In The Middle Of Nowhere

We went for a drive along a very winding firetrail (that's the car at the left) and, coming out of a forest clearing, chanced upon this rather eccentricly located house.

 

It had copious water tank capacity as well as 40 very large solar electric panels (arranged in two arrays of twenty each) for its power and water requirements.

 

There was no-one home.

 

But I just liked the boldness of the idea of building a modern building out of laid bricks in the middle of nowhere.

 

The "red filter" approach emphasizes the clouds, which were a magnificaent panorama that day.

 

This picture is part of an ongoing experiment of mine to use double exposures to capture a fuller range of tones than a single digital exposure. You can only do it with a tripod and it's tedious, but for certain subjects it works perfectly. All the cloud detail is there - to the brightest points - as well as very satisfactory shadow detail.

 

Another problem was that the sun was dippng in and out of the clouds. I'd start the exposures, get half-way and the light would drop by two, or even three stops. So I had to start again. The whole picture took just under an hour. My wife read the paper, then the road map, then the car's instruction manual, finishing off with the tyre pressure label in the driver's side door jam. It was a close run thing. All she had to go after that was the wording on the back of the registration sticker. I would have been in trouble if she'd finished the sticker and had no more to keep her mind occupied.

 

And there were flies... lots of them. Aussie bush flies never give up. In their short lives of just a few days they have to make hay while the sun shines.

 

The original of this picture is 8 feet across by 12 inches tall, and is rich in detail. You can literally count the bricks in the walls of the house.

Since it's a 360 degree image I know the vehicle is parked beside the house, but the way you've cropped it (is that what you call it for 360's? ) it looks like the house is looking across the valley at the truck, and the two are locked forever in a standoff of sorts.

I can never figure out how these rotational shots are going to.. er... pan out... if you get my drift.

 

The house, built at the edge of the ridge, looks directly out over the valley and the car was parked about thirty metres away and slightly behind the ridge line.

 

You can unwrap the photo by swinging the house back and to the right and the Subaru (it's not a truck) back and to the left, both pivoted on the centre of the photo.

 

The most exciting thing about the shot for me is that my double-exposure method worked pretty well here. I set the meter at one stop above the ISO rating (Camera_ISO = 200, meter = 100) and used the bracketing function to make two shots: one at the meter setting (100) and one at 2 stops under that.

 

The merge was done by superimposing the darker on the lighter and then using the lighter shot as a layer mask, allowing its blown out areas (overexposed white clouds) to pass the equivalent areas of the darker shot, while retarding the dark exposure's shadows (the lighter shot's shadows, being overexposed, do fine on their own).

 

Some tweaking of the layer mask is required, as is some tweaking of the dark shot itself. The aim is to produce as near to a fully populated ("0" to "255") histogram as possible, while at the same time trying to make the photograph more than just a technical exercise.

 

One of the traps to avoid is to get the highlights and shadows correct, only to find you've lost the mids. For example the grass can look real ordinary (super flat) if you're not careful.

 

A few other tricks were used to add contrast to the somewhat flat result, without losing either the hard-won shadows or the highlights.

 

A camera that had this function built-in as a macro (without requiring two shake-producing mirror actions) would be sensational. You would still have some differences between the exposures with a moving subject, but the camera shake aspect would be not there.

 

Some surveillance video cameras have this function, combining two exposures to give a less contrasty final result, so it can be done. Providing two images gives the photographer the ability to adjust for the best blend.

 

Come to think of it, this would be a good way to do ANY bracketed set of exposures.

 

I originally used to use three exposures - minus-2, ZERO, and plus-2 stops - but found this far to difficult to manage.

beuatiful shot, wonderful technique

Did you try working in deeper bit depths, 12 or whatever the RAW file is?

Charlie,

 

I took your suggestion an reactivated a long dormant copy of Capture 1 DSLR capture software. It made recignizable improvements over the standard Nikon upload NEF file software.

 

I had tried Capture 1 before but didn't notice any improvements. This time I really made the program work and got some encouraging results.

 

Having said that, I still don't think C1 would bring in all the finest detail - shadows and highlights - that the merged exposure technique does, but it is a definite improvement.

 

Thanks for jogging my conscience and getting me to try C1 again.

I'd love to view the mural sized print of this one Tony. "8 feet across by 12 inches tall" - how/where did you manage that size print? I love that vehicle, although small in relative size to the house, it seems to balance the whole scene, aswell as contribute possible stories.

I came here to ask you a favour actually, as I was looking for your very detailed write-up about the process of panoramic stitching. I know it is somewhere in your portfolio, but scrolling through your history of posts led me to get completely lost [immersed along the way I hasten to add].

Would you be so kind as to direct me to the panostitching article/page please? Many thanks.

ps have you had the opportunity to try Photoshop CS [aka 8] yet? It has a neat shadow/highlight tool I rather enjoy playing with. The first time I used it I thought "I know who else might like playing with this!"

I think this is the link you were looking for.

I think that this truly exemplifies your very narrow view of the world, but it's interesting that you place the house on one isolated corner of the picture!

What an intelligent comment. Professor Jan does "deep" once again.

360º view and that's narrow?? Ho hum.... righto Jan. Seems like some people have nothing better to do than deliberately go out of their way to get under another's skin. Shame, I am positive it was totally wasted effort ... I can hear Tony bellowing with laughter from the other side of the world.

Tony I am embarassed to say that it was only on the second visit to this particular pano, that I realised the car on the left was actually driving toward the house, rather than away - as the vehicle leaving the frame 'told' my mind's eye. I love the 360º pano's which flummox me like this!

Don't be embarassed G... I usually can't tell how they're going to turn out, either.

 

This is a map of the shot (as far as I can remember it).

 

The green area (and the arrows, plus circle) represent the image area of the picture. The grey area is the segment I cropped out.

 

The objects are roughly in their "real life" positions. "ME" indicates the camera position, half-way down the grassy bank.

>> exemplifies your very narrow view of the world

 

Well, I thought it was a pretty funny joke, 12 inches high by 8 feet wide...

Wow, joining those penels together seemlessly must have been a real challenge.

 

Would you be willing to share some tips about how you did it, software used, etc.?

It took a while to stitch (an hour or so in front of a PC screen) because I first had to merge two exposures - one "over" and one "under" exposed - for each panel.

 

I can also remember starting the 12 images a few times as the sun kept going in and out of the clouds behind me.

 

Nevertheless, despite these irritations, the pic was designed more as an exercise in capturing both highlight and shadow detail by using this "double exposure" method. It think it worked quite well. What I liked about it was being able to select the "look" of the image by varying the degree and method of merging the two exposures for each panel. It gave a lot of post-exposure latitude.

 

When I have the technique under control I may write it up one day. At the moment there are a few things I don't fully understand and can't quite repeat reliably in all field conditions.

 

As for the stitching part, it's fairly straightforward when you use a digital camera in conjunction with a pano head on the tripod (a pano head is one where you can adjust the position of the camera longitudinaly and horizontally above the tripod's axis of rotation... the idea is to get the nodal point of the camera's lens directly over this axis so that, as you rotate the camera between images, there is no parallax error between "near" and "far" objects that may be in the overlap area between panels).

 

Digital cameras are perfect for stitched panoramics because there's no scanning involved and hence no slide carrier lineup problems between scans. To use a technical term, a digital camera gives you orthogonal images... their geometry (as opposed to image content) lines up, so the same warping and stitching formula applies to each image, virtually pixel perfect. Scanned images always have slight anomalies in the positioning of each image in the scanner. These anomalies can compound and result in shears along the egdes of objects in the images, as well as ghosting (pixel or grain smear) when images are layered over each other.

 

There's a link a couple of comments above related to another stitched panoramic. This explains pretty-well everything about the process.

 

Someday I'm going to try a complex, challenging panoramic using this technique, that has real aesthetic appeal as well... which is what it's all about, really.

Thanks for the diagram Tony. The relative positions are a bit easier for me to perceive now. I was also wondering how you made a 'digital' version of a double exposure. How did you merge the two exposures in order to preserve the highlights and dark tones?

G., It wasn't an entirely digital technique. There was an actual double-exposure involved: two images for every panel. Then I blended them in PS using layer masks to lessen interference between the two.

 

From memory (but don't quote me until I've checked this out) the exposures were +1 and -1 stop. The "ZERO" exposure would have had blown out highlights coupled with noisy shadows: alright from a distance, but not really any good at 100% zoom (or in a closely inspected large print). Using the "two separate exposures" technique, the overexposed version would have less noisy shadows and the underexposed section would capture more of the fine highlight detail in the clouds.

 

Really what I was trying to do was to eliminate noise in the shadows while not losing extreme highlight detail. Two slightly different things. A simple "ZERO" stop offset exposure would have probably produced adequate detail in the highlights, but I would have had to artificially "sex up" the shadows (as Andrew Gilligan would say, if he were a photographer and not a journalist), and the shadow noise with them.

 

Blending the two exposures is more than just a matter of "averaging them out" by layer mixing in "NORMAL" mode at "50%". The layer mask technique - basically just the "light" image inverted and used as a mask through which the "dark" image is blended - doesn't allow shadow areas of the "dark" image to in any way interfere with with shadow areas of the "light" image... and vice versa.

 

Once you have your most "difficult" panel blended correctly - the one with the highest or most problematic contrast ratio - you need to make sure they're *all* blended using the same blending formula, or else your final stitch will be "lumpy" in its tonality. A PS "action" is the best way to do this with reliable repetitive accuracy. The first version of the "action" might need to be amended (i.e. re-recorded), as I've found that the "problematic" area is often not where I intuitively thought it would be. I found a couple of times that getting the "house" section to a satisfactory blend made, say, the "car" section too hot. As in all things, it's a matter of balance.

 

I can hear you asking.... "Why not do *two* stiches, one light and one dark, and then blend them both in one pass?". The answer is that this pic is many hundreds of megabytes in file size, and I tried that - once. I was just about able to cook a Sunday roast, wash the car, feed the animals and go to the cinema in-between chug-a-lugs of my hard disk, swapping, temping, remembering history, generating this and that dump file and otherwise blowing a gasket every time I hit the "ENTER" button to activate whatever PS function I was using at the time. If you have lots of gigs of RAM, then go ahead. Anything less than two, do your experiments on individual panels, not full 100% versions of your image.

 

Another caveat against the "heroic" mega-blend is that very often the stitching software will latch onto one detail or the other in one of your "light" or "dark" versions and just won't let go, stuffing up the stitch every time, so that the two never quite overlay perfectly. Brrr.... it makes me shudder just to think of the time I've wasted on being a tough-guy, PS-wize.

 

Another thing you might like to note is that the first iteration of your "final" merged, stitched picture will look quite flat. Think of it as similar to a scan of negative film: they always look flattish too, at first viewing. This is because the dynamic range of even a good scan on a properly ICC-profiled, bright PC monitor is a lot less than the dynamic range of "real life". For instance, out in this paddock, I was almost blinded by the brightness of the glarey sunlight coming (it seemed) from almost every direction. If anything, the "light" version was how I felt, immersed in this location with my naked eye: washed-out clouds, darkish shadows, with a bit of non-overwhelming grassy slope in the middle somewhere. But the "light" version would have made a crappy print in a room with subdued lighting, especially if close inspection was the order of the day. The viewer would see the faults straightaway, and no amount of explaining and excuse-making by the photographer would have convinced anyone that it was a technically good picture, or print (this is why many landscape photographers *never* shoot at midday, when this shot was made, by the way).

 

Once I had my blended result, with pretty-much a full black and a full white (say only about 5% off each histogram limit), I needed to tweak the shadows and highlights (and the mids, too) to get the impressionistic version of the scene that I wanted. In other words, the image still needed some good ol' digital darkroom work done on it to make it *convincing* to the viewer (who by definition almost, wouldn't have been there with me as I took the shot). There's nothing strange in having to do this. No technique is a panacaea for every problem in a shot. To get something perfect straight out of a scanner, you need a controlled lighting condition, which - in a paddock out the back of Braidwood NSW, Australia - wasn't available to me. Every picture of a landscape that we see is only the photographer's *impression* of what he or she saw. Whether the photograph is successful (or not) depends on being able to communicate the ambience of the original scene to the initially dispassionate viewer.

 

I look at a landscape photograph as a "found object", a starting point from which the photographer can progress towards a sort of light poem, that truly represents what the scene was all about. Having waxed so purple ("light poem", did I *really* say "light poem"?), there is still a definite boundary between presenting the picture in an impressionistic manner, and outright faking it. That boundary should not be crossed lightly.

 

But I digress. Back to technicalities...

 

Shooting RAW images also helps in any pre-blending tweaks, as RAW images can be manipulated a little as if they were "in camera" exposures (but not as much as some techno naifs would think they can be, basing their predictions solely on reading the manufacturers' over-optimistic blurbs).

 

There are all kinds of practical problems with the double-exposure technique, not the least of which is that clouds (or worse, trees, people anything not nailed down or dead already) might move between exposures, especially if the D100 (or whatever digital camera) decides to take a memory-write "breather" in-between the two exposures. The delay can be several seconds, in which anything fast-moving can... well... move along quite a bit. You end up with "ghost" images along edges. OK, so it's not the end of the World if this happens, it's just that you need to perform some delicate editing with the clone tool to fix it.

 

I've attached a detail (from the top left-hand corner of the verandah roof, at about 60% zoom factor compared to the original) of shadows under the verandah roof, and details in the cloud formation behind it. Note that this was one of those "fast-moving clouds" situations and I've had to be a little creative with the clone tool regarding some of the boundaries between the clouds and the blue sky.

 

I don't think micro adjustments of this kind really take away from the integrity of the image. To the contrary, I think they add to it, in a funny sort of way: they eliminate a trivial distraction from the picture, by removal of a "ghosting" situation (in the swirling clouds) that otherwise might would have deleteriously drawn the viewer's attention to that section of the image.

 

The Subaru Forester car was my favourite piece of exposure-merging in the picture - as any detail in it is basically "white on white" - but unfortunately I don't have the original "dark" and "light" versions of those panels for comparison.

 

Having re-read the last sentence, I'm amazed at what we photographers can wet our knickers over. No wonder it's a solitary art. No self-respecting *normal* society would have us as anything but marginal members, tolerated but definitely in the "leery" class.

A cold beer Peter? Jeez, I'm sending out a Publisher with a crate of champagne! Thanks so much Tony. Another page saved to disk, and another tutorial to enjoy. Cheers!

Regarding wetting your knickers over the Subaru, you're right. Even as a female photographer I have got all excited over trying to reveal detail in the shadows looking up the chassis of Landrover! Are we weird obsessive compulsives, some kind of geeks, or what?

Great work. The comments for the shoot and the process at which you derived the photo makes for very good reading. Well done!