Computer Graphics with Photoshop - Jpeg File size mystery

Author

Jpeg File size mystery

norris_watkins@hotmail.com

2007-06-19, 6:22 pm

I have a canon 6.3 MP DSLR ( Digital rebel ) camera, that produces
jpeg files of the size approx 3-4 Mbytes.
But if I open one of these files in photoshop elements and resave it
as jpeg ( I know I shouldnt resave jpegs ), the new file is about 500
kb. ( in the default good quality of 8 ).
Even if I choose the best jpeg quality ( 10 ) the file is still only
about 700 kb.

When I check the quality of the produced jpegs, by zooming the picture
to huge sizes, I dont see much difference in the quality. ( Even the
exif data - camera model, shutter speed, date etc - are preserved )

Is it possible that the camera didnt have much horsepower to do a good
job with the compression, so that it ended it with a big file that was
not much compressed. If that is true then I can save some hard disk
space by recompressing the images.

thanks
--nw

ronviers@gmail.com

2007-06-19, 10:14 pm

On Jun 19, 5:10 pm, norris_watk...@hotmail.com wrote:
> I have a canon 6.3 MP DSLR ( Digital rebel ) camera, that produces
> jpeg files of the size approx 3-4 Mbytes.
> But if I open one of these files in photoshop elements and resave it
> as jpeg ( I know I shouldnt resave jpegs ), the new file is about 500
> kb. ( in the default good quality of 8 ).
> Even if I choose the best jpeg quality ( 10 ) the file is still only
> about 700 kb.
>
> When I check the quality of the produced jpegs, by zooming the picture
> to huge sizes, I dont see much difference in the quality. ( Even the
> exif data - camera model, shutter speed, date etc - are preserved )
>
> Is it possible that the camera didnt have much horsepower to do a good
> job with the compression, so that it ended it with a big file that was
> not much compressed. If that is true then I can save some hard disk
> space by recompressing the images.
>
> thanks
> --nw

Hi nw,
What are the pictures of? How they compress will depend a lot on
things like noise, spatial detail, exposure, DoF etc. Take a picture
of a piece of concrete, it will not compress much.

Ron

norris_watkins@hotmail.com

2007-06-19, 10:14 pm

On Jun 19, 6:33 pm, "ronvi...@XXXXXXXXXX" <ronvi...@XXXXXXXXXX> wrote:
> On Jun 19, 5:10 pm, norris_watk...@hotmail.com wrote:
>
>
>
>
>
>
>
>
>
> Hi nw,
> What are the pictures of?

These are pretty ordinary pictures like our family members, birds,
bridges etc.
I understand that each of them compresses differently. But I dont
understand why there should be such a big difference between the
camera compression and photoshop.
Actually even MS Paint compresses it comparable to PS.

I m wondering if there is any difference in chroma subsampling etc.

By the way when I right click one of these pictures in windows
explorer and look at the properties, I see that they have the bit
depth as 24 as expected.

But if I look at the properties->meta data->exif data in PS, I see
that the bit depth is 8 bits.
So Im confused even more.

--nw

norris_watkins@hotmail.com

2007-06-19, 10:14 pm

> By the way when I right click one of these pictures in windows
> explorer and look at the properties, I see that they have the bit
> depth as 24 as expected.
>
> But if I look at the properties->meta data->exif data in PS, I see
> that the bit depth is 8 bits.

'Sorry I meant properties->meta data->File properties' here

--nw

tacit

2007-06-19, 10:15 pm

In article <1182294963.583549.181340@u2g2000hsc.googlegroups.com>,
norris_watkins@hotmail.com wrote:

> By the way when I right click one of these pictures in windows
> explorer and look at the properties, I see that they have the bit
> depth as 24 as expected.
>
> But if I look at the properties->meta data->exif data in PS, I see
> that the bit depth is 8 bits.
> So Im confused even more.

Photoshop and Windows Explorer both agree.

Photoshop shows you 8 bits. 8 bits per channel times 3 channels (RGB) =
24 bits. Photoshop shows you the number of bits per channel; Explorer
shows you the number of bits per pixel.

--
Photography, kink, polyamory, shareware, and more: all at
http://www.xeromag.com/franklin.html

Bruce Uttley

2007-06-20, 3:14 am

In article <1182291023.286596.88340@w5g2000hsg.googlegroups.com>,
<norris_watkins@hotmail.com> wrote:
>I have a canon 6.3 MP DSLR ( Digital rebel ) camera, that produces
>jpeg files of the size approx 3-4 Mbytes.
>But if I open one of these files in photoshop elements and resave it
>as jpeg ( I know I shouldnt resave jpegs ), the new file is about 500
>kb. ( in the default good quality of 8 ).
>Even if I choose the best jpeg quality ( 10 ) the file is still only
>about 700 kb.
>
>When I check the quality of the produced jpegs, by zooming the picture
>to huge sizes, I dont see much difference in the quality. ( Even the
>exif data - camera model, shutter speed, date etc - are preserved )
>
>Is it possible that the camera didnt have much horsepower to do a good
>job with the compression, so that it ended it with a big file that was
>not much compressed. If that is true then I can save some hard disk
>space by recompressing the images.
>
>thanks
>--nw

I have the same camera and have noticed that the original Rebel
routinely produces 'bloated' jpg images when shooting Medium
resolution. A couple of years ago I downloaded a copy of
PhotoMan by Keith Sheppard to analyze all the components of
the jpg files produced and found that the images contained a
lot of 'dead wood' -- thousands of bytes of data that wasn't
part of the jpg image or part of the EXIF tags.

You can download a (free) copy of Keith Sheppard's PhotoMan
application from:
http://homepages.tesco.net/~Keith.S...otoman/home.htm
Use it to compare a before and after image.

My solution to remove the 'dead wood' is to use IrfanView's
lossless jpg transformations, selecting the transformation option
of "None". As the option says, it "can be used for optimizing
and cleaning" -- exactly what it does for me.

Jim

2007-06-20, 3:14 am

<norris_watkins@hotmail.com> wrote in message
news:1182294963.583549.181340@u2g2000hsc.googlegroups.com...
> On Jun 19, 6:33 pm, "ronvi...@XXXXXXXXXX" <ronvi...@XXXXXXXXXX> wrote:
>
>
> These are pretty ordinary pictures like our family members, birds,
> bridges etc.
> I understand that each of them compresses differently. But I dont
> understand why there should be such a big difference between the
> camera compression and photoshop.
> Actually even MS Paint compresses it comparable to PS.
>
> I m wondering if there is any difference in chroma subsampling etc.
>
> By the way when I right click one of these pictures in windows
> explorer and look at the properties, I see that they have the bit
> depth as 24 as expected.
Total number of bits is 8 * 3
>
> But if I look at the properties->meta data->exif data in PS, I see
> that the bit depth is 8 bits.
> So Im confused even more.
8 Bits per channel. 3 Channels times 8 bits per channel results in 24 bits
per pixel.
Jim
>
>
> --nw
>
>
>
>
>
>
>

Nicholas Sherlock

2007-06-20, 3:14 am

Bruce Uttley wrote:
> [...] A couple of years ago I downloaded a copy of
> PhotoMan by Keith Sheppard to analyze all the components of
> the jpg files produced and found that the images contained a
> lot of 'dead wood' -- thousands of bytes of data that wasn't
> part of the jpg image or part of the EXIF tags.

This is probably the camera's stored thumbnail of the image, so that it
can browse images quickly on the camera.

Cheers,
Nicholas Sherlock

norris_watkins@hotmail.com

2007-06-20, 3:14 am

On Jun 19, 11:31 pm, Nicholas Sherlock <N.sherl...@XXXXXXXXXX> wrote:
> Bruce Uttley wrote:
>
> This is probably the camera's stored thumbnail of the image, so that it
> can browse images quickly on the camera.

Thank you very much for the responses.
Both dead wood and thumpnails sounds very plausible.

Thanks Bruce for the tip in irfanview. Yes when I did the lossless
compression with 'none', I managed to get the size reduced by about
15%. ( A 2400 kb file became 2000 kb )
But nothing compared to the huge savings I was seeing with the 'save
as' of photoshop. ( 2400 kb file becoming 650 kb )

At this point Im wondering if it is even possible to produce the exact
same quality jpegs with different compression ratios.
I mean how can I verify that the quality has not suffered when I did a
'save as'

Thanks again also for the clarification on 8 bits vs 24 bits
--nw

Toke Eskildsen

2007-06-20, 6:15 pm

wrote:

> At this point Im wondering if it is even possible to produce the
> exact same quality jpegs with different compression ratios.

Not normally. But it's simple to show that it can be done. The trivial
case is a one-pixel image, saved at different compression ratios.

> I mean how can I verify that the quality has not suffered when I
> did a 'save as'

One way is to use the tool compare from the freeware suite ImageMagick.
This will give you a measurement for the difference between two images.

Another way is to put the images on top of each other as layers, then
subtract them and look for noise. You might want to do a levels
adjustment to make the difference more visible.

A third way is simple: Put the images on top of each other in different
windows, so that they align exactly. Then switch back and forth with
alt-tab.

As for your original question: JPEGs are made to benefit from
specialised hardware. Your camera probably has more horsepower that
your computer, when it comes to JPEG compression.

I must admit that I am not familiar with Photoshop Elements, but
compressing 6.3MP down to 700KB sounds like an optimisation towards
screen viewing, rather than archival or big format prints.
--
Toke Eskildsen - http://ekot.dk/

Joe

2007-06-20, 6:15 pm

norris_watkins@hotmail.com wrote:

> I have a canon 6.3 MP DSLR ( Digital rebel ) camera, that produces
> jpeg files of the size approx 3-4 Mbytes.
> But if I open one of these files in photoshop elements and resave it
> as jpeg ( I know I shouldnt resave jpegs ), the new file is about 500
> kb. ( in the default good quality of 8 ).
> Even if I choose the best jpeg quality ( 10 ) the file is still only
> about 700 kb.
>
> When I check the quality of the produced jpegs, by zooming the picture
> to huge sizes, I dont see much difference in the quality. ( Even the
> exif data - camera model, shutter speed, date etc - are preserved )
>
> Is it possible that the camera didnt have much horsepower to do a good
> job with the compression, so that it ended it with a big file that was
> not much compressed. If that is true then I can save some hard disk
> space by recompressing the images.

I don't have PS Element to know what you are doing at you end, so why not
give the *exact* option at your end see if I (we) can be able to figure out
what you do wrong. Also, in general

- CD/DVD is so cheap these days and good family memory can be priceless for
the next generation(s), so I would suggest to keep the original and the
retouched version. And save as HIGHEST

- And *if* you save as HIGHEST QUALITY then the size should either about the
same, and usually *larger* than the original (unless you crop and reduce the
quality).

> thanks
> --nw

Mike

2007-06-20, 10:14 pm

In article <Xns9955D95D18470tokeeskildsen@130.225.247.90>, darkwing@daimi.au.dk says...
> wrote:
>
>
> Not normally. But it's simple to show that it can be done. The trivial
> case is a one-pixel image, saved at different compression ratios.
>
>
> One way is to use the tool compare from the freeware suite ImageMagick.
> This will give you a measurement for the difference between two images.
>
> Another way is to put the images on top of each other as layers, then
> subtract them and look for noise. You might want to do a levels
> adjustment to make the difference more visible.
>
if you put one on top of the other then the difference mode is useful to see where the biggest effect is.

tacit

2007-06-20, 10:15 pm

In article <1182318886.398434.260370@n60g2000hse.googlegroups.com>,
norris_watkins@hotmail.com wrote:

> At this point Im wondering if it is even possible to produce the exact
> same quality jpegs with different compression ratios.
> I mean how can I verify that the quality has not suffered when I did a
> 'save as'

You can't. The quality DID suffer.

If you open a JPEG and then do a Save As, the JPEG is compressed again.
Every time you do this, the quality goes down. The loss of quality is
always cumulative with each successive iteration; if you open the JPEG
and the save it again, you lose more quality; open and save again, you
lose still more quality; and so on.

If you care about quality, don't use JPEG.

--
Photography, kink, polyamory, shareware, and more: all at
http://www.xeromag.com/franklin.html

Joe

2007-06-21, 3:14 am

tacit <tacitr@aol.com> wrote:

> In article <1182318886.398434.260370@n60g2000hse.googlegroups.com>,
> norris_watkins@hotmail.com wrote:
>=20
exact[color=darkred]
a[color=darkred]
>=20
> You can't. The quality DID suffer.
>=20
> If you open a JPEG and then do a Save As, the JPEG is compressed again.=
=20
> Every time you do this, the quality goes down. The loss of quality is=20
> always cumulative with each successive iteration; if you open the JPEG=20
> and the save it again, you lose more quality; open and save again, you=20
> lose still more quality; and so on.
>=20
> If you care about quality, don't use JPEG.

Losing quality may not be the exact word (even I have been hearing this
for ages), or the color of some pixel may chance a little. And it was
nearly a decade ago when I tested a low quality JPEG, and software these
days are better as well as most digital cameras have much higher =
resolution
than decade ago. IOW, few thousands of pixels with color changing a =
little
may not effect 9-12+MP image.

So, I would suggest to find a good graphic program (like Photoshop or
similar) then copy a 8-10MP image over and over several times, then ZOOM =
IN
the original and the last saved 500-600% (or whatever) til you see the
square-blocks, and see for yourself.

Yes, some poor graphic program's may lose lot of quality (you can see =
with
your own eye) but not all graphic programs.

John Boy

2007-06-21, 6:14 pm

Joe wrote:

> Yes, some poor graphic program's may lose lot of quality (you can see with
> your own eye) but not all graphic programs.

JPEG is a standard, and easy to code. I see no reason the results should
be different with any program.

Joe

2007-06-21, 6:14 pm

John Boy <askme@askme.tv> wrote:

> Joe wrote:
>
>
> JPEG is a standard, and easy to code. I see no reason the results should
> be different with any program.

JPEG is standard but the programmer isn't, and that's why different
program gives different result, and that's why some dreamers say PSP is just
as good as Photoshop but 1/6 the price <bg>. And I didn't think it would
make much or any difference til around 7-8+ years ago I messed around with a
cheapie graphic program called something like Magix (something ???) and I
saw it destroyed the image right before my eyes. I couldn't believe it so I
saved different graphic at Max Quality and the screen was filled with
artifacts (didn't even need to zoom in to see them).

Same with RAW converter (even between LightRoom, ARC v3.x, ARC v4.x and
Photoshop) they don't give the same result. And I only talking about some
color, contrast, explosure, noise etc..

Dave

2007-06-21, 6:14 pm

>if you put one on top of the other then the difference mode is useful to see where the biggest effect is.
>

>If you care about quality, don't use JPEG.
>

> IOW, few thousands of pixels with color changing a little
>may not effect 9-12+MP image.

>yes, some poor graphic program's may lose lot of quality (you can see with
> your own eye) but not all graphic programs.

>JPEG is a standard, and easy to code. I see no reason the results should
>be different with any program.

> JPEG is standard but the programmer isn't, and that's why different
>program gives different result, and that's why some dreamers say PSP is just
>as good as Photoshop but 1/6 the price <bg>.

A few years ago I was involved in a discussion whether jpg's also
lose value when getting moved. I said yes, and the other guys said no.
And that is where we left it.

Moving between files may be different, but moving from one drive to
another, is still nothing less than copy to one drive and deleting on
the other.

How many of you guys take the images to the printer on a memory stick?
I mean, it is not practical to use a CD every time, when sometimes
(like my intentions for tomorrow) is to have about 20 jumbos
( 4 x 6 inch or 100 x 150 cm's) and two A4's printed.

If this was for framing or for a newspaper, I would have had it done
from Tiffs, but this is simply for once in a while casual viewing, so
it will be done from jpg's.

Dave

Toke Eskildsen

2007-06-21, 6:14 pm

Joe wrote:

> So, I would suggest to find a good graphic program (like
> Photoshop or similar) then copy a 8-10MP image over and over
> several times, [...]

It's an old experiment, which sounds intuitive, but gives misleading
results: Most, if not all, JPEG-encoders converge to a stable image
when you recompress the same image over and over. By far the biggest
loss in quality happens at the first save, while subsequent
recompressions introduces less and less noise.

If the sun is in the right direction and you diety of choice smiles
upon you, you might be able to contact my webserver and see this
demonstrated at http://ekot.dk/misc/recompress2/

This means that the experiment you outline will apparently demonstrate
that there are no real problem with using JPEG as a working format,
while the truth is (fortunately) what we all are told: That there will
be significant cummulative losses.

If the compression settings are tweaked between recompressions or if
the image is changed between saves (e.g. adjusting brightness and
such), the losses will continue to be significant.

As for the losses themselves, I find that a much overlooked aspect of
JPEG recompression is the dulling of colors. The noise is easy to see,
but we tend not to notice when the colors are just a little less
bright, compared to the previous version.
--
Toke Eskildsen - http://ekot.dk/

sonyantony@hotmail.com

2007-06-22, 6:16 pm

I ve been reading this interesting thread.
Since I too used to use the same camera, I decided to try this out.
Yes there is a 5 fold size reduction when you do a 'save as' in
photoshop, or MS paint.
I have uploaded the pictures to my webserver.
http://bellsouthpwp2.net/s/o/sonyantony/
It will be great if somebody can look at these pictures and explain.

--sony

hoffmann@fho-emden.de

2007-06-22, 6:16 pm

sonyant...@hotmail.com schrieb:
> I ve been reading this interesting thread.
> Since I too used to use the same camera, I decided to try this out.
> Yes there is a 5 fold size reduction when you do a 'save as' in
> photoshop, or MS paint.
> I have uploaded the pictures to my webserver.
> http://bellsouthpwp2.net/s/o/sonyantony/
> It will be great if somebody can look at these pictures and explain.
> --sony

3072 x 2048 pixels = 18 MBytes TrueColor.
Original JPEG: 6 MBytes. Compression ratio only 3.
Photoshop JPEG: 2 MBytes. Compression ratio 9.

Color subsampling delivers already a compression ratio 2.
The color information is averaged for four adjacent pixels.
A further compression by 1.5 is not worth the effort.
Even without color subsampling: a JPEG compression
ratio 3 is too low. Ratio 2..3 can be achieved by
lossless TIFF compression.

With color subsampling (mostly standard):
Ratio = 5: very good quality
Ratio =10: good quality
Ratio =20: good Web quality

Both files contain the sRGB ICC profile, which consumes
only 3kBytes.

http://www.fho-emden.de/~hoffmann/jpeg131200.pdf

Best regards --Gernot Hoffmann

Joe

2007-06-22, 6:16 pm

Dave <d@d.durbs> wrote:

>=20
to see where the biggest effect is.=20[color=darkred]
>=20
>=20
>=20
with[color=darkred]
>=20
should=20[color=darkred]
>=20
is just[color=darkred]
>=20
>=20
> A few years ago I was involved in a discussion whether jpg's also
> lose value when getting moved. I said yes, and the other guys said no.
> And that is where we left it.=20

The JPEG vs other Mighty God has been going on for nearly 2 decades, and
many just pass what they read/heard from other like a word of bible =
without
knowing what has been going on.

> Moving between files may be different, but moving from one drive to
> another, is still nothing less than copy to one drive and deleting on
> the other.

Years ago, some claimed that the quality of his photos on CD got faded
away because the life of CD was getting shorter etc. And yup! I have =
read
some experted passing the value experience to others by saying copying =
JPG
(not saving) from one drive to other will degrade the quality <bg>
=20
> How many of you guys take the images to the printer on a memory stick?
> I mean, it is not practical to use a CD every time, when sometimes
> (like my intentions for tomorrow) is to have about 20 jumbos
> ( 4 x 6 inch or 100 x 150 cm's) and two A4's printed.

My inkjet printer has built-in memory card reader and it can print =
without
computer, but I haven't printed any without going through Photoshop none =
for
me so far. *BUT* these days with many cheap small printer that can print
directly from camera or memory stick, and some stores have machine with
Scanner and Printer they accept memory stick and CD etc.. so the answer =
is
YES
=20
> If this was for framing or for a newspaper, I would have had it done
> from Tiffs, but this is simply for once in a while casual viewing, so
> it will be done from jpg's.=20

Still no TIFF for me, cuz I don't save and save and save the same JPG =
over
and over, and with 9-10+GP a slightly color changing of few pixels won't
make me sweated so TIFF ain't in my dictionary <bg>

=20
> Dave

Joe

2007-06-22, 6:16 pm

Toke Eskildsen <darkwing@daimi.au.dk> wrote:

> Joe wrote:
>
>
> It's an old experiment, which sounds intuitive, but gives misleading
> results: Most, if not all, JPEG-encoders converge to a stable image
> when you recompress the same image over and over. By far the biggest
> loss in quality happens at the first save, while subsequent
> recompressions introduces less and less noise.
>
> If the sun is in the right direction and you diety of choice smiles
> upon you, you might be able to contact my webserver and see this
> demonstrated at http://ekot.dk/misc/recompress2/

I sure can see the difference *but* I don't know what the heck the guy did
at his end. And of course I won't buy his story unless I know he knows what
I is going or I know what he did to the test etc..

Why don't you try it yourself, and it wouldn't take more than some seconds
(to make the copying), or 1-2 minutes to compare the difference

1. Start with a good quality image. Just don't use RAW converter or
anything to blow up any color channel, or chaging the sharpness, noise etc..
Of course unless you know exactly what to do, then it may be another story
(IOW, I am saying to PUBLIC not directly to you or any specific)

> This means that the experiment you outline will apparently demonstrate
> that there are no real problem with using JPEG as a working format,
> while the truth is (fortunately) what we all are told: That there will
> be significant cummulative losses.

You may not want to live with what you are told to live your life, but you
may want to try thing your own.

> If the compression settings are tweaked between recompressions or if
> the image is changed between saves (e.g. adjusting brightness and
> such), the losses will continue to be significant.

If you have problem then you may want to learn some trick or skill to
overcome the problem. And when you say "compression" which doesn't mean
much or anything, because it will depend on your technique, compression
level, and program you use etc..

> As for the losses themselves, I find that a much overlooked aspect of
> JPEG recompression is the dulling of colors. The noise is easy to see,
> but we tend not to notice when the colors are just a little less
> bright, compared to the previous version.

And I am guessing you are telling me about the NOISE based on the samples
you see at the web? Me? just by looking at the samples I can smell
something ain't right <bg>

Toke Eskildsen

2007-06-24, 6:14 pm

Joe wrote:

> Toke Eskildsen <darkwing@daimi.au.dk> wrote:
>
> I sure can see the difference *but* I don't know what the
> heck the guy did at his end.

I am "that guy" and I wrote what I did on the page under the heading
"Methodology". If you read that part, I'd be happy to clarify any
specific questions to it, here as well as on the web page.

(checking...)

I see that the webserver refused to show the Tcl-script I used. That is
fixed now.

Toke:
>
> You may not want to live with what you are told to live your
> life, but you may want to try thing your own.

In your newfound knowledge that I am in fact the author of the page
you're referring to above, do you have any elaborations for your snide
comment above? Maybe "Whoops... I messed up. Sorry about trying to
belittle you, when you in fact did try things out on your own"?

Toke:
>
> If you have problem then you may want to learn some trick or
> skill to overcome the problem.

The JPEG FAQ disagrees with you:
http://www.faqs.org/faqs/jpeg-faq/part1/section-10.html
Could you elaborate, please?

> And when you say "compression" which doesn't mean much or anything,
> because it will depend on your technique, compression level, and
> program you use etc..

Just as "then copy a 8-10MP image over and over several times" doesn't
make sense? It's a matter of detail-level in the discussion.

> And I am guessing you are telling me about the NOISE based on
> the samples you see at the web? Me?

I have no idea about who you are, outside of this thread. Judging by
your exceedingly bad manners, I'll assume that you consider yourself
very knowledgeable in this area.

> just by looking at the samples I can smell something ain't right <bg>

What's wrong with the samples then? Don't you find them representative
for the kind of images JPEG is used for (with the obvious excemption of
course)?

Or is this an "The guy don't know how to take good photos, so it's a
sure thing that he messes up in the technical department too"-thing?
--
Toke Eskildsen - http://ekot.dk/

Joe

2007-06-24, 10:15 pm

Toke Eskildsen <darkwing@daimi.au.dk> wrote:

> Joe wrote:
>
>
> I am "that guy" and I wrote what I did on the page under the heading
> "Methodology". If you read that part, I'd be happy to clarify any
> specific questions to it, here as well as on the web page.

How do you come up with the fancy NOISE that I have can't figure out how
you did it. Also, noise alone already blew my mind away, but sharpest noise
I have ever seen really got me good.

Or anyone who wants to find out the difference just making the dupe of
original 5-10 times (save to BEST) and they can see for themselves.

> (checking...)
>
> I see that the webserver refused to show the Tcl-script I used. That is
> fixed now.
>
> Toke:
>
> In your newfound knowledge that I am in fact the author of the page
> you're referring to above, do you have any elaborations for your snide
> comment above? Maybe "Whoops... I messed up. Sorry about trying to
> belittle you, when you in fact did try things out on your own"?

I still wondering how you come up with those fancy noise. I bet most
folks here can try and have the same question.

> Toke:
>
> The JPEG FAQ disagrees with you:
> http://www.faqs.org/faqs/jpeg-faq/part1/section-10.html
> Could you elaborate, please?
>
>
> Just as "then copy a 8-10MP image over and over several times" doesn't
> make sense? It's a matter of detail-level in the discussion.
>
>
> I have no idea about who you are, outside of this thread. Judging by
> your exceedingly bad manners, I'll assume that you consider yourself
> very knowledgeable in this area.

You mean don't believe what I see is bad manner? I haven't said I am
good/bad yet, but I would suggest other to copy their JPG several times to
see the difference.

>
> What's wrong with the samples then? Don't you find them representative
> for the kind of images JPEG is used for (with the obvious excemption of
> course)?

What wrong? the sample just doesn't smell right, and I can't say it
smells good <bg>

> Or is this an "The guy don't know how to take good photos, so it's a
> sure thing that he messes up in the technical department too"-thing?

Do I have to agree with you to make you happy? You asked to look at the
sample and I can't find anything I can agree, and you are not happy because
I don't agree with you? <bg>

Toke Eskildsen

2007-06-25, 3:14 am

Joe wrote:
> How do you come up with the fancy NOISE that I have can't
> figure out how
> you did it. Also, noise alone already blew my mind away, but
> sharpest noise I have ever seen really got me good.

Are you talking about the normalised difference? Normalised means that
the narrow band of murky colors making up the difference is stretched
over the full spectrum, this making it easy to see where the noise is
concentrated. It does not show how harsh the noise is, just the
distribution - besides that, it acts as a guide of where to look in the
normal pictures.

http://en.wikipedia.org/wiki/Normal...e_processing%29

If that wasn't it, you'll have to be more specific in your questioning.

> Or anyone who wants to find out the difference just making
> the dupe of original 5-10 times (save to BEST) and they can see
> for themselves.

That would demonstrate my point very well: That unchanged (besides the
recompression) JPEG images stabilises. It's also an unrealistic
workflow and thus does not lead to any real-world use.

> You mean don't believe what I see is bad manner?

I am not stupid enough to fall for such a simple trick. Try again.

Now, you skipped my question about recompression with changes:

I would like an answer there, if you have the time, as your
observations are clearly different.
[color=darkred]
>
> What wrong? the sample just doesn't smell right, and I can't
> say it smells good <bg>

If you can't put words to your gut feelings, you won't change anything.

>
> Do I have to agree with you to make you happy?

Sigh... You clearly passed "Dirty tricks in conversations 101" and took
extensive notes, didn't you?

When you don't give me any more data that what you did, I try and
discover what you mean by asking questions. As you see, my first
question was whether the samples weren't representative.

> You asked to look at the
> sample and I can't find anything I can agree, and you are not
> happy because I don't agree with you? <bg>

I an unhappy, because you aren't capable of stating what is wrong with
either my samples, methodology or conclusions. I am unhappy because you
just repeat your proposal for an experiment, even though I have already
tried it and even though I have explained why the experiment isn't
representative for real life working with images.
--
Toke Eskildsen - http://ekot.dk/

Joe

2007-06-25, 6:16 pm

Toke Eskildsen <darkwing@daimi.au.dk> wrote:

<snip>
>=20
> I am not stupid enough to fall for such a simple trick. Try again.

I read people say "three times is the charm" <bg>. So, You mean don't
believe what I see is bad manner?

Toke Eskildsen

2007-06-25, 6:16 pm

Joe wrote:

> Toke Eskildsen <darkwing@daimi.au.dk> wrote:
>
> I read people say "three times is the charm" <bg>. So, You
> mean don't believe what I see is bad manner?

When did you stop beating your wife?

I do believe that you are not interested in reaching a conclusion. This
discussion is over for now for my part. I'll participate again, if you
get back on track.
--
Toke Eskildsen - http://ekot.dk/

Joe

2007-06-25, 10:14 pm

Toke Eskildsen <darkwing@daimi.au.dk> wrote:

> Joe wrote:
>
>
> When did you stop beating your wife?

I guess since I started beating you <bg>

> I do believe that you are not interested in reaching a conclusion. This
> discussion is over for now for my part. I'll participate again, if you
> get back on track.

No, I do interest in something I can learn, but don't want to waste my
time on thing has no value to me. I am guessing most people don't find
anything true from your sample, and I still don't know how you can come up
with that fancy noise.