Soft photos = trash?
/forum/topic/1071354/0



Robin Usagani
Registered: Oct 26, 2010
Total Posts: 2845
Country: United States

I just had this photo session. http://blog.usagani.com/2011/12/bridget-and-rob-winter-family-photos-at.html
I missed focus because I wasnt on the right focusing mode. They suddenly ran toward me. I was going to delete it but I just cant. I love the photo too much. So does the content of this photo trumps the softness?

Here is a link to high resolution so you can see how soft it is.
http://usagani.com/img/s9/v14/p104502094.jpg

I really feel that even if this was printed on a large media, it is still a strong piece. What do you guys think?





AuntiPode
Registered: Aug 05, 2008
Total Posts: 6776
Country: New Zealand

It tells a great story and serves as a sweet memento. Not every image must be an award winner to be worthwhile.



RustyBug
Registered: Feb 02, 2009
Total Posts: 12942
Country: United States

Nice capture ... definitely not trash.

Threw some additional blur at the background and put a canvas texture to it. A little levels/curves adjustment.



AuntiPode
Registered: Aug 05, 2008
Total Posts: 6776
Country: New Zealand

Here's a version with USM at 16. 60, 0 applied to the whole image to bump mid-tone contrast and some smart sharpen applied to just the subject:



Robin Usagani
Registered: Oct 26, 2010
Total Posts: 2845
Country: United States

Hey.. nice edits all. I included it to what I delivered. This is the only one that is soft.



sbeme
Registered: Dec 23, 2003
Total Posts: 17380
Country: United States

Yeah, nice edits.
I like the canvas idea.
And the contrast/sharpness boost per Karen.

Back to the original: totally cute expressions, well-framed, nice shot low to the ground, excellent choice on the shallow DOF and BW.

Scott



cgardner
Registered: Nov 18, 2002
Total Posts: 9376
Country: United States

It's a great candid capture.. What I did below is to dupe the background layer, sharpen the copy, then selectively blend it it only on the foreground subject...







Perceptually what creates the emotional reaction in a portrait are the facial clues around the eyes and mouth. When editing portraits I use an action which creates duplicate surface blur, high pass sharpen, and USM layers with masks. On studio portrait I'll blur everything except the mouth and eyes, then the high pass mask in the areas where I want to add back texture and detail. Then I'll open the USM layer mask in any areas like the eye and lip catchlights where I want a harder more specular appearance. I used the same action here, but since the foreground was already soft from being OOF I didn't have to blur it first



Robin Usagani
Registered: Oct 26, 2010
Total Posts: 2845
Country: United States

nice edit cgardner. I almost never sharpen my photos.. maybe I should. I usually just trash them if it is not sharp SOOC.



cgardner
Registered: Nov 18, 2002
Total Posts: 9376
Country: United States

Given the nature of the RGB matrix on the sensor and how the camera interprets it into RGB pixels on a monitor some amount of USM is needed on any digital image simply to make the reproduction on screen and print match the impression of the scene content seen by eye.

The first thing to fall victim to the image processing are the specular highlights. This is seen in photos of animals with fur and feathers which get their 3D texture visually from the specular highlights of direct sunlight or non-diffused artificial reflecting off the tiny flat facets of the hairs and feathers. An OOC shot will render them flat and matted. USM will restore them to something closer to what is seen in person. Often in shots of animals or water sparkling on a river scene I will dupe the photo and apply overall sharpening to the base layer, then oversharpen the top layer, blending it the additional "sparkle" the USM creates selectively where I want to emphasize 3D texture.

Post what you consider a sharp OOC shot and I'll edit it to show you the difference with the judicious application of USM.



RustyBug
Registered: Feb 02, 2009
Total Posts: 12942
Country: United States

+1 @ ooc can only achieve very good/mediocrity at it's best.

The camera applies a "global" sharpening to compensate for the "global" blurring that the AA filter imparts to the image. Additionally, the camera doesn't change the TYPE of sharpening being applied (to match the scene) but only the AMOUNT of sharpening. Anything OOC will NEVER be at its best that it is capable of. It might be very good, really good, pretty good, good enough for average use, etc. ... but NEVER it's best that it is capable of.

Camera mfr's really never bother to explain that to consumers ... leaving us poor souls to chase our tails and scratch our heads why we can't get an "awesome shot" like others do ooc (with the faulty assumption that theirs were truly sooc also) ... it isn't the camera or the photographer that's the problem ... it's the fact that they failed to to tell you that there is another step involved in the process.

SOOC provides this post-capture step using the "global" processing of the tiny computer in the camera rather than the powerful & selective processing that YOU can apply outside the camera. SOOC is more convenient by far, but it comes at a price. For some, that is a fair price, for others ... not so much.

For me, the camera only produces a "negative" ... then it's simply a matter of do you take the "negative" to the "one hour lab" at the drugstore (akin to sooc) ... or do you take that same "negative" to a professional lab/finisher/printer (akin to PP). The images that come from the pro lab and the drugstore may come from the same "negative" ... but the final product will be vastly different. One might be "Oh, that's nice", while the other is transformed into a "WOW !!! That's fabulous". They were both YOU'RE CAPTURE ... but significanltly different outcomes, due to using significantly different levels of finishing.

So it is with the convenience of the "one hour lab" and "sooc" ... it can be quite nice and certainly more than "good enough" for many applications, but it will never be it's best ... and "throw away's" are much more salvagable (with limits) than one might otherwise first consider. Lots of people (you or others) can provide/improve/alter/correct finishing ... but no one can replicate the capture of a moment such as you did here ... which btw is priceless.

As always S&P to taste.

My .02 ... HTH



Robin Usagani
Registered: Oct 26, 2010
Total Posts: 2845
Country: United States

Ok, here is one that I consider sharp. BTW, what is USM?






Here is high resolution link

http://usagani.com/img/s11/v30/p906532958.jpg

cgardner wrote:
Given the nature of the RGB matrix on the sensor and how the camera interprets it into RGB pixels on a monitor some amount of USM is needed on any digital image simply to make the reproduction on screen and print match the impression of the scene content seen by eye.

The first thing to fall victim to the image processing are the specular highlights. This is seen in photos of animals with fur and feathers which get their 3D texture visually from the specular highlights of direct sunlight or non-diffused artificial reflecting off the tiny flat facets of the hairs and feathers. An OOC shot will render them flat and matted. USM will restore them to something closer to what is seen in person. Often in shots of animals or water sparkling on a river scene I will dupe the photo and apply overall sharpening to the base layer, then oversharpen the top layer, blending it the additional "sparkle" the USM creates selectively where I want to emphasize 3D texture.

Post what you consider a sharp OOC shot and I'll edit it to show you the difference with the judicious application of USM.




RustyBug
Registered: Feb 02, 2009
Total Posts: 12942
Country: United States

USM = UnSharpen Mask
Yes, it sounds like an oxymoron ... but it is used for sharpening.

Here's a screen shot of 100% crop from your large file. The inset box is showing the lip/teeth before USM was applied.

The Amount slider tells how strong to apply the USM
The Radius slider tells how far to consider it's neighboring pixels
The Threshold tells how much difference there must be between the neighboring pixels for it to apply the amount.

In the case of your image ... i.e. skin (or skies) I apply a threshold to keep from sharpening those areas that have smooth transitions, yet still applying sharpening to those that have more contrasty transitions (eyes/lips/etc.).

In the second crop ... I've used a different sharpening approach which probably wouldn't work well for the girls face, but you can see the difference in the thread/button.

So, when it comes to sharpening, there are decisions to be made relative to your subject/scene as to how you may want to approach it. That's why sooc can only get you so far ... it has to play it safe because it doesn't really have the ability to judge which area of the image is the more important area ... that you'll make your sharpening decisions based on.

Just like contrast, saturation, etc. sharpening is a tool that requires judicious/selective utilization to achieve your best ... rather than global sooc which can still be really good, but never BEST. And, just like all aspects of editing ... S&P to taste.

Many ways to go about sharpening ... and it is certainly possible to oversharpen as well as under sharpen. Toss on the fact that your output size has a factor on your sharpening strategy and you can begin to see why sooc sharpening isn't the way to achieve BEST.

HTH



cgardner
Registered: Nov 18, 2002
Total Posts: 9376
Country: United States

The term "unsharp masking" (USM) originated back in the days when color separations where done on process cameras and and later on drum scanners. Been there, done that, in the days before Photoshop.

What it does is change the contrast at tonal boundaries. These are 800% screen shots before and after USM was applied...












RustyBug has explained the basic controls. By way of analogy, think of USM creating a stone wall at the tonal boundary. Amount will make wall higher, Radius will make it wider encroaching on more real estate on either side.

The method I use was learned about 10 years ago from a Dan Margulis article in a magazine. It involves two steps: first applying USM, then using the Edit Fade function to adjust it with an interesting twist. The fade step is applied in Luminosity mode which thanks to some deep Photoshop mojo applies it to the L channel at the color management level.







As a first step I applied 500, .2, 0 sharpening overall. That is the maximum amount, but applied very narrowly to the tonal borders (i.e. a high, narrow wall). Then I used Edit > Fade to scale it back, changing the mode in Fade from "normal" to "luminosity". Due to the way Photoshop works deep in it's bowels changing the mode to luminosity applies the sharpening only to the L channel as if in Lab mode even when the file mode is RGB. It eliminates any color artifacts that might occur when two contrasting primary or secondary colors are sharpened in RGB.

The more useful think the Fade Step does is allow you to adjust the amount if USM interactively while seeing the full image in the edit window by moving the Opacity slider from 0% (no sharpening applied) to 100% (what is seen before the fade step).

The amount of USM needed varies depending on the size of the image and how much detail it has. I use the high amount / low radius method for my small screen images because I like how it restores the crispness without any halos. But I hardly ever apply it at 100%. I move the opacity slider back and forth and usually wind up around 60-70%. The edit above is at 85% which is a bit more than I'd normally use for a portrait. Also I wouldn't apply USM overall for a portrait.

For portraits I have an action I use which creates surface blur, high pass sharpen, and USM layers with masks which I selectively open and then blend the layers to adjust the appearance of the skin and tone down distractions like the buttons by blurring them a bit to take off the "sharp edges". Here's the high res copy with that treatment applied..







BTW - the skin highlights are blown out in the the red channel of that image due to either overexposure at capture or from the 16 bit wide gamut > 8 bit sRGB conversion step. You need to allow "headroom" at capture for the editing workflow changes. In addition to applying the skin treatment I fixed that and cloned out the less than ideally placed nose shadow that was hanging out because your key light was a bit to low...

I apply USM for printing differently than for small screen images. Something to realize about USM for prints is that you really can't judge it on screen because it needs to compensate for variables of printing. The best way to zero in on a good USM baseline for prints is to take an image, apply various amounts and methods of USM, make prints, put them on the wall and then look at them from different distances.

Viewing distance is a variable affecting perception of sharpness. Beyond reading distance of 18" the brain interprets 3D shape based on overall contrast differences between planes, similar to a wireframe rendering. As a general rule as you make a print larger and increase the viewing distance you can increase the amount of sharpening.


RustyBug
Registered: Feb 02, 2009
Total Posts: 12942
Country: United States

+1 @ oversharpen, fade to taste (also at layers/opacity/blend)
+1 @ sharpen in L channel of LAB, K channel of CMYK or Luminosity mode @ RGB



Robin Usagani
Registered: Oct 26, 2010
Total Posts: 2845
Country: United States

Ok , I will read it slowly at home. Nice edit.