Sorting photos

Larry Colen lrc at
Mon Jul 5 18:42:09 EDT 2010

On Jul 5, 2010, at 3:01 PM, paul stenquist wrote:

> On Jul 5, 2010, at 4:51 PM, Larry Colen wrote:
>> On Jul 5, 2010, at 12:58 PM, Doug Brewer wrote:
>>> Larry Colen wrote:
>>>> I did some family portraits yesterday, and am going through and sorting them out.  After making a pass to throw out all the ones that aren't perfectly, or even sufficiently in focus I wonder why I could buy a pocket camera, with a dinky embedded processor that'll find people's faces and focus on them, but I don't have something in lightroom to find people's faces and looking at edge sharpness (eyes, hair etc) rate how well focused that they are.
>>>> While I wouldn't want software to rate the artistic merits of a photo, software that would rate and sort photos by various technical criteria (focus, sharpness, exposure, ...)  would save me a lot of time in post processing. Sure, there are pathologic cases where you're deliberately goofing with sharpness or exposure, and there maybe some great photos that have some technical flaw, but which are still great, but for most of what I do, it would be a huge help.
>>>> --
>>>> Larry Colen lrc at sent from i4est
>>> Sorry, Larry, but a big part of being a photographer is learning how to edit.
>> A big part of being a photographer is knowing how to focus and set the exposure of your camera, how many pros do you think still shoot everything in full manual?  I'm not looking for something that'll edit everything for me, I'm looking for something that'll speed up one of the most time consuming tasks, taking a pass through the photos, pixel peeping to see which ones really are sharp enough to blow up. 
> "Sharp" is a judgement call. No photo is perfectly sharp. And what might be acceptably sharp for an action pic might not be acceptably shapt for a static, posed photo. And that's just the beginning. You gotta make your own calls. Software can't do that for you.
> Paul

"Sharp enough" is a judgement call.  Sharpness, on the other hand, should be quantifiable.  

For example, there is edge detection software that can detect the edges of various shapes. If the values change completely from one pixel to another, that would be "perfectly sharp". If we rate blurriness, as the inverse of sharpness as the width of the transition, we get increasingly blurry images:

 (just showing one dimension)
blurriness 0:
blurriness 2:
blurriness 4:
blurriness 6:

So, in some cases, you might set your threshold somewhere between 0&2, and in other cases between 4&6.  However, if I have a bunch of photos that are all nearly the same,  it would be handy to have the machine rate them in order of blurriness, so that I could then go and look at the sharpest couple of photos and see which ones I like the best, and spend a lot less time looking at the photos below, or well below, the threshold.

I can see why detecting edges in two dimensions would be a lot more work, much less measuring the edge thickness.  But when a $1,000 desktop computer has enough power that it would have given Seymour Cray a priapism, not that many years ago (assuming he was still alive anyways), a two or three pass process that finds the faces, finds the edges in the faces, and measures the blurriness of those edges shouldn't be an insurmountable problem.  As a photographer, what I do with that information is up to me.

Larry Colen lrc at sent from i4est

More information about the PDML mailing list