Don, Noons: TIF v. JPG. More samples.

[Ole-Hjalmar Kristensen]

Yes, I was curious about Scott's scan, since it had remarably little
grain/noise.


SW> He might be talking about my image, where I put up the full thing to
SW> show just what you can do with a 3.5 MB sized file.
SW> http://www.sewcon.com/temp/4000ppi.jpg
SW> In my image the quality level was something like 8 out of 12 using
SW> Photoshop, lower then I ever use unless the image is getting posted on
SW> the Internet.

<snip>

SW> Scott

 

[Scott W]

Yes, I was curious about Scott's scan, since it had remarably little
grain/noise.

Well I cheated just a bit, I wanted the cleanest image possible so I
used one from the 20D, stitched to get the resolution up.

The point was that how good the image is way more important then how it
is saved. The way you hear some people talk a jpeg image is only good
for web pages and will never be as good looking as a tiff.

Scott

 

[Ken Weitzel]

Well I cheated just a bit, I wanted the cleanest image possible so I
used one from the 20D, stitched to get the resolution up.

The point was that how good the image is way more important then how it
is saved. The way you hear some people talk a jpeg image is only good
for web pages and will never be as good looking as a tiff.

Scott

Hi...

For those among us who have psp, there's an easy test that will
quickly and easily show us what those differences are.

Simply scan something worth scanning at your best quality with
twain.

Look at it on your monitor, then get set to "save as" a jpeg.
Click on options on the save page. Next click on run optimizer.
You'll see your original in the left hand pane, and the jpeg
version in the right hand pane.

Put your cursor (hand) on the right hand pane, depress the left
mouse button, and drag it just a few pixels worth. Keep the left
button depressed and look for a sec, then let the button up.
You'll actually be watching as it compresses.

Move the slider to various compression levels, and you'll see the
differing effects.

I respectfully suggest that for the technically minded of us there
will be a huge difference, even at a quality setting of one.

I leave it the artistically minded folks to offer their opinions -
that's just not my thing

Take care.

Ken

 

[Don]

Move the slider to various compression levels, and you'll see the
differing effects.

I respectfully suggest that for the technically minded of us there
will be a huge difference, even at a quality setting of one.

Uh-oh! Now you've done it! ;o)

Prepare to be called names, have your results pooh-poohed and then
challenged to show the difference in their irrelevant "examples"
which, as it turns out:

Well I cheated just a bit

which in turn - and after all the shouting - just confirms what I've
been quietly saying from the start: Any example is totally useless
unless the complete testing environment is known.

Don.

 

[Raphael Bustin]

Uh-oh! Now you've done it! ;o)

Prepare to be called names, have your results pooh-poohed and then
challenged to show the difference in their irrelevant "examples"
which, as it turns out:

>Well I cheated just a bit

which in turn - and after all the shouting - just confirms what I've
been quietly saying from the start: Any example is totally useless
unless the complete testing environment is known.

Frankly, Don, I couldn't really understand what Ken
was talking about so I let it slide. I was hoping you
would, too.

My claim, throughout, has been very clear, constrained,
and consistent. I can only vouch for the applications I use.
Lord knows what Ken's using. "Click on run optimizer."
Heh. It isn't Photoshop he's talking about, or any
application I've run across.

Your claim, on the other hand, was unconstrained and
general and unequivocal -- and yet in about a month's
time (and dozens of requests from several parties) you
have yet to offer a *single* example.

You have no right to lecture any of us on the methods
of science or logical debate or the etiquette of USENET
discussion if you don't understand that.

I will continue to challenge your BS on this topic.


rafe b.
www.terrapinphoto.com

 

[Ken Weitzel]

Frankly, Don, I couldn't really understand what Ken
was talking about so I let it slide. I was hoping you
would, too.

My claim, throughout, has been very clear, constrained,
and consistent. I can only vouch for the applications I use.
Lord knows what Ken's using. "Click on run optimizer."
Heh. It isn't Photoshop he's talking about, or any
application I've run across.

Your claim, on the other hand, was unconstrained and
general and unequivocal -- and yet in about a month's
time (and dozens of requests from several parties) you
have yet to offer a *single* example.

You have no right to lecture any of us on the methods
of science or logical debate or the etiquette of USENET
discussion if you don't understand that.

I will continue to challenge your BS on this topic.

Hi Rafe...

Perhaps I should apologize for not only getting a little a
bit on the old side, but also for being the victim of several
strokes. Good days and bad days, but even on the good not
nearly as eloquent as I once was. Sorry about that.

However, quote from my previous message...

"For those among us who have psp, there's an easy test that will"...

And re-write it for you...

For those among us who have p(aint) s(hop) p(ro), there's an easy test
that will...

 

[rafe b]

Hi Rafe...

Perhaps I should apologize for not only getting a little a
bit on the old side, but also for being the victim of several
strokes. Good days and bad days, but even on the good not
nearly as eloquent as I once was. Sorry about that.

However, quote from my previous message...

"For those among us who have psp, there's an easy test that will"...

And re-write it for you...

For those among us who have p(aint) s(hop) p(ro), there's an easy test
that will...

Sorry to drag you into this. I will have to download the
trial of Paint Shop Pro to see what you're talking about.
Hopefully there's a demo or trial version.

This tiff with Don concerns JPG at the best quality settings
of any given application. I'm not talking about compression
ratios of 3:1 or more. In fact, at the settings I'm talking
about, JPG compression is generally around 60% or so
(ie., a bit over 2:1.)


rafe b
www.terrapinphoto.com

 

[Scott W]

which in turn - and after all the shouting - just confirms what I've
been quietly saying from the start: Any example is totally useless
unless the complete testing environment is known.

Don.

Don, it is clear that no image that anyone can come up with will
satisfy you, and I really don't see the need to argue over the
validity of any of our images as test cases for the loss in image
quality between an original tiff and a jpeg. I am perfectly happy to
use one of your scans as a test case, it only needs to be a small crop
of a tiff say 500 x 500 pixels. That would be a file that is only 1.5
MB in size, this should not be a problem for anybody to deal with.
Even a 200 x 200 crop should be more then large enough to make you
point.

In this way you would now the complete testing enviroment. You don't
need to do a new scan, if jpegs are really as bad as you would have
others believe you must have a lot of scans on hand that would do, just
take a crop from one.

Scott

 

[Raphael Bustin]

Perhaps I should apologize for not only getting a little a
bit on the old side, but also for being the victim of several
strokes. Good days and bad days, but even on the good not
nearly as eloquent as I once was. Sorry about that.

However, quote from my previous message...

"For those among us who have psp, there's an easy test that will"...

And re-write it for you...

For those among us who have p(aint) s(hop) p(ro), there's an easy test
that will...

Yes, I apologize for reading too hastily. I downloaded and installed
Paint Shop Pro just to experience the tool you're talking about.
Quite cool, though a bigger window would be even better. Why
quibble.

I've just done a "study" of JPG file sizes and compression ratios
among four applications: Each app is freshly installed and
up to date. I tested the four *least compressed* JPG steps
from each application. Here's how the sizes stack up.
All the files are here if anyone's curious:

<http://www.terrapinphoto.com/jpg_vs_tif/>

// The original is a Canon 10D RAW capture at
// 3072 x 2048, downsampled to 1500 x 1000 in PS CS2.

05/03/2006 08:43 PM 4,529,036 original.tif

// Adobe Photoshop CS2
// Scale: 12=best, 0=most compressed, Step: 1

05/03/2006 08:44 PM 1,470,466 pscs2_q12.jpg
05/03/2006 08:44 PM 935,719 pscs2_q11.jpg
05/03/2006 08:44 PM 680,192 pscs2_q10.jpg
05/03/2006 08:45 PM 518,272 pscs2_q9.jpg
05/03/2006 08:45 PM 417,044 pscs2_q8.jpg
05/03/2006 08:45 PM 325,989 pscs2_q7.jpg

// PaintShopPro
// Scale: 001=best, 99=most compressed, Step: 10

05/03/2006 09:19 PM 2,043,316 psp001.jpg
05/03/2006 09:19 PM 604,858 psp011.jpg
05/03/2006 09:20 PM 412,090 psp021.jpg
05/03/2006 09:21 PM 253,443 psp031.jpg
05/03/2006 09:22 PM 173,935 psp041.jpg

// FastStone
// Scale: 100=best, 0=most compressed, Step: 10

05/03/2006 08:57 PM 1,448,278 fs100.jpg
05/03/2006 08:57 PM 536,514 fs90.jpg
05/03/2006 08:58 PM 371,597 fs80.jpg
05/03/2006 08:58 PM 299,198 fs70.jpg

// XnView
// Scale: 100=best, 0=most compressed, Step: 10

05/03/2006 08:51 PM 1,396,310 xn100.jpg
05/03/2006 08:52 PM 521,405 xn90.jpg
05/03/2006 08:52 PM 364,162 xn80.jpg
05/03/2006 08:52 PM 295,042 xn70.jpg


Isn't JPG great? Seriously...

My general point about the visual quality of JPGs
holds for these four applications. I don't see any
visual degradation in any of these images until you
get to around 6:1 or 8:1 compression from "original.tif."



rafe b
www.terrapinphoto.com

 

[Ken Weitzel]

Yes, I apologize for reading too hastily. I downloaded and installed
Paint Shop Pro just to experience the tool you're talking about.
Quite cool, though a bigger window would be even better. Why
quibble.

I've just done a "study" of JPG file sizes and compression ratios
among four applications: Each app is freshly installed and
up to date. I tested the four *least compressed* JPG steps
from each application. Here's how the sizes stack up.
All the files are here if anyone's curious:

<http://www.terrapinphoto.com/jpg_vs_tif/>

// The original is a Canon 10D RAW capture at
// 3072 x 2048, downsampled to 1500 x 1000 in PS CS2.

05/03/2006 08:43 PM 4,529,036 original.tif

// Adobe Photoshop CS2
// Scale: 12=best, 0=most compressed, Step: 1

05/03/2006 08:44 PM 1,470,466 pscs2_q12.jpg
05/03/2006 08:44 PM 935,719 pscs2_q11.jpg
05/03/2006 08:44 PM 680,192 pscs2_q10.jpg
05/03/2006 08:45 PM 518,272 pscs2_q9.jpg
05/03/2006 08:45 PM 417,044 pscs2_q8.jpg
05/03/2006 08:45 PM 325,989 pscs2_q7.jpg

// PaintShopPro
// Scale: 001=best, 99=most compressed, Step: 10

05/03/2006 09:19 PM 2,043,316 psp001.jpg
05/03/2006 09:19 PM 604,858 psp011.jpg
05/03/2006 09:20 PM 412,090 psp021.jpg
05/03/2006 09:21 PM 253,443 psp031.jpg
05/03/2006 09:22 PM 173,935 psp041.jpg

// FastStone
// Scale: 100=best, 0=most compressed, Step: 10

05/03/2006 08:57 PM 1,448,278 fs100.jpg
05/03/2006 08:57 PM 536,514 fs90.jpg
05/03/2006 08:58 PM 371,597 fs80.jpg
05/03/2006 08:58 PM 299,198 fs70.jpg

// XnView
// Scale: 100=best, 0=most compressed, Step: 10

05/03/2006 08:51 PM 1,396,310 xn100.jpg
05/03/2006 08:52 PM 521,405 xn90.jpg
05/03/2006 08:52 PM 364,162 xn80.jpg
05/03/2006 08:52 PM 295,042 xn70.jpg


Isn't JPG great? Seriously...

My general point about the visual quality of JPGs
holds for these four applications. I don't see any
visual degradation in any of these images until you
get to around 6:1 or 8:1 compression from "original.tif."

Hi Rafe...

Glad you found it of interest...

I can still muddy the waters a bit more though

Duplicated as best I can your effort... took a pic of my
newest next door neighbor and best friend 5 month old baby in his
Mother's arm... 3264x2448 (near as I can come to your size)

Raw 10 megs; processed without touching, saved as 23 meg tiff.

Guessing by file sizes that your paint shop pro "steps" meant that
your first step was 10. Saved a jpeg at that level.

Then just for the heck of it, I counted the image colours

The tiff was 891,629 colours.
the 10 jpeg was 323,064 colours.

Amazed me that 2/3rd's of the colours were modified, so tried
it again. Saved it at paint shop pro's quality 1 (the very best).
File size was (about) 10 megs. (same as raw) And the colors were
only 351,546 - still about 2/3rd's of each and every pixel changed!

Just for what it's worth

Take care.

Ken

ps - paint shop pro - image > count image colours

 

[Raphael Bustin]

Hi Rafe...

Glad you found it of interest...

I can still muddy the waters a bit more though

Duplicated as best I can your effort... took a pic of my
newest next door neighbor and best friend 5 month old baby in his
Mother's arm... 3264x2448 (near as I can come to your size)

Raw 10 megs; processed without touching, saved as 23 meg tiff.

Guessing by file sizes that your paint shop pro "steps" meant that
your first step was 10. Saved a jpeg at that level.

Then just for the heck of it, I counted the image colours

The tiff was 891,629 colours.
the 10 jpeg was 323,064 colours.

Amazed me that 2/3rd's of the colours were modified, so tried
it again. Saved it at paint shop pro's quality 1 (the very best).
File size was (about) 10 megs. (same as raw) And the colors were
only 351,546 - still about 2/3rd's of each and every pixel changed!

Can you see the difference? That's what matters.

I certainly can't. And I've tried. I can kind of imagine
how PSP's metric might be derived. I don't believe there's
a Photoshop equivalent. Again, I ask: can you *see*
the difference?

I understand that JPG works in part by sub-
sampling chroma information. There's no argument
there. It's one of JPG's many tricks and ploys.

JPG also involves a color space conversion both
in the encode and decode.



rafe b
www.terrapinphoto.com

 

[Surfer!]

In message <[email protected]>, Raphael Bustin

My general point about the visual quality of JPGs
holds for these four applications. I don't see any
visual degradation in any of these images until you
get to around 6:1 or 8:1 compression from "original.tif."

I had problems reducing a JPG - the original wasn't too bad but on
reducing it sprouted jaggies and other undesirable artifacts. So, I
scan into TIF not because I regard it as producing a better imaged
compared to the *same* thing saved as a JPG, but so that if necessary I
can produce a lower resolution file without running into the same
problems again.

 

[Bart van der Wolf]

SNIP

Can you see the difference? That's what matters.

Obviously the JPEG and similar lossy compression schemes are designed
to minimize visual loss. Since the human visual system (HVS) has much
lower chroma resolution than luminance resolution, a number of colors
at the highest spatial frequencies can be sacrificed without too much
visual impact.

I certainly can't. And I've tried. I can kind of imagine
how PSP's metric might be derived. I don't believe there's
a Photoshop equivalent.

I'm not sure what metric you mean, but if it's the color count then
there is a plug-in which will do that on 8-b/ch images:
<http://www.telegraphics.com.au/sw/#countcolours>

Bart

 

[Don]

However, quote from my previous message...

"For those among us who have psp, there's an easy test that will"...

And re-write it for you...

For those among us who have p(aint) s(hop) p(ro), there's an easy test
that will...

I'm afraid I warned you Ken:

Prepare to be called names, have your results pooh-poohed
etc. :-(

I found your message to be quite clear and to the point. As would
anybody who actually bothered to *READ* it instead of lashing out.

Anyway, I'd say the best course of action is just to let the angry,
content-free, assaults roll of your back.

Don.

 

[Bruce]

Don,

Your statement in the previous thread about the high quality jpg
"standing out like a sore thumb" is totally incorrect, yet you keep
pressing the issue.
Others spent the time to actually test this and present their results.
You rejected their conclusions with no evidence what so ever.
You deny accusing them of lying- you just questioned "the veracity of
their examples".
You deny saying that their efforts were inadequate- its just the
documentation of their examples.
Yet you will not produce your own example.

Here you purposefully misquote "well I cheated just a bit"
Cheating in that for this example he used an exceptionally low noise
image from a digital camera composite instead of a scan. So any jpg
artifacts should have been more visible. In this case, he also saved it
at less than top quality. To my view, he beat you with both hands tied
behind his back.

You insist others are wrong, you accuse them of lying, say their
examples are irrelevant, taunt them and then try to spin this as if you
were some poor innocent (and you have done this here on other threads as
well- to the point of driving other knowledgeable people away). As many
people who have tested it can affirm, your assertion about seeing the
jpg effects at high quality settings is wrong, and if you really believe
otherwise, show us an example.

 

[Scott W]

In message <[email protected]>, Raphael Bustin


I had problems reducing a JPG - the original wasn't too bad but on
reducing it sprouted jaggies and other undesirable artifacts. So, I
scan into TIF not because I regard it as producing a better imaged
compared to the *same* thing saved as a JPG, but so that if necessary I
can produce a lower resolution file without running into the same
problems again.

Several comments, I generally scan directly into Photoshop and then
save from there. I trust Photoshop more for generating a good jpeg
compared to the scanner software.

Down sizing a jpeg should not be a problem as long as the quality is
fairly high. You might try this experiment, take a scan that has been
saved to a tiff and load it into Photoshop, save as a quality 12 jpeg
and close the file. Now open both the tiff and the jpeg and resize
both and check to see if you can see any differences.

This is how the jpeg and tiff look at 100%, this is of course a crop
from a much larger photo
http://www.sewcon.com/temp/fullsizecrop.tif
Kind of hard to tell which was the jpeg.

Here I have put the tiff in one layer and the jpeg in another.
http://www.sewcon.com/temp/layers.tif

And here they are resized to 70%
http://www.sewcon.com/temp/70precent.tif

This is an easy test to do for any tiff and I would be more then happy
to do it for anyone's scans.

Scott

 

[Don]

Your statement in the previous thread about the high quality jpg
"standing out like a sore thumb" is totally incorrect, yet you keep
pressing the issue.

Wrong on both counts.

I haven't written anything for *weeks* on the subject! Indeed, the
only people "pressing the issue" are those trying to pick a fight with
name calling and provocations, to which I don't respond.

Anyway, I've asked you twice already to:

!!!===> Please provide a *single* quote *in context* <===!!!

to support your wild and unsubstantiated accusations.

In both cases you go quiet for a few days and then resurface with the
same baseless accusations again which I have already addressed in some
detail in previous responses, even pointing you to exact messages.

Therefore, it appears any future correspondence is pointless since you
repeatedly refuse to *read* the thread (i.e. past messages) and
address that.

Don.

 

[Bruce]

Don,

How can you say "I haven't written anything for *weeks* on the subject!"
when my post was in direct reply to your message from the day before
in which you said that the examples were "irrelevant" and "useless" and
in which you even purposely tried to misquote out of context to imply
that an example was a cheat. If you believe we are wrong, then present
some evidence to support your attacks on our observations. Stop trying
to spin this as if we are trying to pick a fight. You are the one who
stepped in saying that others were wrong and attacking their competence
and "veracity" rather than presenting facts or examples.

 

[Scott W]

Wrong on both counts.

I haven't written anything for *weeks* on the subject! Indeed, the
only people "pressing the issue" are those trying to pick a fight with
name calling and provocations, to which I don't respond.

Anyway, I've asked you twice already to:

!!!===> Please provide a *single* quote *in context* <===!!!

to support your wild and unsubstantiated accusations.

In both cases you go quiet for a few days and then resurface with the
same baseless accusations again which I have already addressed in some
detail in previous responses, even pointing you to exact messages.

Therefore, it appears any future correspondence is pointless since you
repeatedly refuse to *read* the thread (i.e. past messages) and
address that.

Don.

See Don this is the part that is so frustrating you dismiss all other
images with statements like " Any example is totally useless unless the
complete testing environment is known" and yet you will not put forth a
test image of your own. I believe that my image if a very valid one,
but why should this matter when it would take you no more then 5
minutes to post your own image?

Of course the reason you don't is that you have already tried the
experiment and you are seeing the same thing that we are seeing, your
images don't behave any differently then ours. So instead of
admitting that your rather overstated the evils of jpeg images you
would rather try to raise doubt about any tests that show that you were
way wrong.

And when someone points out that this is a lack of integrity on your
part you get all huffy and talk about being attacked.

You keep trying to divert the topic away from your statement "At
100% magnification (i.e. 1:1) even a JPG image at lowest compression
(i.e. highest quality) stands out like a sore thumb when
compared to the original.". You imply that others are getting bad
results because among other things they scanning software might be
mutilating the image to the point that jpeg artifacts would not show
up. It is clear that you can play this game with any image that anyone
posts, you can always raise doubts to its validity.

But if you really believed that this was the case then you could post
your own image, but you can't because you don't have an image that
will show visible degradation when saved as the highest quality
setting.

When this thread started to get back to really looking at the true
effects of jpeg compression in you come saying that the test image is
"totally useless"

Scott

 

[Don]

How can you say "I haven't written anything for *weeks* on the subject!"
when my post was in direct reply to your message from the day before

The key phrase there (which you missed even though you quoted it) is:

"on the subject"

You're confusing procedural matters with the subject matter.

Again, just read the thread. All the answers are there. Repeated
several times.

Don.