Beware Jpeg file format

[Phil Fennessy]

There have been a number of threads of late concerned with the Jpeg in camera function and users deciding to save as jpeg and erase the original RAW files. This raises some VERY serious implications on the perception of the user and their understanding of Jpeg file format that relate not only to Kodak DCS cameras but also to all users of digital files.

There are some very good reasons to store images as jpeg, principally if you have to transmit over a phone line, however it is a very bad idea to use jpeg as the primary file format to store images. Don’t just take my word for this here’s a test for any jpeg software – take a tiff file (create one in Photoshop if you only have jpeg, lots of gradients, primary colours text etc). Open the image and save as a jpeg. Close both the original image and the jpeg. Open both and select all on the jpeg image copy it and paste over the tiff file (use the short cup keys and it should paste exactly over the tiff file). You should now have the two files as layers in the same image. Choose difference as the mode for the top image. If all is well you should now have a very black image. Choose flatten image and auto levels – what you are now looking at is the artefacts/data difference between the two images. EVERY time a jpeg image is opened and saved it degrades the image again.
So some simple rules

1. Always save the RAW file if your camera can (it’s the digital negative) jpeg is like a dupe and saved jpeg is a dupe of a dupe

2. Never sharpen if you want to jpeg a file –it increases both the file size (up to 1/3) and increases the jpeg artefacts – let the receiving end apply sharpening

3. If you work on a jpeg file in a workflow save it as a tiff or you will increase the problems every time you save.

4. If you REALLY REALLY have to store received images as jpegs, don't just hit save in photoshop otherwise you continue to apply maximum compression . Save as and choose jpeg -this means you can set the level of jpeg to a lesser compression

phil
--Phil Fennessy
Kodak Europe

 

[bartP]

Thank you for reminding me to that again!
I'll keep it in mind,

Bart

]

There have been a number of threads of late concerned with the Jpeg
in camera function and users deciding to save as jpeg and erase the
original RAW files. This raises some VERY serious implications on
the perception of the user and their understanding of Jpeg file
format that relate not only to Kodak DCS cameras but also to all
users of digital files.
There are some very good reasons to store images as jpeg,
principally if you have to transmit over a phone line, however it
is a very bad idea to use jpeg as the primary file format to store
images. Don’t just take my word for this here’s a test for any
jpeg software – take a tiff file (create one in Photoshop if you
only have jpeg, lots of gradients, primary colours text etc). Open
the image and save as a jpeg. Close both the original image and the
jpeg. Open both and select all on the jpeg image copy it and paste
over the tiff file (use the short cup keys and it should paste
exactly over the tiff file). You should now have the two files as
layers in the same image. Choose difference as the mode for the top
image. If all is well you should now have a very black image.
Choose flatten image and auto levels – what you are now looking at
is the artefacts/data difference between the two images. EVERY time
a jpeg image is opened and saved it degrades the image again.
So some simple rules
1. Always save the RAW file if your camera can (it’s the digital
negative) jpeg is like a dupe and saved jpeg is a dupe of a dupe
2. Never sharpen if you want to jpeg a file –it increases both the
file size (up to 1/3) and increases the jpeg artefacts – let the
receiving end apply sharpening
3. If you work on a jpeg file in a workflow save it as a tiff or
you will increase the problems every time you save.
4. If you REALLY REALLY have to store received images as jpegs,
don't just hit save in photoshop otherwise you continue to apply
maximum compression . Save as and choose jpeg -this means you can
set the level of jpeg to a lesser compression

phil

--
Phil Fennessy
Kodak Europe

 

[Terry Sessford]

Hmmmm ... a Kodak employee ... I know that you won't be in a position to answer this question, but I'll ask it all the same! There has been much talk of a joint Kodak/Olympus digital SLR with interchangeable lenses, but no one seems to have any recent reliable news on the current situation. Is such a camera really going to appear in the shops and, if so, have you any idea when this may happen? I would be very interested in any news of this exciting possibility. (You can always email me off-line if you don't feel able to disclose this information to the world -- grin!)

Terry.

There have been a number of threads of late concerned with the Jpeg
in camera function and users deciding to save as jpeg and erase the
original RAW files. This raises some VERY serious implications on
the perception of the user and their understanding of Jpeg file
format that relate not only to Kodak DCS cameras but also to all
users of digital files.
There are some very good reasons to store images as jpeg,
principally if you have to transmit over a phone line, however it
is a very bad idea to use jpeg as the primary file format to store
images. Don’t just take my word for this here’s a test for any
jpeg software – take a tiff file (create one in Photoshop if you
only have jpeg, lots of gradients, primary colours text etc). Open
the image and save as a jpeg. Close both the original image and the
jpeg. Open both and select all on the jpeg image copy it and paste
over the tiff file (use the short cup keys and it should paste
exactly over the tiff file). You should now have the two files as
layers in the same image. Choose difference as the mode for the top
image. If all is well you should now have a very black image.
Choose flatten image and auto levels – what you are now looking at
is the artefacts/data difference between the two images. EVERY time
a jpeg image is opened and saved it degrades the image again.
So some simple rules
1. Always save the RAW file if your camera can (it’s the digital
negative) jpeg is like a dupe and saved jpeg is a dupe of a dupe
2. Never sharpen if you want to jpeg a file –it increases both the
file size (up to 1/3) and increases the jpeg artefacts – let the
receiving end apply sharpening
3. If you work on a jpeg file in a workflow save it as a tiff or
you will increase the problems every time you save.
4. If you REALLY REALLY have to store received images as jpegs,
don't just hit save in photoshop otherwise you continue to apply
maximum compression . Save as and choose jpeg -this means you can
set the level of jpeg to a lesser compression

phil

--
Phil Fennessy
Kodak Europe

 

[Phil Fennessy]

Terry
you're assuming that someone tells the staff whats going on - Oh really!
phil
--Phil Fennessy

 

[Davy Balo]

Terry,

Good luck in your quest for "confidential" info from Kodak. (^. )

You should check out the new E-20 Olympus camera with the new CCD (^.^'), don't know for sure that you can change the lens.

http://cf.olympus-europa.com/consumer/digimg/intro.cfm?id=E-20P

http://www.olympusamerica.com/cpg_section/cpg_product.asp?p=16&bc=3&product=714

BTW Phil, I didn't know that you are Jay's evil European twin... lol

Davy

Hmmmm ... a Kodak employee ... I know that you won't be in a
position to answer this question, but I'll ask it all the same!
There has been much talk of a joint Kodak/Olympus digital SLR with
interchangeable lenses, but no one seems to have any recent
reliable news on the current situation. Is such a camera really
going to appear in the shops and, if so, have you any idea when
this may happen? I would be very interested in any news of this
exciting possibility. (You can always email me off-line if you
don't feel able to disclose this information to the world -- grin!)

Terry.

 

[Fred Stucker]

I've seen this kind of test before, and I object to the use of auto-level to expose "defects" in an image. If the largest variation between pixels on separate layers was only one part in 256, then using auto-level will still make the black image look like a snowstorm. The most realistic test is to compare JPEG and TIFF files at actual pixel size on a computer monitor. If you don't see a difference on a monitor, then you certainly won't see any when the image is printed at 200-300 pixels per inch.

You will have to look pretty closely to detect JPEG artifacts in images saved using Photoshop #11 or #12 compression. I also don't feel that zooming in to 400% is fair--such fine details get lost in print. In-camera sharpening doesn't increases JPEG artifacts, because JPEG compression is applied as a last step before the image is saved. Sharpening can introduce it's own artifacts (edge halos and increased aliasing and moiré) but this also happens in Photoshop. I estimate that my D1's Fine JPEG compression is about the same as Photoshop #10 compression--still very close to TIFF output.

I should point out that some cameras, such as my father-in-law's Kodak DC4800 use far more JPEG compression then a Pro SLR in high-quality mode. The 4800 files I've seen sometimes have so many artifacts that it looks like film grain.

I agree that photographers should save their original files, and Nikon NEF RAW files aren't too huge anyway. JPEG can give excellent results as long as minimal compression is applied, and there aren't multiple file saves made in JPEG format.

Fred

There have been a number of threads of late concerned with the Jpeg
in camera function and users deciding to save as jpeg and erase the
original RAW files. This raises some VERY serious implications on
the perception of the user and their understanding of Jpeg file
format that relate not only to Kodak DCS cameras but also to all
users of digital files.
There are some very good reasons to store images as jpeg,
principally if you have to transmit over a phone line, however it
is a very bad idea to use jpeg as the primary file format to store
images. Don’t just take my word for this here’s a test for any
jpeg software – take a tiff file (create one in Photoshop if you
only have jpeg, lots of gradients, primary colours text etc). Open
the image and save as a jpeg. Close both the original image and the
jpeg. Open both and select all on the jpeg image copy it and paste
over the tiff file (use the short cup keys and it should paste
exactly over the tiff file). You should now have the two files as
layers in the same image. Choose difference as the mode for the top
image. If all is well you should now have a very black image.
Choose flatten image and auto levels – what you are now looking at
is the artefacts/data difference between the two images. EVERY time
a jpeg image is opened and saved it degrades the image again.
So some simple rules
1. Always save the RAW file if your camera can (it’s the digital
negative) jpeg is like a dupe and saved jpeg is a dupe of a dupe
2. Never sharpen if you want to jpeg a file –it increases both the
file size (up to 1/3) and increases the jpeg artefacts – let the
receiving end apply sharpening
3. If you work on a jpeg file in a workflow save it as a tiff or
you will increase the problems every time you save.
4. If you REALLY REALLY have to store received images as jpegs,
don't just hit save in photoshop otherwise you continue to apply
maximum compression . Save as and choose jpeg -this means you can
set the level of jpeg to a lesser compression

phil

--
Phil Fennessy
Kodak Europe

 

[Ron Reznick]

Fred, I agree that JPG when using the absolute minimum of compression is a usable format. I of course save all RAW files for later reprocessing if necessary, but after doing extensive testing I have found that the lowest-compression JPG output from Capture is indistinguishable from TIF, and it is my default file-save mode. File size of these JPGs is generally double that of a JPG=12 save from Photoshop 6x, and occasionally exceeds the size of the RAW file if the level of small detail is high. I use these JPGs for creation of the web-images as well as the primary full-size images, and only use TIF if I am going to upsample or crop-and-upsample -- all color/level/curve correction and full-size USM are applied before saving (USM is not applied to TIFs that will be upsampled though).

As a general rule, reworking/resaving a JPG is not a good idea, as losses are cumulative, but with a very low-compression JPG that is to be resized down significantly, or that is to have some modifications done to it and then is resaved at compression=12 in Pshop 6x, the results are so benign as to be indistinguishable at 100% or in print.

Ron Reznick
http://digital-images.net
http://trapagon.com

I've seen this kind of test before, and I object to the use of
auto-level to expose "defects" in an image. If the largest
variation between pixels on separate layers was only one part in
256, then using auto-level will still make the black image look
like a snowstorm. The most realistic test is to compare JPEG and
TIFF files at actual pixel size on a computer monitor. If you
don't see a difference on a monitor, then you certainly won't see
any when the image is printed at 200-300 pixels per inch.

You will have to look pretty closely to detect JPEG artifacts in
images saved using Photoshop #11 or #12 compression. I also don't
feel that zooming in to 400% is fair--such fine details get lost in
print. In-camera sharpening doesn't increases JPEG artifacts,
because JPEG compression is applied as a last step before the image
is saved. Sharpening can introduce it's own artifacts (edge halos
and increased aliasing and moiré) but this also happens in
Photoshop. I estimate that my D1's Fine JPEG compression is about
the same as Photoshop #10 compression--still very close to TIFF
output.

I should point out that some cameras, such as my father-in-law's
Kodak DC4800 use far more JPEG compression then a Pro SLR in
high-quality mode. The 4800 files I've seen sometimes have so many
artifacts that it looks like film grain.

I agree that photographers should save their original files, and
Nikon NEF RAW files aren't too huge anyway. JPEG can give
excellent results as long as minimal compression is applied, and
there aren't multiple file saves made in JPEG format.

Fred

--Ron Reznick http://digital-images.nethttp://trapagon.com

 

[Mike Delaney]

Phil

When I shoot in the .JPG fine mode, I am in the habbit of transferring the contents of the CF card to CD without opening them, this, I believe gives me a base .jpg file, that is now safe from corruption, as, even though it is opened and re-opened from the CD, it is always a new file, this is not the case for the saved results of the session obviosly, but I think it does protect the "original". Am I wrong? Remembering that the images on the CD are never re-written to the CD?
Mike D

 

[Hans Kilian]

Hi Mike.

You are correct that you're saving the original image on your CD.

But you should be aware that your camera is performing some JPEG compression which degrades the image when compared to shooting raw (or uncompressed) images.

You get the best quality with raw images. The downside is that raw images take up a lot of space, so unless you have a large memory card in your camera you won't be able to shoot a lot of images. A raw image from a 4 megapixel camera will occupy 12 MB if the camera uses 8 bits for the red, green and blue colour information. And 18 MB if the camera uses 12 bits. That's a lot of space for a single image...

When you use your camera's 'fine' JPEG compression it probably doesn't compress the image much, but some degradation is taking place...

Hope that helps,
Hans

Phil
When I shoot in the .JPG fine mode, I am in the habbit of
transferring the contents of the CF card to CD without opening
them, this, I believe gives me a base .jpg file, that is now safe
from corruption, as, even though it is opened and re-opened from
the CD, it is always a new file, this is not the case for the saved
results of the session obviosly, but I think it does protect the
"original". Am I wrong? Remembering that the images on the CD are
never re-written to the CD?
Mike D

 

[Jay Kelbley]

Good luck in your quest for "confidential" info from Kodak. (^. )

That is typically why confidential information is marked as "confidential".

BTW Phil, I didn't know that you are Jay's evil European twin... lol

I very much resent the implication that I am not evil...

-Jay--Jay Kelbley-Eastman Kodak [email protected]

 

[Mike Delaney]

Hans,

I agree, there is some loss, but I have found that with the Nikon D series, the .jpg fine settings are very practical, especially for sports applications, which is why I got into the habbit of shooting in this mode with my original D-1, and have continued to do so with the D-1x, and while raw is certainly better, it is a bit of overkill for my current applications. Have a nice Holiday Season
Mike D

 

[Thomas41622]

Phil,

Here's what I am doing with my jpeg images...let me know if I am doing the right thing.

I shoot in jpeg on a Fuji S1. After loading in my computer, I copy the folder of files and only look at the copies. Once I decide what to print, I open the original file (which is backed up) and duplicate it, close the original file, and immediately save the copy file as a .psd format, then proceed to work (rotate, levels, color correct, clone, burn, dodge, etc.) on the image.

After printing, I save it as .psd in case I need to work on it again. I regularly back up my files.

I know some people work in Tiff and I am wondering what the difference is and should I do the same.

Any tips appreciated.

Thomas

--Thomas

 

[Steve Newbold]

As this was originally posted in the 'Pro' forum I'll reply with my 'pro' hat on

I don't understand how any professional photographer or digital artist can save their work as a JPG because it looks almost as good when printed. IMO there is no excuse for an image being 'almost' as good. The camera will let you save in raw format, photoshop lets you work in a raw, uncompressed format so why bother sacrificing image quality just to save a MB here and there.

It's almost the equivalent of an photographer who wants to print some of his work for a gallery etc. He can either get it Lambda printed which costs money, or he can print on his Epson printer which looks 'almost' as good. Proffesionals should never sacrifice quality as it is your work that represents you. If your amatuer or semi-pro then this might sound a little heavy but I'msure that most Pro's will agree with me.

Personally, I'm still gutted that we can't work in 16bit mode to the extent we can inb 8bit mode. That will certainly improve print quality and the amount we can alter images.

Steve Newbold

 

[Paul Rumohr]

I don't understand how any professional photographer or digital
artist can save their work as a JPG because it looks almost as good
when printed. IMO there is no excuse for an image being 'almost'
as good. The camera will let you save in raw format, photoshop
lets you work in a raw, uncompressed format so why bother
sacrificing image quality just to save a MB here and there.

When you're shooting volume like catalog, RAW is too slow. You look will look like a complete idiot while your camera is writing and you are trying to catch up with the models.

And when the art director is looking over your shoulder and you're trying to open 50 NEFs so he can look at them to approve the shot, you had better be an entertaining conversationalist for about 20 minutes.

There are plenty of damn good reasons to shoot JPG and in my opinion, good photography has NEVER been about the technical image quality as much as what the photography communicates.

I was blown away by the JPG quality when I first saw samples from the D1X. And then I started reading about well open the image this way in PS 6 and then that way and then do this and do that and do a "difference thingy" and see the artifacting, etc. etc.

I respect the technical prowess, but I think these guys have way too much post-production time.

I should also mention that a lot of pro photography doesn't end up in museums, but on things like Taco Bell napkins, Nabisco cereal boxes, and crappy newsprint catalogs (and BTW hasn't the Victoria Secret's catalog just taken a major dive in quality?).

Using JPGs in your workflow is FAST. When I work with JPGs I intentionally sacrifice quality.

But I DELIVER the JOB in TIME.

 

[Steve Newbold]

A fair point.

If time is the determining factor and JPG gives you an acceptable quality then perhaps you are right and there is an excuse.

I personally don't think that using JPG saves you that much time in a shoot. In a recent shoot I was present at (although not the photographer) we shot the new Range Rover (out 2002) for the brochure. We used a Leaf digital back for the whole shoot which some might say is over-kill for a simple brochure shot.

The art director approved the previews on screen as we shot, raw 14-bit HDR files were 'processed' in a few minutes on a decent G4, burnt on to a CD and the AD took them back with him.

The best bit is when the agency calls the photographer asking for the shot to be placed on a nation-wide billboard. This is when the quality pays off, although the D1x images saved as jpg's would have been fine for the brochure shots, they just wouldn't cut it for a billboard image. Think of all the £££ the photographer now makes for no extra effort and no extra time.

I don't understand how any professional photographer or digital
artist can save their work as a JPG because it looks almost as good
when printed. IMO there is no excuse for an image being 'almost'
as good. The camera will let you save in raw format, photoshop
lets you work in a raw, uncompressed format so why bother
sacrificing image quality just to save a MB here and there.

When you're shooting volume like catalog, RAW is too slow. You
look will look like a complete idiot while your camera is writing
and you are trying to catch up with the models.

And when the art director is looking over your shoulder and you're
trying to open 50 NEFs so he can look at them to approve the shot,
you had better be an entertaining conversationalist for about 20
minutes.

There are plenty of damn good reasons to shoot JPG and in my
opinion, good photography has NEVER been about the technical image
quality as much as what the photography communicates.

I was blown away by the JPG quality when I first saw samples from
the D1X. And then I started reading about well open the image this
way in PS 6 and then that way and then do this and do that and do a
"difference thingy" and see the artifacting, etc. etc.

I respect the technical prowess, but I think these guys have way
too much post-production time.

I should also mention that a lot of pro photography doesn't end up
in museums, but on things like Taco Bell napkins, Nabisco cereal
boxes, and crappy newsprint catalogs (and BTW hasn't the Victoria
Secret's catalog just taken a major dive in quality?).

Using JPGs in your workflow is FAST. When I work with JPGs I
intentionally sacrifice quality.

But I DELIVER the JOB in TIME.

 

[Terry Sessford]

Good luck in your quest for "confidential" info from Kodak. (^. )

That is typically why confidential information is marked as
"confidential".

Well, I knew that I wouldn't get an answer for the very reason that you state, but the temptation of raising the rumoured Kodak/Olympus digital SLR with a Kodak employee was just too great to resist. So Jay, do you think that the Kodak/Olympus camera will hit the shops before or after Canon's replacement for the EOS D30? :-D

Jay Kelbley
-Eastman Kodak Company
[email protected]

Terry.