Crop #1

Crop #1

By removing distracting elements (fish, translucence), the overall composition is made simpler, yet stronger.  Mostly yellow, with red touches to break up the monotones.  Some smoothing and simplifying of details brings it together.

Crop #2

Crop #2

This required a full rotation (180 degrees) and some work to bring out the details.  From a distance, it’s a landscape.  Up close, strange and compelling details are revealed: the fish in the pond, strangely-upside-down lily pads.  And yet, the rest looks normal.  A feast for the eyes (well, my eyes)…

Koi Pond Reflections

Permit me, if you will, the opportunity to share a failure with you.  The image (to the right) is to what I refer to: abuilding reflected on a koi pond.

This was taken at sunset on our recent trip to Hawaii.  The sun had just passed behind a building and the reflections were strikingly strong.  There is no amplification of the effect here. A quick shot and on I went, chasing a tired 5-year old.

Looking at it later, I noticed that something was… missing.  But what?

A commenter on Flickr pointed it out for me: it’s too busy. There are great elements, but it’s compositionally poorly executed.

Contrast that with this:

Bubble Globe

Here is a good example of a complex subject and presenting it simply. So many details in the bubble, yet they  blend effortlessly.

It’s your post processing.

The thing I’ve noticed over the year & a half since I went to NAPP, is that my photos immediately jumped in quality. It’s not just the ability to introduce contrasts.  It’s also the sharpening techniques, color awareness, painting techniques, dodging and burning.

All of that came together in this, a favorite (recent) image:

Fiery Icicle

Post processing: slight saturation increase in reds and oranges.  Slight luminosity increase in blues, decrease in reds and oranges.  Addition of clarity filters, slight increase in overall contrast.  Reduction of global saturation and luminosity across houses.  Increase in luminosity on icicle. Single-pass sharpening on luminosity layer.

It took less than three minutes.

So, for starters: the two images.

Original Treeline

Moon, correctly exposed

The treeline picture has a pretty decent exposure for the nighttime, but it’s noisy, there’s a little blur, some color cast, and the moon has… funky things going on.  Not only is is pretty much over exposed, which was expected, but you can see a second sub-image.  This is due to me being lazy and it being cold outside – I shot through a window.

The correctly exposed moon is a much faster, higher speed pic.  It’s much smaller, however – there isn’t the over-exposure glow, so it’s really thinner rather than smaller.

So now comes the thought: how to combine these?  First step is to remove the moon in the treeline pic. I did this quickly with the patch tool in Photoshop CS4: select the moon, drag to a solid part of the sky, and presto – no moon! Entirely believable, too.

Now that we have a flat (albeit noisy) sky, it’s a matter of copying the moon from one to the other.  I’m interested in speed, so I’m going to cut corners. I start by duplicating the background layer, then selecting the background via the menu: Select->Color Range. If I stop here, the edges of the moon will be too sharp, so next I click on Select->Refine Edge.  Settings (after playing for less than a minute): Radius=1, Contrast=0, Smooth=3, feather=1.0.  The selection now looks like this:

Selected and smooth

Hit the “delete” key, and you’re left with only the moon on that layer. The next step: click & drag it over to the original image.  If it lands on top, you’re set: just move it around to the position of the original moon. It’ll look OK, but it’s choppy at the edges (still) – needs some smoothing.

Now, I said I was going for quick & dirty, and here’s the quick & dirty for smoothing out those edges: pull up layer styles (double-click on the new layer), and move around the “blend if” options.  Here’s what I ended up with:

"Blend If" settings

Do you see what’s going on there?  We have a black & white image, the moon.  It’s blending (i.e., hiding) from the moon layer all pixels from the upper marker and darker on the upper layer with anything darker on the lower layer.  Since the moon is brighter, a quick drag of those markers gets us right in, and looking sharp.

Moon over the Treeline

The final result, with a little noise still around the original moon:

Continuing with online printers: this is the last company that I’ll review, and my favorite.  You’ll see why:

Color shifting: Want to avoid it altogether?  Download their color profiles (one per paper/print type). Output your images with that color profile, and what you’ll see, you’ll get.  Take that back: one time their printers were out of calibration.  They re-printed free of charge. Thumbs up, obviously.

Crops: for smaller pics, maybe a pixel.  Larger pics (up to 16×20, in my experience): maybe three pixels. As good as it gets.

Paper quality: all quality papers (Kodak, usually), and you get your choice of finish.  Includes metallics, glossy, others… More thumbs up.

Sharpness: I’ve generally had it hit or miss with them until the most recent print job.  I finally found that they’re considered a “contone print shop”, and you do “contone sharpening”. More about that another day, though.

Overall: AdoramaPix has pretty much earned my loyalty.  Some prices might be higher than with competitors, but the quality is there and they cater to the more advanced amateur & pro alike.

Another continuation.  Yesterday was CostCo, today MPIX.  MPIX, in my opinion, has the best Metallic prints of all companies that I’ve used.  I’m not sure it’s appropriate for all outputs, but it seems to work pretty well.  Sometimes it seems to lack smoothness, though.

So, to jump into it:

Color: So long as you follow instructions, no color shifts. That means uploading images in sRGB.  I messed this up once, and the quality was still good – but doing sRGB made it much, much better.  Big thumbs up here.

Crops: no discernible cropping to the edges.  Maybe a pixel or two, but easily acceptable.  Another big thumb up.

Paper quality: decent heft, Kodak endura.  Some may prefer something else, but it’s acceptable.  Still thumbs up.

Sharpening: Oooh, so close.  Big failure here – they flat-out recommend not sharpening the photos. I have a real problem with this – the output needs to be to my spec, not their automated one. Any sharpening at all, in any way, shape, or form seems to give horrible halos

Overall: I’ve used them for the last two years in a row for my personal holiday cards.  After this year, where I deliberately kept sharpening to a minimum, I may not do it again.  They halos are barely there, but… well, they’re there.

Still, the price is right.  And their metallic prints?  Gorgeous.  By mid-year, I may have changed my mind on these guys!

Continuing on from yesterday’s topic, a review of CostCo’s print service. To keep things fair, I’ll use the same criteria: color skewing, sharpness, paper quality, and cropping. And before I go in to trashing the CostCo online printing, let me state: most people that I know like their service & see nothing wrong with it.  My problems could conceivably be due to my own behavior (which works fine with other print shops), or it could be the location nearest me.

Color: good God, possibly the worst coloring I’ve ever seen.  Take a nice, warm scene with green grass and a smiling, pink baby.  Got it in mind?  Now turn the grass dull (light) grayish green, the skin tones more yellowish, and remove any character from dirt in the pic.  That’s what happened to me.  I’d say that it’s like submitting an AdobeRGB photo when they’re expecting sRGB – only I’ve done that, and it’s no worse.  And other printers don’t skew it so badly.

The color problem appears to be specific to large prints, BTW.  But in this day & age, that’s all I do, so it hurts.

Crops: Again, one heck of a crop that I’ve seen.  One year I used their 5×7″ photos for our holiday cards.  There was “Happy Holidays” in script towards the bottom, and made sure there was plenty of padding between the edges and the letters.

They still chopped into it.

Paper quality: decent.  Kodak, pretty good weight.  Felt like a waste of quality paper after seeing the color shift.

Sharpness: to their credit, they didn’t add any sharpening.  The end results were more or less what I expected.

Overall: I won’t touch their photo service with a ten-foot pole. YMMV.

I recently placed an order of photos, and was fantastically pleased with the results. But more on that another day – today I want to place some thoughts into writing about how I got here.  Permit me, if you will, to walk through my memories…

In my memory, SnapFish was one of the early entrants to online print shops.  They were also the first online company that I used, and for a while my favorite.  Having said that, I haven’t touched them for several years now.  Looking back, I think it has been about six years since I’ve done anything with them (wow, how time flies!).  Five at best.

Why? When I’m looking for a print shop, I’m generally looking for a combination of quality and price.  When in question, I usually go for quality. Qualities like: no arbitrary halos around features, no changing of colors, no cropping into the picture, decent quality of paper.

Personally, I only found them to be so-so.  It may have been my lesser-experienced eye back then, but the colors seemed to have greens and warm tones amplified. This is a common trick for landscapes and portrait/people shots, respectively – and it just bugs me to no end. On the plus side, it does look pretty good for shots straight out of the camera.  On the not-so-good side, if you do your own post processing, then it’s going to look radioactive.

Of course, that was a while ago.  They probably have fixed that by now.

Those halos?  That’s from oversharpening.  I didn’t really see that on prints from them, but I also didn’t really sharpen my own results.  I wouldn’t hold it against them, were I to do another run of prints.

Cropping?  Oh yes, they definitely do – or did.  And not necessarily a small amount, either: I seem to recall losing up to 5 pixels off the end of a smallish 4×6″ print.

Paper?  Decent quality, but felt a bit flimsy compared to other papers. Nothing I’d hold against them, but nothing I’d use for photos of any large size.

So to summarize: I’ve found that SnapFish did pretty good work, but I’ll only ever use them for cheap 4×6″ prints. Something for collages or other disposable uses.

Exposing for Digital

First off, the rule of thumb with digital photography is “expose for the highlights, develop for the shadows.” This is because light reactivity on a digital sensor is fairly linear – that is, each nanosecond X amount of light is received on the sensor. Now, the more light that is received, the “whiter” the detail gets.  Eventually it can’t get any lighter, and that pixel is flat-white.  At this point, we call it “blown out.”

So, with digital you shoot to avoid blown out pixels, which leaves shadows slightly underdeveloped.  Then you try to smooth things out after the fact – thus the need for noise reduction software, etc.

Exposing for Film

Now, the rule of thumb with film is “expose for the shadows, develop for the highlights.” This is referring to how emulsions react to light. One might think that emulsion reactivity, like a digital sensor, is linear – for every nanosecond, x amount of crystals react and you get some detail.  That isn’t true, however: it’s curved.  After a certain amount of exposure to light, the reactivity of the crystals slows down.  So, given a highly dynamic range, a film camera might be able to expose for the shadows and not blow out the highlights.

With film it’s all the more important: in digital, you’re virtually guaranteed detail in the shadows.  With film, it takes a small amount of time to start the chemical reaction.  If there’s not enough light, there’s no reaction – and instead of blown-out highlights, you end up with undeveloped shadows.  The same problem as with digital, but on the opposite end of the spectrum.

Shooting Digital with Film’s rules

Now, what Reichmann said was to shoot digital as if it’s film.  This won’t work with JPEG output, mind you: once JPEGs blow out a pixel, that’s it – game over.

But with RAW photos it’s different. See, RAW photos actually have a stop or two (in my experience, almost three with Nikon RAW) of flexibility past what you can see on a screen. So, the argument goes: expose for the shadows.  You’ll end up with a bright picture, but detail is still all around.

In post-processing you can subjectively reduce your exposure, until you get something you like.  Your shadows will be well-defined, noise will be reduced, and you can control parts of the exposure, just like in the old days in the darkroom.

Yesterday’s final pic is a great example of this in action.  The exposure was fine on the foreground, but the sky was blown out.  Subjective exposure tweaks brought it all back together.

Unprocessed Image

Processed

The thing that should jump out at you in the “before” picture is that the clouds have virtually no definition and the sky is washed out.  They are drastically overexposed due to the dark foreground.  Now, some of this may be due to the altitude – I was about 8,000 feet above sea level, something I’m not too used to.  So my mental focus wasn’t that sharp, and I didn’t review the histogram on-site for blown-out highlights.

ACR Settings used for resurrecting and finalizing this photo - click for larger

Move your mouse over the two pictures.  See the red that shows up there?  That’s where there is no detail – in the JPEG-saved version, at least.  The first photo is as it came out of the camera.  The second has just a minor tweak in Adobe Camera RAW to drop exposure by just under a stop.  See how much detail returned in the clouds?

This is because this was a RAW photo – and that means that highlights aren’t actually blown out.  They certainly are if it’s a two or more stops past the exposure of the photo, but if it’s less?  Detail is still there, as shown by the second photo.

An overview of the settings used to resurrect this photo are to the right.  You’re now seeing an underexposed foreground, but properly exposed clouds.  Bringing back up some of those shadows and you end up with as much detail as you could ever want in a photo like this:

The final version