Help: How to determine if a photo capture was sharp

[jwstulin]

One problem with digital photography is that we are always tempted to keep “drilling down” into a photograph on the screen to examine the amount of detail actually captured by our cameras. This is guaranteed to lead to frustration since, regardless of how perfectly sharp the photograph is, we will eventually drill down to a level where the photo will no longer look sharp.

What is your rule-of-thumb in examining photos to determine if a photograph is as sharp as it could be, or if there was some problem with the image capture (e.g. bad camera, bad lens, or, most likely,  poor photo technique by the photographer).

Thanks,

Jeff Stulin

[Schewe]

Sharpen the image at 100% and evaluate at 50%. The basic problem is that display is a low resolution device...if you are printing at 360PPI and viewing the image at 100%, an LCD whose PPI output is about 100PPI will be 3.6 x as large. But when zooming out to get a closer representation of the size, the display is low resolution. The way to evaluate the image sharpness for a print, is make a print.

[JohnTodd]

If my maths are correct, an iPad3 has a resolution of about 270dpi, so theoretically, that would make a reasonable 100% test device?

[Schewe]

Except, in order to see it on an iPad, you have to import the image. If you do so with iTunes, it reprocesses the size of the image (unless you already sized it to the iPad screen display). And iTunes processing also adds a touch of blur. So, no, the iPad (even the iPad HD) would be useless for seeing your original image.

[jwstulin]

I know how to determine if a photo is "good enough" to print at a certain size. What I don't know is
how I can evaluate a photo to determine if it is the techically best that my camera & lens can deliver.
How can I tell if it is "tack sharp" if I can always zoom in enough so that it is a blur?
For a "perfect" photo what should be your examination criteria when the photo is displayed at full resolution?
Or should the evaluation occur at more than full resolution?

I want to know this so that I can better evaluate both my equipment and my image capture technique.

Jeff Stulin

[Schewe]

For a "perfect" photo what should be your examination criteria when the photo is displayed at full resolution?

1:1 zoom...that's the only zoom that will show one image pixel for each screen pixel. Anything else is SciFi (and impacted by the display dither).

[joofa]

Use JIDM for an objective measure of image sharpness. Free download from the website in my signature. For an example usage for determining the sharpest f-stop by a different person please see this:

http://forums.dpreview.com/forums/read.asp?forum=1018&message=38205278

Joofa

[Mike Sellers]

Capture One has a feature that will paint the sharpest areas of the image green. I sometimes vary my focus on the same image then use C1 to see which one has the best focus. You can demo C1.
Mike

[BJL]

If my maths are correct, an iPad3 has a resolution of about 270dpi, so theoretically, that would make a reasonable 100% test device?

Maybe this would work well if you import raw files and convert on the iPad with software like PhotoRaw (by our sandymc!) or piRAWnha?

But how about this: if, for example, your screen is 100ppi and you plan to print at 300ppi, display the image at 100% pixels on screen, and discipline yourself to view from three times the expected print viewing distance. Or a bit less, like twice the anticipated print viewing distance, to allow for some viewers having sharper eyes than yours. I find it fairly natural to view a big screen from about two feet away, more than twice my close print scrutiny distance.


P. S. A minor point: I get 264ppi for the new iPad, confirmed at http://www.apple.com/ipad/specs/

[jwstulin]

I am not so much interesting in determining which part of the photo is the most sharp. I am
more interested in being able to put the photo on the screen (Macintosh), performing a quick visual evaluation,
and then determining if it is technically the best possible. Mostly as a way to self evaluate my photo capture technique.

What do you do to quickly look at a photo and determine if it is technically good?

Jeff Stulin

[PierreVandevenne]

The question in itself is a bit vague. Talking about sharpness and to keep things simple, you have to worry about focus, vibrations and subject movement.

focus: if you are reasonably close to where you want to focus, in a portrait for example, you will obviously have zones in perfect focus (say, the eyes) and zones slightly out of focus (say the nose). What does perfect focus  ( the maximum sharpness you can achieve for your system all other things being perfect) mean? Something will be in perfect focus, but maybe not what you intended. The key thing is to put your focus exactly where you want it and that's both potentially difficult and a creative decision. In landscapes, that's also a creative decision: what you want to emphasize and so on.

vibrations: typically, this would be tripod with mirror lockup in addition to perfect focus point placement. That should give you a reference point to which you can compare images taken in other situations.

There are mathematical methods that estimate the sharpness of an image, but there is no commonly used absolute "sharpness unit". The discrimination of lines on targets and how they "rise" from the background is commonly used but hardly applicable to real world images. Another method is to measure the MTF or eventutally the FWHM of point sources. Again, hardly applicable on real world images (and the MTF doesn't always correlate perfectly with the FWHM). Even in well calibrated scientific instrument one will move the focus around some definition of what the MTF should look like. MTF-50 is commonly used as it correlates well with our perception. Visually, minimizing the FWHM by adjusting focus around the optimum works well.

But, in practical photographic terms, what I do is shoot a few images as perfectly as I can (tripod, MLU, careful focus, etc...) and grow a feeling on what the best performance given a camera lens combo I can aim for is.

[jwstulin]

Pierre:

That is exactly what I am trying to understand. When you view your photo on screen what do you to to determine if it is sharp? Do you examine it at 100% and look at a part of the image with lots of detail (say a pile of small rocks)? Or you you zoom in a certain amount and examine the structure of the pixels?

Jeff Stulin

[PierreVandevenne]

I usually use 1:1 to find the sharpest shot in a serie. Can accasionally zoom to 200% if I feel the urge to pixelpeep. But I think this is personal. That shot is taken with the 300 - 1.4 TC (420mm) - MK II - ISO 400 - hand-held at 1/1250 sec 4.5 The last zoom level is at 200% below and looking at the hair, I can't do much better than that in that setup. How do I know it? By experience...

A mathematical method looking at the whole picture would tell me a picture focused on the background is on the whole sharper than the one focused on the scene. That's why joofa and all were suggesting the patch method.

[bwana]

Except, in order to see it on an iPad, you have to import the image. If you do so with iTunes, it reprocesses the size of the image (unless you already sized it to the iPad screen display). And iTunes processing also adds a touch of blur. So, no, the iPad (even the iPad HD) would be useless for seeing your original image.

i wonder how i can learn these sorts of interesting details? Would the ipad HD still do this if I am viewing an image directly from dropbox? This avoids the iTunes detour.

[hjulenissen]

It would be interesting to compare "active" focus points to image file. How sharp are transitions close to the selected AF points? How steep are edges compared to the steepest found in the library for this lense/sensor/settings? Are there any transitions far from these points that are sharper?

Anyone used to using focus peaking on video cameras? Might be nice to have such an overlay when browsing image in camera (for deleting poor ones).

-h

[Schewe]

Anyone used to using focus peaking on video cameras? Might be nice to have such an overlay when browsing image in camera (for deleting poor ones).

Yes it would...I think both Hasselblad and Phase One have the ability to peaking... I think it's the basis of C1's Focus Masking.