16 bit/channel mode

During what circumstances in photoshop do you convert to 16-bit or 32-bit mode? Since the lower mode is 8-bit by default are the higher modes reserved for images above 8-bits per channel? Since all jpgs are 8-bit (correct?) is there any benefit with working in photoshop with these images in 16-bit mode?--would that benefit remain if saving back as jpg or only if kept in a 16-bit psd, tiff, or equivalent?

I'm unsure if RAW photos from all cameras are higher than 8-bit so I avoid assuming the benefits are only with Tiff or RAW images. My Canon 5DmarkII claims "Its 14-bit analog/digital conversion allows you to create stunning 16-bit TIFF images from RAW 14-bit data.". But when I open images in ACR on the bottom it always says "Adobe RGB (1998); 8 bit ; 5616 by 3744 (21.0MP); 240ppi". At what point am I supposed to realize the supposed fact that these RAW files are 14-bit?

When it comes to HDR imaging I read higher bit becomes necessary. I understand that higher bits per channel allow an exponentially greater number of colors (shades per channel) but as far as when to use it I'm a little confused. Will someone please elaborate on this higher bit imaging principle and practices?

Thank you

sethmarshall said:

Since all jpgs are 8-bit (correct?) is there any benefit with working in photoshop with these images in 16-bit mode?

Click to expand...

Although "it couldn't hurt", there seems to be no practical advantage to converting an 8-bit file to 16-bit for processing. Seven years ago Dan Margulis, author of Professional Photoshop , issued a challenge for anyone to provide an 8-bit real-world color photo where reasonable post-processing gave better results in 16 bits than staying in 8 bits did. It was a number of years before anybody came across even one.

(By the way, Margulis is much more enthusiastic about converting the image to L*a*b* color space for processing.)

When you convert the RAW file, at least in ACR, you can select 8 or 16 bit. And a few color spaces. 16 bit can be helpful if your going to make locallized tonal and color adjustments once in Photoshop, but to be honest, I can't really say I have seen any real improvements by keeping files in 16 bit. Sometimes when I need to deal with critical shadow details it helps, but not an everyday thing

Granted, there's no point in converting an 8-bit file to 16 bits, but by shooting in raw mode, you can get the most out of your camera's 14-bit capture by choosing 16 bits (and also a larger color space like aRGB or ProPhoto) during the raw conversion.

Whether you need 8 or 16 bits really depends on the image captured, the type of subject you are shooting, and the intended output. In many cases you may not need any more than 8 bits, but in other cases you may need 16 bits in order to preserve a wide color gamut or the fine gradations of color withing a gradient. Shooting raw will give you this option, and during the raw conversion you may be able to tell from the type of capture (such as a dawn or sunset sky with color gradients, or vivid greens in a landscape) and from the distribution of colors in the histogram whether you need to choose a wider color gamut such as Adobe RGB or an even wider color space such as ProPhoto. You may find that choosing a wider color space with preserve more of what you captured, and you can often see the histogram spread out to a more normal looking curve when you select the wider color space, versus some colors being clipped at one end.

In these cases, choosing the wider color gamut will preserve these more saturated colors, including the finer gradations of these colors if you also choose 16-bits (which you should if working in a wider color gamut like aRGB or wider and you have captured vivid colors or color gradients). Maintaining these colors during processing by working with 16 bits in a wider color space will let you see the maximum range of colors in your final processed image, either displayed on a wide gamut monitor or printed on a capable printer (many can reproduce colors outside the limited sRGB color space). Even within a limited range of colors, the 16-bit processing can preserve color gradients better, such that you can avoid processing artifacts that may appear as posterization and banding within a color gradient, and may give you more latitude in processing a less-than-ideal exposure.

Whether you want to choose this method routinely or just on a selective basis, depending on the situation or whether you have a problem is up to you, but if you choose to shoot in raw mode then you will always have the option.

-BrianZ

I want to thank you all for your reply and helping me understand this better. My apologies for the late thank you but I had to leave town unexpectedly and haven't had access to my email..... okay, I got engaged and we ran off for a bit !

We went to Telluride, CO and I took a ton of photos. I've been working on developing my skills with HDR and time lapses (separately for now).

With HDR, would I want to make sure I'm working with the 16-bit file of all (3 in most of my cases) my images? Would this be a case to be certain I'm working in 16-bit? I use Photomatrix which as far as I know you can't tell if it's converting my RAW (DNG--it doesn't accept CR2) to 16-bit or not.

I say this because you've stated in ACR with photoshop I have to select if I want my images to be 16-bit or remain the default 8-bit. I'm surprised it's not 16-bit by default since isn't that where the RAW files hide the extra stop of DR?!!

As for time lapses. I'm shooting those RAW to hopefully pull out that extra stop of DR. What is the best way to do this?

As you can see I may be a little confused. I really hope to understand this better and I look forward to hearing your replies.

I do not know of any 16 bit monitors to show the superiority of 16 bit on screen, indeed, it is my belief that even the best of them are having trouble showing 8 bits of tonal variation in any channel.

As far as printing is concerned, I think the situation is worse. I know of no 16 bit printers, and even the with the strictly 8 bit devices we may be lucky to get 5 or 6 bit depth actually available..... (micro-densitometers in reflection mode are the tools to tell us, but we are probably better off not knowing!!)

So, when to use 16 bit, even if you can't ever see it?

Well, what we can see is the lack of 16 bit working when heavy tonal or colour shifts are edited into an image. This is when the vastly greater number of levels helps prevent posterisation, apparently.

On (inevitable) conversion to 8 bit for output, quantisation will throw away the thousands of 16 bit levels, but it is a once-only disposal at the end of the editing process, instead of multiple roundings-up and roundings-down accumulating in the data at every editing move.

This single quantisation is supposed to be better, but people are still undecided about whether it actually is.
--
Regards,
Baz

My epson R1800, Mac, & CS4 give me the option of sending 16 bit data to the printer.
Canon ipf9000 prints 16 bit, as do the other ipf series printers.

Working in 16 bit ProPhoto, and using a wide gamut inkjet, will yield colors beyond anything previously possible.

8 bit and sRGB throw out a huge chunk of the capabilities of your equipment.

dik said:

My epson R1800, Mac, & CS4 give me the option of sending 16 bit data to the printer.
Canon ipf9000 prints 16 bit, as do the other ipf series printers.

Click to expand...

The printers may accept 16 bit data, but I don't think anything more than 8 bit is ever sent to the paper. Leastways, if it was , I would be genuinely amazed.

Even if it turned out to be true, and output was (nominally) at 16 bit, I don't think your eyes or mine would ever see the difference.... and I'm pretty sure Epson and Canon know that, too.
--
Regards,
Baz

I'm not satisfied working to the limits of my own (limited) visual capacity, nor yours.

I expect people with better vision than my own to view my work, so I try to work to the technical limits of the available equipment.
It's not much harder to use the full capabilities of the gear.

A discussion over at luminous landscape -
http://www.luminous-landscape.com/forum/lofiversion/index.php/t9785.html
Will

dik said:

I'm not satisfied working to the limits of my own (limited) visual capacity, nor yours.

I expect people with better vision than my own to view my work, so I try to work to the technical limits of the available equipment.
It's not much harder to use the full capabilities of the gear.

Click to expand...

In tests most people demonstrate an ability to distinguish around 120 different levels of grey, slightly fewer if the levels are coloured. Some individuals can distinguish about 160, that's if they are young, in good health, and in good light!

So obviously, 256 levels is more than we need for photographic output, and likely more than we ever get in print, too.

Still... no point in limiting your creativity to only the tones you CAN SEE , eh... ;-)
--
Regards,
Baz

Sorry, but this is kind of misguided. Human vision doesn't work in discrete steps like imaging devices, and the gray levels, as you call it, are visualized completely differently than colors inside the eye. While there are 3 color receptors (in tetrachromats, 4), they don't correspond to R G and B, or C M Y K. Those are human invention to approximate.

Indeed, nobody is quite sure exactly how many colors can be perceived by humans, but best guesses are in the many tens of millions. And given genetic variations in the pigments of the color receptors, each individual made indeed perceive colors differently than the next. Well beyond 8 bit's capability. Read up on the history of color theory - it's very interesting. A unique blend of art, biology, chemistry, genetics, electronics and computation all factor in...

Also in regards to your earlier post, most newer printers include 16 bit drivers that yes, send 16 bit data to the printer. The newest epsons exceed adobe RGB gamut in many places.

Barrie Davis said:

dik said:

I'm not satisfied working to the limits of my own (limited) visual capacity, nor yours.

I expect people with better vision than my own to view my work, so I try to work to the technical limits of the available equipment.
It's not much harder to use the full capabilities of the gear.

Click to expand...

In tests most people demonstrate an ability to distinguish around 120 different levels of grey, slightly fewer if the levels are coloured. Some individuals can distinguish about 160, that's if they are young, in good health, and in good light!

So obviously, 256 levels is more than we need for photographic output, and likely more than we ever get in print, too.

Still... no point in limiting your creativity to only the tones you CAN SEE , eh... ;-)
--
Regards,
Baz

Click to expand...

nachos said:

Sorry, but this is kind of misguided. Human vision doesn't work in discrete steps like imaging devices, and the gray levels, as you call it, are visualized completely differently than colors inside the eye. While there are 3 color receptors (in tetrachromats, 4), they don't correspond to R G and B, or C M Y K. Those are human invention to approximate.

Click to expand...

Hmmm.... that's all a red herring, I'm afraid. The actual mechanisms of human vision are a side issue here.

What we are talking about is whether people can see a difference between 8 or 16 bit information, after it has been transferred through the medium of ink on paper.

That information is available to us, and the answer is NO, they can't.

If you have a 16 bit "capable" printer you should know this from a side-by-side comparisons of your own... or are you putting off doing the test?! ;-)

In fact, has anybody here ever SEEN a direct comparison? Even a demonstration produced by Epson or Canon themselves, presumably extolling the benefits of their new 16 bit handling, would give us something solid to argue over. Until then...... [???]

[Colour gamut is another matter and similarly not specially relavant here.]
--
Regards,
Baz

see http://www.facebook.com/note.php?created&&suggest&note_id=110352602295#/note.php?note_id=110352602295&ref=nf

Reichmann reviewed one of these a few months back: http://www.luminous-landscape.com/reviews/printers/7900-9900.shtml

It should be noted, that printing in 16 bit mode is a given on these machines. Professional printers that print day in and day out can easily tell the difference. Arguing over this is like arguing whether MF is higher quality than DSLR. You either get it, or you don't...

Barrie Davis said:

nachos said:

Sorry, but this is kind of misguided. Human vision doesn't work in discrete steps like imaging devices, and the gray levels, as you call it, are visualized completely differently than colors inside the eye. While there are 3 color receptors (in tetrachromats, 4), they don't correspond to R G and B, or C M Y K. Those are human invention to approximate.

Click to expand...

Hmmm.... that's all a red herring, I'm afraid. The actual mechanisms of human vision are a side issue here.

What we are talking about is whether people can see a difference between 8 or 16 bit information, after it has been transferred through the medium of ink on paper.

That information is available to us, and the answer is NO, they can't.

If you have a 16 bit "capable" printer you should know this from a side-by-side comparisons of your own... or are you putting off doing the test?! ;-)

In fact, has anybody here ever SEEN a direct comparison? Even a demonstration produced by Epson or Canon themselves, presumably extolling the benefits of their new 16 bit handling, would give us something solid to argue over. Until then...... [???]

[Colour gamut is another matter and similarly not specially relavant here.]
--
Regards,
Baz

Click to expand...

Re: inputting 16 or 8 bit data into HDR software.

If your exposures are close enough, it doesn't matter, because the values overlap. If your exposures are far apart though, using 16 bit may increase quality. I've tested this extensively, and as my exposures are close together and thoroughly contain the scene, no difference is apparent. I would guess that photomatix's internal RAW processor is using 16 bits; I do not use it however.

I find there is more use in outputting from the HDR software into 16 bit following tone mapping. Recall that HDR natively is using 32 bits internally.

As for outputting to 8 or 16 bit really is a matter of:
a. do you need the headroom
b. want to deal with the extra I/O time and storage space that 16 entails

If I shoot a 2000 image catalog, it is a non-trivial amount of time and space to hand over 16 bit tiff files. The client will complain.

So my clients receive 8 bit tiff files from me unless otherwise requested. Sometimes I'll send the retoucher 16 bits if I know she will be doing some extensively tonal shifts.

sethmarshall said:

I want to thank you all for your reply and helping me understand this better. My apologies for the late thank you but I had to leave town unexpectedly and haven't had access to my email..... okay, I got engaged and we ran off for a bit !

We went to Telluride, CO and I took a ton of photos. I've been working on developing my skills with HDR and time lapses (separately for now).

With HDR, would I want to make sure I'm working with the 16-bit file of all (3 in most of my cases) my images? Would this be a case to be certain I'm working in 16-bit? I use Photomatrix which as far as I know you can't tell if it's converting my RAW (DNG--it doesn't accept CR2) to 16-bit or not.

I say this because you've stated in ACR with photoshop I have to select if I want my images to be 16-bit or remain the default 8-bit. I'm surprised it's not 16-bit by default since isn't that where the RAW files hide the extra stop of DR?!!

As for time lapses. I'm shooting those RAW to hopefully pull out that extra stop of DR. What is the best way to do this?

As you can see I may be a little confused. I really hope to understand this better and I look forward to hearing your replies.

Click to expand...

nachos said:

Reichmann reviewed one of these a few months back: http://www.luminous-landscape.com/reviews/printers/7900-9900.shtml

Click to expand...

And, surprise surprise, the 16 bit working in these machines was so significant, that Riechman didn't comment on it at all in his review. I feel the same, that is, if there is a difference, it isn't worth mentioning!

nachos said:

It should be noted, that printing in 16 bit mode is a given on these machines. Professional printers that print day in and day out can easily tell the difference.

Click to expand...

Yeah... I guess I'd take the authority of a pro printer on this matter, if he could demonstrate what he was talking about.
Who did you have in mind? Can you get him to post?

nachos said:

Arguing over this is like arguing whether MF is higher quality than DSLR. You either get it, or you don't...

Click to expand...

The advantages of MF over smaller sensors are easy to see, and not much disputed..... (although there are still many circumstances where smaller sensors deliver sharper images, of course)

However, clear superiority is not the case with 16 bit printing, where I contend that a full 8 bits of information rarely hits the paper, and nobody would be able to tell even if 16 bits was actually achieved, say with new paper and ink technology....

.... because our eyes just aren't that good.

Hey! Did you know that we actually get only 15 bits out of "16 bit" Photoshop? HALF as many levels as people think.
--
Regards,
Baz

Barrie, like I said, using 16 bit is a given when using prophoto rgb. 8 bits while in prophoto rgb is self-defeating at best, visually degrading at worst.

Barrie Davis said:

nachos said:

Reichmann reviewed one of these a few months back: http://www.luminous-landscape.com/reviews/printers/7900-9900.shtml

Click to expand...

And, surprise surprise, the 16 bit working in these machines was so significant, that Riechman didn't comment on it at all in his review. I feel the same, that is, if there is a difference, it isn't worth mentioning!

Barrie Davis said:

It should be noted, that printing in 16 bit mode is a given on these machines. Professional printers that print day in and day out can easily tell the difference.

Click to expand...

Yeah... I guess I'd take the authority of a pro printer on this matter, if he could demonstrate what he was talking about.
Who did you have in mind? Can you get him to post?

Barrie Davis said:

Arguing over this is like arguing whether MF is higher quality than DSLR. You either get it, or you don't...

Click to expand...

The advantages of MF over smaller sensors are easy to see, and not much disputed..... (although there are still many circumstances where smaller sensors deliver sharper images, of course)

However, clear superiority is not the case with 16 bit printing, where I contend that a full 8 bits of information rarely hits the paper, and nobody would be able to tell even if 16 bits was actually achieved, say with new paper and ink technology....

.... because our eyes just aren't that good.

Hey! Did you know that we actually get only 15 bits out of "16 bit" Photoshop? HALF as many levels as people think.
--
Regards,
Baz

Click to expand...

nachos said:

Barrie, like I said, using 16 bit is a given when using prophoto rgb. 8 bits while in prophoto rgb is self-defeating at best, visually degrading at worst.

Click to expand...

Colour spaces is another red herring.... not relevant to the 16 bit printing matter.

Look, if you want to continue the printing discussion with me, then find a point to make that is about that topic. Otherwise, I'm outta here.
--
Regards,
Baz

What the heck? If you are using prophoto rgb you should be in 16 bit from capture to paper. As many modern printers note, the visible gamut (and yes, you can see it...) is higher than what can be resolved with Adobe RGB, so you use ProPhoto RGB. Most sensors exceed Adobe as well. So if you want to maximize the information, which is again, visible, you go to Prophoto RGB and you use 16 bit.

Therefore, 16 bits is highly relevant. Indeed, you could (technically) use 8 bit with it but this would be unwise.

I'm sorry Barrie, but you appear to not have much actual knowledge on this topic.

Our studio has been involved with high level digital imaging for over 13 years now and I believe we may know a thing or two

Barrie Davis said:

nachos said:

Barrie, like I said, using 16 bit is a given when using prophoto rgb. 8 bits while in prophoto rgb is self-defeating at best, visually degrading at worst.

Click to expand...

Colour spaces is another red herring.... not relevant to the 16 bit printing matter.

Look, if you want to continue the printing discussion with me, then find a point to make that is about that topic. Otherwise, I'm outta here.
--
Regards,
Baz

Click to expand...