2
\$\begingroup\$

I was watching a video about frequency separation techniques for skin retouching in Photoshop. The subject is irrelevant, but I mention it because this comes from a decent source and a high-end technique.

First thing I see, the person (a knowledgeable photographer / editor / retoucher who works all the time with Photoshop and has many deep tutorials on his site and on YouTube) transforms his RAW file into 8 bit TIFF files.

RAW files are 12 ~ 14 bits. I'm pretty sure he knows that. Why an 8 bit TIFF? This was a given for him, so I'm puzzled.

\$\endgroup\$
1
  • 2
    \$\begingroup\$ Not a single answer saying anything about gamma, just wow. \$\endgroup\$ Nov 12, 2017 at 17:42

4 Answers 4

3
\$\begingroup\$

RAW files are 12 ~ 14 bits. I'm pretty sure he knows that. Why an 8 bit TIFF? This was a given for him, so I'm puzzled.

The higher bit depth is certainly safer for major corrections of exposure, contrast etc. I would especially be cautious when using ProPhoto RGB that may have tendency to posterization in 8 bit. But 8 bit may be enough for his type of photography and it is usually enough as a final format for print and obviously for web presentations.

From what I've read, and from the fact that both our monitors AND the CMYK printers we use (therefore, "anything in which we can see our pictures these days") cannot go beyong an Adobe RGB color space.

Some inkjet printers go beyond.

\$\endgroup\$
1
  • \$\begingroup\$ Thanks! I did notice, while comparing data on color spaces in various places in the web, that some printing devices did print outside of AdobeRGB. \$\endgroup\$ Aug 18, 2014 at 6:09
5
\$\begingroup\$

The only advantages to saving your RAW files as 8-bit is for memory conservation or if certain tools only work with 8-bit images. There is no advantage from a quality point of view, if you're going to do a lot of editing especially in a wide colour space then you may get posterisation when working with only 8 bits.

Regarding colour spaces, it is advisable to use the widest space possible when editing, even if it is wider than your target colour space. This is because you might do something that moves the colours outside the narrow gamut only to perform a later step that moves them back in. If you were editing in that narrow gamut space the colours would get clipped and the final result would be different.

\$\endgroup\$
1
  • \$\begingroup\$ ... and therefore it is indeed a good practice to always use ProPhoto as it is the widest space commonly available and also a good remaping choice while converting from RGB to CMYK (another issue, though). \$\endgroup\$ Aug 20, 2014 at 2:45
5
\$\begingroup\$

The reasons why 8 bit TIFF is acceptable are:

  • raw files are typically linear and most used profiles (including AdobeRGB, sRGB, ProPhoto and whatnot) use gamma-encoding. Read more about it here. 8 bit gamma 2.2 encoding is roughly same as 16 bit linear when taking human vision as a reference
  • debayered/demosaiced image is having redundant information compared to the raw file, it never has the information which a full-colour sensor like Foveon would deliver
  • image looses precision upon colour conversion - when a colour conversion matrix is applied to transform camera data into some common space like ProPhotoRGB, the calculations amplify noise thus making higher bit depth excessive

If you do not need to edit image heavily (change exposure several stops for example or apply agressive curve, do colour mixing) you loose virtually nothing if you convert to 8 bit. Historically, many image editors are making computations at the colour depth of the source file so switching from 8 bit to 16 bit after opening the source image can improve quality of edited image almost like the source was 16 bit gamma-encoded in most cases even if the source was 8 bit (GIMP was better than Photoshop CS5 in this regard last time I checked. Photoshop would save PSD as 16 bit while GIMP allows to set precision in runtime and still save as 8 bit).

Geometrical transformations, patching and compositing are good examples of operations which do not require 16 bits at all.

If space savings are significant it is reasonable to use 8 bit output from raw developer. However, you might still want a 16 bit exported image if you resize it down if you will be printing (there is rarely a reason to print downsized image of course) it or displaying on more than 8 bit displays.

\$\endgroup\$
3
  • \$\begingroup\$ "8 bit gamma 2.2 encoding is roughly same as 16 bit linear" No, it's not. A 16 bit linear image is considerably better than a gamma encoded, 8 bit image. It is, after all a deeper depth than the original RAW. And it produces fewer artifacts when resize sampling in Photoshop which can sometimes produce Moire effects with gamma encoded images. \$\endgroup\$
    – doug
    Mar 15, 2019 at 23:07
  • \$\begingroup\$ @doug thanks for pointing that out, I need to be more specific. I rephrased it. Resizing artifacts are the defect of Photoshop though. \$\endgroup\$ Mar 16, 2019 at 9:58
  • 1
    \$\begingroup\$ Likely much of Photoshop's algos date back to when the math operations to avoid resizing artifacts and properly handling gamma doing so were slow. Not much of a problem these days. Luckily, it's an infrequent occurrence with most images. Good answer. \$\endgroup\$
    – doug
    Mar 16, 2019 at 15:10
1
\$\begingroup\$

Since the destination for many photos is the web, conversion to 8-bit does not really hurt anything. Even most printers (even many pro printers) have an 8-bit pipeline.

The main concern in editing with an 8-bit file is that you may introduce banding in smooth areas (usually the sky) but for skin smoothing there's probably not much danger.

As for using ProPhoto, the argument there is that even though it exceeds what most monitors will display and most printers can produce, it's still closest to what the camera is actually capturing and therefore by using it you are not throwing away color information the camera captured, which may be useful someday:

http://www.luminous-landscape.com/tutorials/prophoto-rgb.shtml

Some printers can print outside even the Adobe RGB space in some dimensions, so there can be a point to that in use with printing even today.

\$\endgroup\$
2
  • \$\begingroup\$ Interesting - he did quote a Luminous Landscape article while he said why he was using ProPhoto. I have read many articles there, but not this one. Thanks for the link. \$\endgroup\$ Aug 18, 2014 at 6:04
  • 1
    \$\begingroup\$ Regarding 8 bits, it gets weird there. As I go from LR (general adjustments) into PS, where I use ... well, PS + NIK + Topaz, mostly... I prefer to keep as many colors as I can, as those changes, layers, whatever, won't ever go back into LR. And, since nothing is settled regarding "24 bit monitors", whatever we will come to have, or even how many colors the human eye + brain can discern (this one is a wild goose chase where the goose is a ghost) ... I'd keep it to "max". \$\endgroup\$ Aug 18, 2014 at 6:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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