The significance of pixel peeping

[Ray]

I like my prints big. Paintings are big. Why not photos?

For this reason, I have an Epson 24" wide 7600. It's an ideal size for photos to be hung in one's living room.

But such large images need contrast enhancement and extra sharpening to look good.

The following image was converted in RSP with a moderate amount of 'detail enhancement' and a 'negative' sharpening. It was cropped to a more panoramic aspect ratio, saved in the STN Genuine Fractals format, then resized to 23.5"x40", resulting in a file size of 340mb at 360dpi.

[attachment=1559:attachment]

Applying Focus Magic to this large file size resulted in further sharpening. Zooming in 100% on the 340mb image, I noticed the following sharpening halos.

[attachment=1560:attachment]

Dear me! This is not good. I wonder if this will be noticeable on a large print (23.5"x40").

It's not noticeable in the slightest degree. Even with a jewelers' headband magnifying glass in conjunction with my most powerful reading glasses, these halos are 'hardly' noticeable.

Phew! That's a relief!  

[bjanes]

I like my prints big. Paintings are big. Why not photos?

For this reason, I have an Epson 24" wide 7600. It's an ideal size for photos to be hung in one's living room.

But such large images need contrast enhancement and extra sharpening to look good.

The following image was converted in RSP with a moderate amount of 'detail enhancement' and a 'negative' sharpening. It was cropped to a more panoramic aspect ratio, saved in the STN Genuine Fractals format, then resized to 23.5"x40", resulting in a file size of 340mb at 360dpi.

Applying Focus Magic to this large file size resulted in further sharpening. Zooming in 100% on the 340mb image, I noticed the following sharpening halos.

Dear me! This is not good. I wonder if this will be noticeable on a large print (23.5"x40").

It's not noticeable in the slightest degree. Even with a jewelers' headband magnifying glass in conjunction with my most powerful reading glasses, these halos are 'hardly' noticeable.

Phew! That's a relief! 
[a href=\"index.php?act=findpost&pid=95914\"][{POST_SNAPBACK}][/a]

Ray,

Your observations are consistent with what Bruce Fraser noted in his Real World Sharpening book: sharpening for the best print appearance can look scary on screen. I remember the discussions we had previously on pixel peeping and your observations reinforce my opinion that pixel peeping can be counterproductive if it does not correlate with what is seen on the final product, the print (in most cases).

A very nice shot, BTW.

Bill

[Brian Gilkes]

This has been said before, but here it is again for new-comers.
You cannot softproof sharpening. An approximation is observation at around 30% (NOT 100%)\
The only accurate proof for sharpening is a hard proof.
Cheers
Brian
www.pharoseditions.com.au

[gdeliz]

I like my prints big. Paintings are big. Why not photos?

For this reason, I have an Epson 24" wide 7600. It's an ideal size for photos to be hung in one's living room.

But such large images need contrast enhancement and extra sharpening to look good.

The following image was converted in RSP with a moderate amount of 'detail enhancement' and a 'negative' sharpening. It was cropped to a more panoramic aspect ratio, saved in the STN Genuine Fractals format, then resized to 23.5"x40", resulting in a file size of 340mb at 360dpi.

[attachment=1559:attachment]

Applying Focus Magic to this large file size resulted in further sharpening. Zooming in 100% on the 340mb image, I noticed the following sharpening halos.

[attachment=1560:attachment]

Dear me! This is not good. I wonder if this will be noticeable on a large print (23.5"x40").

It's not noticeable in the slightest degree. Even with a jewelers' headband magnifying glass in conjunction with my most powerful reading glasses, these halos are 'hardly' noticeable.

Phew! That's a relief! 
[a href=\"index.php?act=findpost&pid=95914\"][{POST_SNAPBACK}][/a]

I apply sharpening to a duplicate layer and then use the CS2 blend-if controls to suppress light and dark halos. A standard setting when I capture sharpen with Focus Magic is 16-36 on the left hand sliders and about 205-235 on the right. This does a good job of suppressing halos and allows further rounds of sharpening without building up too many artifacts.

George Deliz

[feppe]

I apply sharpening to a duplicate layer and then use the CS2 blend-if controls to suppress light and dark halos. A standard setting when I capture sharpen with Focus Magic is 16-36 on the left hand sliders and about 205-235 on the right. This does a good job of suppressing halos and allows further rounds of sharpening without building up too many artifacts.

George Deliz
[a href=\"index.php?act=findpost&pid=95964\"][{POST_SNAPBACK}][/a]

According to PhotoKit Sharpener documentation - and I'm sure Fraser's book on sharpening - these halos are often desirable, as they increase the perceived sharpness without looking like halos _in print_. The point of the top post is that he tested this and came to the same conclusion: halos are often (usually?) required for top-notch results, and that pixel-peeping should not be used in lieu of looking at the final print.

[mcanyes]

Here is something that I don't see much, but it has been very helpful to me. Establish a fixed method of viewing your sharpening results. I check the sharpening at 100% and at 50%. Checking at odd percentages doesn't work well because Photoshop displays the best detail at 25% 50% 75% 100% etc.

At 100% the image should look over sharpened. At 50% it should, to borrow Bruce Frasier's term, look crunchy (slightly oversharpened).

I agree that you have to print the image to really see the sharpening, but you can develop your eye so that you can make a pretty good guess if the image is ready for a test print.

[Ray]

Here is something that I don't see much, but it has been very helpful to me. Establish a fixed method of viewing your sharpening results. I check the sharpening at 100% and at 50%. Checking at odd percentages doesn't work well because Photoshop displays the best detail at 25% 50% 75% 100% etc.

At 100% the image should look over sharpened. At 50% it should, to borrow Bruce Frasier's term, look crunchy (slightly oversharpened).

I agree that you have to print the image to really see the sharpening, but you can develop your eye so that you can make a pretty good guess if the image is ready for a test print.
[a href=\"index.php?act=findpost&pid=95990\"][{POST_SNAPBACK}][/a]

There seems to me another problem in assessing the right degree of sharpening and contrast for large prints. Large prints seem to work best when the rendition is bold, punchy and contrasty. I find myself using the unsharp mask 'local contrast enhancement' technique much more if the print is intended to be large. This not only exaggerates the halos, but renders a broad, dark edge to to image parts set against a light background, such as foliage against sky.

Any simple technique of avoiding this effect? My photoshop skills are not in proportion to the number of postings I have on this forum   .

[Tim Gray]

Use the blend-if dialog box to constrain the sharpening towards the mid tones (or less contrast adjustment on the bright side of the edge).  There's an explanation of this technique here:

http://ronbigelow.com/articles/sharpen4/sharpen4.htm

also - it helps to use a mask to take the sky out of the equation for local contrast enhancement.

[Ray]

Use the blend-if dialog box to constrain the sharpening towards the mid tones (or less contrast adjustment on the bright side of the edge).  There's an explanation of this technique here:

http://ronbigelow.com/articles/sharpen4/sharpen4.htm

also - it helps to use a mask to take the sky out of the equation for local contrast enhancement.
[a href=\"index.php?act=findpost&pid=96160\"][{POST_SNAPBACK}][/a]

Thanks for that link, Tim. There's a lot to keep me occupied there. I'm particularly impressed so far with the effects of a duplicate layer in luminosity blend mode and 50% opacity. It's simple and very effective in reducing the halo and color fringe effects.