. There's nothing wrong with that principle, but in practice you should be able to do everything you need in the raw converter, so it's a moot point.

  • Hermann Klecker in his answer to "How much post processing advantage is gained..."

    However, there are significant advantages in the further processing. After opening a file in photoshop or whatever, the next step should be the conversion into a 16-bit format. Any interim result should be saved in a non-destructive 16-bit format too. If the final result is to be handed over or uploaded in any 8-bit (sometimes jpeg is even required) format, then it should be the last and very final step converting the file back to 8 bit.

Is it really a must? Do I notice the difference if I don't do the work in 16 bit?

share|improve this question
Honestly, without knowing what processing you do, this question cannot be answered. Now, if you do not know, go for 16-bits. In the best case you'll have better results, in the worst case, you'll have the same. –  Itai Apr 16 '13 at 13:46
I don't know, it was those two answers to other questions that brought me a memory from years ago when I did this conversion up and back down with the job done in between. I had completely forgotten about it, and the reasons too, but now I'm asking what good would it do, really. My editing is always quite light, most common things to do are lightness, contrast, hues (when WB went wrong), sharpness and crop (which has nothing to do with bit depth). Anyway, an answer could also be about what kind of processing would benefit from larger bit depth and what processes would not. –  Esa Paulasto Apr 16 '13 at 15:04
add comment

3 Answers

up vote 2 down vote accepted

It depends on both the image and how much editing you are planning to do. Let's say you're only going to adjust levels. Then

8-bit JPEG -> 16 bit PSD -> levels adjustment -> 8-bit JPEG

Will give identical results to:

8-bit JPEG -> levels adjustment

This is because an 8-bit image converted to 16-bit is identical to the 8 bit version (the lower order bits will all be zero) and there's no difference between doing the levels on a 8 bit image (which will be computed internally at a higher precision and then rounded) and doing the levels command on a 16 bit image and then converting to lower precision.

However, if you do several steps, then keeping a high precision intermediate version of the image will give you better results. If you keep rounding to 8-bit after every single operation, you're very quickly going to lose tonal range and end up with a heavily posterized image.

Now I said it depends on the image content, a very contrasty image with posterization will generally go unnoticed, whereas a low contrast image with subtle graduations in the sky or in skintones will be heavily degraded by posterization.

share|improve this answer
Sounds convincing and makes me realise how little I know of inner workings of image processing, i just use the editor without understanding what goes on inside. –  Esa Paulasto Apr 16 '13 at 16:58
add comment

I've been working professionally with photos for ages. Can I tell the difference between working in 16-bit and 8-bit environment? Depending on the process, yes, I can. Can the client tell the difference? No, they cannot (unless if they are really photography-savvy.

In other words, I wouldn't worry too much about it. There are several things (several plugins, for instance) that don't even work in a 16-bit environment. The most important thing is JPEG compression in the save/load cycle. In other words, do NOT save as JPEG unless and until the final product is ready. The most destructive thing to do is to work on an image, save as JPEG, then resume later, repeating. Even at full quality jpeg compression settings, there is bound to be some degradation after repeated load/processing/save/load cycles. Work with lossless formats, then convert to JPEG when you're ready

share|improve this answer
Whatever the reasons to work with JPEG, people just do it. The difference would be seen in the finished photos of the same original between one that got worked in 16 bit and the one that got the same work done in 8 bit. Or would not? –  Esa Paulasto Apr 16 '13 at 11:32
For example Nikon Coolpix P510, released 14 months ago, can not do RAW. –  Esa Paulasto Apr 16 '13 at 11:41
@EsaPaulasto -- even if the camera can only produce JPEG images, your editing "works in progress" ought to be maintained in a lossless format. The editor's native format such as, say, PSD, is usually a good choice. Then you can go back to JPEG. Staying in JPEG is only safe if you know the image will not be closed and later re-opened. Multiple saves while the image is "live" in the editor make no difference, but every time you open a saved JPEG, the artifacts from the last save become part of the data for the new session. –  user2719 Apr 16 '13 at 16:03
@StanRogers - yeah, that makes sense. In my editing I always do whatever I do at one go, not saving/opening in between, and I never ever overwrite the original. Later, if I feel some more editing is needed, I start over again from the original and again no overwriting the original. –  Esa Paulasto Apr 16 '13 at 16:45
add comment

I'd be suspect of how much you could gain by going from 8 bit to 16 bit only to go back to 8 bit again. All you could possibly do is to adjust an 8 bit image to try to expand the dynamic range, but since you are going to convert it back to 8 bit again at the end of the process, that information would just be lost again.

If you could start with 16 bit, then there is certainly a huge advantage to manually adjusting the image to make it fit 8 bit better, but if the file is already 8 bit, then the damage is done and information is lost. There could be some minor gains if you are using something destructive since you'd have to be careful about your order of operations to not peak portions of the image you work with later, but careful order of operations could avoid that. You also have some gains in rounding error reduction if you are doing heavy processing as Stan Rodgers pointed out in the comments, but I'm not sure how significant of an overall impact it will have.

The biggest thing is going to be to use lossless formats while editing so that you avoid further generations of compression. If you have the space, software and processing power to work with 16 bit, it isn't going to hurt you, but you might not have super huge gains compared to what lossless formats and/or native 16 bit shooting would get you.

share|improve this answer
Suppose he converts to grayscale and then he edits the curves to colorize only some shades of grey... Wouldn't it be better with 16 bits? –  comocomocomocomo Apr 16 '13 at 14:36
At the very least, it makes the difference between posterization and dithering as quantization error artifacts in the final 8-bit image, particularly where adjustments/tool use overlap. True, you don't have any more data to work with from the original or in the final image, but you're not committing to rounded/truncated results until all of the calculations are done. –  user2719 Apr 16 '13 at 15:44
@EsaPaulasto - if nobody else does, I will. I want people who don't have 20K+ reputations to feel like maybe they can participate here as well. –  user2719 Apr 16 '13 at 16:14
@StanRogers - thanks for the comment about rounding error. I hadn't thought about that as being a potentially significant factor. I've updated my answer to include that. How significant of an impact do you think it has? I always work in RAW so I don't have much direct experience dealing with rounding issues. –  AJ Henderson Apr 16 '13 at 16:49
As Matt Grum pointed out in his answer, it really depends how complex the adjustments are. "Complex" can involve non-linear curves, dodging/burning with soft-edged brushes and the like, where each of the tools in an 8-bit space can handle its own side-effects, but then when you add the effects of one tool to the effects of another, it'll posterize badly on you. And you really don't need 16 bits to get around the problem; even 10 bits gives you four intermediate values in each colour at each pixel (which is why you'll never see the problems in the RAW space — my 2004 D70 is 12-bit). –  user2719 Apr 17 '13 at 4:14
show 2 more comments

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.