No Noise is Good Noise

Rainbow Falls, Watkins Glen SP, NY

I just returned from a very short visit to New York’s Finger Lakes region (I had a doctor’s appointment down there and it seemed like a good opportunity to visit a few of the outstanding state parks in the area).  We managed to “hit” Taughannock Falls, Buttermilk Falls, Robert Treman, and Watkins Glen state parks.  In case I haven’t mentioned it recently, these places are under-advertised and under-appreciated within the state.  On any given day, you’re more likely to see people who don’t live in NY than people who do.  That’s too bad, because some of them are real gems.

But I digress.  What I really wanted to comment on concerns one of the strengths, I think, of the Canon 5D Mark II.  And that is it’s performance at high ISO’s.  I am genuinely impressed, to put it mildly.

The image above was taken at one of my favorite spots in Watkins Glen SP called Rainbow Falls.   I’ve posted images of it previously as well, so forgive me if this looks very familiar.

On this particular visit, however, I was unable to use my tripod (I need an L – bracket that fits the battery grip for the 5D MK II  – the bracket on the old 5D doesn’t fit).  Anyway, it meant that everything would be “handheld”.  Which further meant that I needed exposure times that were at least 1/20 second for the lens I was using.  Well, for spots like this one that include deep shadows, the only hope was to use a high ISO.

For this image, that meant going to ISO 3200.  On my old 5D, I’d tried 3200 a couple of times but always had to deal with at least a little noise, especially in the shadows.  So I was a little nervous about pushing it that much, but I really had no choice.

I can find no noise in this photograph.  None.  I’ve looked at it at the pixel level and I’ve printed it.  There just isn’t any there.  As I said, I’m really, really impressed.

Actually, I’ve made very few adjustments to this image.  I added a little “fill light” to unblock some of the deep shadows and I pushed the “clarity” slider just a bit, but little else.  I made no contrast adjustments, no luminosity adjustments, and no saturation adjustments.  Fundamentally, it’s what came out of the camera.

Have I mentioned that I’m impressed?

Toys I have that make me Crazy……

Drawbridge, Mystic River, Mystic, CT

One of the things that intrigued me about my new Canon 5D mark II camera was it’s ability to create HD video.  I wasn’t sure how much I’d use it, but it seemed like a nice feature to have.  I’d looked at a lot of the videos made with this camera on the web and was suitably impressed.  And since I have grandsons (who are currently playing football) and since we love to travel, I thought I’d probably have opportunities to try it out.

I should point out that long ago I owned a pretty good camcorder (I used it mostly for my daughter’s basketball and soccer games when she was in high school).  That camera used VHS tape as the recording medium.  It was kind of heavy, but it did a reasonably good job.  Most importantly, playback was simple – just pop the tape into a VCR and off you went.  Well, that was then and this is now.

Early last week I made it a point to learn how to use the video recording function on the 5D MK II.  Technically, that’s not hard to do.  You have to activate “Live View” and do a few other things in the menus, but that only takes a few minutes.  Push a couple more buttons and the thing is recording HD video.  Then you start to realize how much you don’t know about this process.

For all intents and purposes, focus needs to be done manually.  There are autofocus options, but they seem to me to be a little slow.  One thing you have to realize is that during Live View mode the mirror is in the “Up” position.  So focus isn’t achieved in quite the same way.  So if you leave things up to the camera, stuff can go out of focus pretty quickly if it’s moving (like kids running).  Especially since it seems to want to shoot with the lens wide open.

A recent firmware update has improved that situation, however.  You can now set exposure manually by setting the camera to – you guessed it – “M”.  If you do that, setting ISO, aperture, and shutter speed is straightforward.  So you could set the lens to f/8 or f/11 and DOF will improve significantly (if that’s what you want).

But the biggest problem, in my opinion, is playback and editing.  And that’s something I honestly didn’t expect.  The camera produces files that are in MOV format (Apple Quicktime).  You can’t change this format – MOV is the only option.  If you have a PC (like me), you soon learn that these files don’t play all that well on Quicktime using a Windows platform.  They’re jerky and they stutter.  From what I’ve read, part of the reason is that Quicktime was never fully optimized for use on Windows systems.  In most cases it works just fine, but for some files – like these – it chokes a little bit.

You can use Canon’s Zoombrowser (supplied with the camera), and that plays these files just fine.  But that software doesn’t allow you to do much else in terms of editing or changing formats.  You could also, of course, hook the camera up to an HD TV and watch the raw video that way.  Again, though, it would be unedited.

So what you really need to do is find software that will play the raw video and allow you to transcode it (turn it into a file format that Windows “likes” and that Windows based editing software will recognize).  One that looks very promising is called Neoscene.  Many Windows users who own 5D Mark II’s recommend it highly, although I haven’t tried it yet.  But I’m about to.

Damn, I hate being an absolute “Newbie”.  But that’s what I am as far as this stuff goes.  Maybe if I figure it out, I’ll actually post some video.  We’ll see. 

Oh – if anyone reading this knows anything about 5D Mark II video, feel free to set me straight!  I promise I won’t be offended.

M9 Madness?

Baptist Hill Church, Mystic, CT

On a personal note, I have to say I’m loving my new 5D Mark II.  As I said before, the color depth seems to me to be outstanding (which helps in B and W conversions).  More importantly, the dynamic range of this camera is clearly better than that of my old 5D.  I haven’t yet done any side by side comparisons, but I can say that scenes that used to cause me problems – especially with highlight blowouts – are now much more manageable.  For whatever reason, the “Recovery” slider in Lightroom’s RAW converter seems to ”work”  in more instances than it did with 5D images.  And the 5D was good in terms of dynamic range potential.  The Mark II is just better.

Maybe it’s because I just bought a new camera that I’ve noticed all the hoopla over the new Leica M9.  As photographic gear frenzies go, this one appears to be right up there in terms of intensity.  Even though the darned thing is going to cost around $7000 (U. S.), everybody seems to want one.  As all the positive reviews continue to come in, I’m guessing the demand will go up exponentially.

Now, I have to say that I personally have no interest in the M9 or any other Leica camera.  I also have to admit that I don’t know much about rangefinders in general.  I know a little about how they work, but I’ve never used one.  I’ve always been an “SLR” person.  But this isn’t about rangefinders versus SLR’s.

What really got my attention yesterday was a post by Michael Reichmann on The Luminous Landscape.  In this 18 minute video, Michael Reichmann describes a visit to the Leica factory, including partial descriptions of the manufacturing and testing processes.  As someone who has spent the better part of 35 years working in manufacturing and product testing, this video raised a number of red flags in my mind.  If I were a wealthy person and thinking I might want one of these new toys, what I saw might well have changed my mind.

Why?  Well, the potential problem (and I strongly emphasize the word “potential”) is actually summarized in one of Mr. Reichmann’s statements that introduced the video: 

“It’s also clear that demand for the M9 is turning out to be very high, and so it may be some time until the initial feeding frenzy can be satisfied. When you’ve watched my video of what’s involved in final assembly and testing you’ll understand why. No robots, no mass production – just careful and therefore slow hand work and meticulous testing. No wonder Leicas cost what they do.” 

It was, in fact, this statement that induced me to watch the video.  I watched it not because I had an interest in the camera, but because I was professionally interested in any insight into how Leica produces one of their top – of – the – line products.  In particular, I was interested in the implication that both the manufacturing and testing processes were manually intensive.  Most people would probably say that that was a good thing, that relying on highly skilled craftsman is better than relying on automated, computer driven processes.  When someone advertises that a product is “handmade”, the implication is that it’s going to cost more because the quality will automatically be superior.

Well, the part about costing more is certainly true.  But the notion of “superior quality” is a bit of a myth.  More often than not, manually intensive processes yield lower quality products.

How do I know this?  For any given product or process, I don’t know that product quality will be inferior if it’s a manually intensive operation.  It may very well be that the M9 is, in terms of quality and reliability, the best camera in the world.  The only way I (or anyone else) could make that judgement would be to study the process close-up.  And that, of course, isn’t likely to happen.

But I can say that I’ve studied many such “manually intensive” processes over the years.  I can also say that if the process includes manual (or visual) inspection at the end, outgoing product quality is almost always marginal.  In some cases, it’s actually flat out unacceptable.  As a general rule, people are very bad at visual inspection.  To put it more succinctly, we suck at it.

Here’s some “hard” evidence:  When people who do visual inspection are given product of known quality (that is, they are fed “pieces” that are either not defective at all or have known defects on them), the average for culling out the defective pieces is about 60%.  If the test, for example, involves 100 pieces, and 20 of them have “major” defects, the average inspector will find 12 of them.  The remaining 8 defective pieces will pass through the sytem (and potentially to the consumer).  In case you haven’t thought of it, there’s also the other side of the coin – some inspectors will “reject” pieces that are not defective.  A perfectly good piece, in other words, will be thrown away or sent back for “rework”.  And that means more production costs.

In my career, I’ve done dozens of these kinds of tests and they always come out the same.  The average is always about 60%.  Some people are a little better and many are worse.  It’s also true that people tend to be inconsistent.  Inspectors are not as effective at the end of their shift as they are at the beginning.  They get tired, or start thinking about what they’re going to do when they get out of work, or maybe they don’t feel very well on a particular day.  And, by the way, inspectors are usually at the lowest end of the company’s pay scale.  I’ve never figured that one out.

The lesson in all of this is simple.  Automation trumps “manually intensive” operations every time.  When maintained properly, an automated system works faster, never gets tired, and most importantly, never makes mistakes. 

Again, I have no idea if any of this affects Leica.  I only know what I saw in Michael Reichmann’s video.  But what I saw looked an awful lot like processes I have seen up close and personal.  So if it were me looking to buy one of these cameras, I’d be nervous.  Think about it:  What if some of what I’ve alluded to does affect this process?  And what if demand does increase significantly?  Leica will either have to spend less time on each camera or hire more operators.  Either way, the chance of an error (on your camera) goes up.

Any potential M9 buyers out there?