Digital BW, The Print

It's amazing how quickly you can fill a CD disk if you store several
versions of an image. This is especially true if you are stitching a
composite made of, say, a set of three photos across married to
another set above it and a third set below. So despite the cost of
even good CDs, I'd like to choose a *lossless* compression scheme: I
can see storage space will be the real problem, not raw material
cost. ZIP takes a long time, but *Real World PShop* likes it. After
reading
the specs on .PNG, however, I'm inclined to adopt this format
instead. It's much faster than .ZIP, and just as protective of the
image. What
I don't know is how many of the image archivers support it. Does
anyone use .PNG on a regular basis?

Bob Bollini

Bob,

I am guessing you are working on a PC. What program are you saving your 
images from? If it is Photoshop, why PNG over, say, compressed TIFF? 
Tiff is a popular enough format, but I would guess that as long as Photoshop 
can read it, any lossless, compressed format should do the trick.
Also, Genuine Fractals offers a lossless option that may have merit.

Maybe if you give us an idea of what you are trying to do with your images, we 
can offer more specific advice. Are they grayscales? what MB sizes? are you 
trying to create an archive segmented  across several discs? will you be using 
Stuffit?  etc etc.

Also, since this list specializes in monochrome digital printing, is your 
question ultimately about making such prints? If not, there may be more 
specialized lists that deal with archive management issues.

Antonis


--- In DigitalBlackandWhiteThePrint@y..., Robert Bollini <rbollini@n...> wrote:

> It's amazing how quickly you can fill a CD disk if you store several
> versions of an image. This is especially true if you are stitching a
> composite made of, say, a set of three photos across married to
> another set above it and a third set below. So despite the cost of
> even good CDs, I'd like to choose a *lossless* compression scheme: I
> can see storage space will be the real problem, not raw material
> cost. ZIP takes a long time, but *Real World PShop* likes it. After
> reading
> the specs on .PNG, however, I'm inclined to adopt this format
> instead. It's much faster than .ZIP, and just as protective of the
> image. What
> I don't know is how many of the image archivers support it. Does
> anyone use .PNG on a regular basis?
> 
> Bob Bollini

>
> I think it was Austin that pointed out here, or someplace else, that
> compressed TIFF often ends up with a *larger* image than
> uncompressed TIFF. I've proved this to myself on numerous
> occasions. PNG seems to be gaining popularity, but I've never
> checked its compression.

Really? of about 20,000 images saved as tiff so far, I can't say I've ever
noticed this.

I'll look when I do the next 30,000 soon, but my experience so far is
compressed tiff work fine and actually does compress - so far.



Tim A

I think it was Austin that pointed out here, or someplace else, that 
compressed TIFF often ends up with a *larger* image than 
uncompressed TIFF. I've proved this to myself on numerous 
occasions. PNG seems to be gaining popularity, but I've never 
checked its compression.

I'm skeptical of archiving anything in any format but those most 
common - even in PS native format; who knows what you'll have to 
open it with in ten years. I've got some FD Painter files in native 
format on some CD somewhere - those may well become lost 
images.

Tim Spragens

> I am guessing you are working on a PC. What program are you saving
> your images from? If it is Photoshop, why PNG over, say, compressed
> TIFF? 


--
Tim Spragens
http://www.borderless-photos.com

I've been scanning images with Vuescan as compressed TIFF, 48-
bit RGB. If I open these in Pshop and do anything, then resave, 
they grow. This made me stop to take a look. If I save either as 
compressed or non- with TIFF in Pshop, they're bigger, and the 
compressed can be larger than non-compressed. I don't think 
Adobe must be using the full compression possibilites of the 
format. 

Very likely it is image dependent, and with high-detailed photos it 
is worse than those with smooth areas. 

I'd be happy to be proved wrong here, or shown a switch I haven't 
flipped.

Tim Spragens

> Really? of about 20,000 images saved as tiff so far, I can't say I've
> ever noticed this.
> 
> I'll look when I do the next 30,000 soon, but my experience so far is
> compressed tiff work fine and actually does compress - so far.
> 


--
Tim Spragens
http://www.borderless-photos.com

> I think it was Austin that pointed out here, or someplace else, that
> compressed TIFF often ends up with a *larger* image than
> uncompressed TIFF. I've proved this to myself on numerous
> occasions. PNG seems to be gaining popularity, but I've never
> checked its compression.

> I've been scanning images with Vuescan as compressed TIFF, 48-
> bit RGB. If I open these in Pshop and do anything, then resave,
> they grow. This made me stop to take a look. If I save either as
> compressed or non- with TIFF in Pshop, they're bigger, and the
> compressed can be larger than non-compressed. I don't think
> Adobe must be using the full compression possibilites of the
> format.

A few weeks ago we started to look into image compression for e-mailing large files to clients, and we
specifically started researching 'Stuffit'.  It was pointed out (on colorsync) that Stuffit uses the same
compression method as LZW compression of tiffs in Photoshop.  And when a file is already compressed in tiff,
and then 'stuffed', it ends up larger, because of the stuffit artifacts *added* to the already compressed
tiff.

This is what is, perhaps, happening here, when an already compressed file is altered and re-compressed (or
just saved as a tiff) there is much more information there than when it started..  I imagine that it is
probably not a good idea to compress any image until it is in it's final form.   And further, that *any*
compression method actually discards some quality information in a file.

Oh, we finally decided that the best method of moving large files to clients was to post them (*not*
compressed) to an ftp site for them to download directly.

> Very likely it is image dependent, and with high-detailed photos it
> is worse than those with smooth areas.

Isn't that how compression works...a complex algorithm of 'averaging'?  An image that has more 'plain' or
'broad' areas of the same tone, compresses more than an image of 'unique' tonal ranges.


Harvey Ferdschneider
partner, SKID Photography


[Non-text portions of this message have been removed]

>
>
>A few weeks ago we started to look into image compression for 
>e-mailing large files to clients, and we
>specifically started researching 'Stuffit'.
>
>Harvey Ferdschneider
>partner, SKID Photography

Last week I tried the trail version of Genuine fractals print pro to 
up size an 35mm neg to 10x8, it took an hour to process on a G4 Mac 
with 1 gb memory. The original scan was 20 mg and the final print 
over 100mg! So I have decided to go for the MrSid photoshop plugin. 
It is claimed that a 20 mg can be reduced to 1 mg , opened in 
Photoshop without detail loss, worked on and saved either in MrSid or 
TIFF.

To use MrSid for compressing files for emailing or downloading, the 
receiver will have to use the free MrSid browser to open the file. 
And, the actual Photoshop plugin is only available for Mac but the 
files are cross platform.

If all this sounds confusing then blame LizardTech for a very dry 
presentation of their product!  More info on: 
http://www.lizardtech.com/

Jerry.

>
> Last week I tried the trail version of Genuine fractals print pro to
> up size an 35mm neg to 10x8, it took an hour to process on a G4 Mac
> with 1 gb memory. The original scan was 20 mg and the final print
> over 100mg! So I have decided to go for the MrSid photoshop plugin.
> It is claimed that a 20 mg can be reduced to 1 mg , opened in
> Photoshop without detail loss, worked on and saved either in MrSid or
> TIFF.

Did you try the 'compressed' option in Genuine Fractals?  It would make sense that bumping a small image up to
8x10 would increase the files size.  You don't mention what dpi you saved the 8x10 to.

Also, you could have saved that 100 MG image as a '12' (top) quality jpg, or as a LZW compressed tiff, to
knock down the size.

We too, have found that when you bump an image way up in genuine fractals, it take a lot of computing power.
We assumed that was due to the complex algorithms used to effectively 'fill in the dots' between the original
pixels.

Harvey Ferdschneider
partner, SKID Photography, NYC


[Non-text portions of this message have been removed]

Hi Jerry,

I wasn't aware of the MrSID product. From the looks of the Lizartech site, 
Genuine Fractals is toast! If you do get it, would you do a side by side test and 
let us know how the two compare? 

Thanks for bringing this to our attention here.

Antonis



PS: If you decide to have your name instead of your email  in the author  
column of the list, here is how to do it: Go to our files section  > List-related 
matters > Your name instead of a partial email address on posts. It gives you 
the details there. 


--- In DigitalBlackandWhiteThePrint@y..., meander@m... wrote:

...

So I have decided to go for the MrSid photoshop plugin. 

.....

More info on: 

> http://www.lizardtech.com/
> 
> Jerry.

>Hi Jerry,
>
>I wasn't aware of the MrSID product. From the looks of the Lizartech site,
>Genuine Fractals is toast! If you do get it, would you do a side by 
>side test and
>let us know how the two compare?
>
>Thanks for bringing this to our attention here.
>
>Antonis
>

Actually, to test it for free download their Photo Solo:

>http://www.lizardtech.com/cgi-bin/products/desc.pl?tsb=25730

This is aimed at digicam users and has a max file size of 2100x1600 pixels.

Jerry.

on 8/26/01 5:09 PM, SKID Photography at skid@... wrote:

>> Very likely it is image dependent, and with high-detailed photos it
>> is worse than those with smooth areas.
> 
> Isn't that how compression works...a complex algorithm of 'averaging'?  An
> image that has more 'plain' or
> 'broad' areas of the same tone, compresses more than an image of 'unique'
> tonal ranges.

It depends very much on the compression algorithm. See a long discussion on
the Filmscanners list and/or Digital Silver a couple of months ago.

LZW compression typically does not do well on data that does not have
consistently repeating data patterns, because it relies on replacing
repeating patterns with smaller indexed markers. It works extremely well on
e.g. text because certain patterns like 'the' or 'and' recur. It does not
work well on detailed image data like 4000 dpi Tiffs for obvious reasons. I
oversimplify but you get the point.

RLE (run length encoding) works well on images that have large identically
flat areas, like bitmapped solid-color graphics. Again it is not very
helpful for information-rich images like photographic TIFFs.

JPG does well on images that have smooth tonal transitions, because the
frequency analysis it does on each 8x8 area works best with relatively
high-frequency data. It does not do well with large flat areas because the
'chunking' of the image becomes obvious, and it does not do well with *very*
high frequency random data like fields of wheat unless you crank the quality
all the ways up. It is lossy of course so out of the running here.

Genuine Fractals is an interesting case. Its lossy compression option does
seem to do well with most images. It claims the recovered file is visually
indistinguishable and I have to agree in the tests I have done it seems to
be. One thing GF is particularly good at doing is preserving sharpness and
the impression of high-frequency detail. Its lossless compression doesn't
achieve great savings but seems to do better than LZW on photo TIFFs in my
experience.

MrSID I haven't tried, but Lizardtech have just acquired Genuine Fractals so
they will either kill it off or (hopefully) incorporate the technology into
future versions of their own software.


-- 
John Brownlow

http://www.pinkheadedbug.com

ICQ: 109343205

>Hi Jerry,
>
>I wasn't aware of the MrSID product. From the looks of the Lizartech site,
>Genuine Fractals is toast! If you do get it, would you do a side by side
>test and
>let us know how the two compare?
>
>Thanks for bringing this to our attention here.
>
>Antonis

I thought MrSID was only a compression program and that the resizing was
only up to the original file size, but I really haven't checked it out
thoroughly. I also did not know it was lossless.

Nonetheless it is amazing. A great place to check it out is the online
Library of Congress Panoramic Maps Collection (part of their American
Memory site).
http://memory.loc.gov/ammem/pmhtml/panhome.html

These maps are available for free download in the MrSID format. I
downloaded a map of my town and got the MrSID viewer, plug-in, or whatever
the free download is. I was stunned when the little file opened up in
Photoshop at a high resolution at full size (and in RGB). I converted to
greyscale and did a lot of restoration, then printed out with Piezography.
I just wish I had a larger printer than the 1160 for this one. I did print
the legend seperately so that I could go full width on the 1160 with the
map portion.

Several people continually sell prints of these files on eBay (to the
consternation of the Library).

Mark