What do they say about archiving photo images in the newsgroups

red bullet gif image Stan The Man (macho@mac.com) wrote February 10 Year 2001 in comp.graphics.apps.photoshop
I would use TIFF rather than JPEG because JPEG is a lossy format. It throws away data whenever you open it, do something, and save it again.

Only if, after opening, you re-save it as a JPEG, which will indeed lead to progressive degradation. But you don't lose any more data if you resave a JPEG as a TIFF or other lossless format. This is what publishers would do.

In fact, in prepress terms, maximum quality JPEG compression results in images which are indistinguishable from uncompressed files. In a survey of print professionals who were shown the printed results of various hi-res Photoshop images printed from uncompressed files and from max quality JPEGs of the same image, none could tell the difference. So JPEG compression is an extremely good choice when trying to get the maximum number of files on a CD-ROM, e.g. a photo archive.

Some people in this newsgroup still prefer JPEG. This is the answer from Stan from the same discussion as the answer above
Granted. In all honesty, I've even seen JPEGs come through my shop, from just such sources. But I prefer to avoid them.

And it's also true that whether or not the degradation can be spotted depends on large part on the image itself. Darker images fare better than lighter images (the luminosity information is compressed separately from the color information in a JPEG), and images with no fine detail (such as clouds or skies) fare better than images with lots of fine detail (such as cityscapes). On an image which lacks fine detail, JPEG compression may indeed be unnoticeable.

I'd take survey results with a grain of salt. Often, such surveys are easy to rig. If, for example, I were to show a JPEG image of a sky scene to a number of prepress professionals, it's very unlikely that anyone, myself included, would be able to spot the compression. But JPEG compression is so dependent on the type of image being compressed that blanket statements about its quality should be approached with suspicion.

Kris Bolling wrote September the 5th Year 2000 in alt.comp.periphs.scanner.
I've been scanning family photos, and working with computers in general, for while. Here's what I've learned.
1) Scan your archives at the highest OPTICAL resolution possible and with the highest color resolution possible at that optical setting Most imaging/painting software will let you re-scan a file to lower resolutions (loosing information in the process) to produce a smaller file for distribution, but you want your original file to be as faithful to the photograph as possible.
Some scanning software provide resolutions greater than is physically possible by the hardware. This is done by inserting pixels that are "between" the values of the adjoining optical pixels - and actually degrade the image, making them fuzzy.

2) Do not use lossy compression when saving the images. e.g.: If saving as JPEG files set the compression factor to 1. (Yes, you are going to have very big files!)

3) Save your achieves on at least two physically independent media. aka: Backup, backup, backup!

4) Media goes bad, even when not used. Verify you can view every picture annually or semi-annually. Slideshow software is great for this. Setup a slideshow that includes all pictures and let it run. Go walk the dog, trim the hedges, whatever and when you get back if there's no "error reading file" nasty-gram you're OK.

5) Roll with technology. As new storage media is introduced remember to move your archives to it. Anybody still using 8 1/2" floppies or Bernoulli drives? Typically there's a 3 year window when old and new technology overlap - take advantage of the time.

6) Use the most common format to store your pictures; JPEG, TIFF etc. Also heed #5, what's common today may not be so in the future, roll your format with the times.

7) Keep an external log/database describing the pictures. Don't depend on embedded comments in the file.

8) "Retouched" images are not archive. Once you've got the picture the way you want it save it as a different version. Never, ever, delete your original scan.

(now getting of soap box) Hope this helps,
-- Kris

Per Inge Oestmoen (pioe@coldsiberia.org) answered February 28, 2002 Re: TIFF vs JPEG
As I often say; try to go to National Geographic and try to sell them JPEG files for publication. They will not even consider it.

Moreover; in serious digital work, it is well known that almost all images need tweaking and optimization. The use of Curve Tools, Saturation, Color Balance and finally Unsharp Masking are all routine procedures. Try to take photos from any digital camera or from a scanner for that matter. Take some images in TIFF, and capture some parallel images in JPEG, preferably in a low compression. Now test and find out which files are most successfully optimized.

There is no question that the files that have never been compressed, consistently yield the better results, whereas the files that have once been loosely compressed with JPEG can be modified too, but to a much smaller degree, and the result is nowhere near what is obtainable when you work with Tiff's or other non-compressed formats.

The more information there is, the easier and more successful subsequent modifications and corrections will be. It is almost tragicomic when proponents of the use of JPEG as a general storage format fail to understand the ramifications of lossy JPEG compression.

In many instances the difference between images that started out as Tiff's and those who have undergone low compression JPEG-conversion is hard to tell, in particular on screens. Bearing in mind all the other parameters like exposure, focus and aberrations one may superficially make a case that the use of lossy JPEG is equally justifiable as TIFF. However, all things being equal, lossy compression invariably leads to loss of information that may manifest in reduced quality on a high-quality print depending on the motive. For people and portraits JPEG seems to be satisfactory most of the time provided the photo was 100% optimally captured. This is frequently not the case, and for foliage, conifers and other subtle details of nature there will be a loss of quality. There is no way around it.

When it comes to modifiability and latitude for tonal and other adjustments, there simply is no comparison. The potential of any picture is dramatically reduced by lossy compression.

Lastly; a widespread assumption is that one can "improve" a JPEG by subsequent conversion into a non-lossy format. This is of course not true, even if there is a deceptive increase in file size the information that is thrown away can never be recreated, and even though our eyes fail to see the difference on today's monitors, any attempt at extensive editing of Tiff's and JPEG's respectively will quickly reveal the superiority of the uncompressed files. Having one's master files in uncompressed form is a prerequisite to maximum quality. Unfortunately, one might say, because the storage and energy problem with digital cameras is very real and is no doubt what has led so many digital photographers to use lossy compression for master files because there is often no other realistic alternative given the space and battery energy constraints. Sad to say it as it is, that does not improve upon the nature of lossy compression.
.........There is no debate that a large JPEG with many pixels is much preferable to a small TIFF (with fewer pixels) since the pixels are what forms the image. That is a matter of course. Thus there is no question that a 5 Mb JPEG is a better file than a 5 Mb TIFF. But that is hardly an argument against storing in pristine TIFF (or another uncompressed format) whenever and if you can, because a fifteen-Mb TIFF will be superior to a five-Mb JPEG of the same pixel size. That is the relevant comparison.

-- Per Inge Oestmoen, Norway

Tony Spadaro (tspadaro@ncmaps.rr.com) gave is comment Re: TIFF vs JPEG
But all else being equal - which would you choose? And all else will Not be equal if the jpeg format was used through 2 or more saves. The jpeg will not be printable.

A test of jpeg vs. tiff quality on a computer monitor is, from the get-go, invalid. You cannot put a tiff on a website. High quality jpegs can indeed look pretty good --- if they are first or at most second generation jpegs. You need only look around on the web to see the results of people who edit pictures in jpeg mode. The artefacts destroy the picture pretty quickly

Lets take the typical picture though, where after saving and before printing you want to crop, correct colour, spot out a few noise artefacts etc using image editing software. 90 percent of all pictures are going to need some work done - otherwise you might as well have Wal-Mart do the printing - it would be cheaper anyway.
The second save will be at much lower quality than the first - even if the jpeg quality level is still kept on the high setting. This is going to make a mess out of edges. The jpeg artefacts are now being averaged into the compression - so you have artefacts of artefacts.
As I said before you can find all sorts of websites where the jpegs were saved and re-saved. The pictures look as if they have leprosy.
Tony Spadaro at http://home.nc.rr.com/tspadaro/

From JPEG FAQ I quote: Does loss accumulate with repeated compression/decompression?
The bottom line is that JPEG is a useful format for compact storage and transmission of images, but you don't want to use it as an intermediate format for sequences of image manipulation steps. Use a lossless 24-bit format (PNG, TIFF, PPM, etc) while working on the image, then JPEG it when you are ready to file it away or send it out on the net. If you expect to edit your image again in the future, keep a lossless master copy to work from. The JPEG you put up on your Web site should be a derived copy, not your editing master.

File Formats, Digital Still Images. References to standards, technical literature, and image processing software (utilities) are listed below for electronic still photography file formats in use or being reviewed for use in Harvard's digitization initiatives.

From Art Mosher (Mail) [mail@artmosher.ca] I got following mail June 7 Year 2002.
I noticed your web site in my search through comp.periphs.printers; while trying to figure out what photographic image printer to buy to go along with my new digital camera (Fuji S602).

Your web site is very well done & offers many insights into digital photo topics.

I read the article on TIFF as an archival image storage, and fundamentally agree with your conclusions. However, in the practical world, many ORIGINAL images are taken in JPG format. Most cameras are set up to deliver JPG format in all but the highest resolution modes. As TIFF or raw forms consume monstrous amounts of storage and take considerable times to store & retrieve, most routine photography I suspect is done into JPG originals !!

With the result that one often STARTS with a 'lossy' file format !!

When they put mainframe processing power in cameras AND memory is sold in GIGABYTE chunks instead of MEGABYTES; AND at an affordable price, THEN TIFF will make more sense for routine use.

I DO understand that to capture original source material, such as your artwork, in the highest possible detail, one would START with a TIFF image. - Art

From Dugphoto (dugphoto@aol.com) December 17 Year 2002.
Saving JPEGs as TIFF
.I've been exploring my new Fuji S2 Pro and somehow downloaded images shoot as JPEGS into TIFF on my computer. I thought they would be JPEGS with a TIFF suffice. Then I downloaded the JPEGS as JPEGS and noticed that the JPEGS were about 1,400 KB while the TIFFS were about 9,700 KB. Obviously not the same. Does this mean I'm saving more data and therefore a better print by saving the JPEG images as TIFFS. Any relevant feedback would be appreciated.

Bernard Hill answered:
There is no loss in saving as tiff. Ever.
When you look at a picture, it's just a picture. An arrangement of pixels.

When you save a picture you have to save it on disk in a picture format. The nearest to what you are seeing is for Windows users, BMP format. Each dot on the screen is a value in the file. However this gives a big file and is not readable by other than Windows.

So the nearest universal format is uncompressed TIFF. There's also compressed TIFF whereby the data is compressed but not lost. For instance (I'm being simplistic here) if there is a bunch of pink pixels in the middle of the screen the compressed format says "100 pixels, all pink, occurring here") and this takes less than 100 individual identical values in the file. But you can see the data has not changed or quality lost by saving in this type of "non-lossy" compression.

This is called file compression and there are lots of different ways of compressing files without any loss.

However photographs are rarely compressed by very much by this non-lossy process because patterns or blocks are quite rare in pictures, so compression does not help much. Here is an actual list example:

   Uncomp TIFF    Comp TIFF  PNG format
A sunset     5.49Mb     1.75Mb     1.73Mb
Cornfield     5.49Mb ;  5.52Mb   4.06Mb

(PNG is an alternative compressed format)

I chose these 2 pictures because a sunset has very little detail and so compression works well. The cornfield has lots of detail and compression works badly - in fact it fails in one case as the compressed file is larger than the original!

Now enter jpg format. When the data is saved **and only when it is saved** some data is thrown away from what you see on the screen. The file on disk is less accurate that what you are seeing in front of you. (Until you re-load the file on the disk, in which case you see the imperfections).

This is because colour changes have been thrown away and averages made in places where it probably doesn't matter much. But this loss occurs only in the actual save to jpg on disk. And with jpg you can define roughly how much detail you throw away, or how much smaller you want the file to be. However I can say that best quality jpgs of the files above are about this size:

Sunset 0.6Mb
Cornfield 1.9Mb

So that's at most a third the size, and at least a tenth. Again, note the detail on the original dictates the file size.

But the important thing to realise is that image deterioration only occurs **when you save as a jpg**. There is no quality penalty for changing formats from jpg to tiff: and the only quality loss is when you save as a new jpg file, e.g. converting tiff to jpg.

So the advice normally given is:

1 Shoot so that your camera either saves jpg or tiff, depending on your card capacity and quality requirements.
2 When editing, save intermediate files in TIFF as there is no loss of quality

However my own opinion is that "best jpg" is indistinguishable on paper from TIFF that I always save intermediate files in that format because of the lower file size. But no doubt others disagree... :-)

Bernard Hill (bernard@braeburn.co.uk)
Braeburn Software
Selkirk, Scotland

Tim Gray answered Dec. 30 2002 in the forum; http://www.luminous-landscape.com/cgi-bin/forum/ikonboard.cgi?s=f81a8cd9b2dfcf9fed752e5711efcb3e;act=SF;f=5:
Auuestion about archiving.
I don't see a problem with media. The issue is to respond to the changes in technology when they reach maturity. How many still have any archived documents on 5.5 floppies? 3.25 floppies? CDs? (lots here - but give the DVD RW a year or so to mature and all my files will be on DVDs). I'm a bit more concerned on the software side than the media side. Even if I update my media regularly - will I be able to read the RAW files 20 years from now? Even JPG 2000 gives a file size more than twice that of a RAW file. I don't archive the 18mb 16bit TIFF files I get from the lowly D30 - I archive the RAW files.

And what do I say myself about archiving photo images in TIFF format?

Top of page

This page was last updated: June 15 Year 2003

back home