JPEG quality settings often too high?

Class A · 07-31-2010, 06:43 PM - **1 Like**

Jeffrey Friedl has created an interesting page entitled "An Analysis of Lightroom JPEG Export Quality Settings".

His examples demonstrate how well JPEG compression works below 100% levels. Towards the 100% end of the quality scale, with many images there is a considerable increase in file size with very marginal to no increase in visual precision.

Although he discusses JPEG compression in the context of Lightroom JPEG export settings, the examples are very interesting on their own.

dragonfly · 07-31-2010, 08:58 PM

Thanks for posting that. I wish Adobe would have explained that in the export window.

imtheguy · 07-31-2010, 11:50 PM

I certainly can't tell the difference among JPGs saved at 92, 95 and 98 but always saved higher than needed hoping it would help if had to go back and edit/resave. Guess thats an experiment for another night. Two copies of one shot, saved at 90 and 98 quality. Then open and resave each at that quality as copy1, copy2, copy3 checking the size as an indication that they really are being compressed differently. Now do the 2 copy3 versions still look exactly alike as the originals did? We have experts that could answer that but its more instructive to do it myself.

Thanks for the article post!

rawr · 08-01-2010, 06:24 AM

JPG compression levels can make a big difference to file size and yet can have little apparent impact on the way a image looks? Hold the front page!

But seriously, duh. Anyone who has ever saved a JPG before in their whole life would know this already.

Class A · 08-01-2010, 06:36 AM - **1 Like**

Originally posted by rawr

Anyone who has ever saved a JPG before in their whole life would know this already.

No. Saving "one" JPG isn't enough to "know" this. You need at least two to be able to compare.

Note that I didn't provide a link to the article -- which many seem to find helpful -- for the breaking news, but because the author went through some length to actually illustrate the differences for the various LR JPEG quality settings.

Did you know that LR only supports 13 quality levels and just maps 0-100% to these 13 levels? I guess you didn't so the article could be useful even to those who are knowledgeable about JPEG compression in general.

hcc · 08-01-2010, 07:32 PM

I do not use Ligthroom,. But both PTlens and Noiseware software recommend explicitely 90% as default:
"At 90% no difference can be seen when the image is compared with the original and viewed at 400% magnification. At 80% and below image quality deterioriates. (default = 90)." (PTLens Help file)

Food for thoughts...

mysticcowboy · 08-07-2010, 05:55 PM

It's so true about JPEG quality settings and file size. If you want the best possible image for archiving, a 16 bit TIFF in Pro RGB, or at least Adobe RGB, is the way to go. The 8 bit JPEG file will never hold the information of 16 bit one. JPEGs for working files like those you send to a printer or backup online will be fine at 90%. For web display, 70% usually will equal monitor resolution.

When I used to teach Photoshop and web design classes, I had dozens of students, over dozens of classes, save their files at different settings. Nobody could consistently pick a 90% from a 100% JPEG file in a blind comparison test, with either 300 dpi prints or 100% monitor display. Forget Photoshop settings of 11 or 12.

imtheguy · 08-07-2010, 09:06 PM

Originally posted by mysticcowboy

Nobody could consistently pick a 90% from a 100% JPEG file in a blind comparison test, with either 300 dpi prints or 100% monitor display. Forget Photoshop settings of 11 or 12.

As your student, my question would have been does a 98% jpg that has to be reopened and saved twice more (I know, poor planning) degrade to a much lesser extent than one saved at 90% three times. Do i end up with obvious differences in quality between the two final editions? thanks.

mysticcowboy · 08-18-2010, 09:30 AM

Originally posted by imtheguy

As your student, my question would have been does a 98% jpg that has to be reopened and saved twice more (I know, poor planning) degrade to a much lesser extent than one saved at 90% three times. Do i end up with obvious differences in quality between the two final editions? thanks.

There will be some degradation in quality. Some people say that any degradation is too much. Some won't notice the difference. I would, but I look at a lot of images. If you don't then there's no problem. It also depends on the image. If there is a lot of very fine detail or very subtle gradients, then re-saving will be noticeable in prints. If you're only posting them online its less likely that you'd notice.

By using jpeg you have already degraded the image from what's possible.

Class A · 08-18-2010, 02:32 PM

Originally posted by mysticcowboy

There will be some degradation in quality. ...

The question was: "While higher quality settings do not produce visible differences for the first generation JPEG, are they perhaps beneficial for 2nd, 3rd generation versions of a JPEG?". Good question which you didn't address in your response.

Originally posted by mysticcowboy

By using jpeg you have already degraded the image from what's possible.

JPEGs allow for less latitude in post processing compared to RAW or 16-bit TIFF files, but I'd be surprised if you were able to visually distinguish prints made from high-quality JPEGs vs RAW files.

Why do you think it becomes next to impossible to distinguish the topmost quality settings for JPEGs? Because a certain amount of degradation cannot be removed, no matter how high the quality setting, or rather because at a certain point there is nothing to improve upon anymore? I don't know about the existing implementations but in theory the principle behind JPEG compression allows a 100% reconstruction of the image as long as you store enough data. Progressive levels of compression are achieved by increasingly throwing away data pertaining to the highest remaining frequencies in the image.

simico · 08-19-2010, 01:42 AM

Originally posted by imtheguy

As your student, my question would have been does a 98% jpg that has to be reopened and saved twice more (I know, poor planning) degrade to a much lesser extent than one saved at 90% three times. Do i end up with obvious differences in quality between the two final editions? thanks.

Why don't you try it yourself? I'm sure you have an image editor on your pc.

imtheguy · 08-19-2010, 07:22 AM

Originally posted by simico

Why don't you try it yourself? I'm sure you have an image editor on your pc.

Of course I have tried it

but if I had a good eye for this I wouldn't be using JPG in the first place, right? I am not even proficient enough in PP to make RAW conversions look any better than the JPG the k20d spits out on its own. Thus, vote of no confidence in my ability to analyze JPG differences.

JohnBee · 08-19-2010, 07:31 AM

Originally posted by Class A

Jeffrey Friedl has created an interesting page entitled "An Analysis of Lightroom JPEG Export Quality Settings".

His examples demonstrate how well JPEG compression works below 100% levels. Towards the 100% end of the quality scale, with many images there is a considerable increase in file size with very marginal to no increase in visual precision.

Although he discusses JPEG compression in the context of Lightroom JPEG export settings, the examples are very interesting on their own.

Wow great post!!!
I love the webpage design and functions too(very intuitive).
I don't know if the image samples are coming across intact in FF but I was still able to experience the details across the difference examples(quite fun).

A funny thing about image grain is where I've been working in RAW for so long that I've grown hyper sensitive to JPG's now

Though for straight out of camera experiences, I think JPG is an excellent solution as a small, reliable and accurate medium.

Anyways, I just want to say great post once again.
Thanks!

imtheguy · 08-19-2010, 08:02 AM

Originally posted by Class A

Jeffrey Friedl has created an interesting page entitled "An Analysis of Lightroom JPEG Export Quality Settings".

That is the article I ran across a while back that first got me thinking about using less than max JPG setting, but ONLY for snaps I had no expectation of ever editing again. Thus my question about multiple saves. Guess I can take a clue from J Friedl but instead of just a 1-12 group in PE8, make my own whereby the same shot saved at setting 9,10,11 and 12 is resaved (with a tiny change) at the same setting. Maybe successive saves should go up a step? This is the question I started with.

....so much to learn to catch up with the rest of you.

07-31-2010, 06:43 PM - 1 Like	#1
Class A Pentaxian Join Date: Aug 2008 Location: Wellington, New Zealand Posts: 11,251	JPEG quality settings often too high? Jeffrey Friedl has created an interesting page entitled "An Analysis of Lightroom JPEG Export Quality Settings". His examples demonstrate how well JPEG compression works below 100% levels. Towards the 100% end of the quality scale, with many images there is a considerable increase in file size with very marginal to no increase in visual precision. Although he discusses JPEG compression in the context of Lightroom JPEG export settings, the examples are very interesting on their own.

08-19-2010, 07:31 AM	#13
JohnBee Veteran Member Join Date: Sep 2008 Location: Newrfoundland Photos: Albums Posts: 4,667	Originally posted by Class A Jeffrey Friedl has created an interesting page entitled "An Analysis of Lightroom JPEG Export Quality Settings". His examples demonstrate how well JPEG compression works below 100% levels. Towards the 100% end of the quality scale, with many images there is a considerable increase in file size with very marginal to no increase in visual precision. Although he discusses JPEG compression in the context of Lightroom JPEG export settings, the examples are very interesting on their own. Wow great post!!! I love the webpage design and functions too(very intuitive). I don't know if the image samples are coming across intact in FF but I was still able to experience the details across the difference examples(quite fun). A funny thing about image grain is where I've been working in RAW for so long that I've grown hyper sensitive to JPG's now Though for straight out of camera experiences, I think JPG is an excellent solution as a small, reliable and accurate medium. Anyways, I just want to say great post once again. Thanks! Last edited by JohnBee; 08-19-2010 at 08:19 AM.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
K-x vs K-m JPEG quality	deadwolfbones	Pentax DSLR Discussion	8	11-21-2009 08:47 PM
RAW + JPEG with JPEG on One Star quality	laissezfaire	Pentax DSLR Discussion	58	12-10-2008 02:42 PM
Jpeg compression quality	Cambo	Pentax DSLR Discussion	4	06-01-2008 03:19 AM
JPEG gallery quality	Duncan J Murray	Site Suggestions and Help	1	04-27-2008 02:38 PM
K10d JPEG quality	rwa	Pentax DSLR Discussion	35	11-29-2007 09:34 AM

07-31-2010, 08:58 PM	#2
dragonfly Veteran Member Join Date: Jun 2009 Location: on the wall Photos: Gallery Posts: 715	Thanks for posting that. I wish Adobe would have explained that in the export window.

07-31-2010, 11:50 PM	#3
imtheguy Site Supporter Join Date: Mar 2009 Location: Virginia Beach Photos: Gallery \| Albums Posts: 2,950	I certainly can't tell the difference among JPGs saved at 92, 95 and 98 but always saved higher than needed hoping it would help if had to go back and edit/resave. Guess thats an experiment for another night. Two copies of one shot, saved at 90 and 98 quality. Then open and resave each at that quality as copy1, copy2, copy3 checking the size as an indication that they really are being compressed differently. Now do the 2 copy3 versions still look exactly alike as the originals did? We have experts that could answer that but its more instructive to do it myself. Thanks for the article post!

08-01-2010, 06:24 AM	#4
rawr Veteran Member Join Date: Jun 2009 Posts: 11,913	JPG compression levels can make a big difference to file size and yet can have little apparent impact on the way a image looks? Hold the front page! But seriously, duh. Anyone who has ever saved a JPG before in their whole life would know this already.

08-01-2010, 06:36 AM - 1 Like	#5
Class A Pentaxian Join Date: Aug 2008 Location: Wellington, New Zealand Posts: 11,251 Original Poster	Originally posted by rawr Anyone who has ever saved a JPG before in their whole life would know this already. No. Saving "one" JPG isn't enough to "know" this. You need at least two to be able to compare. Note that I didn't provide a link to the article -- which many seem to find helpful -- for the breaking news, but because the author went through some length to actually illustrate the differences for the various LR JPEG quality settings. Did you know that LR only supports 13 quality levels and just maps 0-100% to these 13 levels? I guess you didn't so the article could be useful even to those who are knowledgeable about JPEG compression in general.

08-01-2010, 07:32 PM	#6
hcc Pentaxian Join Date: Nov 2009 Location: Brisbane, Australia Photos: Gallery \| Albums Posts: 4,002	I do not use Ligthroom,. But both PTlens and Noiseware software recommend explicitely 90% as default: "At 90% no difference can be seen when the image is compared with the original and viewed at 400% magnification. At 80% and below image quality deterioriates. (default = 90)." (PTLens Help file) Food for thoughts...

08-07-2010, 05:55 PM	#7
mysticcowboy Veteran Member Join Date: Apr 2009 Location: port townsend, wa Photos: Albums Posts: 968	It's so true about JPEG quality settings and file size. If you want the best possible image for archiving, a 16 bit TIFF in Pro RGB, or at least Adobe RGB, is the way to go. The 8 bit JPEG file will never hold the information of 16 bit one. JPEGs for working files like those you send to a printer or backup online will be fine at 90%. For web display, 70% usually will equal monitor resolution. When I used to teach Photoshop and web design classes, I had dozens of students, over dozens of classes, save their files at different settings. Nobody could consistently pick a 90% from a 100% JPEG file in a blind comparison test, with either 300 dpi prints or 100% monitor display. Forget Photoshop settings of 11 or 12.

08-07-2010, 09:06 PM	#8
imtheguy Site Supporter Join Date: Mar 2009 Location: Virginia Beach Photos: Gallery \| Albums Posts: 2,950	Originally posted by mysticcowboy Nobody could consistently pick a 90% from a 100% JPEG file in a blind comparison test, with either 300 dpi prints or 100% monitor display. Forget Photoshop settings of 11 or 12. As your student, my question would have been does a 98% jpg that has to be reopened and saved twice more (I know, poor planning) degrade to a much lesser extent than one saved at 90% three times. Do i end up with obvious differences in quality between the two final editions? thanks.

08-18-2010, 09:30 AM	#9
mysticcowboy Veteran Member Join Date: Apr 2009 Location: port townsend, wa Photos: Albums Posts: 968	Originally posted by imtheguy As your student, my question would have been does a 98% jpg that has to be reopened and saved twice more (I know, poor planning) degrade to a much lesser extent than one saved at 90% three times. Do i end up with obvious differences in quality between the two final editions? thanks. There will be some degradation in quality. Some people say that any degradation is too much. Some won't notice the difference. I would, but I look at a lot of images. If you don't then there's no problem. It also depends on the image. If there is a lot of very fine detail or very subtle gradients, then re-saving will be noticeable in prints. If you're only posting them online its less likely that you'd notice. By using jpeg you have already degraded the image from what's possible.

08-18-2010, 02:32 PM	#10
Class A Pentaxian Join Date: Aug 2008 Location: Wellington, New Zealand Posts: 11,251 Original Poster	Originally posted by mysticcowboy There will be some degradation in quality. ... The question was: "While higher quality settings do not produce visible differences for the first generation JPEG, are they perhaps beneficial for 2nd, 3rd generation versions of a JPEG?". Good question which you didn't address in your response. Originally posted by mysticcowboy By using jpeg you have already degraded the image from what's possible. JPEGs allow for less latitude in post processing compared to RAW or 16-bit TIFF files, but I'd be surprised if you were able to visually distinguish prints made from high-quality JPEGs vs RAW files. Why do you think it becomes next to impossible to distinguish the topmost quality settings for JPEGs? Because a certain amount of degradation cannot be removed, no matter how high the quality setting, or rather because at a certain point there is nothing to improve upon anymore? I don't know about the existing implementations but in theory the principle behind JPEG compression allows a 100% reconstruction of the image as long as you store enough data. Progressive levels of compression are achieved by increasingly throwing away data pertaining to the highest remaining frequencies in the image.

08-19-2010, 01:42 AM	#11
simico Veteran Member Join Date: Jan 2009 Location: Budapest Posts: 821	Originally posted by imtheguy As your student, my question would have been does a 98% jpg that has to be reopened and saved twice more (I know, poor planning) degrade to a much lesser extent than one saved at 90% three times. Do i end up with obvious differences in quality between the two final editions? thanks. Why don't you try it yourself? I'm sure you have an image editor on your pc.

08-19-2010, 07:22 AM	#12
imtheguy Site Supporter Join Date: Mar 2009 Location: Virginia Beach Photos: Gallery \| Albums Posts: 2,950	Originally posted by simico Why don't you try it yourself? I'm sure you have an image editor on your pc. Of course I have tried it but if I had a good eye for this I wouldn't be using JPG in the first place, right? I am not even proficient enough in PP to make RAW conversions look any better than the JPG the k20d spits out on its own. Thus, vote of no confidence in my ability to analyze JPG differences.

08-19-2010, 08:02 AM	#14
imtheguy Site Supporter Join Date: Mar 2009 Location: Virginia Beach Photos: Gallery \| Albums Posts: 2,950	Originally posted by Class A Jeffrey Friedl has created an interesting page entitled "An Analysis of Lightroom JPEG Export Quality Settings". That is the article I ran across a while back that first got me thinking about using less than max JPG setting, but ONLY for snaps I had no expectation of ever editing again. Thus my question about multiple saves. Guess I can take a clue from J Friedl but instead of just a 1-12 group in PE8, make my own whereby the same shot saved at setting 9,10,11 and 12 is resaved (with a tiny change) at the same setting. Maybe successive saves should go up a step? This is the question I started with. ....so much to learn to catch up with the rest of you.