The Anatomy of a Prank

I recently wrote a post describing a new camera I (along with 3 other photographers) have been testing.  Naturally, I've had a lot of questions about the camera.  So with that in mind, I'd like to write this follow up post.

First off, the camera doesn't exist.

I made it up.

Sorry.  Don't hate me.

Here's how it went down:

I decided I wanted to do a photography-related April Fools joke.  However, I've never been a fan of ridiculous 'jokey' pranks like this or the way-over-the-top stuff like this.  No; I wanted it to be semi-believable.  Something that would make the reader say, "That's cool and I can absolutely see it happening, but it does seem a bit out there".  And no one was going to believe me if I said that I'd been testing a new version of Kodachrome or a digital back for 35mm cameras, or whatnot.

I began to think about what people might really want in a digital camera; especially those who shoot film.  What, if anything, would make them consider abandoning film?  Do people who shoot film do so strictly for the 'look', or is there something else?  If there were a digital camera that perfectly emulated the look and feel of film (including the dynamic range), would people be tempted to switch?

The answer, it seems, is: it depends.

While a few people immediately caught on to the joke, some people seemed genuinely interested in the camera.  Frankly, I would be too.  One of the major 'features' of my imaginary camera was its ability to compress a wide dynamic range into a small range of tones on the sensor, which would then be expanded into a final image in a computer.  This is more-or-less how negative film is able to capture such a huge scenic dynamic range, and the linear response of current digital sensors is one of the key reasons that they still cannot match the DR of modern negative films.

Then there was the issue of making the output look like film.  That's when I came up with the idea of a digital camera that mimics film through the (very real) process of convolution.  I knew about convolution through my experiences in the pro audio world (where it is already routinely used to emulate halls and rooms in reverberation algorithms), and through my day job, where we use the inverse process (deconvolution) to improve the resolution and acuity of images from fluorescence microscopy.  Deconvolution is also used in photography (Photoshop's 'Smart Sharpen' is based on deconvolution, for example).

Frankly, I think this is a pretty damn good idea.  Shoot some film under a set of known conditions.  The response of that film is a known quantity that could, in theory, be mathematically modelled.  Then, apply that to the output of a camera sensor with a known output, and voila.  Frankly, I'm amazed that no camera company has ever thought of this.  Or maybe they have, and it's a ridiculously impractical idea.  Who knows?

Anyway, once I had those two ideas, the post basically wrote itself.  It was absolutely critical that I not make the camera too over-the-top flawless.  Hence the caveats about the requirement for a powerful computer, the limited availability of film stocks, and the 'just ok' high ISO performance.  Again, believable, but kinda crazy.  That was the goal.

In retrospect, two things almost killed the prank before it got off the ground: first, saying that 3 other photographers were testing the camera naturally raised the question of where their previews were.  I have to admit that I totally overlooked this.  On the plus side, Johnny Patience, who was one of the first people to immediately recognize this as a joke, volunteered to go along with the plan, and confirmed the accuracy of my review.  This was a huge step for the joke, because Johnny has a much larger following than I do, and thus lent the whole story far more legitimacy.

The second thing that almost killed the prank was my statement that I would consider buying this camera and giving up film.  I wouldn't do that.  Ever.  To his credit, Tom Welland picked up on this (since Tom has actually met me and knows my affinity for the process of shooting film, rather than just the end result).

Anyway, I am sorry to disappoint you all.  I hope you had fun with this and you won't hate me.  It was interesting to read your questions and comments, and hopefully it got everybody thinking a bit about why they choose to shoot film (or not), and what they truly value in a camera and in photography.

Happy April!



The Coming Revolution

I have news.

Big news. 

News that I can't discuss in full - yet.

For the last five years or so, I've slowly transitioned away from digital photography, and embraced film almost completely.  Yes, I still own a couple of digital cameras (a Nikon D800 and a Leica M-E), but these get comparatively little use.  For me, film is 'it'.  I've produced several tutorials on shooting, developing, and scanning film, along with a popular YouTube video on scanning colour negative films.  Let there be no mistake: I #believeinfilm.

So it was somewhat odd when, about 2 months ago, a rather large camera company (you've heard of them and likely used their products) approached me with an offer to try out a new digital camera they were developing.

"You know I'm primarily a film photographer, right?", I told them straight away. 

"Yes.  That's exactly why we want you to test the camera.  This camera is not like anything you've ever shot with.  The images look and feel like film, and offer very similar qualities to what you're used to with the best negative films." 

I'm not sure how or why they chose me (I'm aware of at least 3 other photographers trying out this camera, all of whom are far more widely known than I am), but choose me they did.

Well, to say I was intrigued (and flattered) would be an enormous understatement.  I immediately agreed to test the camera, under a strict NDA, of course.  Once the paperwork was signed, they overnight'ed me the camera, a couple of batteries, some software on a USB drive, and two lenses.  The instructions I was given were simple: shoot whatever you want, for one month, and then send the camera back with your comments and critiques about the experience.  I was free to write whatever I wanted, although for obvious reasons I was not permitted to discuss the name of the company or the finer details of camera itself (most of which I don't know about, anyway).

But I was permitted to discuss the camera's gestalt.  So I'm going to do that.  Now.

This camera is absolutely, without question, the best digital camera I've ever used.  And it's not even really close.  This isn't just kinda like shooting film.  This is exactly like shooting film, but, of course, without the film.  I always said that even if a company brought out a camera that produced files that look exactly like Portra 400 or Provia 100, I'd still keep shooting film.

Well, let me tell you: that day has come, and I am seriously questioning that logic.  This camera is that good.

So, what *can* I say about it?

  1. Dynamic range is enormous.  Like, Portra 400 enormous.  I don't have the capacity to measure these types of things, but based on tests I've seen on other sites, I'd wager this camera offers upwards of 16 stops of DR, with most of that coming above middle grey.  Yes, this is the first digital camera I've ever used that actually has film-like highlights. Gentle transitions to white, and very recoverable top end.  Basically, you can shoot this camera as you would with any negative film (expose for the shadows, for example), with no concern.
  2. Regarding point 1, apparently the company has achieved this through two key processes.  First, a totally new sensor design (I *can* tell you that the sensor has quite a high MP count).  Second, a response curve that much more closely resembles that of negative film, whereby the dynamic range of the scene is compressed into a smaller density range on the sensor.  The result of this is a very low contrast Raw file, which requires a good degree of post-processing to get good results.  Fortunately, the hardware company has partnered with the right software company to make this a trivial matter.
  3. The look of the processed images is out of this world.  I've used VSCO, Alien Skin, DxO Film, and all the rest.  NONE of them look remotely like this.  This is digital film.  And not just one kind; the aforementioned software folks have modeled a number of modern films, and the results are CRAZY good.  How have they done this?  Convolution.  Remember that word; you're going to be hearing a lot of that word in the next 2-3 years.  Essentially, the software folks have produced characteristic "impulse responses" for each film stock; that is, mathematically modeled how a given film performs under a given set of conditions.  Once that model is produced, the output of the new sensor - a known quantity - can be "convolved" with the impulse response, to generate a final file that looks nearly identical to the modeled film.  This is crazy stuff, folks.  The 'convolution revolution' is here.
  4. The upshot of all of this is that the camera itself does not have the compute power to do this in anything close to real time.  There is an LCD, but the image you see is a rough approximation (a good approximation, but an approximation nevertheless).  The real business end of this camera system is in your computer.  Post processing is an integral part of the process, making this very much a "pro" level camera.
  5. At launch, there will be a reasonable selection of impulse responses made available with the camera (I'm told 7-10 films), with more in production, along with a user tool to generate your own (and, apparently, a forum for sharing user impulse responses).  I had the opportunity to test 4 - two colour, two black and white (I can't say which films, of course) - and they blew my mind.  I tested them against some actual film of the same variety, scanned the film, and then had my wife give the files a random name for a blind test.  Well, I honestly could not tell which was digital and which was film.  That's the first time that has ever happened.  In particular, I was impressed with the B&W models, since B&W film is, IMHO, enormously different than 'classical' digital B&W.  And before you ask whether your favourite everyday "be there" B&W film is modeled (you know which one I'm talking about), the answer is 'yes', and it's really, really good.
  6. The final image quality is outstanding.   Resolving power is terrific, and of course dynamic range is ridiculous.  High ISO performance is good, although this kinda misses the point of the camera.  This is designed to look and feel like film...including at high ISO.  Fortunately, you can (if you wish) set the camera up to give a more traditional digital look at high ISO, in which case the camera is very good.  Not the best I've seen, but still excellent.
  7. The camera does not have a reflex mirror, has a lens mount you've likely used before, and offers a very short flange distance.  And you wont need to do any 'equivalent' focal length conversions.  Oh, and image composition is what I would term 'traditional, with a modern twist'.  Make of all of that what you will.
  8. The design of the camera is in keeping with the current trend toward 'retro' design (although I wouldn't call it that with this company).  There are very few external controls, and the controls that are there are very well thought out.  Build quality is "pro" in every way.

Sadly, after a month of shooting, I had to send the camera back, along with my critiques (which, to be clear, were very minor).  Will I be buying it when it comes out (I'm told to expect an announcement next month, with release mid year)?  I don't know.  Frankly, I'm scared.  I've got a LOT invested in film and film cameras.  But this just might be the system that finally convinces me to go 100% digital.

I have zero information regarding price, so no need to ask.  Judging by the history of this company, however, I expect the price to be reasonable, though not dirt cheap.  This is some serious camera, and the price will reflect that, but it won't (I hope) be astronomical.

I'm very, very excited for the coming revolution.

EDIT: Last night I found out that I could post one low-res image from the camera for the purposes of this blog.  I initially posted this photograph of my daughter on Twitter, and said it had been shot on Kodak 250D motion picture film (which I was developing at the time).  I lied.  It's this camera.

Pretty good, ain't it?