Camera and Lens tests

[tom mann]

So over the last few months I've been researching cameras.  I have a lot of requirements, but in regard to model train photography I wanted to see if I could improve on two things:

1.  Take photos that have more perspective (vanishing sight lines). I currently use a 50mm macro on a 1.5x crop sensor, so the 75mm end result ends up making things look slightly isometric.
2.  Get better depth of field without diffraction. Some of my photos (especially of Z scale items) use Helicon to focus stack 12 images at f8, but this is tedious to do every time. So most of my photos are at f16 or f22, which is ok when resized for the web but not optimal for print. 

I ended up getting the Sony RX1R which isn't the ideal product-photography camera.  However it has some advantages.

So with my current set up, taking a photo at the lens' sweet spot (f8) results in the following image.  Nice and sharp only where I focused:



The cab number is a crop at 100%.

Changing to f22, I can focus in the middle of the model to get the whole model in focus:



But look at how the cab number is now slightly blurry, as a result of diffraction and the fact that I moved the focus plane.  Also notice that the hight of the long hood doesn't get smaller the further away it gets from the vantage point. This is the isometric effect that bugs me (recall the Arnold SW 1 images that the manufacturer provided compared to the photographs of Ken Patterson).

So in looking at cameras I learned that diffraction effects are not just based on the smaller apertures, but the pixel size of the sensor.  See: http://www.cambridgeincolour.com/tutorials/diffraction-photography.htm

So a larger sensor with a larger pixel size will allow you to have a smaller aperture (and greater depth of field) before the resolution is adversely affected.

to be continued...

[peteski]

Using shorter focal distance lens (for example 24mm instead of 50mm) will give you more exaggerated perspective (but also wider angle), if that's what you're after.  That more exaggerated perspective in model photos will make them look more like the 1:1 items (because IMO the human eye associates the vanishing sight lines with larger objects, especially if most of the photo is in focus).

[tehachapifan]

Nice weathering on that unit!

[tom mann]

So based on my research, what I believe would be the optimal model train camera would be composed of:

1. A full frame sensor, with a moderate pixel count to avoid diffraction issues.  But a high enough pixel count to allow for cropping and 300dpi printing.
2. A lens with <1ft focusing distance at ~35mm. This would be good enough unless 1:1 macro images are needed (and they aren't, really). The 35mm field of view would allow for perspectives similar to real life train photography.
2a. Bonus: an extra 50mm macro lens (for product-style photography).
2b. Extra bonus: a 35mm macro lens (which I haven't seen for full frame systems).
3. In-camera automated focus stacking for extreme depth of field (when f22 isn't enough).

Since this would be my only camera, another 'want' comes to mind: The smaller, the better.

[wazzou]

I thought you had the same setup I do. 
Nikon DX with 60mm f2.8 Macro.  I currently have the D7100, but still have my D90 as well.

[tom mann]

So I can ramble on but let me cut to the chase: I bought a Sony RX1R. It's a 35mm fixed lens camera, and it has a mechanical slider that moves the lens out a few mm like an extension tube would. The result is not quite true 1:1 macro, but a close distance focus of .2m.

If I swap cameras, the scene now looks like this:



However, the 24MP sensor gives me plenty of headroom to crop, so I can get this:



Note that the 100% crop of the numbers is still sharp at f22.  The image is still big enough to print at 11" @300 dpi. It's almost the same view as above, but there is a little more perspective.

But getting lower and closer allows for the "real life" perspective (this is what I've been wanting):



Again, this is at f22 with focus on the right-most louver by the cab.  The extreme front and rear are slightly out of focus, but it's maybe noticeable only on the handrail. I could probably fiddle with and focus on the 2 in the road number to improve it.

It took a lot of research to get here, and it would have been nice to test things out in a camera shop. 

[Chris333]

Sells models to buy new camera.




Has nothing left to take photos of.

 

[Cajonpassfan]

No kidding, isn't that like three grand?!
Pics are awesome though...
Otto

[tehachapifan]

Tom, what are you using for your base/background? I really like how it looks. Also interested to see how your new camera handles N scale models.

[tom mann]

No kidding, isn't that like three grand?!
Pics are awesome though...
Otto

There is a huge used market for this - look on ebay and you pretty much have you pick for less than half the price.  The one I got was used for 5000 photos and is in good shape cosmetically.

There are other options - but I'm convinced of the importance of the full frame and that will put you above $1000 right now.  Another thing full frame gets you is the ability to shorten the exposure time due to the fact that the sensor is absorbing more light.  In my tests, I can go from 1.6s to 1s.

Tom, what are you using for your base/background? I really like how it looks. Also interested to see how your new camera handles N scale models.

I bought a light tent and backdrop from Table Top Studio a few years ago. 

I've been testing with n scale and either more cropping is needed (resulting in a smaller image) or a lot of perspective distortion is visible (when photographed at .2m closest distance). I'll get some samples up next.

[Ed Kapuscinski]

Ok, so these look like good product or "look at my model" photos. But what about putting it in a setting.

How does this work out on the Hinshaw Valley?

[nscaleSPF2]

Tom, I am really glad that you found a camera that works for you.  This serves as a benchmark for the rest of us: If to want to maximize the detail in a photo, the way to go is a full frame sensor (i.e. one that is the same size as a 35mm slide, 1.5x1.0 inches) and combine this with an outstanding lens.  It is hard to beat Zeiss optics.

As far as your subject, I am blown away by how real the locomotive looks.  I have a hard time convincing myself that this is N scale.  The size of the coupler doesn't even give it away.

If I may offer a couple of additional comments...

As far as perspective, as Peter indicated, a shorter focal length lens decreases the "isometric" nature of the photo and will add to a more realistic perception of the locomotive.  However, when you cropped the first RX1R shot, you increased the effective focal length of the lens.  I would guesstimate that the cropped version resulted in an effective focal length of about 60 to 70mm.  That's why, perspective wise, it looks a lot like the shots taken with your A580.  If you can move the camera closer to the subject, the perspective will be even better (here's hoping you can get the same depth of field).

Ed's comment is a good one, too.  If your results are giving you a depth of field of about 3 to 4 inches, then that is the same amount of the layout that will be in focus.  Pardon me for stating the obvious.  The only way to get more depth of field is to go with a smaller sensor, which is not going to give you the kind of spectacular results that have, so far.  Unless you use the RX1R with focus stacking.

These comments are meant to be positive, and I hope that they help the rest of us understand these photo issues better.

[nscaleSPF2]

Oh, one other lesson we can learn from Tom:  If the results don't give you the photo you want, then experiment with a different "angle".  As the last photo above shows.

[tom mann]

Here is a shot on the Hinshaw Valley as requested by @Ed Kapuscinski .



The angle isn't as low as I would like because the ground slopes up where the camera was resting.

[peteski]

As far as your subject, I am blown away by how real the locomotive looks.  I have a hard time convincing myself that this is N scale.  The size of the coupler doesn't even give it away.

That's what I thought at first (silly me, you can't really have that level of detail, and such thin realistic handrails in N scale) but Tom's later comment made me realize that DUH, this is a larger scale loco (H0 or even larger). Tom is a multi-scale guy.