dslr quandries / composite-image meddling
As most Nikon “enthusiasts” are aware, Nikon has been delaying … or just not bothering … to release a replacement for the venerable D700, their only full-frame DSLR that retails under $5000. The gap is pretty large – the D700 runs around $2500ish canadian, and the next step up, the D3S, is $5200. Both are 12 megapixels, incidently, the only DSLR body in Nikon’s line surpassing that – 24 MP – being the D3X, which is around $8000.
This seems like a problem to me, and most retailers have seen it as such for some time – not only is there a huge “megapixel” gap, there’s a giant price range which Nikon hasn’t filled – actually two gaps, really; The D300s, their high-end crop-sensor DSLR, runs around $1300, the D700 clocks in at $2500, and the D3S, again, $5200 – all 12 MP cameras, by the way.
This becomes a ‘worse’ problem when you look at the Canon line – their 7D, at 18 MP, with a 1.3x crop sensor sits nicely between the D700 and D300s price wise, and their D700 equivilent (speaking fairly arbitrarily here), the 5D MkII, is about the same cost, and is a massive 21 MP full frame sensor. Canon fills in all the gaps nikon has left, and doesn’t have this bizarre fixation with 12 MP in their full-frame cameras.
Why am I bitching about this? Well, basically, I’ve been reaching the limitations my S5 Pro can manage for print-size, and I’ve wanted to make the jump to a full-frame DSLR. But it seems somewhat redundant to go from a 12 MP camera to another 12 MP camera, so I’ve been hoping, for a good year now, to invest in the replacement for the D700 when it came out, but Nikon has been dragging its heels. They say “summer” now, which, I guess, is better than nothing, but meanwhile I have prototyping and planning to do … not to mention exhibition proposals and grants … so I’ve been forced to improvise.
(and, yes, heresy, but if more delays or some kind of shocking price point occurs with the D800, I will switch to Canon and get a 5D Mk II)
With composite imagery, the problem of parallax tends to be the chief obstacle in stitching multiple horizontal layers of images together, particularly when there’s the existence of horizontal and vertical lines to match up … or extreme angles … or big long-ass panos. Being a significant distance from your subject tends to help – a 2 gigapixel or some nonsense image of the grand canyon comes to mind from a few years back – so big, expansive landscapes lend themselves to extremely “high megapixel” composite images. Interior spaces, anything within 15′ of the camera, problems arise.
So the current mess of a project I’m thinking through right now requires an extreme amount of ‘readable’ detail, and will be printed fairly large (I’m thinking 60′ x 90′ ish right now, and I’m fishing for suggestions on mounting such monstrosities…), so a high-res composite image is pretty much a must. I really wanted a 20MP, full-frame camera to execute this with, because it’d require less photos to stitch together, but that just doesn’t seem to be happening any time soon.
So I’ve been futzing around with methods for achieving this on my own with my trusty S5 Pro. My first successful test involved a garage wall, standing about 10′ away from it, using a 50mm lens to help offset the parallax issue, and I was pretty shocked at how well it stitched together. The resulting image, ironically, covers about the range of a 20mm lens from the same distance – this isn’t really compressing space like a wide-angle composite pano.
Anyway, what follows (click the ‘read more’ below) is the final image, and then some details with reference image from where I pulled them. (… as I’m inserting these into the post, I’m noticing they’re fairly dark and not as sharp as I thought. I didn’t really do any post-processing apart from the composite stitching, so .. that’s likely why. It’s just proof of concept anyway.)
The details here are “100% pixel” level, so they’re not actually “print size”. I’m pretty happy with the result given I executed it in like 10 mins of shooting which was pretty slapdash, unorganized, using a ball head which can’t measure degrees or stay level between adjustments. The real test will come when I move this, well, into the garage, and see how this works with an interior. I suspect the lack of need for a wide angle of view will help a great deal, we shall see I guess.
On reflection, this is pretty counter to photography as a whole – there’s little or no composing “through the viewfinder” at this stage, and the camera becomes less a tool for composition, and more a tool for recording components of a larger image, quite literally. The lens ceases to be a composition element or frame-guide, and more a mechanism to reduce parallax and distortion. It’s an odd space to work in, but the final product is still intrinsically photographic. I hope.