The brushes functionality will allow a user to apply different brush techniques and even mix and smudge “wet” paint.  These types of natural painting features have so far only been available in dedicated digital painting applications like Corel Painter.

The other amazing feature is the new warping tool, which basically allows you to stick “pins” in places that you don’t want to move while you fiddle with the surrounding areas via a mesh grid overlay.  It’s the same idea as the current Mesh Warp tool, but it looks to provide much, much more control.

These features were first unveiled at the recent Photoshop World event as sort of a sneak peek into the Adobe Labs.  There has been no promise from Adobe that any of these features will be in the upcoming release of Photoshop CS5, but one can only hope.  I’m not as concerned with the painting stuff (since I’m not a painter) but that warping tool looks amazing.  Want.  Now.

Check out Adobe’s Facebook video to see these tools in action.

This feature is based on a “randomized correspondence algorithm” and something about “structural image editing” that can “quickly find approximate nearest neighbor matches between image patches.”  To me, it looks and sounds like the Healing Patch tool and Photoshop CS4’s Content Aware Scaling on steroids.  And crack.  It will be interesting to see what the final flavor of this feature ends up looking like in Photoshop CS5, which is expected to come out April-ish 2010.

For those that want to geek out on the underlying algorithm of seam carving/content-aware scaling, here you go:

The inspiration

One day several months ago my girls were climbing all over me, so I pulled out my mobile phone and snapped this quick self portrait:

The photo quality was terrible, but as a dude in a house full of girls, it was a great visual metaphor of my life.  My mission: recreate this as a better quality photo.  Here’s what I was able to come up with:

The setup

I didn’t spend a lot of time on the wardrobe pickings, but I did have some direction in my head: I wanted a drab background, I wanted me to be wearing drab casual clothes, and I wanted my girls to be dressed in pink.  This would reinforce the “story” that I was trying to capture: drab ol’ dad surrounded by his frilly little girls.

As for the lighting and background, I placed a single strobe/umbrella at about 2 o’clock and up as high as the 12’ ceiling would allow (I was going for “moody” and “reflective”), and hung my standard DIY backdrop cloth as a background.

The shot

With my wife as the trigger-man, we shot a dozen or so shots of me doing my best to hold or otherwise support my three daughters.  I really didn’t want to do any Photoshopping of faces – you know, where you aggregate the best faces from multiple photos into one artificially “great” photo?  That totally has its place, but not here.  Not this photo.  Regardless of all of the other edits I knew I was going to do, I wanted us to be real.  We ended up with a couple good shots and selected this one as the best:

Now, as you can see, there are multiple problems with this photo, but our faces and postures looked good to me.  With that, believe it or not, the hard part was over.

The Photoshop edits

As I mentioned before, the problems with the original photo are numerous: it’s off center and crooked (not at all a criticism to my lovely wife/assistant; I couldn’t have done this shot without her), the backdrop doesn’t extend to the top of the frame, and the card table used to increase the height of the strobe light is visible.  Let’s take care of these critical issues first before worrying about the “higher order” problem of making this look like cool, moody softbox lighting.

First, using a couple guides as a reference, I rotate the image slightly to straighten it up and expand the canvas to give me some leg room.

Using the Clone and Healing tools, I get rid of the clutter on the right, extend the backdrop on the top, and add to the background on the left.  There are also these pesky seams in my backdrop, so I’ll get rid of those too while I’m at it.  I don’t feel like thinking about final aspect ratio yet (5 x 7?  8 x 10? Who knows) so I just clone the background out a little more and do a quick crop to get rid of the distracting tilt (I’ll do a final crop as one of the last steps).

One of the first issues I want to address is normalizing the pink hues in my girls’ outfits.  That is, I want the pinks to be the same pink.  By carefully selecting their clothing, I apply a Hue/Saturation adjustment layer (with Colorize selected) and dial in a single hue of pink.

The most important adjustment I need to make with this image has to do with the lighting, and it’s really to compensate for me not having the right equipment.  I wish I had a softbox that would have directed the light on me and my girls and restricted the light from hitting the rest of the room.  But I don’t have a softbox (yet); all I have are umbrellas which gush light all over the place.  I’ll have to fake the softbox effect by darkening the edges and creating a vignette.

First I draw a shape with the Elliptical Marquee Tool that represents the general shape of light that I’m shooting for—I want our upper torsos and faces to be well-lit but then everything else should fall off into the shadows.

Next I hit the Q key to hop into Quick Mask Mode, as this will allow me to build the right selection visually.

Initially the mask is inverted—it’s protecting the areas that I want exposed and exposing the areas that I want protected.  I hit Ctrl+I to invert it and make it right.

Next, I apply a ton of Gaussian Blur to soften the edges of what will be my selection.

Note, however, that this makes  some of our faces partially masked (as noted by being partially red).  Since I don’t want our faces to be darkened at all, I protect them by touching each face with a big, soft-edge brush loaded with black paint (remember, I’m painting on the mask in Quick Mask Mode, not the actual image!).

With the mask done, I can exit out of Quick Mask Mode by hitting the Q key and admire my new selection.

With the selection still active, I create a Curves adjustment layer and dial in a nice vignette by adjusting the output level white point and the overall midpoint curve.  Nothing scientific about those settings…just eyeballing it.

Lastly, I need to do something about that background.  It needs to be more drab and it currently has a reddish color cast to it.  Fortunately, I don’t need to do any complicated color correction; I just need to desaturate it a bit.  By doing a careful selection around my girls and me, and then inverting that selection, I create an adjustment layer that will just affect the background.

If I desaturate it all the way, it looks weird; leaving some of its color makes it look real but still neutral.

I decide to crop it as a 5 x 7, and apply some final Levels adjustments and mild sharpening.

With that, I call it done.  I love this photo.  Mission accomplished.

Lessons learned

• Having the right lighting and backdrops can eliminate a ton of Photoshop work, but sometimes you have to make do with what you have.  Things can still turn out great; it’ll just take some extra work.

Full size image

The inspiration

I love those Where’s Waldo-type books where there is a dizzying amount of detail in the photo and you have to find Waldo and various other things.  I thought it would be fun to somehow create our own scene where we acted out a few family highlights from the year in a way that felt like a Where’s Waldo scene.

The setup

The strategy boiled down to this: 1) put the camera on a tripod. 2) take a bunch of a photos of us in different places, and 3) mash them up in Photoshop with selective reveals using layer masking.  If that last part totally doesn’t make sense, hang with me…I’ll explain.

Fortunately we had access to a large room with lots of tables and chairs.  We needed a big room to create a big scene, and all of the tables and chairs would help make the scene extra “busy.”  Even better, this room had a balcony where I could set up the camera nice and high to overlook the action below.

Problem #1: Even with being up in a balcony, I still had to use the widest focal length of my zoom lens (18mm).  Ugh.  I really wanted to be far enough away so that I could zoom in.  This would have minimized the difference between the closest subjects and the farthest subjects. (In the Where’s Waldo scenes, all those little guys are the same size regardless of where they are on the page).  At 18mm, the stuff nearest to the camera will look a lot bigger than the stuff at the back of the room.  Bummer, but onward.

Problem #2: The lighting was really poor in the room, so I knew I would have to use my strobe lights.  I don’t own the best lights in the world, and they certainly don’t have the muscle to light up a room that size.  My test shots showed a serious vignette—the center of the frame was well-lit but the corners of the frame were much darker:

Normally, I find that there is no easier way to sexy up an image than to add or exaggerate a vignette.  It not only adds depth, but it also directs the viewer’s eye to the main subject of the photos.  But this is exactly why I did NOT want any vignette; I don’t want the viewer’s eye to be drawn anywhere.  To the contrary, I want it to be quite confusing!  Bummer again, but I had to keep moving.  I’d just have to fix it later in Photoshop.

The shoot

The rules were simple: don’t move any furniture, do something random, and leave a Post-it note where you stood.  The Post-it note procedure would prevent us from being in a spot that had been occupied in a previous photo and it allowed us to evaluate how “full” the room was getting.  I had a remote control that would start the timer on the camera, giving us a few seconds to get in place before the camera snapped.  Here’s a quick time lapse of the roughly 100 photos we shot over the course of a couple hours.

Reviewing the photos

After tossing out all of the near-duplicates and those that just weren’t that interesting, I had 45 great images to work with.  Cool!  5 family members times 45 shots equals 225 people filling up the room.  Nice.  However, now that I was viewing the images on my monitor I noticed another problem.

Problem #3: The image quality in the corners and outer edges of the frame showed some serious distortion and softness.  The center of the frame looks good with nice detail:

…but the edges look terrible.  Softness, noise, distortion, darkness, plague, pestilence:

I don’t know whether this was because I don’t have a million dollar lens (it’s just the standard 18mm – 55mm that came with my camera), or whether all lenses do this at 18mm, but this meant I’d have to do some selective Photoshop sharpening in those areas.

Problem #4: Although I had set my white balance to match my lights as I normally do, something about all the wood in that room just made everything seem way yellow and muddy.

I decided to tackle all of these problems—the vignette, the soft corners, the muddy color—as a final step once everything else was done.  For now, it was time to start mashing up these photos in Photoshop to see if this was even going to work.

Using layer masks in Photoshop

I decided that the last shot—the posed family group shot—was the most important one, so I used this as the starting point.

Knowing that this photo would be the bottom layer in what would be my master Photoshop file, I did a Save As to a Photoshop .PSD file.  I opened the next photo and immediately did a Shift+drag from the its Layers pallet and dropped it on the master file Photoshop file.

This creates a second layer in the master Photoshop file.  Holding the Shift key down while doing this makes the edges of the dragged layer snap to the edges of the photo where it’s dropped.  And because these two photos were taken using a tripod, everything lines up perfectly.

So now I have a 2-layer Photoshop file, with one layer merely covering the layer underneath it.  The first thing I do with this new layer, however, is completely hide it using a layer mask filled with pure black:

Making sure the mask is selected (just click on the mask in the layers pallet), I use a soft-edge white paintbrush to paint on the mask which reveals the parts of the image that I want to show.

Here is the 2nd layer shown by itself so you can see what I’ve revealed by painting on the mask:

As you can see, I don’t even have to be particularly accurate at this point with my paintbrush.  Because the photos were all taken using a tripod, everything lines up perfectly:

I repeat this same process for all subsequent photos: 1) open the next photo, 2) Shift+drag it to the master Photoshop file, 3) hide the layer with a black mask, 4) paint white on the mask in the spots that I want to show.  The only time I have to be careful is when I’m revealing something on a layer that touches or overlaps another spot that I’ve revealed on a previous layer.

The more photos I add, the more “crowded” it gets, and the more careful I have to be with my mask painting to make it look real.  It’s time consuming, but the effect is pretty cool, I think:

Finishing touches

To finish up the image, I correct the vignette and softness around the edges…

…and neutralize the muddy yellow color cast…

…and added a whimsical candy cane frame around the whole thing:

Full size image

There’s a lot of ground to cover in just those last few steps that I’ll save for another tutorial.  The goal of this tutorial was to show how tripod photography and selective “reveals” via layer masking can create an interesting illusion.  In this case, the illusion of a family of 5 looking like a crowd of 225.

Lessons learned:

• My two studio strobe lights are not bright enough to light up a huge room.
• I should have bumped the exposure up a few ticks.  For example, to fix the vignette, I brightened the edges.  It would have been better if the edges were properly exposed and instead I reduced the brightness of the an over-exposed middle.  The end result would have been the same, with the exception of less grain/image noise.
• Once this type of photo shoot starts, don’t move anything!  There were a couple of chairs that got bumped between photos at some point, and this required some delicate Photoshop surgery to fix.  Fortunately, it was just in a few spots, but it added extra work that could have been avoided.
• Not necessarily learned but confirmed: feed the kids directly before a photo shoot.  Normally, they quickly lose their patience with me and I end up going Christian Bale on them (ok, maybe not that bad).  I was amazed at their level of cooperation, and for that I thank the pizza.

Lingering questions:

• Do all wide angle lenses create optical distortion around the edges, or just cheapo ones?