PKR
Veteran
8, 12, 14 vs 16-Bit Depth: What Do You Really Need?!
By Greg Benz
PART 2 OF 3
Because Lightroom only allows +5 stops of exposure, I also adjusted the curve to bring in the top-right point to 80% for the both of the versions below. The first version (on top) is the processed 14-bit image. As you can see, there is tremendous shadow detail. There is of course noise in the image, but this is actually a printable file (though certainly not ideal). Lightroom’s white balance tool was easily able to use the gray card to get proper white balance.
This next variant is the processed 12-bit image. It is also surprisingly useful for such an extreme adjustment but has some clear issues. I used the exact same +5ev and curve adjustments. Lightroom was unable to get a proper white balance from the gray card, there is simply too much color noise at the pixel level in this file. So I copied the white balance from the above image, which resulted in an image which was slightly warm and definitely a bit too green.
I then manually corrected the image the best I could, but there were no white balance settings which looked fully correct nor matched the 14-bit file. The final image shows a residual color cast and greater contrast (with the shadow behind the towel being most notable). Much more concerning though is the splotchy color noise (which you can see in the lighter part of the towel shadow below). Also, tweaking the white balance just slightly more than I have here started to show some large grey splotches in the wood of the door. So while this result is “ok”, it is just shy of a disaster.
So there is an advantage to shooting with a 14-bit file on the Nikon D850, but it is relatively slight under extreme conditions. Even if portions of your shadows are this underexposed, I can’t see a scenario where you would fully correct them to a middle gray. 12-bit files are a very reasonable option. (I have not posted the Capture One results here, but both are worse, with the 12-bit file being truly terrible for this extreme underexposure.)
What if you have a fancy camera that captures 16-bit RAW files – should you be worried about Photoshop’s 15-bit quality? No. For several reasons:
The limiting factor is your RAW conversion software, not Photoshop. I don’t know if Lightroom uses 15+1 or true 16-bit math internally, but I suspect the latter. I don’t have a 16-bit camera to test. No matter which camera or RAW conversion software you use, it is best to do white balance and tonal adjustments in RAW before Photoshop for best results.
As noted above, 14-15 stops is plenty.
Camera companies can claim any bit depth they want, it does not mean that you are getting better quality. In other words, precision (the number of bits) and accuracy (the quality of the numbers stored with those bits) are not the same. Noise is a very good example of this discrepancy. I wouldn’t be surprised if you are not getting 16-bit accuracy from a 16-bit file, but that is speculation on my part. [Note that I’m not saying these aren’t excellent cameras that produce better images, they probably are – I’m just saying that I don’t think Photoshop’s 15+1 bit depth design is something to worry about when processing files from these cameras].
That said, using 16-bit capture should give you at least an extra bit in Photoshop and may be beneficial.
In summary:
Do not shoot JPG (8-bits)
A 12-bit RAW file is excellent for most work and offers significant space savings over 14-bit RAW. This is the best choice if you care about file size.
If you want the absolute best quality in the shadows, shoot 14+ bit RAW files (ideally with lossless compression to save space). This is the best choice if you don’t care about larger files and shoot scenes with wide dynamic range (deep shadows).
If you can shoot 16-bits, that’s fine but probably overkill. Worth testing your camera to see if you can use a lesser setting to save on file size.
How many bits should you use in Photoshop?
Based on the discussion above, it should be clear that 8-bits is not enough. It is possible to see banding immediately in 8-bits. And if you don’t see it right away, even modest adjustments can expose it. So go with 16-bits.
That holds true even if you are using an 8-bit source file (such as a stock image downloaded in JPG). Even if the source has been degraded, processing in 16-bits will still yield better results as it will minimize the compounding of rounding errors in the math with multiple adjustments.
There is no reason to use 32-bits for photography unless you are processing an HDR file.
How many bits do you need for sharing on the Internet?
The benefits of 16-bits are largely in the ability to manipulate the image without causing issues. Conversion of the final edited image to 8-bits is perfectly fine and has the advantage of creating much smaller files on the Internet for faster uploads/downloads. Be sure that Photoshop’s dithering is enabled. Go to Edit / Color Settings and make sure “Use dither (8-bit/channel images)” is checked. If you are using Lightroom to export to JPG, dithering is used automatically (you don’t have a choice). This helps add a bit of noise that should minimize the risk of any banding being exposed with the final conversion to 8-bits.
How many bits do you need for printing?
If you print at home, you can just create a copy of your 16-bit working file and finalize it (flatten, sharpen, change color space if needed, etc). But what about if you are sending your images over the internet to be printed by a pro lab? Many will accept 16-bit TIF files, and that’s a great way to go. However, if the vendor requires a JPG or you want to send a smaller file, you might be faced with questions about converting to 8-bits.
If your print lab accepts 16-bit formats (TIFF, PSD, JPEG2000), that’s probably the way to go – but ask your vendor what they recommend if you are unsure.
If you have to send a JPG, it will be in 8-bits, but that shouldn’t be a concern. In reality, 8-bits is fine for final print output. Remember that most issues with 8-bits are caused by making changes to 8-bit data, not the initial conversion. I have printed hundreds of very high-quality images that were uploaded to my vendor as 8-bit JPGs and the final images look amazing (exported from Lightroom with 90% quality and Adobe RGB color space). I would recommend making all other changes (flattening, color space conversion, sharpening, etc) before conversion to 8-bits.
If you don’t see banding on your monitor after conversion to 8-bits, you should be ok to print. However, you can help guard against potential issues by ensuring that Photoshop is using dithering for the conversion to 8-bits (see previous section).
What is the difference between Bit-Depth and Color Space?
Bit-depth determines the number of possible values or increments. Color Space determines the maximum values or range (commonly known as “gamut”). If you were to use a box of crayons as an example, greater bit-depth would be like having more shades (more crayons) and greater gamut would be like having the most saturated color be bolder (regardless of the number of crayons).
To see out the difference, consider the following simplified visual example:
As you can see, increasing bit-depth reduces the risk of banding by creating more increments, while expanding color space (wider gamut) enables the use of more extreme colors. But the two do interact because the jumps will get bigger if you use the same bit-depth with a wider gamut. And it is those jumps that relate to banding.
How does Color Space impact Bit Depth?
As you can see, increasing bit-depth reduces the risk of banding by creating more increments, while expanding color space (wider gamut) enables the use of more extreme colors. But the two do interact because the jumps will get bigger if you use the same bit-depth with a wider gamut. And it is those jumps that relate to banding.
How does Color Space impact Bit Depth?
After all that discussion, it really comes down to a few simple rules.
Camera settings:
A 14+ bit RAW file is a good choice if you want the best possible quality, particularly if you expect you may need to do extreme tonal adjustments (such as increasing shadow exposure by 3-4 stops).
A 12-bit RAW file is excellent under most conditions and should be used if you want to save file space or shoot more quickly. For my D850, a 14-bit RAW file is roughly 30% larger than a 12-bit one, so that’s an important consideration. And the larger files may impact your ability to shoot long continuous sequences as the camera’s buffer fills.
Never shoot JPG if you can avoid it. If you shoot live events, you might be the exception to the rule (to quickly upload and send images). Even still, you might consider using a JPG+RAW setting if you need a higher quality file too. I’d probably stick with sRGB as your camera color space if you do shoot JPG, as your work is probably just going on the web and a smaller gamut reduces risks of banding with 8-bits. If you are shooting RAW, you can ignore the color space setting (RAW files don’t really have a color space, it isn’t set until you convert the RAW file to another format).
Lightroom and Photoshop (working files):
Always save your working (layered) files in 16-bits. Only use 8-bits for your final output to JPG for sharing smaller files on the web (and printing if that’s what your vendor requires/prefers). It is ok to use 8-bits for final output, but it should be avoided at all costs prior to final output.
Be sure to zoom in to 67% or closer to make sure that any banding you see is not due to the way Photoshop previews layered file. This is a very common issue that causes the photographer to falsely believe there is banding in the image.
Be careful when using HSL in Lightroom and Adobe Camera RAW, as this tool is prone to color banding. This has very little to do with bit-depth but is a source of banding.
If your source file is only available in 8-bits (such as a stock JPG), you should immediately convert the layered working document to 16-bits. Subsequent edits on 8-bit images will not degrade as badly if that math is performed in a 16-bit mode.
Skip the 32-bit working space, unless you are using it as a way to combine multiple RAW files and then multi-process them as 16-bit layers (HDR workflows). There are massive feature limitations in the 32-bit space, workflow challenges, and the files are twice as big. I would generally recommend merging to HDR in Lightroom instead of using 32-bit Photoshop files.
Lightroom’s HDR DNG format is perfectly fine to use. (You may be aware that it uses 16-bit floating point math in order to cover a wider dynamic range with a similar number of bits. Keeping in mind that we only need to expand dynamic range a few stops with HDR and that we really only need 12-14 bits in a single RAW file, this is an acceptable format that increases quality without creating enormous files.) Of course, be sure to export from this RAW as a 16-bit TIF/PSD when you need to continue on to Photoshop.
If you are one of the few people who need to use an 8-bit workflow for some reason, it is probably best to stick with the sRGB color space. With a 16-bit workflow, I see no reason to worry about banding/posterization with ProPhoto RGB and I use ProPhoto RGB as my primary color space these days. I believe the concerns with ProPhoto are probably driven by theoretical concerns that not found in the real work, banding caused by use of HSL in RAW (ie, not related to the color space), false perception of banding when viewing layered files without zooming in, or using ProPhoto with 8-bit test files (because any loss of quality at 8-bits is a big deal). Others may disagree with me on this, but I have yet to push a file and find banding issues related to ProPhoto in 16-bits. You should always use 16-bits when working with ProPhoto, which makes the minor waste of bit-depth a non-issue.
When using Photoshop’s gradient tool, checking the “dithering” option creates the perception of 1 extra bit of detail. This may be helpful if working in an 8-bit file. For a 16-bit file, it is unnecessary and increases the size of the saved file (assuming you are using compression to save your files).
A better generalized solution for removing banding is described below.
Exporting to the web:
JPG with 8/bits and sRGB color space is ideal/standard. While some monitors are capable of displaying greater bit depth, the increased file size is probably not worth it. And while more and more monitors are capable of wider gamuts, not all browsers properly support color management and could display your images incorrectly. And most of those larger gamut monitors have probably not been color calibrated by their owners either. So, sadly, the lowest common denominator rules the internet for now.
Printing:
8-bits is fine for the final output, but go for 16 if your vendor supports it.
Monitor:
A standard monitor is fine. But be aware that you may potentially see some banding due to an 8-bit display that is not truly in the image.
If you can afford it, a 10-bit display is ideal if you aren’t on a budget. A wide gamut (ie Adobe RGB) monitor is also ideal. But neither is really necessary, and I’ve done plenty of high-end work on a standard monitor. Be sure to calibrate the monitor though if you are sending files out for print. I do critical work on a 27″ Eizo (CG2730).
By Greg Benz
PART 2 OF 3
Because Lightroom only allows +5 stops of exposure, I also adjusted the curve to bring in the top-right point to 80% for the both of the versions below. The first version (on top) is the processed 14-bit image. As you can see, there is tremendous shadow detail. There is of course noise in the image, but this is actually a printable file (though certainly not ideal). Lightroom’s white balance tool was easily able to use the gray card to get proper white balance.
This next variant is the processed 12-bit image. It is also surprisingly useful for such an extreme adjustment but has some clear issues. I used the exact same +5ev and curve adjustments. Lightroom was unable to get a proper white balance from the gray card, there is simply too much color noise at the pixel level in this file. So I copied the white balance from the above image, which resulted in an image which was slightly warm and definitely a bit too green.
I then manually corrected the image the best I could, but there were no white balance settings which looked fully correct nor matched the 14-bit file. The final image shows a residual color cast and greater contrast (with the shadow behind the towel being most notable). Much more concerning though is the splotchy color noise (which you can see in the lighter part of the towel shadow below). Also, tweaking the white balance just slightly more than I have here started to show some large grey splotches in the wood of the door. So while this result is “ok”, it is just shy of a disaster.
So there is an advantage to shooting with a 14-bit file on the Nikon D850, but it is relatively slight under extreme conditions. Even if portions of your shadows are this underexposed, I can’t see a scenario where you would fully correct them to a middle gray. 12-bit files are a very reasonable option. (I have not posted the Capture One results here, but both are worse, with the 12-bit file being truly terrible for this extreme underexposure.)
What if you have a fancy camera that captures 16-bit RAW files – should you be worried about Photoshop’s 15-bit quality? No. For several reasons:
The limiting factor is your RAW conversion software, not Photoshop. I don’t know if Lightroom uses 15+1 or true 16-bit math internally, but I suspect the latter. I don’t have a 16-bit camera to test. No matter which camera or RAW conversion software you use, it is best to do white balance and tonal adjustments in RAW before Photoshop for best results.
As noted above, 14-15 stops is plenty.
Camera companies can claim any bit depth they want, it does not mean that you are getting better quality. In other words, precision (the number of bits) and accuracy (the quality of the numbers stored with those bits) are not the same. Noise is a very good example of this discrepancy. I wouldn’t be surprised if you are not getting 16-bit accuracy from a 16-bit file, but that is speculation on my part. [Note that I’m not saying these aren’t excellent cameras that produce better images, they probably are – I’m just saying that I don’t think Photoshop’s 15+1 bit depth design is something to worry about when processing files from these cameras].
That said, using 16-bit capture should give you at least an extra bit in Photoshop and may be beneficial.
In summary:
Do not shoot JPG (8-bits)
A 12-bit RAW file is excellent for most work and offers significant space savings over 14-bit RAW. This is the best choice if you care about file size.
If you want the absolute best quality in the shadows, shoot 14+ bit RAW files (ideally with lossless compression to save space). This is the best choice if you don’t care about larger files and shoot scenes with wide dynamic range (deep shadows).
If you can shoot 16-bits, that’s fine but probably overkill. Worth testing your camera to see if you can use a lesser setting to save on file size.
How many bits should you use in Photoshop?
Based on the discussion above, it should be clear that 8-bits is not enough. It is possible to see banding immediately in 8-bits. And if you don’t see it right away, even modest adjustments can expose it. So go with 16-bits.
That holds true even if you are using an 8-bit source file (such as a stock image downloaded in JPG). Even if the source has been degraded, processing in 16-bits will still yield better results as it will minimize the compounding of rounding errors in the math with multiple adjustments.
There is no reason to use 32-bits for photography unless you are processing an HDR file.
How many bits do you need for sharing on the Internet?
The benefits of 16-bits are largely in the ability to manipulate the image without causing issues. Conversion of the final edited image to 8-bits is perfectly fine and has the advantage of creating much smaller files on the Internet for faster uploads/downloads. Be sure that Photoshop’s dithering is enabled. Go to Edit / Color Settings and make sure “Use dither (8-bit/channel images)” is checked. If you are using Lightroom to export to JPG, dithering is used automatically (you don’t have a choice). This helps add a bit of noise that should minimize the risk of any banding being exposed with the final conversion to 8-bits.
How many bits do you need for printing?
If you print at home, you can just create a copy of your 16-bit working file and finalize it (flatten, sharpen, change color space if needed, etc). But what about if you are sending your images over the internet to be printed by a pro lab? Many will accept 16-bit TIF files, and that’s a great way to go. However, if the vendor requires a JPG or you want to send a smaller file, you might be faced with questions about converting to 8-bits.
If your print lab accepts 16-bit formats (TIFF, PSD, JPEG2000), that’s probably the way to go – but ask your vendor what they recommend if you are unsure.
If you have to send a JPG, it will be in 8-bits, but that shouldn’t be a concern. In reality, 8-bits is fine for final print output. Remember that most issues with 8-bits are caused by making changes to 8-bit data, not the initial conversion. I have printed hundreds of very high-quality images that were uploaded to my vendor as 8-bit JPGs and the final images look amazing (exported from Lightroom with 90% quality and Adobe RGB color space). I would recommend making all other changes (flattening, color space conversion, sharpening, etc) before conversion to 8-bits.
If you don’t see banding on your monitor after conversion to 8-bits, you should be ok to print. However, you can help guard against potential issues by ensuring that Photoshop is using dithering for the conversion to 8-bits (see previous section).
What is the difference between Bit-Depth and Color Space?
Bit-depth determines the number of possible values or increments. Color Space determines the maximum values or range (commonly known as “gamut”). If you were to use a box of crayons as an example, greater bit-depth would be like having more shades (more crayons) and greater gamut would be like having the most saturated color be bolder (regardless of the number of crayons).
To see out the difference, consider the following simplified visual example:
As you can see, increasing bit-depth reduces the risk of banding by creating more increments, while expanding color space (wider gamut) enables the use of more extreme colors. But the two do interact because the jumps will get bigger if you use the same bit-depth with a wider gamut. And it is those jumps that relate to banding.
How does Color Space impact Bit Depth?
As you can see, increasing bit-depth reduces the risk of banding by creating more increments, while expanding color space (wider gamut) enables the use of more extreme colors. But the two do interact because the jumps will get bigger if you use the same bit-depth with a wider gamut. And it is those jumps that relate to banding.
How does Color Space impact Bit Depth?
After all that discussion, it really comes down to a few simple rules.
Camera settings:
A 14+ bit RAW file is a good choice if you want the best possible quality, particularly if you expect you may need to do extreme tonal adjustments (such as increasing shadow exposure by 3-4 stops).
A 12-bit RAW file is excellent under most conditions and should be used if you want to save file space or shoot more quickly. For my D850, a 14-bit RAW file is roughly 30% larger than a 12-bit one, so that’s an important consideration. And the larger files may impact your ability to shoot long continuous sequences as the camera’s buffer fills.
Never shoot JPG if you can avoid it. If you shoot live events, you might be the exception to the rule (to quickly upload and send images). Even still, you might consider using a JPG+RAW setting if you need a higher quality file too. I’d probably stick with sRGB as your camera color space if you do shoot JPG, as your work is probably just going on the web and a smaller gamut reduces risks of banding with 8-bits. If you are shooting RAW, you can ignore the color space setting (RAW files don’t really have a color space, it isn’t set until you convert the RAW file to another format).
Lightroom and Photoshop (working files):
Always save your working (layered) files in 16-bits. Only use 8-bits for your final output to JPG for sharing smaller files on the web (and printing if that’s what your vendor requires/prefers). It is ok to use 8-bits for final output, but it should be avoided at all costs prior to final output.
Be sure to zoom in to 67% or closer to make sure that any banding you see is not due to the way Photoshop previews layered file. This is a very common issue that causes the photographer to falsely believe there is banding in the image.
Be careful when using HSL in Lightroom and Adobe Camera RAW, as this tool is prone to color banding. This has very little to do with bit-depth but is a source of banding.
If your source file is only available in 8-bits (such as a stock JPG), you should immediately convert the layered working document to 16-bits. Subsequent edits on 8-bit images will not degrade as badly if that math is performed in a 16-bit mode.
Skip the 32-bit working space, unless you are using it as a way to combine multiple RAW files and then multi-process them as 16-bit layers (HDR workflows). There are massive feature limitations in the 32-bit space, workflow challenges, and the files are twice as big. I would generally recommend merging to HDR in Lightroom instead of using 32-bit Photoshop files.
Lightroom’s HDR DNG format is perfectly fine to use. (You may be aware that it uses 16-bit floating point math in order to cover a wider dynamic range with a similar number of bits. Keeping in mind that we only need to expand dynamic range a few stops with HDR and that we really only need 12-14 bits in a single RAW file, this is an acceptable format that increases quality without creating enormous files.) Of course, be sure to export from this RAW as a 16-bit TIF/PSD when you need to continue on to Photoshop.
If you are one of the few people who need to use an 8-bit workflow for some reason, it is probably best to stick with the sRGB color space. With a 16-bit workflow, I see no reason to worry about banding/posterization with ProPhoto RGB and I use ProPhoto RGB as my primary color space these days. I believe the concerns with ProPhoto are probably driven by theoretical concerns that not found in the real work, banding caused by use of HSL in RAW (ie, not related to the color space), false perception of banding when viewing layered files without zooming in, or using ProPhoto with 8-bit test files (because any loss of quality at 8-bits is a big deal). Others may disagree with me on this, but I have yet to push a file and find banding issues related to ProPhoto in 16-bits. You should always use 16-bits when working with ProPhoto, which makes the minor waste of bit-depth a non-issue.
When using Photoshop’s gradient tool, checking the “dithering” option creates the perception of 1 extra bit of detail. This may be helpful if working in an 8-bit file. For a 16-bit file, it is unnecessary and increases the size of the saved file (assuming you are using compression to save your files).
A better generalized solution for removing banding is described below.
Exporting to the web:
JPG with 8/bits and sRGB color space is ideal/standard. While some monitors are capable of displaying greater bit depth, the increased file size is probably not worth it. And while more and more monitors are capable of wider gamuts, not all browsers properly support color management and could display your images incorrectly. And most of those larger gamut monitors have probably not been color calibrated by their owners either. So, sadly, the lowest common denominator rules the internet for now.
Printing:
8-bits is fine for the final output, but go for 16 if your vendor supports it.
Monitor:
A standard monitor is fine. But be aware that you may potentially see some banding due to an 8-bit display that is not truly in the image.
If you can afford it, a 10-bit display is ideal if you aren’t on a budget. A wide gamut (ie Adobe RGB) monitor is also ideal. But neither is really necessary, and I’ve done plenty of high-end work on a standard monitor. Be sure to calibrate the monitor though if you are sending files out for print. I do critical work on a 27″ Eizo (CG2730).
