16bit photoshop files: why do they bloat?

Here's something I've noticed: a 16 bit file with adjustment layers
bloats really badly compared to 8 bit.

I created a 720x576 8 bit image and drew some stuff on it. Then I added
an adjustment layer and drew on the layer mask. The resulting file is
shown in PS as 1.19/1.56M.

I did the same but created as 16 bit. Now it shows as 2.37/6.33M

So the 8 bit image with adjustment layer grows 50%, but the 16 bit
grows roughly 3-fold.

This is an example. I sure anyone who uses 16 bit files would notice
similar effect with real files. Any ideas why?

Cheers!

DG

 

One advantage of using 16 bit color is that the wider data base allows for
finer manipulation of histograms. These are mathematical adjustments, not
adjustments done solely by eye. The fact is that when that image is printed
it will be down converted to 8 bits by an arbitrary algorithm in the printer
driver over which you have little or no control, other than trial and error
manipulations to see how the final print turns out.

Many, probably most, people cannot see the entire 16 bit gamut, their
monitors cannot accurately reproduce that range of colors, and printing
processes live in an 8 bit world.

There are good reasons for using 16 bit color. However if you do not clearly
understand why you need to use 16 bit color then 16 bit color may actually
hinder the creation of a final printed image. It is very difficult to
develop a color management work flow that reliably matches the monitor to
the printed image even when you follow all the cookbook recipes. I think
most people should stick to and understand 8 bit color and not worry about
what they may be missing with 16 bit color, because it isn't all that much.

 

thanks for the intro to 16 bit files, I'm OK with all that already

but none of its answers why 16 but files bloat so much

 

Hi...

OK, I'll give it a quick shot

24 bit color is three 8 bit channels; one for each color.
Total of 3 bytes.

48 bit color requires two bytes per channel, for a total
of 6 bytes.

But you said you already knew that, so I'd suggest doing
your experiment without that added layer, or after flattening
it.

Ken

 

the 3 bytes per pixel in 8 bit and 6 bytes for 16 bit means the file in
16 bit will be twice as large. Which we see it is because the base file
is 1.19Mb in 8 bit and 2.37Mb 16 bit.

fair enough.

so I add an adjustment layer and the 8 bit goes to +50%, while the 16
bit goes to +260% (approx)

if I change the 16 bit file (with adjustment layer) to 8 bit then
surprise surprise it goes to the same size as the 8 bit.

but why the bloat when you add an adjustment layer?

 

Hi...

Dunno, can't help you with that one... don't use
Photoshop (use PSP).

But I wonder if PS is doing some kinds of tricks with
that added layer. Perhaps keeps track of what portions
are actually different? Throws away transparent chunks?
Keeps a little more undo info?

You experiment with ps, maybe try adding a layer that's
totally unrelated to the "real" layer. (paste a whole
different picture onto it, or something)

If you like I'll try it with psp and see what happens.

Ken

 

If I take a 16 bit file and simply add another layer (ie not adjustment
or layer effect/style) it pretty much doubles in size, which is much
what you would expect. Adding a layer style doesn't change the size at
all.

You have a go with PSP and see what happens

 

Hi...

Arghhh, should have warned you. I'm an old guy; stroke
damaged, easily confused and forgetful

We'll have to depend on your experiments; psp doesn't
support 48 bit.

Sorry about that.

Ken

 

Probably the difference is compression. Photoshop
seems to do some lossless compression to photoshop
files, e.g. in the layers. (I am not 100% certain
of this, but it is my observation as some photoshop
files are smaller than a tif equivalent, and
layers with transparent regions are particularly
smaller than if the layers are full of data.)

16-bit data is very hard to compress.
So you might be seeing an efficiency difference.

(I work with Photoshop-CS and 16-bit a lot.)
Roger

 

In message <>,
"Roger N. Clark (change username to rnclark)" <>
wrote:

It isn't even trying, for the main layer. A solid-color 12MP image
saves as 72M bytes in 16-bit, and 0.6M bytes in 8-bit, as .psd, but they
should both be equally compressible. This is without layers.

--

 

I once read in a photoshop manual that it RLEs the PSD file, so blank
areas will cause a smaller file than busy areas.