Doesn't the Color Picker work in 16-bit mode?

[Lee James]

When I switch Photoshop to 16-bit mode, the color picker seems to stay in 8-bit mode (the values only go up to 255 instead of 65535).

Doesn't the Color Picker work in 16-bit mode? Or is there some way to switch it to 16-bit mode, so you can see the actual color values?

[c.pfaffenbichler]

Click on the eyedropper icon in the Info Panel and set it to 16-bit.

[c.pfaffenbichler]

And keep in mind that Photoshop 16-bit amounts to 15-bit + 1.

[Lee James]

Thanks for the answers.

I was aware you could do it in the info panel, but I just find it confusing why the Color Picker doesn't show the true values? I mean why not? Seeing 8-bit values kind of makes me question whether I'm truly in 16-bit mode and if everything is working properly.

I often use to Color Picker to do two things at once:

1. See the color of the pixel I've picked

2. Compare the colors of different pixels and see if they're actually different.

Not having 16-bit values makes this tool kind of imprecise and clumsy - though it's not a big deal.

Also, could anyone explain what is meant by "15 bit +1" ? I don't get that.

[D Fosse]

Suicide  wrote

I often use to Color Picker to do two things at once:

1. See the color of the pixel I've picked

2. Compare the colors of different pixels and see if they're actually different.

16 bit depth isn't for appearance. It's for data integrity through repeated editing operations.

Visually separating 8-bit values is difficult enough, and in many cases you can't. Don't forget that 8 bit depth is all you get out of your monitor (at best 10 bit if you have a display system capable of that).

Try this one to give you a little perspective:

Free Online Color Challenge and Hue Test; X-Rite

I managed a perfect score on my second try, a fact I take some pride in because only a few % manage that. The first attempt gave two errors.

[Lee James]

https://forums.adobe.com/people/D+Fosse  wrote

16 bit depth isn't for appearance. It's for data integrity through repeated editing operations.

The word "repeated" there makes me a little nervous. All of Photoshop's key editing operations are now non-destructive (smart objects, adjustment layers etc) so the only people performing repeated destructive editing any more are amateurs.

Visually separating 8-bit values is difficult enough, and in many cases you can't.

That's kind of my whole point. You can't see the difference with your eyes which is why you need the Color Picker to confirm that two colors are indeed different.

Try this one to give you a little perspective:

Free Online Color Challenge and Hue Test; X-Rite

Hmmm, I expected that to be somewhat tricky but I got a perfect score very quickly and easily.

[Chuck Uebele]

You can work non-destructivly and still need to use 16 bit for reducing banding and other artifacts. Just one adjustment layer curve can mess things up, but if you have several adjustment layers, it can get bad quickly.

As far as being able to tell if values are the same color, that goes back to if you're really able to perceive the difference. Does it really matter if you can't?

[Lee James]

https://forums.adobe.com/people/Chuck+Uebele  wrote

You can work non-destructivly and still need to use 16 bit for reducing banding and other artifacts. Just one adjustment layer curve can mess things up, but if you have several adjustment layers, it can get bad quickly.

Surely when Photoshop calculates the effects of multiple adjustment layers, it only rounds down to 8-bit or 16-bit at the very end of its calculations? It would be crazy to round down to 8 or 16 every step of the way.

It therefore shouldn't make any difference whether you have 1 adjustment layer, or a thousand, Photoshop only shifts the values once and converts them back to 8-bit once.

In other words, banding is not created by working in 8-bit. Banding is only a result of having a source image with too low a bit depth.

As far as being able to tell if values are the same color, that goes back to if you're really able to perceive the difference. Does it really matter if you can't?

I guess not. But when I first opened a 16-bit image in Photoshop the first thing I did was to open the Color Picker to test that the image really was 16-bit. Seeing 8-bit values threw me. It's not a big deal though.

[D Fosse]

Suicide  wrote

It therefore shouldn't make any difference whether you have 1 adjustment layer, or a thousand, Photoshop only shifts the values once and converts them back to 8-bit once.

That's not correct. Each layer is calculated separately.

Banding is cumulative. It gets exaggerated by working in 8-bit depth.

[Lee James]

https://forums.adobe.com/people/D+Fosse  wrote

That's not correct. Each layer is calculated separately.

Surely that cannot be right? If that were true I would be stunned!

That would destroy images unnecessarily! That would make adjustment layers almost completely pointless! That would mean I would have to start saving all my work in 32-bits, and worrying about how many adjustment layers I use.

Why would Adobe do that? Why would Photoshop crudely change color values down to 8-bit at every step of the calculations and introduce banding to the image? Why not just use the exact, precise values = no banding?

With the greatest of respect, you have to be wrong. I really hope you are wrong!

[D Fosse]

So what would you do with one Curves layer, and one Hue/Sat layer - and do the final calculation only once?

Generally, it sounds like you worry too much. Just work in 16 bits and you'll be fine. Photoshop won't destroy your images - there is more than enogh data depth to handle any practical situation.

[D Fosse]

That's right. Destructive or non-destructive is not the issue. You still need 16 bit depth to preserve full data integrity.

In fact the whole term "non-destructive" is misleading and I wish people would stop using it. As long as the adjustment layer is there, you intend to apply it at some point, right? Then it gets destructive. You're just postponing the destruction. The same goes for Smart Objects.

I much better name would be "re-editable" or "reversible", because that's what it really is.

[Lee James]

https://forums.adobe.com/people/D+Fosse  wrote

You still need 16 bit depth to preserve full data integrity.

Surely the sole reason you need 16-bit depth is to work with high bit-depth source images? Data integrity is irrelevant since Photoshop's calculations aren't limited to any bit depth.

In other words, if you take the same 8-bit image and work on it in both 8-bit and 16-bit modes, you will not get any difference at all in the end results.

In fact the whole term "non-destructive" is misleading and I wish people would stop using it. As long as the adjustment layer is there, you intend to apply it at some point, right? Then it gets destructive. You're just postponing the destruction. The same goes for Smart Objects.

The term "non-destructive" refers to the editing process not to the final output. It is taken for granted that parts of your final image will have been changed once. Their original color values will be re-calculated and it will be translated back into 8-bits for display and saving. But that only happens once.

I think "non-destructive" is an incredibly important term. It tells you that you can play with an image hundreds or thousands of times and the quality will never degrade - you will only get one generation of degradation at the very end when you output the image. Thus, non-destructive editing is essential, and I had been longing for features like adjustment layers and smart objects prior to their introduction. (I am also still eagerly waiting for smart masks, but that's a different issue.)

[rayek.elfin]

https://forums.adobe.com/people/D+Fosse  wrote

Try this one to give you a little perspective:

Free Online Color Challenge and Hue Test; X-Rite

I managed a perfect score on my second try, a fact I take some pride in because only a few % manage that. The first attempt gave two errors.

Wow, I got a perfect 0 score on my first attempt. As far as I am aware, I have some colour vision issues, but somehow I managed to ace this particular colour test.

[D Fosse]

Someone's missing my point with that online test. It's not a competition and I wasn't trying to brag. I didn't feel that I "aced" it as rayek apparently did.

The point I was trying to make is that visually separating nearby colors isn't nearly as easy as some people here seem to think. That test illustrates it very well. They are probably not even single 8 bit value steps, but several (I haven't measured it). Using 16 bit values for input - which is what the color picker is for - is absolutely ridiculous.

That was the point, in the context of the original post.

[rayek.elfin]

I did understand your point about the difficulties of nearby colour separation. Just out of curiosity I visited the colour test page, and expected to do very badly, because I have rather poor colour vision. Somehow I did quite well, and I was merely expressing my surprise here. The last thing on my mind was comparing myself with anyone else, or bragging. I was just very, very surprised.

Web writings unfortunately do not include tone of voice, body language, or facial expressions. 99% of interpersonal cues are lost. If we had had this conversation in real life I think you would have gotten the gist of my response. Anyway, it goes to show that I should just not express these personal musings online, and keep them to myself. Because these often lead to misunderstandings and have nothing much to add to a thread.

[Zesty_wanderlust15A7]

"I wasn't trying to brag"

> Indeed you weren't trying. You actually bragged...

"I managed a perfect score on my second try, a fact I take some pride in because only a few % manage that."

Me, I "aced" the test in my first time ever seeing it.

Seems the only people whose eyes are going are the Adobe Community Professionals...

It also explains why you need the JUMBO layer thumbs and never see the generic icons.

[c.pfaffenbichler]

You seem to accuse someone else of bragging, then brag yourself, so I am not sure what your position on bragging actually is. 

[J E L]

@D Fosse That test was a great bit of fun. I got a score of 2 on my first attempt. Not bad! (P.S. I also scored 0 the second time, once I understood where my weakness was.)

[Noel_Carboni]

Suicide  wrote

Also, could anyone explain what is meant by "15 bit +1" ? I don't get that.

When you put an image in 16 Bits/Channel mode, Photoshop uses values 0 through 32768 (inclusive) to represent no luminance (black) to full luminance (white) for a given color channel.  Since all the values between 0 and 32767 can be represented by a 15 bit value, the "+ 1" is often mentioned to indicate there is one more value in the set, 32768.  This choice of this specific number range dates back to the original design of 16 bit data handling in Photoshop.  This data depth proves more than sufficient for virtually all pixel-changing operations.

As D Fosse points out, you can edit quite a lot on an image in 16 Bits/Channel mode using operations that change pixels and never see any visible data inaccuracy.  Astrophotographers, for example, often edit images in 16 Bits/Channel mode and do some quite severe operations (e.g., strong Curves) without causing visible posterization.  "Don't worry about it" is pretty good advice, but you should certainly experiment to see it for yourself.

Developing the best workflow that takes you from raw data to finished product optimally for you with results that suit your needs and with data accuracy you're happy with is up to you.  Photoshop provides the tools.  Combine them creatively and see the results for yourself.  I guess what I'm trying to say is that there's no "one right way" to do things. 

[Lee James]

noel_carboni  wrote

Suicide   wrote

Also, could anyone explain what is meant by "15 bit +1" ? I don't get that.

When you put an image in 16 Bits/Channel mode, Photoshop uses values 0 through 32768 (inclusive) to represent no luminance (black) to full luminance (white) for a given color channel.  Since all the values between 0 and 32767 can be represented by a 15 bit value, the "+ 1" is often mentioned to indicate there is one more value in the set, 32768.  This choice of this specific number range dates back to the original design of 16 bit data handling in Photoshop.  This data depth proves more than sufficient for virtually all pixel-changing operations.

Thanks for the explanation but I still don't get it. Are you saying that Photoshop's 16-bit mode actually uses 15-bit, with the final bit reserved for pure white? I don't get why it would be done that way when 8-bit doesn't work that way.

[rayek.elfin]

Suicide  wrote

Also, could anyone explain what is meant by "15 bit +1" ? I don't get that.

To amend Noel's answer:

Photoshop is the only image editor on the market today with a pseudo integer 16-bit mode. All other image editors with support for 16bpc images work with a value range from 0 to 65535, while Photoshop works with a 0 to 32768 range (which is 0-32767 +1).

The lead / senior Photoshop programmer at the time when the 16bpc image mode was first introduced into Photoshop, Chris Cox, has explained the reasoning behind this decision: it simplified the calculations used in the algorithms. He always seemed to treat it as a feature, rather than as an issue.

Chris in his own words:

0..32768 (NOT 32767) has a middle value of 16384.

0..65535 has a middle value of 32767.5 (oops, not an integer).

And representing 32768 puts it out of 15 bits and into 16.  0..32767 would be 15 bits.

Dividing by 32768 is a very quick shift. Dividing by 65535 is either a long divide, or a long multiply, shift, add operation.

Source: Does Photoshop CC support true 16-bit color editing?

(Btw, I have the utmost respect for Chris Cox as one of Photoshop's core leading developers throughout the years; he has retired a few years ago.)

My own interpretation of this is that it was decided to speed up operations due to the CPU processing and memory limits at the time 16bpc mode was introduced. Don't forget, we are talking about version 2.5(!), which goes way back to 1992! Many users were still working on 286 machines, up-to-date offices on 386 machines which ran at an amazing 25 -33mhz! Ram-wise, people talked about megabytes of memory: a 4mb machine was pretty good.

Also, for real world scenarios, the full range of 16pbc sounded like complete overkill at the time (recall Bill Gates' remark that no-one would ever need more than 640kb?).

Hence, the decision was made to optimize things by stunting Photoshop's core 16bit image mode and limit the value range to a 0-32768 one. A very understandable and logical decision (at that time!).

Which brings us to today's 15bit+1 so-called "16 bit image mode" as implemented in Photoshop, which surprisingly, up to this very day 26 years later, STILL is based on that same core legacy code.

Now, the question to ask is whether this poses any real world problems today. The answer varies on whom you ask and what kind of industry you work in. For the photographer and average user, "no, probably not". Generally camera's don't go that high, but more will so in the future.

For HDR work it poses some problems, though. Even combining a couple exposures into a 16bit HDR image results in half the possible values, and half your values will be thrown out of the window. This may become quite problematic for 3d rendering or, for example, 16 bit height maps used in 3d applications, which have less steps, and which will be causing problems in 3d landscape rendering due to half the possible resolution used. Only half the steps are used for landscape generation...

A couple of years ago a HDR photographer who sells 16bit HDRi images for 3d lighting noticed all his 16bpc HDR images were clipped by Photoshop, and he was losing the detail in those higher ranges, which are important for high fidelity 3d rendering. And Photoshop happily removed that information without a single warning. As you can imagine, that guy wasn't a happy camper when he found out what was happening, and he had to resort to alternative software, and avoid using Photoshop for 16bpc work.

It becomes problematic when 16bpc 3d renders are processed in Photoshop. Suppose you'd like to improve and/or edit some things in that render, save it (losing the full 16bit range of values), and then use that image in a visual FX shot which requires exposing the image at various ranges throughout. Because Photoshop just destroyed half your values, and the condensed exposures no longer contain all of that rendered information. The only solution is to work at 32 bit at that point.

And let's not mention scientific, astronomical, and medical 16bpc integer imagery that utilizes a full 0-65535 value range. Oops. Half the data gone after opening and saving in Photoshop.

In short, it is legacy 16 bpc code that will probably need a complete rewrite of Photoshop's core code. And I would not be surprised if it touches most of Photoshop's code base in some way, so it is understandable that the Photoshop developer team hasn't tackled this yet.

But they will have to at some point. Because we may pretend this is not a problem, but still: Photoshop's 16bpc mode is a stunted creature. PhotoLine, Affinity Photo, Krita, even Gimp, ... All support full 16bpc integer modes, and some even float modes in 16bpc. There is a realistic and practical need for these in real world scenarios, and Photoshop just can't deal with them very well.

Oh, PS (in case people weren't aware: Photoshop's gradient tool creates 12bit gradients (4096 values) when working in 16 bit mode. Another legacy 90s optimization, I guess. PhotoLine's gradients works with the full range of 16bpc values, for example.

Of course, it very much depends on the type of work you do whether or not this is an acceptable situation. My stance is that in 2018 this is a defect in Photoshop, and ought to be fixed rather sooner than later. Requirements have changed, and we no longer live in the nineties of the previous century.

[D Fosse]

I have a feeling this is intentional. I can't think of any situation where you would need to input exact 16 bit values. And input is what the color picker is for.

For readout, use the Info panel as c.p. showed.

[c.pfaffenbichler]

Ah, I do apologize, the OP had asked about the Color Picker specifically but I guess I had misinterpreted

so you can see the actual color values

and assumed they wanted the readouts.

My mistake.

[Chuck Uebele]

I seem to recall an Adobe engineer explaining that things like the color picker and input and output values on curves are only in 8 bit values, as the user can't perceive 16 bit values, and it just complicating the PS code. As mentioned above, 16 bit is to give extra range when making corrections to avoid banding and such artifacts.

BTW, Dag, I too have scored a perfect score on that test, several times.