Hi quality jpg vs tiff (part deux)

On Mon, 03 Nov 2003 22:02:54 GMT
in alt.graphics.photoshop
"Jan Kohl" muttered something like this:

Thanks for all the answers, I didn’t realise it was going to cause somewhat of a ruckus! 🙂
I must say, I do somewhat agree with Bob’s and Al’s commentary, as I could not see a difference between my jpg and tiff.

Here’s an example that I used:
http://www.castlegraphics.com/files/bn_9993_E9_Grand_forks_N D_jun1970.jpg http://www.castlegraphics.com/files/bn_9993_E9_Grand_forks_N D_jun1970.tif
I challenge the "team" to find a visible difference between the two photos…and this is very similar to what I will be
archiving, and some will be black and white, and of course, the original is much larger.
However, both Outpatient and Mr3 both brought up valid points, Outpatient with his excellent website and Mr3 said something that
really made me think…"optimized for human vision which is pretty forgiving and easily tricked".

Outpatient, what version of PS did you use for your exhibit?
Mr3, here’s my answer to your other questions…
Early photo and negatives – the source material may be fragile and only available for one scanning session.- Yes
Are the scanned images for academic use? – No, not planned for, but could be. How are the scans going to be cataloged/stored/distributed/viewed/printed? – most likely on CD/computer monitor
Will the images be published? paper or electronic? – not planned, but should be able to be published
Which image format is expected by the publisher/end user/owner? – whatever I decide… 🙂 Will the scanned images be subjected to specialized image processing at a later date? – not that I know of

Based on the excellent interaction of this group, I’m starting to lean back towards tif, even with the large amount of images.
Many of these old photos will be scanned, and then you never know what could happen to paper photos. But that brings up a few
more questions…

*If I scan with the possibility of publishing in future high quality "coffee table" books, what should the minimum (pixel)
resolution be? Most of my work is done for the web, and I usually try to hit about 18-20000 pixel width images.
*I see that bmps are typically a couple of hundred K less that tiffs, since they both seem to produce high quality (and
editable) images, what are the differences between the two? I had someone tell me that Photoshop does a poor job on bmp files,
however, again, I can see no differences between the two, even when zoomed in on a high res scan.
Thanks again!

»» Jan Kohl

::: computer security consultant :::
the pits – http://www.theuspits.com
castle graphics – http://www.castlegraphics.com

(I used Photoshop 7.0 for that experiment on that web page.)

Granted the differences between your TIF and JPG are extremely subtle, the differences are there. To see it for yourself, simply load both images into Photoshop and them combine them to a single image with 2 layers, each photo on it’s own layer. With the layers perfectly aligned you should be able to toggle the top layer on/off to aid in looking for differences.

Now, zoom in to max zoom (1600%). If you toggle the top layer on/off you should see very subtle color shifts in many of the pixels. I found it most evident on the left windshield of the train. And even though it’s easiest to see on that dark windshield, I’m sure you understand that the subtle changes are taking place all over the image.

I understand this difference is VERY subtle and at normal zoom levels not even perceivable, but is IS there. Once you archive these images, if that’s going to be their final resting spot (IOW, no more image editing for them) then high quality JPGs might be just fine. But if you think there’s a chance you’ll still need to work with these photos, then I’d stick with a lossless format.

Have you considered PNG format? I really like PNG, and it is lossless. You 2.1mb TIF file saved as a PNG is about 1.2mb. Saved as a BMP it was still approx 2mb. Here’s what I found:

Size with lossless formats:
TIF – 2.17mb
PNG – 1.21mb
TGA – 2.07mb
BMP – 2.07mb
JP2* – 1.11mb
(* – JPEG 2000 format set to lossless)

Size with lossy formats:
JPG @ 100% quality – 1.08mb (using ThumbsPlus ver.6)
JPG @ 100% quality – 0.85mb (using Photoshop 7)

Of the lossless formats obviously JP2 is the smallest, but it’s still a rather obscure format. PNG is much more widely accepted.

As for the discrepancy between the two "100% quality" JPGs, that’s due to Photoshop’s lackluster JPG compression. What PS considers to be "100%" could really be a little better. The "100% quality" JPG produced by ThumbsPlus, although a bit larger, is much more accurate and true to the original.

So, I don’t know if that helps or confuses you more. 8^) If it were me and I didn’t have the space (or just didn’t want) to save in .PSD format, then I’d go with PNG, with the exception of any images that needed to retain their multi-layer format, then TIF or PSD is really your only choice.

Sorry so wordy, but this is a subject I enjoy kicking around. 8^)

—
OuTpaTienT / outpatient°AT°rocketmail°DOT°com
http://www.0utpatient.com
http://www.oeyec.com

Based on your answers and intentions, you may want to steer away from trying to scan for a specific end use; i.e., CD, coffee table book, etc.

I would consider your project more of a data capture than a task specific step. Since you plan to multi-purpose the data, the file format is not as important as the capture resolution. With greats loads of pixels, lots of uses can be accommodated.

Points to consider:
1. Most modern publishing submissions are 300dpi uncompressed TIFF. For an 8" x 10" photo, your image dimension will be fixed at 2400 x 3000 pixels. On the scanning side, 300 dpi image requires a minimum of 600dpi scan, more is better.

2. For comfortable image viewing on a computer screen, no image should exceed a single screen . Most surveys put this at 800 x 600, up from 640 x 480, and on the way to 1024 x 768. This is image resolution independent; no how many millions of pixels you started with, only 480,000 pixels are going to fit on the screen

3. Native resolution for modern desktop printers is 360/720 dpi. Those higher numbers in the ads are the ink dots, not pixel data. Printing two, three, or four tiny ink dots to compose one pixel gives them incredible control per pixel, but it’s still 360 or 720 dpi. With good image data and a decent printer, 200 dpi looks as good as 600 dpi to the untrained eye.

4. Cataloging – With lots of images to be scanned and accessed, some type of database will need to be created. Image classification, file naming, data dictionaries, dates, sources, captions,etc…

5. File formats – TIFF is a type of BMP. Both files are raster based and reference image data as an X,Y pixel grid with color information associated with each data point. TIFF was formalized in the early nineties by the major players of the day in an effort to create ‘smarter’ files. The intent was to embed metadata in the file to assist image processing, reproduction, and distribution. The embedded "tag information" (TIFF – Tagged Information File Format) can be as simple as an author’s name and page number, or as complex as colorimetry/gamma data and multiple pages and resolutions within a single file. This information makes TIFF files inherently larger than plain BMP files.

From the official TIFF Specification ver. 6…
http://partners.adobe.com/asn/developer/pdfs/tn/TIFF6.pdf "The purpose of TIFF is to describe and store raster image data. A primary goal of TIFF is to provide a rich environment within which applications can exchange image data. This richness is required to take advantage of the varying capabilities of scanners and other imaging devices."

Data dump from a simple Photoshop TIFF file.
xetven~1.tif: Warning, unknown field with tag 37724 (0x935c) ignored. TIFF Directory at offset 0x8
Subfile Type: (0 = 0x0)
Image Width: 5010 Image Length: 3792
Resolution: 600, 600 pixels/inch
Bits/Sample: 8
Compression Scheme: None
Photometric Interpretation: RGB color
Samples/Pixel: 3
Rows/Strip: 3792
Planar Configuration: single image plane
Photoshop Data: <present>, 5502 bytes

5. Check out JPEG2000. Brought to you by the people that tune image data for human perception.
From the official JPEG website
http://www.jpeg.org/JPEG2000.html
"JPEG 2000 is a new image coding system that uses state-of-the-art compression techniques based on wavelet technology. Its architecture should lend itself to a wide range of uses from portable digital cameras through to advanced pre-press, medical imaging and other key sectors"

6. Bottom line – Every use and application has specific and mostly singular file format and resolution requirements.

Scan big and process as required.
Save early and often.
Incrementally backup and store off-site.
Beware of cheap CD-Rs and fast burners.

Good Luck,

Mr3

"Jan Kohl" wrote in message

Thanks for all the answers, I didn’t realise it was going to cause

somewhat of a ruckus! 🙂

Early photo and negatives – the source material may be fragile and only

available for one scanning session.- Yes

Are the scanned images for academic use? – No, not planned for, but could

be.

How are the scans going to be

cataloged/stored/distributed/viewed/printed? – most likely on CD/computer monitor

Will the images be published? paper or electronic? – not planned, but

should be able to be published

Will the scanned images be subjected to specialized image processing at a

later date? – not that I know of

*If I scan with the possibility of publishing in future high quality

"coffee table" books, what should the minimum (pixel)

resolution be? Most of my work is done for the web, and I usually try to

hit about 18-20000 pixel width images.

*I see that bmps are typically a couple of hundred K less that tiffs,

since they both seem to produce high quality (and

editable) images, what are the differences between the two?

Sorry so wordy, but this is a subject I enjoy kicking around. 8^)

Not at all. I always like learning more!

However…if a png is lossless, yet it loses about 1/2 the size of a tiff, obviously something is being dropped…what?

Cheers!

Beware of cheap CD-Rs and fast burners.

I get around cheap CD-Rs by always burning 2 copies of everything I have. I have yet to lose anything (permanently), as of the
few CD’s that I’ve lost, I had it’s "brother" to copy from. Errrr…you’re not going to tell me that there is something else I
should worry about, are you?

For one, this archival thing I’ll be doing is very important. I want to insure when all these photos are struck to disk that
they can accurately take over as the "master" if the originals get wet, lost, burned, or whatever. Nor do I wish someone 50
years from now going "why the heck did that bozo do THIS???" (Hence my thinking I should go ahead and stay with tiff/bmp, no
matter what the storage cost…)

I use an HP burner, and even though it will burn at 24x, I always burn at 16. I’ve been burning long enough to realize that
doing CDs in 1 minute flat is rather dangerous to your data. As of yet, this HP (2 years old now) has never made a coaster.

Cheers!

People have no prob being "tricked" by the massive compression in MP3’s these days…

For well made jpegs, if a file is massively smaller and ultimately serves it’s intended purpose – that should be the goal. If the difference in appearance is negligable while the bandwidth required to view the art is 1/5th of what it "should" be for unnoticeable perfection, I’d go for the smaller file size. If the brain can’t sense the difference, then is there really a difference worth caring about, especially with the limitations of printing such as the reduced gamut?

JD

However, both Outpatient and Mr3 both brought up valid points, Outpatient

with his excellent website and Mr3 said something that really made me think…"optimized for human vision which is pretty forgiving and easily tricked".

On Tue, 04 Nov 2003 03:37:25 GMT
in alt.graphics.photoshop
"Jan Kohl" muttered something like this:

Sorry so wordy, but this is a subject I enjoy kicking around. 8^)

Not at all. I always like learning more!

However…if a png is lossless, yet it loses about 1/2 the size of a tiff, obviously something is being dropped…what?

Cheers!

»» Jan Kohl

::: computer security consultant :::
the pits – http://www.theuspits.com
castle graphics – http://www.castlegraphics.com

Well unfortunately (or fortunately) it’s not quite that simple. Just because one file is smaller than another you can not assume there is loss of information. You could assume, however, that there is a different method for storing the information. I completely understand this concept because I’m one of those people that can stuff 3 days of clothes into an overnight bag. 😉

Technically I can’t explain why PNGs are smaller, but I do know they are lossless. If you are so inclined then I’m sure you can find the answer somewhere at this site:

http://www.libpng.org/pub/png/

Like maybe here:
PNG (Portable Network Graphics) Specification, Version 1.2 http://www.libpng.org/pub/png/spec/1.2/png-1.2-pdg.html

The abstract from that page:

"Abstract

This document describes PNG (Portable Network Graphics), an extensible file format for the lossless, portable, well-compressed storage of raster images. PNG provides a patent-free replacement for GIF and can also replace many common uses of TIFF. Indexed-color, grayscale, and truecolor images are supported, plus an optional alpha channel. Sample depths range from 1 to 16 bits.

PNG is designed to work well in online viewing applications, such as the World Wide Web, so it is fully streamable with a progressive display option. PNG is robust, providing both full file integrity checking and simple detection of common transmission errors. Also, PNG can store gamma and chromaticity data for improved color matching on heterogeneous platforms.

This specification defines the Internet Media Type "image/png". " —
OuTpaTienT / outpatient°AT°rocketmail°DOT°com
http://www.0utpatient.com
http://www.oeyec.com

In fifty years those CD’s will be more obsolete than 8 track audio cartridges.

Remember Kodak Instamatic and Disc cameras, 8" & 5-1/4" floppies, 256KB memory chips, cellular phone transceivers installed in car trunks, dot matrix printers, PearlCorder audio recorders, Pulsar LED watches, Word Star, Princess rotary dial telephones, and the list goes on….

"Jan Kohl" wrote in message
Nor do I wish someone 50 years from now going "why the heck did that bozo do THIS???" (Hence my thinking I should go ahead and stay with tiff/bmp, no matter what the storage cost…)
<<<<<<<<<<<<<<<<

A quick aside.

In the basement of CBS Television City facility in Los Angeles is the Jurassic park of video technology.

"CBS possesses the largest 2-inch copying facility in the country. It has ten working 2-inch players with six additional machines that may be used for spare parts to keep the other ten running. CBS Entertainment has about 20,000 uncopied 2-inch tapes. The Sports Division has about 10,000. In addition, CBS will perform transfer work for outside organizations. A recent large customer, Martin-Goodson Productions, ordered the transfer of about 32,000 kinescopes and 2-inch tapes to
Digital Betacam. "Eventually," as CBS’s Dan Sullivan reported, "it will not be financially possible for CBS to keep these machines running and when that point is reached large scale transfer projects will no longer be possible simply because there will be no machines to do them on."

In video history, the 2" broadcast video format was the undisputed king. Now an organization the size of CBS will have tens of thousands of hours of programming marooned in the can because there will be no hardware to read the storage media.

Even though Outpatient made a very sophisticated case against JPEG compression in general, I think he missed the big picture.

JPEG is designed for the very specific task of variable compression of natural images that are to be viewed by humans.

In that vein, his sample image is inappropriate for analysis of JPEG artifacts. Assumptions made about JPEG compression artifacts need to include the target image and measurement protocol. JPEG exploits the brain’s interpretation of visual information which shifts the burden of testing from absolute image data (pixels) to perception of data (objects). As Outpatient’s graphics clearly show, JPEG is not very good at compressing artificial images. An entirely different result would be expected given an appropriate target image, and subjective evaluation of the full sized compressed image by a trained observer.

Just for the record, I only use JPEG as an output format; I capture, save, and archive in TIFF, RAW, or PSD, depending on the image and source.

From the official JPEG web site…
http://www.faqs.org/faqs/jpeg-faq/part1/

"JPEG is designed for compressing either full-color or gray-scale images of natural, real-world scenes. It works well on photographs, naturalistic artwork, and similar material; not so well on lettering, simple cartoons, or line drawings. JPEG handles only still images, but there is a related standard called MPEG for motion pictures.

JPEG is "lossy," meaning that the decompressed image isn’t quite the same as the one you started with. (There are lossless image compression algorithms, but JPEG achieves much greater compression than is possible with lossless methods.) JPEG is designed to exploit known limitations of the human eye, notably the fact that small color changes are perceived less accurately than small changes in brightness. Thus, JPEG is intended for compressing images that will be looked at by humans. If you plan to machine-analyze your images, the small errors introduced by JPEG may be a problem for you, even if they are invisible to the eye."

"Jeff H." wrote in message

People have no prob being "tricked" by the massive compression in MP3’s these days…

For well made jpegs, if a file is massively smaller and ultimately serves it’s intended purpose – that should be the goal. If the difference in appearance is negligable while the bandwidth required to view the art is 1/5th of what it "should" be for unnoticeable perfection, I’d go for the smaller file size. If the brain can’t sense the difference, then is there really a difference worth caring about, especially with the limitations of printing such as the reduced gamut?

JD

However, both Outpatient and Mr3 both brought up valid points,

Outpatient

with his excellent website and Mr3 said something that really made me think…"optimized for human vision which is pretty forgiving and easily tricked".

On Tue, 04 Nov 2003 03:46:07 GMT, "Jan Kohl" wrote:

Beware of cheap CD-Rs and fast burners.

I get around cheap CD-Rs by always burning 2 copies of everything I have. I have yet to lose anything (permanently), as of the
few CD’s that I’ve lost, I had it’s "brother" to copy from. Errrr…you’re not going to tell me that there is something else I
should worry about, are you?

A CD could be dead in 5 years or less. DVD OTOH is supposed to last around 100 years.

However, what you should remember is that if you want this stuff kept for posterity, then you should be rearchiving on whatever is the latest format.

So, you can do CD for now, but you’d be much safer using either an external hard disk (what I do) or DVD or both. And when the "next big thing" comes along, you need to wait till it[s reasonably mature, then archive to that, and so forth. otherwise, no matter if those CDs last the next 50 years, they’ll be no more use than coasters because nothing will be able to read them.

—

Hecate

veni, vidi, relinqui

Coming from the audio side of things where the quality issue rages on, IE: 192 kHz sampling rates… I added my 2 cents by saying that I’d rather hear a great song on a worn-out, hissy cassette tape than a Britney Backstreet tune made with Reason in 30 minutes, recorded perfectly to SACD.

All the hairs audio engineers split using esoteric tube gear are lost when converting to 128/44.1 mp3’s anyway.

Meaning: a great image saved as a well-made jpg will move more people emotionally than a technically perfect digital file of something random.

All the hairs we split as image experts are lost when worrying about tiffs/jpegs are lost when printing at 175 lpi, 4 color offset anyway.

The biggest problem with the bad rap jpegs unfairly get is that for so long people used 50% compression, as a "compromise" , and of course that always looks like crap.

JD

Powered be the spirited discussions of the "team", I’ve had to go back to the books several times.
From those explorations, my vague recollection of the technical side of file formats and data compression has been updated and expanded.

From a practical point, I have been following this discussion for my own application; archiving 3,200 digital captures.

For those that are interested, this is what I discovered…

TIFF is BMP with metadata Tags added to the file header. Lossless – Nothing more than a X,Y bit bucket with internal post-it notes. For special applications, some of those Tags are very useful. http://partners.adobe.com/asn/developer/pdfs/tn/TIFF6.pdf 385KB PDF file

PNG is GIF on steroids.
Lossless – Based on the algorithm family that powers ZIP. The PNG spec was created in less than a month in response to Unisys asserting patent rights over the GIF compression scheme. http://www.libpng.org/pub/png/pngintro.html Intro to PNG http://www.libpng.org/pub/png/pnghist.html History

JPEG is Your eyeball as seen by a computer.
Lossy – Created by Photo Experts to take advantage of how the human brain processes visual stimuli.
Using mathematical magic, images are cut into 8x8pixel squares, evaluated, and the least significant visual info is discarded. What’s left is encoded with more math magic.
Little known facts:
If you open a JPEG and make alterations to an area, only that area is recompressed on file save. The rest of the image shouldn’t take a hit. If you can help it, never save JPEG at the highest Q. Start at 10% less than the highest value.
http://www.jpeg.org Official web site
http://www.faqs.org/faqs/jpeg-faq/part1/ Frequently Asked Questions – Part 1

JPEG 2000 is JPEG on steroids
Lossy and Lossless – uses a new compression scheme that is more efficient; 3000KB->19KB (99% reduction) is reasonable but not great. On-The-Fly variable compression rate keeps text in an image readable. No more JPEG blues when compressing a text-on-a-picture image.
http://www.jpeg.org/JPEG2000.html Official web site
http://www.eetimes.com/story/OEG19991228S0028 Overview with sample image

For those that are interested, this is what I am going to do….

1. Capture the image at the highest practical resolution available. Probably won’t have a choice of file format for the capture. For non-scientific purposes, desktop scans at 600dpi are ok, higher dpi for smaller items. Your mileage may vary.
2. Save the capture as a PNG format file. Using the sample image of the train, the PNG file was 30% smaller. The PNG file becomes the master file for the image. PNG is a lossless format; eliminating the issue of data degradation.
3. Save the capture as a JPEG format file. This becomes the For-Position-Only (FPO)/ID image. Using the sample image of the train, a JPEG file created in PS-Save for Web, Q80, 720pixels wide was 90% smaller. This file will suffice for all computer related applications; i.e., web, image browsers, CD catalog, etc.
4. With the PNG and JPEG files, I have the ‘capture data’ locked up and the ‘visual representation’ for everything else.

DONE.

"Jeff H." wrote in message

Coming from the audio side of things where the quality issue rages on, IE: 192 kHz sampling rates… I added my 2 cents by saying that I’d rather

hear

a great song on a worn-out, hissy cassette tape than a Britney Backstreet tune made with Reason in 30 minutes, recorded perfectly to SACD.
All the hairs audio engineers split using esoteric tube gear are lost when converting to 128/44.1 mp3’s anyway.

Meaning: a great image saved as a well-made jpg will move more people emotionally than a technically perfect digital file of something random.
All the hairs we split as image experts are lost when worrying about tiffs/jpegs are lost when printing at 175 lpi, 4 color offset anyway.
The biggest problem with the bad rap jpegs unfairly get is that for so

long

people used 50% compression, as a "compromise" , and of course that always looks like crap.

JD