Sign in with
Sign up | Sign in
Your question

Does editing a JPEG reduce its quality?

Last response: in Digital Camera
Share
Anonymous
June 18, 2005 7:09:04 PM

Archived from groups: rec.photo.digital (More info?)

Just editing it, *no* saving, just playing with the stuff in photoshop
or paintshoppro.

Does it reduce the quality of a JPEG?
Anonymous
June 18, 2005 8:29:36 PM

Archived from groups: rec.photo.digital (More info?)

Nostrobino wrote:
> "Mike Henley" <casioculture@gmail.com> wrote in message
> news:1119132544.066322.297460@g44g2000cwa.googlegroups.com...
> >
> >
> > Just editing it, *no* saving, just playing with the stuff in photoshop
> > or paintshoppro.
> >
> > Does it reduce the quality of a JPEG?
>
> No. As long as you *do not* save whatever image you've played with to the
> same filename (overwriting the original) you haven't changed the original in
> any way. You can play with it all you like, as often as you like.
>
> N.

Thanks all. I guess good practice though would be to "save as" BEFORE
one makes any changes or edits, just in case one accidentally saves
them.
Anonymous
June 18, 2005 10:59:19 PM

Archived from groups: rec.photo.digital (More info?)

"Mike Henley" <casioculture@gmail.com> wrote in message
news:1119132544.066322.297460@g44g2000cwa.googlegroups.com...
>
>
> Just editing it, *no* saving, just playing with the stuff in photoshop
> or paintshoppro.
>
> Does it reduce the quality of a JPEG?

No. As long as you *do not* save whatever image you've played with to the
same filename (overwriting the original) you haven't changed the original in
any way. You can play with it all you like, as often as you like.

N.
Related resources
Anonymous
June 19, 2005 2:15:19 AM

Archived from groups: rec.photo.digital (More info?)

Yes it does! Can you measure it? Only with a very large magnifying glass
unless you save to low quality.

"Mike Henley" <casioculture@gmail.com> wrote in message
news:1119132544.066322.297460@g44g2000cwa.googlegroups.com...
>
>
> Just editing it, *no* saving, just playing with the stuff in photoshop
> or paintshoppro.
>
> Does it reduce the quality of a JPEG?
>
Anonymous
June 19, 2005 2:15:20 AM

Archived from groups: rec.photo.digital (More info?)

"Pete D" <no@email.com> wrote in message
news:X11te.22033$F7.10005@news-server.bigpond.net.au...
> Yes it does!

No, it does not.


> Can you measure it? Only with a very large magnifying glass unless you
> save to low quality.

He specifically said, "*no* saving." No matter how large a magnifying glass
you use you won't see any change in the original image, which is what I
believe he's asking about. As long as he doesn't *save* the results of
playing with it to the same filename, he hasn't changed the original one
tiny bit.

N.


>
> "Mike Henley" <casioculture@gmail.com> wrote in message
> news:1119132544.066322.297460@g44g2000cwa.googlegroups.com...
>>
>>
>> Just editing it, *no* saving, just playing with the stuff in photoshop
>> or paintshoppro.
>>
>> Does it reduce the quality of a JPEG?
Anonymous
June 19, 2005 2:22:02 AM

Archived from groups: rec.photo.digital (More info?)

Mike Henley wrote:

>
> Just editing it, *no* saving, just playing with the stuff in photoshop
> or paintshoppro.
>
> Does it reduce the quality of a JPEG?


Hi Mike...

Maybe get just a little clearer for you if you consider
and accept that once it's loaded into photoshop or psp
it's not a jpeg. It's just something you're looking at.

Let's try this. You're hungry. Also tired, so you order
some let's say kfc delivered.

When it arrives, it matter not a whit whether it came by
bicycle, moped, car, semi-trailer, or fighter jet. It's
just food on your table, right? Play with it all you like.

The same applies to tiff or jpeg or bmp or whatever.

However, in our silly example, turns out you ordered
a little too much. So, you send a couple of pieces over
to me.

Now when you "deliver it", the manner does matter.
The bike will be too slow, it'll get cold. The fighter
jet will probably drop it from the sky - that'll wreck it:) 

So it won't arrive in the same condition as you sent it; but
once again after it gets here it's just food again, right?
And I can play around with it all I like, it won't change.

A little clearer?

Ken
Anonymous
June 19, 2005 2:28:52 AM

Archived from groups: rec.photo.digital (More info?)

Pete D wrote:

> Yes it does! Can you measure it? Only with a very large magnifying glass
> unless you save to low quality

Hi Pete.

With the greatest of respect, he can't edit a jpeg. None of
us can. (excluding hex editing and such wierdness)

PSP doesn't display a jpeg or tiff or any other format.
Once it's loaded it's a dib, which he can play with to his
hearts content.

Only if and when he decides to save it again can anything be
"damaged", and only if he decides to save the edited
pic in a lossy format.

Ken
Anonymous
June 19, 2005 3:49:04 AM

Archived from groups: rec.photo.digital (More info?)

In article <QaCdndglY9vUOinfRVn-hQ@comcast.com>,
"Nostrobino" <not@home.today> wrote:

>You can play with it all you like, as often as you like.

LOL, and who says digi - idiots are impotent twits.
--
Would thou choose to meet a rat eating dragon, or
a dragon, eating rat? The answer of: I am somewhere
in the middle.
Anonymous
June 19, 2005 4:06:42 AM

Archived from groups: rec.photo.digital (More info?)

"Ken Weitzel" <kweitzel@shaw.ca> wrote in message
news:Ee1te.1738035$6l.1703881@pd7tw2no...
>
>
> Pete D wrote:
>
>> Yes it does! Can you measure it? Only with a very large magnifying glass
>> unless you save to low quality
>
> Hi Pete.
>
> With the greatest of respect, he can't edit a jpeg. None of
> us can. (excluding hex editing and such wierdness)
>
> PSP doesn't display a jpeg or tiff or any other format.
> Once it's loaded it's a dib, which he can play with to his
> hearts content.
>
> Only if and when he decides to save it again can anything be
> "damaged", and only if he decides to save the edited
> pic in a lossy format.
>
> Ken

Yes of course, should read the post more carefully.
Anonymous
June 19, 2005 4:07:32 AM

Archived from groups: rec.photo.digital (More info?)

"Nostrobino" <not@home.today> wrote in message
news:k-Gdndkp25X8NCnfRVn-gg@comcast.com...
>
> "Pete D" <no@email.com> wrote in message
> news:X11te.22033$F7.10005@news-server.bigpond.net.au...
>> Yes it does!
>
> No, it does not.
>
>
>> Can you measure it? Only with a very large magnifying glass unless you
>> save to low quality.
>
> He specifically said, "*no* saving." No matter how large a magnifying
> glass you use you won't see any change in the original image, which is
> what I believe he's asking about. As long as he doesn't *save* the results
> of playing with it to the same filename, he hasn't changed the original
> one tiny bit.

You are correct of course, completely self evident.
Anonymous
June 19, 2005 8:44:00 AM

Archived from groups: rec.photo.digital (More info?)

On 18 Jun 2005 16:29:36 -0700, Mike Henley wrote:

> Thanks all. I guess good practice though would be to "save as" BEFORE
> one makes any changes or edits, just in case one accidentally saves
> them.

You could also copy JGPs and other image files to a PRACTICE
directory or folder. Then you not only don't have to concern
yourself with using "save" vs. "save as", but if you come back to it
several months later and want to practice again with a fresh copy,
due to the identical name, it'll be quite easy to locate the
original source. And if you lose track of where all of the
"practice" copies are stored (for example, to delete them) there are
many file utilities available that can search hard drives and show
all of your duplicate files. This isn't as easily done if the
copies have been given altered names using "save as".
Anonymous
June 19, 2005 11:19:39 AM

Archived from groups: rec.photo.digital (More info?)

no, it already has been done. Note, when you apply too much sharpening
to jpeg, you'll see it/ But there is no more damage than the one
already done.


On 18 Jun 2005 15:09:04 -0700, "Mike Henley" <casioculture@gmail.com>
wrote:

*
*
*Just editing it, *no* saving, just playing with the stuff in photoshop
*or paintshoppro.
*
*Does it reduce the quality of a JPEG?
Anonymous
June 19, 2005 11:30:33 AM

Archived from groups: rec.photo.digital (More info?)

Mike Henley wrote:

>
> Just editing it, *no* saving, just playing with the stuff in photoshop
> or paintshoppro.
>
> Does it reduce the quality of a JPEG?
>

The responses so far have considered only one aspect of photo
editing: the compression of the jpeg. This correct logic
goes as follows (since I don't think it was stated well,
so I'm trying another way). When you open the jpeg
and your software decompresses it and reads the image into
memory, the 8-bits/channel in memory are uncompressed.
All the processing you want will not change or add to
the jpeg compression. The loss in a jpeg is only
done when the file is saved. Once in memory, a jpeg
or 8-bit (uncompressed) tif are the same.

But compression losses are NOT the only losses in photo
editing. 8-bit tifs (or 8-but psd, or any other 8-bit
uncompressed file format) is a large limitation to
accuracy. Doing 8-bit math (only 256 levels) loses information
with each editing step!

Example. Say you have an image that needs to be brightened
a few percent. Say you have sky values in one area that
read a DN (data number) = 200 and nearby it is 199. Then
say you brighten by 7% using the levels tool. Your 2 pixels
then become: 200 * 1.07 = 214.00 and 212.93. But this is 8-bit
data, and depending on the software, 212.93 becomes either
213 or 212. You can have values that are near the mid-point,
like 163.47 versus 163.51 and nearest integer you change these
to 163 and 164, effectively adding noise.

So any integer data, when processed loses information.

I recently was preparing an image for a gallery showing,
doing some corrections and then applying an ICC profile
for Fuji Crystal Archive paper on a Lightjet.
I had saved the file, written the CD to take to the photo lab.
The image was a 24x30 inch print from a drum scan (high end stuff),
saved as tif (supposedly lossless).
I then read the CD to be sure the CD was readable, and noticed
the sky looked a little funny. I zoomed in and the sky was
noticeably posterized! I redid the processing by reading in the
8-bit/channel tif image, and my first step was converting to
16-bits/channel, doing the processing, and converting to 8-bits
just before saving.

Even 16-bits/channel loses information when processing due to
integer truncation. But the losses are about 256 times less
than with 8-bit processing, which is enough increased precision
not to encounter integer truncation problems that are visible in
images.

Roger
Anonymous
June 19, 2005 1:31:29 PM

Archived from groups: rec.photo.digital (More info?)

Roger N. Clark (change username to rnclark) wrote:
> The responses so far have considered only one aspect of photo
> editing: the compression of the jpeg. This correct logic
> goes as follows (since I don't think it was stated well,
> so I'm trying another way). When you open the jpeg
> and your software decompresses it and reads the image into
> memory, the 8-bits/channel in memory are uncompressed.
> All the processing you want will not change or add to
> the jpeg compression. The loss in a jpeg is only
> done when the file is saved. Once in memory, a jpeg
> or 8-bit (uncompressed) tif are the same.
>
> But compression losses are NOT the only losses in photo
> editing. 8-bit tifs (or 8-but psd, or any other 8-bit
> uncompressed file format) is a large limitation to
> accuracy. Doing 8-bit math (only 256 levels) loses information
> with each editing step!
>
> Example. Say you have an image that needs to be brightened
> a few percent. Say you have sky values in one area that
> read a DN (data number) = 200 and nearby it is 199. Then
> say you brighten by 7% using the levels tool. Your 2 pixels
> then become: 200 * 1.07 = 214.00 and 212.93. But this is 8-bit
> data, and depending on the software, 212.93 becomes either
> 213 or 212. You can have values that are near the mid-point,
> like 163.47 versus 163.51 and nearest integer you change these
> to 163 and 164, effectively adding noise.

This is why you really want to use adjustment layers when working with
your photos. If I use a adjustment layer to change the brightness and
then decide that it needs a little more adjustment later on the
underlying data is still the same, I am not added noise for every time
I make the adjustment.

Scott
Anonymous
June 19, 2005 4:03:53 PM

Archived from groups: rec.photo.digital (More info?)

"Roger N. Clark (change username to rnclark)" <username@qwest.net> wrote in
message news:42B57379.2010605@qwest.net...
> Mike Henley wrote:
>
>>
>> Just editing it, *no* saving, just playing with the stuff in photoshop
>> or paintshoppro. Does it reduce the quality of a JPEG?
>>
>
> The responses so far have considered only one aspect of photo
> editing: the compression of the jpeg. This correct logic
> goes as follows (since I don't think it was stated well,
> so I'm trying another way). When you open the jpeg
> and your software decompresses it and reads the image into
> memory, the 8-bits/channel in memory are uncompressed.
> All the processing you want will not change or add to
> the jpeg compression. The loss in a jpeg is only
> done when the file is saved.

And you could stop right there as far as the original question is concerned,
because that's all the OP wanted to know. :-)

But the remainder of your post is well worth reading anyway.



> Once in memory, a jpeg
> or 8-bit (uncompressed) tif are the same.
>
> But compression losses are NOT the only losses in photo
> editing. 8-bit tifs (or 8-but psd, or any other 8-bit
> uncompressed file format) is a large limitation to
> accuracy. Doing 8-bit math (only 256 levels) loses information
> with each editing step!
>
> Example. Say you have an image that needs to be brightened
> a few percent. Say you have sky values in one area that
> read a DN (data number) = 200 and nearby it is 199. Then
> say you brighten by 7% using the levels tool. Your 2 pixels
> then become: 200 * 1.07 = 214.00 and 212.93. But this is 8-bit
> data, and depending on the software, 212.93 becomes either
> 213 or 212. You can have values that are near the mid-point,
> like 163.47 versus 163.51 and nearest integer you change these
> to 163 and 164, effectively adding noise.
>
> So any integer data, when processed loses information.

Understood as far as that example goes, but what happens when an image is
altered by relocating pixels rather than changing their DNs? For example: if
I shoot something in vertical format, and then rotate it 90° in
viewing/editing software so that it's correct on the computer monitor, I
notice that the file size has been increased when I save it (this is JPEG of
course). I don't quite understand why. It is the same image, the same x
pixels by y pixels, the pixels just rotated to new locations. I should think
that with the DNs already adjusted to approximations as you've described
above, that work wouldn't have to be done over again and the image could
still compress to the same file size.

I'm not sure this happens with all software. But the fact that it does with
some, suggests that the image is undergoing further processing and therefore
further accuracy loss through approximations (even though file size
increases), isn't that so? Of course I don't make prints from such rotated
images, and on the monitor screen I can't see any difference in image
quality--but that's too coarse to make any such judgements anyway,
obviously.

Is this just because the viewing/editing software uses a different algorithm
for compression than the camera does? Or would the same software saving the
same image again and again through successive generations, without any
deliberate alteration being done to the image, produce a steadily increasing
and more noticeable loss in image quality?


>
> I recently was preparing an image for a gallery showing,
> doing some corrections and then applying an ICC profile
> for Fuji Crystal Archive paper on a Lightjet.
> I had saved the file, written the CD to take to the photo lab.
> The image was a 24x30 inch print from a drum scan (high end stuff),
> saved as tif (supposedly lossless).
> I then read the CD to be sure the CD was readable, and noticed
> the sky looked a little funny. I zoomed in and the sky was
> noticeably posterized! I redid the processing by reading in the
> 8-bit/channel tif image, and my first step was converting to
> 16-bits/channel, doing the processing, and converting to 8-bits
> just before saving.
>
> Even 16-bits/channel loses information when processing due to
> integer truncation. But the losses are about 256 times less
> than with 8-bit processing, which is enough increased precision
> not to encounter integer truncation problems that are visible in
> images.
>
> Roger

Most interesting.

N.
Anonymous
June 19, 2005 4:19:09 PM

Archived from groups: rec.photo.digital (More info?)

Scott W wrote:

> This is why you really want to use adjustment layers when working with
> your photos. If I use a adjustment layer to change the brightness and
> then decide that it needs a little more adjustment later on the
> underlying data is still the same, I am not added noise for every time
> I make the adjustment.

This is partly true. Each adjustment layer would do a separate
operation and thus would suffer from truncation losses.
But it is correct that if you are simply modifying an
existing adjustment layer, it would work on the original
data. A second problem with adjustment layers is that when you
change space, such as to lab mode, adjustment layers are discarded.
Another problem with adjustment layers is they increase your file
size. A 10 meg file with one adjustment layer becomes a 20
meg file, etc. Similarly, in memory requirements go up with
each layer. I work on 1 GByte large format files. A couple of
adjustment layers and the operating system croaks. Even a DSLR
raw output file could easily become hundreds of megabytes in size
with multiple adjustment layers.

16-bit processing avoids all these problems, and keeps file size
and memory size constant while maintaining sufficient precision,
even with multiple processing steps and file saves.

Roger
Anonymous
June 19, 2005 6:07:52 PM

Archived from groups: rec.photo.digital (More info?)

Roger N. Clark (change username to rnclark) wrote:
> Mike Henley wrote:
>
>>
>> Just editing it, *no* saving, just playing with the stuff in photoshop
>> or paintshoppro.
>> Does it reduce the quality of a JPEG?
>>
>
> The responses so far have considered only one aspect of photo
> editing: the compression of the jpeg. This correct logic
> goes as follows (since I don't think it was stated well,
> so I'm trying another way). When you open the jpeg
> and your software decompresses it and reads the image into
> memory, the 8-bits/channel in memory are uncompressed.
> All the processing you want will not change or add to
> the jpeg compression. The loss in a jpeg is only
> done when the file is saved. Once in memory, a jpeg
> or 8-bit (uncompressed) tif are the same.
>
> But compression losses are NOT the only losses in photo
> editing. 8-bit tifs (or 8-but psd, or any other 8-bit
> uncompressed file format) is a large limitation to
> accuracy. Doing 8-bit math (only 256 levels) loses information
> with each editing step!
>
> Example. Say you have an image that needs to be brightened
> a few percent. Say you have sky values in one area that
> read a DN (data number) = 200 and nearby it is 199. Then
> say you brighten by 7% using the levels tool. Your 2 pixels
> then become: 200 * 1.07 = 214.00 and 212.93. But this is 8-bit
> data, and depending on the software, 212.93 becomes either
> 213 or 212. You can have values that are near the mid-point,
> like 163.47 versus 163.51 and nearest integer you change these
> to 163 and 164, effectively adding noise.
>
> So any integer data, when processed loses information.
>
> I recently was preparing an image for a gallery showing,
> doing some corrections and then applying an ICC profile
> for Fuji Crystal Archive paper on a Lightjet.
> I had saved the file, written the CD to take to the photo lab.
> The image was a 24x30 inch print from a drum scan (high end stuff),
> saved as tif (supposedly lossless).
> I then read the CD to be sure the CD was readable, and noticed
> the sky looked a little funny. I zoomed in and the sky was
> noticeably posterized! I redid the processing by reading in the
> 8-bit/channel tif image, and my first step was converting to
> 16-bits/channel, doing the processing, and converting to 8-bits
> just before saving.
>
> Even 16-bits/channel loses information when processing due to
> integer truncation. But the losses are about 256 times less
> than with 8-bit processing, which is enough increased precision
> not to encounter integer truncation problems that are visible in
> images.

Hi Roger...

Indeed, the sole disadvantage of digital. I guess what
separates the film and digital afficioados. Even though
a single grain of film can be only on or off, never
halfway - so it's essentially digital too :) 

The point I was trying (apparently unsucessfully) to
emphasize was the difference between the picture itself
(good or bad) and the storage format.

To try once more - to blame lossless *storage* for pic
degradation would be imho like trying to blame your
car for failing to get your picnic cooler safely to the
park when *you* failed to provide a proper quantity of
ice :) 

Or - to blame your isp's hard drive for not making this
message as eloquent as I'd like it to be :) 

Still begs the question... in the photo described above,
why would you convert it back to 8 bit after processing?

Take care.

Ken

medium.
Anonymous
June 19, 2005 6:07:53 PM

Archived from groups: rec.photo.digital (More info?)

"Ken Weitzel" <kweitzel@shaw.ca> wrote in message
news:Y_ete.1742491$6l.1382153@pd7tw2no...
>
[ . . . ]
>
> The point I was trying (apparently unsucessfully) to
> emphasize was the difference between the picture itself
> (good or bad) and the storage format.

Ken, it's not so much that your attempt was unsuccessful (you did make an
interesting point) as that it just wasn't responsive to the original
poster's question, as far as I can see.

He only wanted to know whether opening an image file in some photo-editing
software and tinkering with it, BUT NOT SAVING the result, would reduce the
quality of the original. That's all.


>
> To try once more - to blame lossless *storage* for pic
> degradation would be imho like trying to blame your
> car for failing to get your picnic cooler safely to the
> park when *you* failed to provide a proper quantity of
> ice :) 

I don't know whether that's better than your KFC analogy, or not. :-)

But it's still getting into an area not relevant to the OP's question.

N.
Anonymous
June 19, 2005 6:28:37 PM

Archived from groups: rec.photo.digital (More info?)

Mike Henley <casioculture@gmail.com> wrote:
>
> Just editing it, *no* saving, just playing with the stuff
> in photoshop or paintshoppro.
>
> Does it reduce the quality of a JPEG?

Certainly some alternations reduce image quality.

Everybody else chose to answer your question, but I'm going to
write about how to edit JPEG with minimal damage.

First, there are certain lossless transformation, such as
90/180/270 rotation, flipping, optimizing, and comment cleanup.
Use Irfanview if you want to do just these things.

Second, you can minimize the damage of editing by saving at
the same JPEG parameters as before (quality, chroma subsampling).
Use the "jpegdump" program to determine JPEG parameters.

Especially with low-quality JPEG, the safest resampling ratio
is 50%. PaintShopPro has some artifact-smoothing tools, but
they don't work very well.
Anonymous
June 19, 2005 7:26:22 PM

Archived from groups: rec.photo.digital (More info?)

In article <Y_ete.1742491$6l.1382153@pd7tw2no>,
Ken Weitzel <kweitzel@shaw.ca> wrote:

>
> Still begs the question... in the photo described above,
> why would you convert it back to 8 bit after processing?

More importantly, if the file was originally 8 bits why would one think
that converting it to 16 bits is an advantage if captured at 8 bits.

& Most importantly; Why would one think it an advantage to convert to 16
from eight then back to 8,...something does not compute.

--
LF Website @ http://members.verizon.net/~gregoryblank

"To announce that there must be no criticism of the President,
or that we are to stand by the President, right or wrong,
is not only unpatriotic and servile, but is morally treasonable
to the American public."--Theodore Roosevelt, May 7, 1918
Anonymous
June 19, 2005 7:26:23 PM

Archived from groups: rec.photo.digital (More info?)

Gregory Blank wrote:

> In article <Y_ete.1742491$6l.1382153@pd7tw2no>,
> Ken Weitzel <kweitzel@shaw.ca> wrote:
>
>
>>Still begs the question... in the photo described above,
>>why would you convert it back to 8 bit after processing?

Because the Lightjet requires an 8-bit/channel file.
I produce files that are ready to send to the lightjet
with no processing by the photo lab. That reduces my cost.

> More importantly, if the file was originally 8 bits why would one think
> that converting it to 16 bits is an advantage if captured at 8 bits.
>
> & Most importantly; Why would one think it an advantage to convert to 16
> from eight then back to 8,...something does not compute.

When processing in 8 bit, you are
limited to 8-bit math. Because of integer truncation, after
processing one step, you effectively have reduced you data
to 7-bits precision. Do several editing steps and you quickly
find areas of the image posterized due to integer truncation
at EACH editing step. That includes levels, curves, unsharp
mask. If you only did those three steps, you are sub 7-bits
precision. Many images you don't NOTICE the degradation,
but it is there.

Here are things in my digital work flow:

read image.
convert to 16-bits/channel if not already 16-bit.
levels adjustment.
curves adjustment.
select areas and feather
curves adjustment.
if ( fast work ) then
convert to LAB.
unsharp mask on luminance channel.
convert back to RGB.
else
write 16-bit tif file.
open in ImagesPlus.
Richardson-Lucey image restoration.
read back into photoshop.
end if
check out of gamut.
convert to CMYK to fix details in out of gamut areas.
any final adjustments, like color balance,
saturation enhancements, levels.
write 16-bit/channel image.

read 16-bit image.
resize image for desired print size.
convert to LAB.
unsharp mask, radius ~ 0.9, amount ~ 66%, threshold ~ 3.
convert RGB.
convert to ICC printer profile.
convert to 8-bits/channel.
write file for printer.

If I stay in 8-bit the whole time, the data becomes more and more
compromised with 8-bit. Sometimes this is obvious, sometimes not.
But as I've learned, especially with photoshop CS because it has
more 16-bit tools, I now see more and more 8-bit artifacts
than I thought were there. Note: photoshop actually does 15-bit
math on 16-bit data to increase speed. ImagesPlus does
full 16-bit math.

I can see noticeable differences in image
quality between photoshop's 15-bit math and ImagesPlus 16-bit
math. Photoshop's math shows more posterization, even when working
on a 16-bit file. For example, see Figure 7 at:
http://www.clarkvision.com/imagedetail/image-restoratio...
and compare transition levels. For example, see the area to the
upper right of the fox's eye, where it looks smoother. The
ImagesPlus image has subtle gradations and the photoshop
processing looks more pasty. It is that pasty look that
is caused by imprecise computations, whether 8-bit, or
photoshop 15-bit processing.

Roger
Photos at: http://www.clarkvision.com
Anonymous
June 19, 2005 9:28:59 PM

Archived from groups: rec.photo.digital (More info?)

"Mike Henley" <casioculture@gmail.com> writes:

> Just editing it, *no* saving, just playing with the stuff in photoshop
> or paintshoppro.
>
> Does it reduce the quality of a JPEG?

It's the act of *saving* in jpeg format that loses information. You
can open a jpeg, edit forever, save in TIFF uncompressed, and you
haven't lost any quality compared to when you started (well, other
than any damage your editing did :-)). You can also save in TIFF
every 5 minutes, close your file, open the copy you saved, and
continue editing without damaging the information.

It's specifically the JPEG compression algorithm that loses
information. It was deliberately designed as a "lossy" compression
algorithm, but also designed to achieve very high levels of
compression with *little* visible degradation.

Basically, it's a really brilliant bit of algorithm design. It does
what it was intended to do, create remarkably small files that unpack
to images that look excellent to people, exceptionally well. One of
the limitations, though, is that repeatedly modifying, saving, and
reopening a file will cause damage to *accumulate* to the point where
it's a problem.

JPEG once at the beginning of the data stream may be acceptable
(high-quality JPEG camera originals), though some insist on shooting
RAW only. JPEG once at the end of the data stream, to create a file
for screen display, or even to create a file to send to your lab
(different compression levels appropriate in those two cases!), is
often acceptable. But JPEG is NOT a good *intermediate* format.
--
David Dyer-Bennet, <mailto:D d-b@dd-b.net>, <http://www.dd-b.net/dd-b/&gt;
RKBA: <http://noguns-nomoney.com/&gt; <http://www.dd-b.net/carry/&gt;
Pics: <http://dd-b.lighthunters.net/&gt; <http://www.dd-b.net/dd-b/SnapshotAlbum/&gt;
Dragaera/Steven Brust: <http://dragaera.info/&gt;
June 20, 2005 3:39:48 AM

Archived from groups: rec.photo.digital (More info?)

Mike Henley wrote:

>
>
> Just editing it, *no* saving, just playing with the stuff in photoshop
> or paintshoppro.
>
> Does it reduce the quality of a JPEG?


The other point no one mentioned, you can edit -and- save a jpeg with no
quality loss if you save it in a non-lossy format like tiff.
--

Stacey
!