Difference in apparant sharpness when JPEGs are viewed in Photoshop and in a Browser

Lester Wareham wrote in post #18764826
Yes the second image is not heavily cropped so starts from a better place, seemed to be the new workflow partly compensated for difference in sharpness, but it is hard to be sure. I am wary of overcooking.

Second dog looks better and not overcooked. I used LR only on this and I use Safari. I have noticed on other sites files don't look as good.

Lester Wareham wrote in post #18764798
So looking at remedial action I found increasing JPEG output sharpening did not really help and to my looked worse. What might look better is some additional sharpening prior to down sampling and the existing output sharpening.

Standard first, followed by additional pre-DS sampling.

What do people think, worth adopting?

Hosted photo: posted by Lester Wareham in
./showthread.php?p=187​64798&i=i57334943
forum: RAW, Post Processing & Printing

Hosted photo: posted by Lester Wareham in
./showthread.php?p=187​64798&i=i189950497
forum: RAW, Post Processing & Printing

#2 is slightly sharper but much less sharp than many bird shots on POTN. The same goes for the wild-dog pair. I'm viewing on a good screen (Retina) with Firefox.

I looked at these earlier, quite a bit of compression artifacts around the neck and shoulder in the original online sample linked to on the first page. Have you posted your compression settings and how you are saving them?

Sorry if I missed it, kinda being pulled different directions today, as usual.

Honestly, I would pick a better image to begin to gauge this problem. Something without all the white area and super high contrast edges.

OhLook wrote in post #18764865
#2 is slightly sharper but much less sharp than many bird shots on POTN. The same goes for the wild-dog pair. I'm viewing on a good screen (Retina) with Firefox.

I though the wild dogs were OK, if I look in on the originals all the hair detail is there, seems visible in the JPEG.

Here are a couple of 100% crops of the sharpened TIFF secondary master.

The sharpening level seems OK here to me, what do you think?

My inflight shots are often not as good as they should be, perhaps down to skill levels, others with identical kit are much better.
100% shots of the sharpened TIFF of the Red Kite and another (static bird shot), The second seems ok to me.

Left Handed Brisket wrote in post #18764870
I looked at these earlier, quite a bit of compression artifacts around the neck and shoulder in the original online sample linked to on the first page. Have you posted your compression settings and how you are saving them?

Sorry if I missed it, kinda being pulled different directions today, as usual.

Honestly, I would pick a better image to begin to gauge this problem. Something without all the white area and super high contrast edges.

Yes the photoshop setting a level 6 baseline, a compromise with quality and memory usage.

Lester Wareham wrote in post #18764877
Yes the photoshop setting a level 6 baseline, a compromise with quality and memory usage.

So, Save as > JPG?

Yes.

Lester Wareham wrote in post #18764873
I though the wild dogs were OK, if I look in on the originals all the hair detail is there, seems visible in the JPEG.

Here are a couple of 100% crops of the sharpened TIFF secondary master.

The sharpening level seems OK here to me, what do you think?

Hosted photo: posted by Lester Wareham in
./showthread.php?p=187​64873&i=i106947388
forum: RAW, Post Processing & Printing

Hosted photo: posted by Lester Wareham in
./showthread.php?p=187​64873&i=i155523628
forum: RAW, Post Processing & Printing

When I make everything on the screen 5 or 6 steps smaller, the detail in those attachments is great, showing that they have the wherewithal to yield sharp final versions–but reducing digitally doesn't have the same effect. I asked about softness in this thread and got lots of information, some of which I understood.

If anyone is using PS the PixelGenius group made PK Sharpener free. It used to be $100. Similar export process as LR (they designed it) which takes the guesswork out of sharpening. Of course you can tweak it after a bit for screen. Not having to guess for export is one main reason I stuck with LR.

http://www.pixelgenius​.com

Lester Wareham wrote in post #18764915
Yes.

I would seriously, seriously suggest never using that method unless you are just looking for a quick, no fuss export. Save for Web and Devices is a much more powerful tool.

To start, choosing an image file size that you think is reasonable for the number of images on that page and that allows a reasonable page download speed for your viewers is a must. Do some testing on your own, but I bet you see that even half meg (500KB) images load very quickly. If I remember the file you posted was about 200K, the file digital paradise posted was 550K, why throw away all that detail if it doesnt really impact most viewer? You can either get close to a preferred file size with the first screen, or you can choose the dropdown menu that allows you to optimize for file size you have chosen ... Ps will choose the settings for you.

You also have the ability to preview before and after and even use 4-up to see the effect of different settings on compression. Some images will compress much more easily than others. Always choose baseline optimized.

To back up a little, uploading files to the POTN server can put them through another compression process. It makes it really hard for us to know if we are seeing the same thing you see with the original files. If you can upload them to dropbox we can see the original unaltered files.

I don't see a difference between my desktop or uploading here otherwise I wouldn't use it.

Left Handed Brisket wrote in post #18764947
I would seriously, seriously suggest never using that method unless you are just looking for a quick, no fuss export. Save for Web and Devices is a much more powerful tool.

To start, choosing an image file size that you think is reasonable for the number of images on that page and that allows a reasonable page download speed for your viewers is a must. Do some testing on your own, but I bet you see that even half meg (500KB) images load very quickly. If I remember the file you posted was about 200K, the file digital paradise posted was 550K, why throw away all that detail if it doesnt really impact most viewer? You can either get close to a preferred file size with the first screen, or you can choose the dropdown menu that allows you to optimize for file size you have chosen ... Ps will choose the settings for you.

You also have the ability to preview before and after and even use 4-up to see the effect of different settings on compression. Some images will compress much more easily than others. Always choose baseline optimized.

To back up a little, uploading files to the POTN server can put them through another compression process. It makes it really hard for us to know if we are seeing the same thing you see with the original files. If you can upload them to dropbox we can see the original unaltered files.

I did use that when I first used PS, but my current process just runs as an action, the save for web is interactive AFAIK. Also the resampling algorithm is I believe the same. The advantage in save for web is trading off compression level and memory usage I believe.

But the problem is not the quality of the files but how different applications present them.

I do wonder if it is photoshop that is out of step here.

OhLook wrote in post #18764932
When I make everything on the screen 5 or 6 steps smaller, the detail in those attachments is great, showing that they have the wherewithal to yield sharp final versions–but reducing digitally doesn't have the same effect. I asked about softness in this thread and got lots of information, some of which I understood.

I’ll read through your thread Oh Look, but my understanding of what you discribe is the unavoidable loss of micro contrast when downsampling an image.

When I last looked into this the bicubic sharper method was ment to be the best normally available. (There is probably a mathematically optical method that might be better I don’t know; typical those sort of algorithms are not practical).

@LHB

So as I as JPEG save method is "another issue", but to just close the look I saved both these test files using SFW in two versions, one quality 60 and the other with the quality set to give about the same file size as my standard workflow.

These don't seem to be any better or worse than the standard workflow to my eye.