bleirer wrote:
Very similar to Perry's. Great minds.... He adds shutter speed awareness or maybe I glossed over that part of your laundry list.
I'm more of an IS/VR kind of guy nowadays, being so technology forward
coupled with good shooting technique.
DirtFarmer wrote:
I find a sharp edge in the photo. I then take the photo and load it into a numeric array in Python using luminance values. Find the sharp edge in the photo and select a sequence of points that goes over that edge. I then take the first derivative of the numbers going across the edge which should yield a peak (could be either positive or negative). I fit the peak to a Gaussian curve to find the width of the curve. A sharp edge will have a curve with a narrow width. Tack sharp would be a width close to 1.0 (the pixel spacing).
I find a sharp edge in the photo. I then take the ... (
show quote)
I'm gonna do what Dirtfarmer does... huh?
On second reading, I more or less grasp the concept of your technique. It could prove useful for an objective measure of sharpness.
Human perception can detect sharpness, too. I rely on this organic approach.
DirtFarmer wrote:
I find a sharp edge in the photo. I then take the photo and load it into a numeric array in Python using luminance values. Find the sharp edge in the photo and select a sequence of points that goes over that edge. I then take the first derivative of the numbers going across the edge which should yield a peak (could be either positive or negative). I fit the peak to a Gaussian curve to find the width of the curve. A sharp edge will have a curve with a narrow width. Tack sharp would be a width close to 1.0 (the pixel spacing).
I find a sharp edge in the photo. I then take the ... (
show quote)
DirtFarmer
Loc: Escaped from the NYC area, back to MA
I see you use your method to determine sharpness for technical reasons, apart from other considerations, such as shutter speed and mirror slap.
DirtFarmer
Loc: Escaped from the NYC area, back to MA
anotherview wrote:
I see you use your method to determine sharpness for technical reasons, apart from other considerations, such as shutter speed and mirror slap.
Yes. Of course the test images were taken with a static subject, camera on a tripod and shutter delay 2 seconds.
It was a fun project and an excuse to get more proficiency in Python. Got a pretty good graphing library out of it. (The graphing routines got started in Fortran around 1963, got translated into Pascal in the '80s and into Python this year).
I wouldnโt want to sit on one, but why settle for โtack sharpโ when a razor is even sharper?
bleirer wrote:
Coincidence! I just viewed this video. Not exactly on topic, but related. Steve Perry secrets for sharper wildlife photos...
https://youtu.be/TLe6mXjFRdAThank you, I enjoyed this video!
Have you tried locking up the mirror of a DSLR? This setting has nothing to do with a technical evaluation. It eliminates the minute vibration from mirror slap.
If you set a bubble level in the hot shoe, then you can see the slight vibration by disturbance of the level fluid when the mirror locks up. In my experience, this setting coupled with others produces the sharpest image possible from the given optics.
It might prove useful to your method (if I understand it) by removing another variable affecting sharpness.
BTW: I defer to your dedicated measurements and process over the decades.
DirtFarmer wrote:
Yes. Of course the test images were taken with a static subject, camera on a tripod and shutter delay 2 seconds.
It was a fun project and an excuse to get more proficiency in Python. Got a pretty good graphing library out of it. (The graphing routines got started in Fortran around 1963, got translated into Pascal in the '80s and into Python this year).
DirtFarmer wrote:
I find a sharp edge in the photo. I then take the photo and load it into a numeric array in Python using luminance values. Find the sharp edge in the photo and select a sequence of points that goes over that edge. I then take the first derivative of the numbers going across the edge which should yield a peak (could be either positive or negative). I fit the peak to a Gaussian curve to find the width of the curve. A sharp edge will have a curve with a narrow width. Tack sharp would be a width close to 1.0 (the pixel spacing).
I find a sharp edge in the photo. I then take the ... (
show quote)
Sounds like a quick and easy fix.
R.G. wrote:
That's the best way to do sharpening and denoise because it's the best level for optimising sharpness and noise removal. Rather than thinking about whether a photo is tack sharp or not you should concentrate on optimising what you have, regardless of whether it falls into the category of "tack sharp" or not. What matters most is what it looks like at normal viewing distance, and if you've optimised it at 100% you'll know it's as good as it can be.
I agree, if one needs to use a loupe to tell if an image is sharp then do not look at digital images. Coming from film days I say digital is not as sharp film was. (Launch slings and arrows.)
I like raptors. If the eye is not tack sharp, I delete. Since I always burst, there are usually some keepers along with discards.
If you want to reply, then
register here. Registration is free and your account is created instantly, so you can post right away.