ISO 50, ND Filters and Diffraction Limits

[zigzag]

I have a 30D and often use AEB and blending for landscapes (using my EFS-S

10-22mm). I find myself often wanting longer exposures (e.g. those misty

sea-wave shots). I have read a bit about ISO 50 on photo.net (5D has it as an

option) but am unclear why it is not offered on most cameras (i.e. mine). My

questions are:<br>

1. Is the lower ISO 100, (200 on Nikons) lower limit due to the sensor noise

floor or due to the danger of blowing highlights (you mean I can't blow

highlights at ISO 3200?)<br>

2. With landscapes, wide angle, foreground subject my first option was to

increase f/stop. I understand I would be diffraction limited around f/14 for

16x20" prints: (<a

href="http://www.cambridgeincolour.com/tutorials/diffraction-photography.htm">Cambridge</a>

so maybe this is not the best option.<br>

3. So, the best option is to set ISO 100, use ND or GND filters, set the CF for

post exposure noise cancellation, MLU. IS there a noise 'sweet spot' for ISO

with dark frame subtraction? (Always lower = best is the rule I believe) i.e.

should I use an ND 4 or ND 2 if I can achieve the same extended shutter speed

with either. But is adding [cheap] Cokin P filters likely to degrade IQ more

than higher f/stops? I would maybe want to stack them but this is not a good

idea on wide angle lenses.<br>

4. Would my time be better spent worrying about composition than technical limits?

[colinsouthern]

My understanding of ISO 50 is that, electrically, it's just ISO 100 down-shifted 1 stop. So although it may give you an increased length of exposure, it comes at the expense of decreased dynamic range (albeit in linear gamma, so not as limiting as one might first think). At the end of the day it's just another marketing feature.

 

In the situations you're talking about I just LOVE my Singh-Ray Variable ND filter. For landscape (most often also involving water) I can dial in my desired F-Stop (usually F11) Dial in my desired shutter speed, and then just dial in the required amount of attenuation to give me a perfect exposure - best Signal to Noise ratio is always with the histogram at or slightly beyond the right-hand end.

 

If you're having DOF issues and find yourself diffraction limited, give some thoughts to the likes of Helicon Focus.

 

Hope this helps :)

[zigzag]

Wow! I didn't know those existed: <a href="http://www.singh-ray.com/varind.html">Singh Ray</a>. That looks like exactly what I want. Some questions on these: <br>Is the '"crossbar" effect' a common issue? <br>Is there likely to be an issue with vignetting using the standard mount to stack a second filter rather than the wide mount (no additional filter). i.e. is the wide mount a safer bet?<br><br>

 

Another reason to get the 40D: Helicon Focus. Using Live View and image magnification it should be possible (?) to select multiple focus points throughout the required DOF to get the required DOF. Useable for landscpaes as well as macro.

[gdanmitchell]

<p>A few things, in somewhat random order.

 

<p>I'm pretty sure that you'll be "defraction limited" well before f/14 on a crop body,

especially if you want crisp focus and you want to make a print as large as 16x20. On a

crop body you'll see (according to my own tests and according to others I've read)

defraction softness in the plane of focus at apertures much smaller than f/8. I'd be willing

to shoot a crop body at a smaller aperture if a) there was no other choice, and b) I did not

plan to make a very large print.

 

<p>Since you mentioned the 5D, I'll point out that defraction effects don't become

noticable until somewhat smaller apertures on full frame. When I moved to full frame I did

a set of tests on my 5D and each of my lenses (17-40mm f/4 L, 24-105mm f/4 IS L, 70-

200mm f/4 L, 50mm f/1.4) to find out their "personalities" on this camera - in particular

their "corner fall-off," corner sharpness, and defraction characteristics. What I found is

that f/11 is as sharp as and perhaps sharper than (depending on the situation) f/8 on FF

and that f/16 is barely different - I don't hesitate to shoot at f/16 at all on FF, and I'll even

do f/22 on rare occasions. (<a

href="http://www.gdanmitchell.com/2007/04/12/sharpness-and-aperture-selection-

on-full-frame-dslrs/">Some examples from the testing are posted here.</a> and <a

href="http://www.gdanmitchell.com/2007/08/23/from-here-to-infinity/">a sample

photo (as opposed to a lens test) is discussed here.</a>)

 

<p>Regarding ISO 50... My understanding is that this is not a "real" ISO setting, but

instead essentially post-processes a 100 ISO frame as you might in photoshop if you had

underexposed a bit. This means that there won't be any noise reduction advantage and

that there could be a dynamic range loss. I never use this setting.

 

<p>Rather than messing with ND Grad filters and holders, I use an alternative for wide

dynamic range landscape photographs. (Including the one in the second link above.) I

typically shoot two frames of the scene with one exposed for the darker areas (the

foreground in my example) and the other for the brighter areas (the peaks and sky in the

example). I then use one of the Russel Brown plugins to bring the two exposures into PS

as Smart Objects, thus allowing me to optimally adjust the two RAW images separately

before using masks to blend them in PS.

 

<p>I can't answer your question about optimal settings for low noise with dark frame

subtraction, though I definitely do use this for longer exposures. I don't think about it

much for exposures of a second or so, but for longer ones - sometimes measured in

minutes - I would always use it.

 

<p>Regarding the "composition versus technical limits" question... yes and no. It is

possible to worry more than necessary about things like noise. Noise is not necessarily a

totally bad thing: viewers generally won't even notice a little noise in a great print, in fact

they may even expect it. On the other hand, photography is a medium that is intrinsically

linked very closely to the technology used to create photographs, so mastering the

technology (while avoiding a situation where the technology becomes more important than

the image) is a good thing.

 

<p>Dan

[arie_vandervelden1]

1. I think most cameras don't have iso 50 because of the danger of blowing highlights - once the photo cells are saturated then they can bleed into adjacent cells with nasty results.<P>

2. With your 30D you get max sharpness around f/11, a little bit less at f/16. Beyond f/16 things start to degrade pretty quickly. I've found f/22 to be barely useable and f/32 is quite bad.<P>

3. In a pinch you can use a circular polarizer for a 2-stop reduction in light. The cheap Cokin P filters won't cause blur; rather, they may reduce contrast a tiny bit and cause a little bit of color cast. Not disastrous when shooting digital. Preferable to shooting f/22 IMO.<P>

4. You should try to get both composition AND technicals right. Screw up any one of the two and your photo is just a snapshot.

[gdanmitchell]

"With your 30D you get max sharpness around f/11..."

 

Actually, with almost all lenses it will be around f/8 on the crop sensor body.

 

Dan

[colinsouthern]

Hi Glen,

 

To answer your questions ...

 

"Is the '"crossbar" effect' a common issue?"

 

The only issue I've ever noticed is when I tried to operate it past 8 stops (ND8!) - at all other times it's fine.

 

"Is there likely to be an issue with vignetting using the standard mount to stack a second filter rather than the wide mount (no additional filter). i.e. is the wide mount a safer bet?"

 

I grabbed the standard version and haven't had any problems, but then again I only use 1.3 and 1.6 crop-factor cameras. If you're shooting wide-open on a FF camera I would expect it to be an issue, albeit a corractable one. If in doubt just get the slim one - it's only $50 more :)

 

Just to give you a "positioning statement" on the Variable ND ... I'm a picky/fussy bastard - I only buy the best (1D3 camera - only L series lenses - tripod that could prop up a car while you change a tyre etc) - and the quality of the Vari ND more than meets my standards - it's expensive, but it just "oozes" quality!

 

"Another reason to get the 40D: Helicon Focus. Using Live View and image magnification it should be possible (?) to select multiple focus points throughout the required DOF to get the required DOF. Useable for landscpaes as well as macro."

 

Yes, this should work well. I've actually written to Canon to ask if they would consider "Focus Bracketing" in the firmware for use in situation like this (got a favourable reply, but still waiting for the feature!). As a work-around you can also try to position the AF points on the important bits and then just select different AF points. You can also download a trial version of Helicon Focus (and the developers play an active part in the forums - you can get bug fixes in days, not years!).

 

Cheers,

 

Colin

[zigzag]

Thanks, with an optimal f/stop near to f/11 it seems I will be needing the ND filter more often than I realised. As I have GNDs already the standard mount vari-ND would be the one to get, followed by the Cokin P mount. Those Singh-Rays are expensive, time to try a few things in the field.

[zigzag]

Dan: Is the Russell Brown plugin for CS3 only? I see a couple of Smart Object tutorials from him: <a href="http://www.russellbrown.com/tips_tech.html">Russell</a>

[colinsouthern]

Glen Flower wrote: "Those Singh-Rays are expensive"

 

... and worth every cent :)

[aaron_fuqua]

ISO on digital cameras is basically just a signal gain control so there truely is a "sweet spot". Evidently Canon sets 0 gain at 100 ISO and Nikon at 200 ISO this is the reason that Nikon finally just got rid of the 100 ISO option with the new 3D and D300 (and made it a "boost" option to better indicate what it truly is).

 

Unlike film it won't help to turn the ISO lower than 100 ISO (200 ISO with Nikon) because it is simply turning the gain below optimum. It is not going to decrease noise it will just decrease the dynamic range.

[drfl]

I just have a 3-ND, but I have tried the Vari-ND and I recommend it (but not for more than about 7 or 8). As far as diffraction, I have battled the same as you with water. I have found on my 20d that I can print large (16x24) up to about f-16. For the most part a 3-5 ND will serve your purpose and keep costs down. I would experiment with that before purchasing the Vari-ND. It's always good to have one of those around anyway. Additionally, although they are not supposed to, Singh-Ray does have a slight ruddy color cast. It isn't too bad, and it can be nice at times. In this example, I set the camera to f-16, used a 3-stop Grad ND and a 3-stop ND. It allowed me to capture some motion in the water while not going too far into diffraction.

 

<a href="http://www.photo.net/photo/4648206">http://www.photo.net/photo/4648206</a>

[mike_davis]

Hi Glen,

 

I'd like to address only the second question of your original post.

 

The 30D sensor size is 15.0 x 22.5mm.

 

The Enlargement Factor to produce a 16x20-inch print with a nominal crop = 16-inch * 25.4mm/inch / 15.0 mm = 27.1x

 

The f-Number at which diffraction will inhibit a desired print resolution for a given enlargement factor can be calculated with this formula:

 

Diffraction-Limit N = 1 / desired print resolution / enlargement factor / 0.00135383

 

To prevent diffraction from inhibiting a very modest print resolution goal of only 2 lp/mm in a 16x20-inch print produced from a nominal crop of the 30D sensor:

 

Diffraction-Limit N = 1 / 2 / 27.1 / 0.00135383 = 13.6

 

f/13.6 is roughly the same as the f/14 obtained by the original poster from the CambridgeInColour diffraction calculator.

 

The thing to remember here is that not everyone has the same requirements for a desired print resolution and the print resolution goal can be relaxed as viewing distance is increased.

 

Many people would not find print resolutions of 2 lp/mm to be very satisfying at a viewing distance of 10 inches. A print resolution of at least 4 lp/mm would be found by most people to possess sufficient subject detail at 10 inches. Print resolutions as high or higher than 8 lp/mm can not be appreciated by most people. If you're not concerned about people viewing your prints any closer than 20 inches (instead of 10 inches) you can cut your desired print resolution in half.

 

If you learn nothing else from what I've written, I hope you (and the rest of the community) will grasp the fact that the f-Number at which diffraction becomes a problem varies with enlargement factor and your personal preference for print resolution, which itself can vary with anticipated viewing distance.

 

There is no one f-Number at which diffraction becomes a problem for any camera. You simply have to grab a four-function calculator and do the math for every combination of enlargement factor and desired print resolution you require. At the very least, for a given camera, you should end up calculating one Diffraction-Limit f-Number for every print size you anticipate producing. As print size decreases, your Diffraction-Limit f-Number will increase (allowing you to stop down for additional DoF (or a slower shutter speed) without inducing diffraction sufficient to inhibit your desired print resolution.)

 

Mike Davis

http://www.AccessZ.com