16 Bit vs 8 Bit Final Output?
toadlet
Registered Users Posts: 192 Major grins
I would like to know if there is really that much tonal differnece in saving a TIFF file as 8bit?
My typical workflow is to convert RAW image to 16bit TIFF, then open and work on it in Photoshop as 16bit layers, and then when I am done, save it as a 16bit layered TIFF.
Now as you can well imagine, with layers etc some of my images are 700mb+ in size! I know I can save these images to DVD media, but opening them again from DVD is painfully slow even on my fast PC. Would I really be doing a dishonour to my images in the quality department if I just worked on my images as 16bit, and then saves the layerd TIFF files as 8bit to save space?
Thanks in advance.
My typical workflow is to convert RAW image to 16bit TIFF, then open and work on it in Photoshop as 16bit layers, and then when I am done, save it as a 16bit layered TIFF.
Now as you can well imagine, with layers etc some of my images are 700mb+ in size! I know I can save these images to DVD media, but opening them again from DVD is painfully slow even on my fast PC. Would I really be doing a dishonour to my images in the quality department if I just worked on my images as 16bit, and then saves the layerd TIFF files as 8bit to save space?
Thanks in advance.
0
Comments
Now take my opinon for waht you paid for it:D.
Good Luck
I apologize for throwing a question at your question, but I'm only working in PSE5 which doesn't allow the use of layers with 16 bit files.
Mike
http://www.digitalphotopro.com/articles/2007/janfeb/bitdepth.php
Author "Color Management for Photographers"
http://www.digitaldog.net/
Results on actual output are a harder thing to evaluate, often there is no difference, sometimes one is liked better than the other by some and not others etc.
Are there perceptable changes to tone in a 8 bpc version compared to the 16 bpc version? No. How about with inkjet or press output? Press, no...inkjet, perhaps with one mono quad process in theory, but in results? The jury is out... Is this what you mean by tonal difference? If one takes the theory/math view, sure, there is a huge difference in 'tonal differences' (discrete values). Some care about the theory more than the results. Others for the results only. Some a mixture of both.
As with all these things, test it yourself and see how your images and workflow relate to both theory and practice.
Not for today's common output, no you would not be doing the image a dishonour. Most output is 8bpc, although there may be some less common methods that are higher. Many output processes have a smaller gradation of tone than the 256 levels in 8bpc images. Something to keep in mind.
If you are done editing the images, then no real harm (and little by all account even if you were editing in 8 bpc). This is one hotly debated topic that is not worth opening on this forum.
There are four basic options as I see it:
1. Archive the 16 bpc, layered files
2. Archive 16 bpc, flat files
3. Archive 8 bpc, layered files
4. Archive 8 bpc, flattened files
Hybrid Option:
5. Archive 8 bpc, layered files (PSD or TIFF) - Archive 16 bpc flat OpenEXR dupe.
What is more important to you, 16 bpc data or layers? Or is it both? Or none of the above?
Option 1: You really only have Photoshop (or Large Document format) and TIFF options for the layered high bits (high bit PDF support being lost in CS2). When using TIFF, ZIP is the best compression, but least supported compression format for other software (LZW file compression has wider support, but even that can break many applications that only read uncompressed TIFF). TIFF with ZIP file and layer compression should give slightly smaller sizes than PSD.
Option 2: I have an option for you that should give you disk space savings - but I hazard to mention this for archival purposes, although that may be unfounded. That option is OpenEXR. You may prefer to stick with TIFF or PSD, despite their larger file size.
Option 3: Here again, layered TIFF ZIP is smaller than PSD, but least compatible. Many apps can read flat PSD files or layered PSD that have a flat composite version of the layers included in the file (maximise compatibility option). Keep in mind that in many cases only Photoshop can access the layers, other apps see the file as flat.
Option 4: Same as option 3.
One can also save the transparancy data from each layer as an alpha channel and save the alphas with the layered file or as a separate grayscale file with alphas. One can then load the selections into the flattened file to isolate objects, thus preserving some of the benefits layers (obviously one only has limited control with this hack). This can save space in the layered file, while still giving you some flexibility. Probably not very useful, but worth keeping in mind for a rainy day!
Option 5: Sacrifice the high bits, keep the layers. Before dumping the high bits, save a dupe in flat 16bpc format, OpenEXR or TIFF or PSD if space is less of a concern (and you prefer to stick to easier/common formats than OpenEXR). You still have the original camera raw/DNG file and have a flat high bit file of the layered working file. Limited insurance, not as good as high bit layers, but if you really value high bit data so much, then it is better than nothing.
Despite the smaller file sizes of layered TIFF, I personally prefer to keep "working layered files" in PSD format, it is easy to spot the PSD files and I know that they are not final images, that is what TIFF files are for (my background is print/prepress and I was using TIFF files before they had layers, so I prefer to keep them simple more often than not).
One can't avoid the flattened composite in TIFF files, but in PSD one can choose to increase the file size by including a full resolution composite, so that other apps can handle the PSD file as a flat object. Some apps can read the Photoshop layers in a PSD file, but many can't and need the flat version included. This is also insurance, one can extract the flat version of the image from the last save, if say a later version of Photoshop changes layer blending behaviour (for hypothetical example). One has to decide if including the flat copy is worth the file size increase vs. increased compatibility with other applications.
Hope this helps,
Stephen Marsh.
http://members.ozemail.com.au/~binaryfx/
http://members.ozemail.com.au/~binaryfx/
http://prepression.blogspot.com/
Yes Racoon Dog?, this is so! PSD, being the native format, supports all of the bells and whistles of Photoshop, including layers, alpha channels, guides, annotations, colour samplers, paths and meta-data such as file info text and ICC profiles etc.
No, it uses an inferior lossless compression method than ZIP found in TIFF (LZ, I believe, but don't quote me, this is my guess why TIFF LZW is slightly smaller than PSD in file size).
Even without the maximize compatibility feature of PSD, it is larger than ZIP TIFF.
Not a problem Mike!
Regards,
Stephen Marsh.
http://members.ozemail.com.au/~binaryfx/
http://members.ozemail.com.au/~binaryfx/
http://prepression.blogspot.com/
Thanks, Stephen!
I am wondering if the popularity of 16 bit files are part of the reason that Adobe started offering unsharp mask and other commonly used features as layer masks in CS3 (this is true, isn't it?), since layer masks presumably take up much less file space that image layers. Perhaps one could take steps to minimize the use of image layers when working in 16 bit and therefore reduce the layered 16 bit file size. As mentioned above, one method would be to use layer masks whenever possible. Another idea would be to replace image layers with a blank layer whose title or text box gives instructions on how to recreate that layer (e.g., "Layer 1/2/3 merged and filter X applied with settings Y & Z"). Am I thinking too far out of the box here?
I know it might seem strange that I show so much interest considering that I am only running PSE5, but I've been drooling over the full version of PS for about 5 years now. One day soon I'll take the plunge.
Mike
P.S. Racoon Dog is correct... two points to Gryffindor!
Tiff should be far more reliable for many users as its an open format that doesn't, like PSD require a fee to use it. That doesn't however mean that the tons of software that can handle a Tiff does so correctly. Adobe owns Tiff, they update its functionality (as they do with PSD) and its up to the 3rd party applications to update their products to provide continued compatibility. So one might work with all options, another may choke with Zip compression.
IF you're working with documents larger than 4 gigs, then the only game in town is the Adobe PSB.
The Tiff with Zip will be smaller than PSD but is a bit slower to save and open.
Author "Color Management for Photographers"
http://www.digitaldog.net/
High bit support has been on-going since around Photoshop 5 or 6 because lots of users want to apply tone and color adjustments on more than 8-bits per color.
Smart Filters are more about flexibility in changing your mind about an edit than anything else. A layer can take up a tiny or a large amount of storage size based on the amount of transparency. An Adjustment Layer is nothing more than a set of instructions (one could say its metadata instructions).
Note that when you flatten the image OR print it, the effects of the edits ARE applied to the data. So doing this on 8-bit data doesn't save you from data loss when you output the document. These layers are 'semi-non destructive" in that there is always some data loss when you alter one set of numbers to another set of numbers, which is all Photoshop really does anyway. In reality, there's no such thing as non destructive editing outside handling raw data. If you edit the data, there's loss.
Author "Color Management for Photographers"
http://www.digitaldog.net/
Muench Workshops
MW on Facebook