Tech
Photographic and website coding info
Copyright — All rights reserved
Unless noted otherwise, everything on this site is the work of Andrew Nemeth, Blue Mountains Australia. This includes all photography, text, sound, and web programming.
Therefore everything on this site is and remains the sole copyright of © Andrew Nemeth Australia 2008. You are obviously permitted to view this material online, but no other use or reproduction is allowed without Mr Nemeth's express prior written consent.
The web might be free, but intellectual property isn't.
Stills photography
Over the last twenty-five years I have used a large variety of cameras, lens focal lengths, film types (35mm and medium format) and even digital (Olympus 4:3).
My stills photography is currently done using Leica rangefinder cameras (lately a 2000 M6TTL), along with a 2003 Hasselblad 501cm 6x6 or Olympus 4:3 / Leica combination. Lenses are mainly a 50mm Summilux-M and a 35mm Summicron-M ASPH on the Leica(s); a Zeiss Planar 2.8/80 CF or Distagon 4/50 CFi-FLE for the 'Blad; a 50mm Summicron-R or 24mm Nikkor ƒ2.8 AI for the Olympus 4:3.
No exposure or focus automation is used. Every scene is hand-metered using a Gossen incident light-meter. Focus is mostly done by scale and I typically work at a distance of only 2-3m when taking candids. When working with the camera away from my face, a variety of optical aids are used to help with framing and composition. For digital I always shoot RAW. For 35mm film: either 200 ISO Kodak C41 for sunshine or 800 ISO Fuji C41 for interiors and night. For 120 film: Kodak Portra 160VC and 400VC. Other than an 80A for indoor use, all colour corrections are done during scanning or RAW processing, or else with curve manipulations in Adobe Photoshop® CS.
Shooting in colour
The decision to use colour since '96 hasn't made my life easy. Aside from being sniped at by "Monochrome über alles" purists, I find my film costs are more than double because I cannot bulk-load nor home-develop the professional quality C41 films I like to use.
Then there is the endless struggle with colour balance. Shooting film indoors or outside is easy enough, but working in mixed light is a PITA. If it isn't an outright temperature clash (eg. open shadow vs. sunlight vs. Q-H vs. fluorescent), then it's way too cyan in the shadows or blue in the highlights or too magenta everywhere else. No amount of filtration or fiddling with Curves can fix it either — it's purely a result of the film and scanner's imperfect colour response.
So for every image you see here, assume a raft of "weird colour" rejects. I find this incredibly frustrating, especially when everything else (expression, composition, framing, focus and mood) was exactly what I was looking for.
This probably explains why magazines were so gung-ho about only accepting transparency film submissions. Despite being a hassle to get the filtration and exposure right when taking the shot, doing the scans is trivially easy. With a calibrated scanner it's literally hands off and WYSIWYG for every E6 frame. With C41 however, you have to re-calibrate for each image, and sometimes you can spend days fiddling with the colour balance before it is "right"!
Mind you, I don't think I'm Robinson Crusoe in this…
In the A Day In The Life Of Australia book (1981) , only 58 of 367 shots are colour images taken under difficult conditions (ie. indoors in mixed light). Of these, only 34 have a believable colour balance, with the remainder suffering from run-away orange, cyan or green casts. This works out to 9 % of the total.
The ratios are even worse in the more recent Magnum Degrees book (2000): 617 photographs, 55 colour interiors, 25 with a realistic colour balance — only 4 % of the total…
If these people had so much technical difficulty using colour film indoors, then little wonder I also find it a pain!
Thankfully I don't face this kind of issue with my VR immersive panoramas, mostly because I always shoot a grey-step calibration card, which can then be used as a colour target when scanning and doing corrections in Photoshop. Can't do that for candid people stills of course. "Er, excuse me, hold this calibration card for a moment will you? (Snap) Thanks, now stand there and act naturally… (Snap)" …Somehow I don't think so :?)
Luckily, adopting a RAW workflow for my DSLR images from June 2006 onwards has made things a little easier. What you lose in blown highlights and a narrower dynamic range, you gain in much easier after-the-fact colour temperature correction. Most of the time that is!…
VR photography
Unless noted otherwise, all my people-VRs were unstaged and unposed. No-one's permission was ever asked. Over the years I have developed a compact and inconspicuous VR rig, which lets me work in-close without drawing too much attention. Heck, most of the time people don't even notice I was there!
Each spherical scene is assembled from seven or eight 35mm photographs, which are stitched together in Photoshop to create 10000 x 5000 pixel 360° masters. These are then scaled for QTVR, print or Java Applet use. My 16mm Leica Fisheye-Elmarit-R lens and Kodak 200 film can obviously deliver much higher resolutions than this: 16000 x 8000 masters, via bureau drum scans, would be easily achievable if ever the need arose.
While waiting for digital technology to evolve, I remain one of the few VR authors who still prefers film. Although I routinely use digital for sound recording (MiniDisc), I feel digital VR still has a way to go for before it can match film's deep dynamic range; true 16-bit capture; low cost; uncropped full-frame coverage; sun-in-frame without blow-outs; camera mechanical reliability and ease of repair; noise-free time-exposures and long-term archival capabilities.
I know many believe digital quickly recovers its cost from money saved on film. Yet my cameras paid for themselves years ago, and still have a forty-year life ahead of them. I am aware digital shooters can work around the lack of dynamic range by multi-bracketing shots, but this only works for "empty room" VRs and not photojournalism. Finally, I am also aware that film development, scanning and colour correction is time-consuming, but then so too is RAW-format post processing. You always have to retouch stitched panoramas anyway, so cloning-out a few extra dust spots won't kill you either :?)
That said, I'm no retro-diehard in this. When digital improves to a point where it can deliver (much) better quality VR images at a (much) cheaper price, then I'll also make the switch. Meanwhile…
The joys of still using film for VR
- Images have a unique look compared to the me-too DSLR crowd
- Battery independent, compact, robust mechanical cameras
- No problem doing time-exposures (minutes, hours…)
- Can shoot straight into the sun on a cloudless day
- Can handle mixed colour temperatures with ease
- At least eight stops of dynamic range per shot
- 12 millisecond shutter lag (with the Leica M)
- Compact, non-obsolescent archival storage
- No sensor-dust or sensor-cleaning issues
- Shoot once — acquire at any resolution
- True 16-bit workflow (if you need it)
- Full frame, no lens cropping factors
- No marketing-driven digital churn
I'm hardly alone in thinking the current generation of affordable DSLRs aren't the be-all and end-all for high-quality photography. See Scott Highton's remarks at VRPhotography; Erik Goetze at VRLOG; Ken Rockwell's general remarks on Film vs Digital and "Large Format Cinematography" by Sal Aridi. Finally for those who are puzzled about my "marketing-driven digitial churn" remark, see "The Shape Of Things That Came" at The Online Photographer blog.
Broadband VR
For a long time I held back from displaying high resolution QTVRs because I knew how great a strain they could be for dial-up users. Thing is, net connectivity has vastly improved over the last few years. Nowadays people think nothing of downloading 4 MB songs or 30 MB movie trailers. So clearly it's time to adjust our notion of just what the "maximum acceptable download size" is.
Although my scenes — at an average of 3 MB — are quite large for modem users, for the broadband crowd they are only a two-minute pull. And for that you get a high-resolution image with 90 seconds of DVD-quality stereo. By comparison an MPEG (linear) video with similar spec's would bloat out to 40 MB (or more)!
Looking for VR panorama creation advice?…
For a concise introduction, see <panoramas.dk/panorama>.
For more detailed tutorials and articles, see <wiki.panotools.org>.
Website coding and design themes
When the site was revamped in late 2004, I implemented a few ideas I had been experimenting with for years:
Flash-free zone
Unlike a lot of photo websites, I deliberately avoided using Macromedia Flash here. It may be great for animations, navigation widgets or cartoons, but I believe Flash has too many limitations to make it a practical development tool for entire websites. Also, due to its overuse in advertising, a significant number of users block all SWF content using Firefox's Adblock or Flashblock extensions, preventing them from seeing any Flash content at all. Finally, let's face it, most Flash you see nowadays is at best superfluous, and at worst an irritating waste of bandwidth and time.
W3C compliant XHTML
All XHTML and CSS is certified W3C 1.0 Transitional. So hopefully everything will display in any kind of browser, without reliance on browser-specific features or tricks.
This also applies to the QuickTime content. An unintended side-effect of the E(b)olas Patent fiasco is that the QT workaround I use is also Web Standards "friendly".
PHP separate from HTML code
All PHP scripts are stored separately from the HTML content and are only called via single-line class methods (for example this page has only five lines of PHP code). This way web pages can remain almost pure HTML, and not become a Barrel-O-Monkeys tangle of scripts and mark-up code. It also lets you develop a class structure (PHP5) and keep it isolated from content which regularly gets edited and changed.
It's also a sneaky way of Google indexing your database or XML content. The web-bot comes along, sees a standard parent-level PHP page and indexes it. It has no idea that most of the page was dynamically fetched from a DB/XML when the URL was loaded :?)
Object-oriented PHP scripts
Okay I admit it, I'm a recovering C++ propeller-head. Thing is, I've found OOP has just as many advantages for website development as it does for apps. For example: code reuse; encapsulation; inheritance; not polluting the global name-space; implementation hiding etc. Also by using OOP, I can easily move code around or completely rewrite the "private" parts of it, without Breaking The Whole Bloody World.
In 2007 I upgraded the site to use PHP5 OOD with all the bells and whistles (exceptions, SimpleXML, public/ protected/ private inheritance etc.) With PHP6 just around the corner, hopefully the upgrade will make things more future-friendly than the PHP4 I used from 2004-7.
On-again, Off-again MySQL
For a while this site was database-free, then for a couple of years it used MySQL, now it's back to being 100% database-clear again.
Although MySQL works brilliantly in a dedicated server environment, you face a lot of problems when you use it on a shared-server host. I found most of the snafus came from having to share the DB server with hundreds of other web-site developers, often newbies who used free DB scripts riddled with bugs. Or else they kept opening persistent DB connections and forgot to close them, resulting in "too many connection errors" for the rest of us. Either way DB server always performance suffered.
Well in April 2007 I had enough. I terminated the MySQL connection and rewrote the site's back-end to run off local XML data files, which are parsed at runtime via SimpleXML-based classes. Site reliability and page rendering speed have since increased remarkably. An additional bonus is that things are easier to maintain because you only have to edit an appropriate XML file once, not edit a CSV file and update/ rebuild/ horse-around with the DB. In fact the only thing I miss from MySQL is its sorting abilities. Of course you can also do it in PHP, but it requires more code and is much slower.
Meaningful site usage metrics
One of the goals of the 4020 upgrade was an ability to track and display finely-detailed usage statistics. Accessed via a password protected login form, I can now check traffic at any time without any CPanel or AWStat limitations.
Surprise-free navigation
A small but important point. Since bandwidth is neither infinite nor free, I have taken care to ensure visitors are never trapped into downloading things they don't want. Hence the large thumbnails, plenty of tool-tips, prompts, and warnings of how many kilobytes to expect whenever you click on a link. This is also why I avoid the one-at-a-time "peek-a-boo" slideshows (with their meaningless "next" or "back" buttons) you see on so many other photo websites.
Parasite repellant
The bane of anyone with original web content is that sooner or later someone will try to rip it off.
Outright theft is impossible to prevent of course, since in an online environment if you can see something then you can just as easily steal it. Luckily other forms of unauthorised usage, such as direct downloading (ie. "deep-linking" or "leeching") can at least be frustrated.
Thus all non-HTML content is htaccess protected. Similarly the web-app runs multiple checks at runtime, to ensure only calls from my server(s) are let through. (Yes I know it is easy to fake referrers. Thing is, every visitor through a hotlinker's site would also have to fake them — incredibly unlikely.)
The feedback form is spammer-proof. All recent images are water-marked. The content engine deploys a simple click-disable script, which at least warns users about copyright infringement. Finally, more sensitive information in the scripts are MD5 encrypted, so even if someone hacks the server, they'll have an amazingly intricate time trying to figure things out.
None of this 100% bullet-proof of course, but every little bit helps.
Why "4020"?
Initially it had something to do with "seeing beyond the ordinary" (ie. 40:20 vision). More importantly, "4020" is much easier than "nemeng" to spell over the phone to Keyboard Challenged Suits.
It was only years later that "4020" became popular as the name of a US Bluegrass Band, or as a possible solution to a weird Xbox Challenge (!)