Macro Field Studio
Moderators: rjlittlefield, ChrisR, Chris S., Pau
Macro Field Studio
I used this modified wood clamp...
...to take this shot at 2.5x...
...and this shot at 5x.
I'm also carrying a small pair of scissors in my bag to cut grass stems. When I was finished I carried the dragonfly back to where I found it. Short tutorial here.
...to take this shot at 2.5x...
...and this shot at 5x.
I'm also carrying a small pair of scissors in my bag to cut grass stems. When I was finished I carried the dragonfly back to where I found it. Short tutorial here.
It's a single frame at F13 for the 2.5x shot, and F10 for the 5x image. I have yet to find a reason to focus stack -as long as I can take control of the motion in the scene then I won't lose detail to motion blur and I can stop the lens down. What a lot of people blame on diffraction is really motion blur due to a flash duration that's too long and too much motion in the scene. With the critter in the wood clamp, and with me bracing the camera either on my thigh or the park bench, it was easy to get a lot of detail.kwhunter wrote: I should have imagined that... It is not a single shot, is it? Too deep the FOV; how many stacked shots if I may ask?
- rjlittlefield
- Site Admin
- Posts: 23625
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
Any chance we can see an actual-pixels crop of the sharpest part of that 5X shot?
I'd like to compare the results you're getting, against the bench tests reported at http://www.photomacrography.net/forum/v ... =9757#9757.
Thanks,
--Rik
I'd like to compare the results you're getting, against the bench tests reported at http://www.photomacrography.net/forum/v ... =9757#9757.
Thanks,
--Rik
- rjlittlefield
- Site Admin
- Posts: 23625
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
Thanks, Dalantech.
From my standpoint, this crop nicely illustrates what's going on.
At the level of actual-pixels, this image is not very sharp. To quantify "not very sharp", I pulled a copy of it into Photoshop, grabbed a 660x468 crop, and ran a series of comparisons. The first comparison was to resize the image to 50% (330x234), then expand it back to the original size, layer that over the original, and click between the two so I could see what changed. What I saw were differences in pixel noise, no visible difference in anything I'd call subject detail. The second comparison was the same drill, but resized to 33.3333% (220x156). The result there was the same, except for a barely detectable softening of some irregularities on the edges of a couple of facets. The third comparison was resized to 25% (165x117). That time there was obvious blurring of those irregularities and a general softening of the edges of the facets.
So, in terms of blurring from whatever source, this image is oversampled by about 3X --- the same amount of subject detail is revealed by an image that has 1/3 as many pixels on each axis, 1/9 as many pixels total.
The flickr description says that this image was shot on a Canon 40D. I believe that's 10.1 Mpixels, 3888 x 2592 in a crop-factor 1.6 sensor, 22.2 x 14.8 mm.
Dividing these numbers by the 3X determined from the resizing test, I conclude that this image would reveal more detail up to about 1296x864, but not beyond that. In fact, when I go looking for those facet details I mentioned above, I can find all of them even in the 800 pixels version posted above.
What's my point, then? Well, it's certainly not to beat up on Dalantech -- the image is gorgeous and I agree that f/10 was a very good setting to make it with. But in my view it's a good setting not because it makes an image that is sharp, but because it makes an image that is sharp enough.
This concept of "sharp enough" is a tradeoff that used to be made all the time, back when you could get only one focus plane per image even for bench work. I wonder if it's been pushed aside -- to our collective detriment -- by the ease of seeing how fuzzy things really are at the level of individual pixels on a high resolution sensor.
About stacking, it's easy to be too optimistic regarding the benefits. Significant portions of this image are OOF, far outside the DOF slab defined by "sharp". If one were to open the aperture to say f/3.5, so as to make the image sharp at the pixel level, then the blur circles in those OOF regions would get 3 times larger -- very blurred indeed! Even with most cooperative subject, that difference cannot really be made up by stacking because everything behind the last focus plane persists in going OOF at the rate determined by the aperture, no matter how deep the stack is. If you open up and shoot this subject as a short stack, what comes out is not a picture that is sharper overall, but rather a picture that is sharper in selected regions and much less sharp everywhere else. It would be a very different image, and I think one that I would not be inclined to print big and stick on a wall.
Interesting tradeoffs.
Dalantech, thanks again for showing the crop.
--Rik
From my standpoint, this crop nicely illustrates what's going on.
At the level of actual-pixels, this image is not very sharp. To quantify "not very sharp", I pulled a copy of it into Photoshop, grabbed a 660x468 crop, and ran a series of comparisons. The first comparison was to resize the image to 50% (330x234), then expand it back to the original size, layer that over the original, and click between the two so I could see what changed. What I saw were differences in pixel noise, no visible difference in anything I'd call subject detail. The second comparison was the same drill, but resized to 33.3333% (220x156). The result there was the same, except for a barely detectable softening of some irregularities on the edges of a couple of facets. The third comparison was resized to 25% (165x117). That time there was obvious blurring of those irregularities and a general softening of the edges of the facets.
So, in terms of blurring from whatever source, this image is oversampled by about 3X --- the same amount of subject detail is revealed by an image that has 1/3 as many pixels on each axis, 1/9 as many pixels total.
The flickr description says that this image was shot on a Canon 40D. I believe that's 10.1 Mpixels, 3888 x 2592 in a crop-factor 1.6 sensor, 22.2 x 14.8 mm.
Dividing these numbers by the 3X determined from the resizing test, I conclude that this image would reveal more detail up to about 1296x864, but not beyond that. In fact, when I go looking for those facet details I mentioned above, I can find all of them even in the 800 pixels version posted above.
What's my point, then? Well, it's certainly not to beat up on Dalantech -- the image is gorgeous and I agree that f/10 was a very good setting to make it with. But in my view it's a good setting not because it makes an image that is sharp, but because it makes an image that is sharp enough.
This concept of "sharp enough" is a tradeoff that used to be made all the time, back when you could get only one focus plane per image even for bench work. I wonder if it's been pushed aside -- to our collective detriment -- by the ease of seeing how fuzzy things really are at the level of individual pixels on a high resolution sensor.
About stacking, it's easy to be too optimistic regarding the benefits. Significant portions of this image are OOF, far outside the DOF slab defined by "sharp". If one were to open the aperture to say f/3.5, so as to make the image sharp at the pixel level, then the blur circles in those OOF regions would get 3 times larger -- very blurred indeed! Even with most cooperative subject, that difference cannot really be made up by stacking because everything behind the last focus plane persists in going OOF at the rate determined by the aperture, no matter how deep the stack is. If you open up and shoot this subject as a short stack, what comes out is not a picture that is sharper overall, but rather a picture that is sharper in selected regions and much less sharp everywhere else. It would be a very different image, and I think one that I would not be inclined to print big and stick on a wall.
Interesting tradeoffs.
Dalantech, thanks again for showing the crop.
--Rik
No problem Rik
In a large print both shots, at the distance that you'd view them, would look just fine.
I'd also argue that the 5x shot that I posted has little value from a compositional standpoint -it's almost an abstract (and not a good one). I took it because I had that composition stuck in my head
In a large print both shots, at the distance that you'd view them, would look just fine.
What that causes Rik is a noticeable dividing line between the area that's sharp and the area that's out of focus and it looks very unnatural and it's one of the reasons why I don't stack -we just don't see that way. Even if you blend the exposures to try to hide the dividing line between the sharp and out of focus region it will still look odd. We also don't see everything in sharp focus, and getting a lot of detail (and along with it depth) in an image can be visually confusing -the eye doesn't know where to settle in the image.rjlittlefield wrote: ... If you open up and shoot this subject as a short stack, what comes out is not a picture that is sharper overall, but rather a picture that is sharper in selected regions and much less sharp everywhere else. It would be a very different image, and I think one that I would not be inclined to print big and stick on a wall.
I'd also argue that the 5x shot that I posted has little value from a compositional standpoint -it's almost an abstract (and not a good one). I took it because I had that composition stuck in my head
- rjlittlefield
- Site Admin
- Posts: 23625
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
I agree.Dalantech wrote:No problem Rik
In a large print both shots, at the distance that you'd view them, would look just fine.
But I feel compelled to point out that what you just said is roughly equivalent to "1296x864 pixels is enough to make a good-looking print".
This is exactly the effect that I was talking about HERE, when I wrote that "Modern cameras have too many pixels anyway." Crop away 3/4 of the total pixels in a 10 Mpixel camera, and you still have more than you need to look good with a small subject, stopped down to optimum DOF in a single shot. Whether your customers would believe this is of course a different matter, so from a marketing standpoint, huge pixel counts are a good thing even if most of them aren't actually needed to capture the image content.
--Rik