8 Bit or 16 bit , sRGB or Pro Photo

Started Sep 11, 2019 | Discussions
OP Ken60 Senior Member • Posts: 2,438
Re: The difference is ...

NAwlins Contrarian wrote:

The visible difference there is caused by ProPhoto RGB versus sRGB. Even the with Bill's Balls test image that you used,* I'd bet that $100 that to our eyes the differences between 16-bit and 8-bit files will not show,

OK this is great if you had the decency, you should have watched the video I linked.

The DigitalDog himself runs the test on his printer , made by Epson, using 16 bit in both, and the results are similar- nay the same !  So then if you read his article my first posting he states. " if you are going to have a bigger space, you might as well put finer data into it !"   After all many of his findings and mine indicate that as the visual increase in decipherable tones within the space offer more delineation.

Frankly the rest of the bilge you spout is in mitigation that presented with an image, on both Canon and Epson , there is a point in offering greater gamut and depth of bit data.

As I said before IF you care to argue, do it with print as I have , after images talk for themselves. So the next move by you is to download the test image and print it .

Despite all your claims and knowledge I am yet to see a print from you to back anything up.

-- hide signature --

Gear ... what I need to get the job done , after all you don't see mechanics listing their brand of spanner as a qualification .

elliotn Senior Member • Posts: 2,308
Re: The difference is ...
1

Ken60 wrote:

NAwlins Contrarian wrote:

The visible difference there is caused by ProPhoto RGB versus sRGB. Even the with Bill's Balls test image that you used,* I'd bet that $100 that to our eyes the differences between 16-bit and 8-bit files will not show,

OK this is great if you had the decency, you should have watched the video I linked.

The DigitalDog himself runs the test on his printer , made by Epson, using 16 bit in both, and the results are similar- nay the same ! So then if you read his article my first posting he states. " if you are going to have a bigger space, you might as well put finer data into it !" After all many of his findings and mine indicate that as the visual increase in decipherable tones within the space offer more delineation.

Frankly the rest of the bilge you spout is in mitigation that presented with an image, on both Canon and Epson , there is a point in offering greater gamut and depth of bit data.

As I said before IF you care to argue, do it with print as I have , after images talk for themselves. So the next move by you is to download the test image and print it .

Despite all your claims and knowledge I am yet to see a print from you to back anything up.

It is very nearly impossible to make a pair of prints that demonstrates the superiority of a 16-bit print pipeline. And it is totally impossible if you are not printing synthetic gradients.

gscotten
gscotten Senior Member • Posts: 2,020
Re: The difference is ...
1

I printed out the test file from Lightroom to my Pro-100 first using the 8 bit driver and then the 16 bit XPS driver. Then, because I don't know how Lightroom's print engine works, I printed it again with Qimage using the XPS driver. Aside from the slightly better sharpening/scaling from Qimage, the three prints are virtually identical.

The iPhone photo below shows a difference in ambient light across the three prints as indicated by the whites of the backgrounds, but in the same light they look the same.

I have never seen a difference between the 8 bit and 16 bit outputs from the same file.

Ken's obviously visible differences are clearly due to the differing color spaces chosen, not the bit depth nor the driver. There may be some banding in images caused by different bit depths, but that is not apparent in his images.

The synthetic ball test contains colors that are significantly out of gamut on every printer I know of.

-- hide signature --

George

 gscotten's gear list:gscotten's gear list
Canon EOS RP Canon RF 24-240mm F4-6.3
technoid Senior Member • Posts: 2,295
Re: The difference is ...
1

gscotten wrote:

I printed out the test file from Lightroom to my Pro-100 first using the 8 bit driver and then the 16 bit XPS driver. Then, because I don't know how Lightroom's print engine works, I printed it again with Qimage using the XPS driver. Aside from the slightly better sharpening/scaling from Qimage, the three prints are virtually identical.

The iPhone photo below shows a difference in ambient light across the three prints as indicated by the whites of the backgrounds, but in the same light they look the same.

I have never seen a difference between the 8 bit and 16 bit outputs from the same file.

Nor have I. On  exactly one occasion I've seen a difference between 8 or 16 bit printed tiff files and JPEG but that's because jpeg doesn't actually render a full 8 bits of resolution. It compresses and throws out information that exists in an 8 bit tiff file. Even so I've only see one file where the difference was visible using highest quality jpegs. Lower quality jpegs frequently suffer banding but lets not go there.

Ken's obviously visible differences are clearly due to the differing color spaces chosen, not the bit depth nor the driver. There may be some banding in images caused by different bit depths, but that is not apparent in his images.

Yep. Exactly so.

The synthetic ball test contains colors that are significantly out of gamut on every printer I know of.

Bill's balls in ProPhoto have many colors that can not only never be printed by any theoretical printer, they also have lots of colors that are "imaginary colors" and are mathematical constructs that don't exist in the human color gamut. They are most useful in evaluating how smoothly a profile's perceptual intent maps these out of gamut "colors" to something pleasing.

The balls were originally designed by Bill Atkinson to be printed from device space, not ProPhoto, in order to visually examine how smoothly RGB printers mapped their gamut boundaries. The source for this was an email from Bill that Andrew Rodney helpfully shared on this forum some years back.

OP Ken60 Senior Member • Posts: 2,438
Re: The difference is ...

Ok lots of replies and comment , I too have been printing with the Pro Photo and 8 / 16 bit. Yes the differences are small , however I have one or two things I want to eliminate before drawing a final conclusion.

When many speak of a colour space they almost give it colour. It has none , it is only a container into which we place data and allow the space to define the interrelationship between that data and its use. So an empty container ....... As this space grows from sRGB to ProPhoto so with the same data content (8 bit) , the transitions must show differently to a higher, more intricately defined amount of data.

I read that the advantages of the Pro Photo space are evident in the darker and more saturated shades it allows. So what does more defined ( greater numeric detail 16 bit ) offer. I feel that it is almost akin to the old jpeg/tiff debate. The tiff has all the tones with no discard to simplify ............ it is at this point I started to look at a monochrome print from Iceland, the black beach and basalt cliffs against a very heavy sky, with white froth running against the black sand. I don't believe my old eyes , or desire to see something achieved deceive me. I can start to see in the darkest tones and in the lighter froth, more contrast in the 8 bit. Is it contrast or did the 8 bit run out of numbers and simplify the ramp , causing a more noticeable transition.

I am not ready to draw a final conclusion for myself, we all have ours, however I have a few lids to lift before I can finalise my thoughts.

One is 8 bit Pro Photo the other is 16 bit Pro Photo ............Canson Platine , Custom profile, perceptual

-- hide signature --

Gear ... what I need to get the job done , after all you don't see mechanics listing their brand of spanner as a qualification .

technoid Senior Member • Posts: 2,295
Re: The difference is ...

Ken60 wrote:

Ok lots of replies and comment , I too have been printing with the Pro Photo and 8 / 16 bit. Yes the differences are small , however I have one or two things I want to eliminate before drawing a final conclusion.

When many speak of a colour space they almost give it colour. It has none , it is only a container into which we place data and allow the space to define the interrelationship between that data and its use. So an empty container ....... As this space grows from sRGB to ProPhoto so with the same data content (8 bit) , the transitions must show differently to a higher, more intricately defined amount of data.

There are areas where the finer granularity of sRGB compared to ProPhoto can result in smaller dE's between a high bit RGB value and an 8 bit RGB value. The difference can be significant if the rounding of the high bit values happens at the most sensitive RGB values. Specifically:

High bit sRGB (31.51 29.49 27.51) and 8 bit (rounded) sRGB (32,29,28) Produces a dE2000 of 1.07.

ProPhoto RGB is more sensitive:

High bit ProPhoto RGB (19.49,16.51,9.49) and 8 bit (rounded) ProPhoto RGB (19,17,9) Produces a dE2000 of 2.44.

So patches of these RGB values in 8 and 16 bits, side by side, will show visible differences. For sRGB, just barely. For ProPhoto RGB, more distinctly visible. You should be able to see the differences with a good, 30 bit (3x10) monitor by creating side by side patches in Photoshop.

Virtually all pixels aren't anywhere near this bad and the noise intrinsic in a photo image, as opposed to a synthetic image, makes these anomalies invisible as they are subsumed in the dithering that naturally occurs by the printer as it spreads tiny differences over larger areas because of the limited number of inks.

But it should be clear that the workflow and especially, any editing, should be done in 16 bits. Printing photos will be fine if done in 8 bits if necessary but one should try to stay in high bit prior to that.

knickerhawk Veteran Member • Posts: 7,064
Re: The difference is ...

technoid wrote:

Ken60 wrote:

Ok lots of replies and comment , I too have been printing with the Pro Photo and 8 / 16 bit. Yes the differences are small , however I have one or two things I want to eliminate before drawing a final conclusion.

When many speak of a colour space they almost give it colour. It has none , it is only a container into which we place data and allow the space to define the interrelationship between that data and its use. So an empty container ....... As this space grows from sRGB to ProPhoto so with the same data content (8 bit) , the transitions must show differently to a higher, more intricately defined amount of data.

There are areas where the finer granularity of sRGB compared to ProPhoto can result in smaller dE's between a high bit RGB value and an 8 bit RGB value. The difference can be significant if the rounding of the high bit values happens at the most sensitive RGB values. Specifically:

High bit sRGB (31.51 29.49 27.51) and 8 bit (rounded) sRGB (32,29,28) Produces a dE2000 of 1.07.

ProPhoto RGB is more sensitive:

High bit ProPhoto RGB (19.49,16.51,9.49) and 8 bit (rounded) ProPhoto RGB (19,17,9) Produces a dE2000 of 2.44.

So patches of these RGB values in 8 and 16 bits, side by side, will show visible differences. For sRGB, just barely. For ProPhoto RGB, more distinctly visible. You should be able to see the differences with a good, 30 bit (3x10) monitor by creating side by side patches in Photoshop.

Virtually all pixels aren't anywhere near this bad and the noise intrinsic in a photo image, as opposed to a synthetic image, makes these anomalies invisible as they are subsumed in the dithering that naturally occurs by the printer as it spreads tiny differences over larger areas because of the limited number of inks.

But it should be clear that the workflow and especially, any editing, should be done in 16 bits. Printing photos will be fine if done in 8 bits if necessary but one should try to stay in high bit prior to that.

Nicely explained. The image below might help to illustrate what you're saying. This is constructed from screen grabs (8-bit on my iMac) of a synthetic gradient in Photoshop. The top half consists of two strips (8-bit and 16-bit) of the same gradient in ProPhoto. The bottom half consists of two strips (also 8-bit and 16-bit) of the same gradient in sRGB. Note that when viewed in Original Size, you should be able to see a visible difference in the top ProPhoto renderings due to the visible banding in the 8-bit version. Virtually no banding is visible in the 8-bit sRGB rendering compared to the 16 bit rendering. The finer "granularity" of the tonal steps in this region of the sRGB color space means less risk of tonal banding. Of course, the naturally occurring dithering effect of noise in real photos usually makes this a non-issue.

Top quarter= Prophoto 8 bit; second quarter= ProPhoto 16 bit; third quarter = sRGB 8 bit; bottom quarter = sRBB 16 bit

OP Ken60 Senior Member • Posts: 2,438
Re: The difference is ...

Sure , despite all this pixel , or dot , peeping .... I truly think the 16 bit workflow all the way to the printer allows the ability to keep the image on screen in photoshop and make little changes, aesthetic adjustments , and go right out to print without having to reduce the bit depth.

Another little note from this set of prints is the blue ball. If you peek at the first set of charts I posted, look at the top right blue ball. Specifically at the outer area of darker tones.... horrid in the 16 bit Pro Photo and quite clean in 8 bit sRGB ! So much for the dedicated blue cartridge of the Pro 1000.

Anyone care to say which black ball four posts or so back , is the 8 bit Pro and which the 16 bit Pro ?

-- hide signature --

Gear ... what I need to get the job done , after all you don't see mechanics listing their brand of spanner as a qualification .

OP Ken60 Senior Member • Posts: 2,438
Re: 8 Bit or 16 bit , sRGB or Pro Photo
1

Canon's Pro1000 XPS driver does not print using 16 bits in Windows 10 on my system. And Canon's Print Studio Pro doesn't print 16 bits either even though there is an option that supposedly does that.

After your declaration I posed this question to Canon CPS, asking about the XPS driver and the data sent from the software to the printer. I asked if the data was read as 16 bit into the software , and then sent as 8 bit to the ImagePrograf Pro 1000. I cut and paste the reply from Canon CPS today :-

Thank you for your request and for contacting Canon.

The 16-bit XPS driver will send 16 bit data to the printer when used with a 16 bit capable software such as Adobe Photoshop.

-- hide signature --

Gear ... what I need to get the job done , after all you don't see mechanics listing their brand of spanner as a qualification .

technoid Senior Member • Posts: 2,295
Re: 8 Bit or 16 bit , sRGB or Pro Photo

Ken60 wrote:

Canon's Pro1000 XPS driver does not print using 16 bits in Windows 10 on my system. And Canon's Print Studio Pro doesn't print 16 bits either even though there is an option that supposedly does that.

After your declaration I posed this question to Canon CPS, asking about the XPS driver and the data sent from the software to the printer. I asked if the data was read as 16 bit into the software , and then sent as 8 bit to the ImagePrograf Pro 1000. I cut and paste the reply from Canon CPS today :-

Thank you for your request and for contacting Canon.

The 16-bit XPS driver will send 16 bit data to the printer when used with a 16 bit capable software such as Adobe Photoshop.

Strange isn't it? I've read similar claims from Canon re the Pro 1000 printing in 16 bits with the xps driver and was looking forward to it since my Epson can't.

I make a number of specialized charts, some of which could benefit from the higher resolution of a higher bit driver. So I was rather disappointed when I tested it and it produced the same banding as the 8 bit driver. Banding that is completely consistent with 8 bit printing.

However, since banding between single rgb steps is not visible,  as gray scale dEs are small and typically less than .5, identifying whether a printer claim to use 16 bits is true requires scanning (or photographing) a gradient print and increasing contrast sufficiently to boost the dEs between bands into visibility.

I posted images of this result in this thread.

On the positive side, given how hard it is to actually test, it's easy to see why printing with 8 bits or 16 bits makes no practical difference in print quality.

OP Ken60 Senior Member • Posts: 2,438
Re: 8 Bit or 16 bit , sRGB or Pro Photo

I have to say that the only time I feel I have almost found the subtle changes in gradient is when it is a long monochromatic range.

As I said before, all I can say is that there is not as much contrast. I feel this is because the 8 bit (256)  shade ramp leaves fudged steps of contrast , where the 16 bit would provide a more even tone grading.   Again the changes could almost be analogous to the debate between the semi medium format digital and the modern full frame sensor rendering.    In my "black beach" cliffs print there is a feel that the deep areas of near black, almost open slightly where the greater tonal range allows.

-- hide signature --

Gear ... what I need to get the job done , after all you don't see mechanics listing their brand of spanner as a qualification .

technoid Senior Member • Posts: 2,295
Re: 8 Bit or 16 bit , sRGB or Pro Photo

I just ran another test with a gradient at 1" wide at 600 PPI from RGB 254 to 255 in 16 bits in Photoshop. It's surrounded  by a thin border so I could find it.

I then printed in in 16 bits using the highest rez settings possible on the Pro1000.

Next, I converted it to 8 bits in Photoshop making sure that the settings->color>dither option was off. This option dithers the conversions of 16 bits to 8 bits and I want RGB 255 on the left and RGB 255 on the right. Then I and printed it.

I can't see anything different unless looked at with a loupe where the light gray dots show up on the left half but not on the right halves of both prints.

The dots are sparse. To capture it I scanned it at 2400 PPI.  To enhance visibility of the light-light gray dots the image contrast was increased.

As you can see, the "16" bit print shows exactly the same pattern as the 8 bit one and it looks like it just rounds the 16 bits to 8 bits exactly the same as converting the image to 8 bits does.

Since the forum doesn't allow posting of 16 bit tifs, I can't post the originals used to print these but they are easy enough to create in Photoshop with the gradient tool setting the range from 254 to 255 using 16 bits.

Here's the contrast enhanced two images scanned at 2400 DPI (to see the dots) .

OP Ken60 Senior Member • Posts: 2,438
Re: 8 Bit or 16 bit , sRGB or Pro Photo

Go into photoshop and go to colour settings..... look for "dither 8 bit" and uncheck it.

Close Photoshop and download the chart I linked. open it as a 16 bit Prophoto  and call  it 16 bit. Now convert it to 8 bit and save as 8 bit. With both open and select two windows side by side , zooming in on the black and white ball.

Now look at the two balls in 16 & 8 bit together each covering half your screen. What you will start to see is the real 8 bit , with all the horrid things that it can deliver, not the faked version of 8 bit you are claiming is as good as 16 bit ...... but dithered to fake it.

-- hide signature --

Gear ... what I need to get the job done , after all you don't see mechanics listing their brand of spanner as a qualification .

technoid Senior Member • Posts: 2,295
Re: 8 Bit or 16 bit , sRGB or Pro Photo

Ken60 wrote:

Go into photoshop and go to colour settings..... look for "dither 8 bit" and uncheck it.

Close Photoshop and download the chart I linked. open it as a 16 bit Prophoto and call it 16 bit. Now convert it to 8 bit and save as 8 bit. With both open and select two windows side by side , zooming in on the black and white ball.

Now look at the two balls in 16 & 8 bit together each covering half your screen. What you will start to see is the real 8 bit , with all the horrid things that it can deliver, not the faked version of 8 bit you are claiming is as good as 16 bit ...... but dithered to fake it.

Ok, but I need some specific info. I presume the image you are referring to is Andrew Rodney's Gamut_Test_File_Flat.tif, which is a 16 bit tif in ProPhoto RGB.  Is this correct? I have downloaded that image.

There are 14 balls in the image two of which are black and white,  So which one are you referring to? I'm surprised you see any difference at all. Could you post a screen shot of the comparison?" I'll then try to duplicate your image.

Thanks.

knickerhawk Veteran Member • Posts: 7,064
Re: 8 Bit or 16 bit , sRGB or Pro Photo

Ken60 wrote:

Go into photoshop and go to colour settings..... look for "dither 8 bit" and uncheck it.

Close Photoshop and download the chart I linked. open it as a 16 bit Prophoto and call it 16 bit. Now convert it to 8 bit and save as 8 bit. With both open and select two windows side by side , zooming in on the black and white ball.

Now look at the two balls in 16 & 8 bit together each covering half your screen. What you will start to see is the real 8 bit , with all the horrid things that it can deliver, not the faked version of 8 bit you are claiming is as good as 16 bit ...... but dithered to fake it.

The dither on/off switch doesn't make that much of a difference here even with a purely synthetic image as is the case with the balls. The banding will just look a little more jagged. With a naturally dithered image, which is usually the case with photographs that aren't hugely edited, the dithering setting will be even less relevant.

More importantly, your experiment is failing to address a fundamental question: what is the bit-depth of your monitor? At most, your monitor is no more than 10-bits deep (and it may not even be working at a true 10-bit depth even if theoretically capable of doing so). Nevertheless, you're able to see even on an 8 bit monitor a difference between the original 16 bit image and the 8-bit truncated version of the image. The 16 bit version looks better (at least it does on my 8-bit iMac monitor). You can see this for yourself by looking at the screen captures in my post here. Those are screen grabs from my 8-bit monitor, yet the 16 bit ProPhoto strip clearly has less noticeable banding than the 8-bit ProPhoto strip.  Bottom line: viewing a 16 bit image on an 8 bit monitor is not the same thing as converting the 16 bit image to an 8 bit image and viewing the 8 bit image on an 8 bit monitor. Similarly, sending a 16 bit image to a printer through an 8 bit driver is not necessarily the same thing as converting the 16 bit image to an 8 bit image and sending that 8 bit image to a printer through an 8 bit driver.

OP Ken60 Senior Member • Posts: 2,438
Re: 8 Bit or 16 bit , sRGB or Pro Photo

Sorry been a bit busy today......... Yes Rodney's Balls !

Simply in Adobe Photoshop colour settings there is a check box for dithering 8 bit images, and this is always checked.... so uncheck it to see the real 8 bit without PS trying to make it look more like 16 bit.

Now restart and open the test file , which you can open in 16 bit Prophoto and then duplicate. Now convert one to 8 bit and  select , window arrange 2 up , and fill your screen with enlarged view of either of the black and white balls .  As you zoom in you will see ( window arrange match all to keep even)  the more real 8 v 16 image.   Of course dithering is never 100% off , your screen might not be optimal , but its an idea that when we talk of 8 bit we often dither to mimic 16 bit.

-- hide signature --

Gear ... what I need to get the job done , after all you don't see mechanics listing their brand of spanner as a qualification .

technoid Senior Member • Posts: 2,295
Re: The difference is ...
1

Ken60 wrote:

Sure , despite all this pixel , or dot , peeping .... I truly think the 16 bit workflow all the way to the printer allows the ability to keep the image on screen in photoshop and make little changes, aesthetic adjustments , and go right out to print without having to reduce the bit depth.

Another little note from this set of prints is the blue ball. If you peek at the first set of charts I posted, look at the top right blue ball. Specifically at the outer area of darker tones.... horrid in the 16 bit Pro Photo and quite clean in 8 bit sRGB ! So much for the dedicated blue cartridge of the Pro 1000.

Anyone care to say which black ball four posts or so back , is the 8 bit Pro and which the 16 bit Pro ?

Hi Ken,

Just got some time and loaded Gamut_test_file_flat.tif in PS. It's 16 bit, ProPhoto RGB. I deselected Edit-Color Settings->Use Dither.

I duplicated it then selected Windows->Arrange->Match All so the two tabs are in the same position. This makes it easy to switch back and forth from one image to another and see even the most subtle changes. Much better than side by side.

Then I converted the duplicate to 8 bit using: Image->Mode->8 Bits/channel. It's still in ProPhoto RGB of course.

Switching tabs back and forth from the 16 bit to 8 bit versions I see no difference at all. Not even subtle changes. Just nothing at all.

Next, I set up View->Proof Setup->Custom selecting my printer profile and Rel. Col. with BPC and applied it to both tabs. There is, as expected, a big change. However, after applying soft proof to both the 16 bit and 8 bit images and toggling the tabs back and forth there was again, no visible change.

However, my monitor was set to sRGB emulation so I switched to the wide, native monitor gamut and repeated the above tests. Colors were more saturated of course as the image is something of a stress test (for printers actually) but once again there were no observable differences between the 16 bit and 8 bit image.

Further info: My main monitor is an Eizo 319CG running in 30 bit mode, Quadro K2000D using ColorNavigator to switch display gamuts.

So if you are seeing changes on your monitor from 16 bit to 8 bit mode something in your system setup is messed up.

OP Ken60 Senior Member • Posts: 2,438
Re: The difference is ...

I am not running a 30 bit screen, my graphics card is a Quadro M2000.

Viewing the balls in 8 and 16, enlarged to fill the screen width , one above another , I can plainly see banding in the 8 bit image , as I would expect with such a tone rich ramp.

-- hide signature --

Gear ... what I need to get the job done , after all you don't see mechanics listing their brand of spanner as a qualification .

Keyboard shortcuts:
FForum MMy threads