A thought:
If using a 10x 0.25 objective, and a fixed 200mm tube lens, without a separate aperture to adjust -
if you have a 4 x 0.1 or similar objective, you may find you "run out of pixels" when you enlarge the detail you want. If you have a teleconverter, you'd get the same image onto more pixels. The TC would go at the camera end.
Not ideal, but probably OK if you just want to see what happens.
I suppose that would also work with a finite arrangement - I've never tried.
Reduced NA gives more detail
Moderators: rjlittlefield, ChrisR, Chris S., Pau
-
- Posts: 3439
- Joined: Sat Nov 20, 2010 10:40 am
- Location: Santa Clara, CA, USA
- Contact:
You make a good point here. Spreading the image over more pixels allows a potentially better image at 2:1 than you could have achieved at 1:1, especially for very small crops. It also eliminates potential IQ reduction due to the AA filter, ie an image taken with 2x teleconverter, reduced 2x and compared pixel for pixel with original 1x image, should have better IQ. I will test this...ChrisR wrote:A thought:
If using a 10x 0.25 objective, and a fixed 200mm tube lens, without a separate aperture to adjust -
if you have a 4 x 0.1 or similar objective, you may find you "run out of pixels" when you enlarge the detail you want. If you have a teleconverter, you'd get the same image onto more pixels. The TC would go at the camera end.
Not ideal, but probably OK if you just want to see what happens.
I suppose that would also work with a finite arrangement - I've never tried.
Going one step further, for higher mag imaging you are limited by the objective NA, but if you are operating inside the sensor DLA, then expanding the image across more pixels would improve overall system sharpness. Let's say you are imaging at 1:1, with a nominal f/2.8 lens (effective f/5.6). Let's also say your sensor pixel pitch is around 6um (Canon XS, XTi, 5D, 6D, etc) such that your DLA is in the range of f/10. That f/5.6 1:1 image is then expanded across 4 times the pixels, with a feff of f/11.2, just starting to limit the sensor resolution. Details that were too small for the sensor array to resolve at 1:1 are now able to be resolved at 2:1. Something more to try.
Edited to add:
I gave the second concept a try first. Using the 105PN, and of course using a Cent as the subject, I checked the center performance of the bare lens at around 0.8:1 versus a 2x downsized image at ~1.6:1. Here are the images, no post-processing:
Bare Lens
With Teleconverter and downsized 2x
Very, very interesting. Critical focus was on the top surface details of the chin. The 2x image is significantly sharper (!!). There is also a significant shift in surface colors. Those reddish and bluish areas may be false colors that are eliminated by spreading the light cones over a bigger area.
Now I need to test this combination in competition with the other lenses in my Shootout at 2.4x.
...Ray
- rjlittlefield
- Site Admin
- Posts: 23626
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
That's pretty compelling, isn't it?
But see http://www.photomacrography.net/forum/v ... 164#101164 and compare the third and fourth images. Even at f/11, there's a lot more detail than is captured in 15 megapixels APS-C. The calculation there is that it would take something like 47 megapixels on APS-C to fully capture the f/11 image. Even more extreme, scroll down to http://www.photomacrography.net/forum/v ... 180#101180 and look at the f/5.6 optical image.
I think we've discussed before, for subjects in the size range of coins it's very easy for an optical image to out-resolve the sensor. As a result, you'll get best results by grabbing the highest megapixels sensor you can afford, be that a D800 or a Hasselblad or the virtual sensor provided by stack-and-stitch, then optically enlarging the image as needed to fill that many-megapixels sensor.
For smaller subjects, it's much more challenging for a single optical image to out-resolve the sensor. I can do it on APS-C by pushing down a Nikon MRL00102 to 5X NA 0.25 (f/10 effective), but that's a special case. More generally the trick with smaller subjects is to stack-and-stitch using a higher NA lens.
Of course none of this has anything to do with "Reduced NA gives more detail", so we've gone seriously off-topic. I second ChrisR's suggestion of starting a new thread to address this new topic.
--Rik
But see http://www.photomacrography.net/forum/v ... 164#101164 and compare the third and fourth images. Even at f/11, there's a lot more detail than is captured in 15 megapixels APS-C. The calculation there is that it would take something like 47 megapixels on APS-C to fully capture the f/11 image. Even more extreme, scroll down to http://www.photomacrography.net/forum/v ... 180#101180 and look at the f/5.6 optical image.
I think we've discussed before, for subjects in the size range of coins it's very easy for an optical image to out-resolve the sensor. As a result, you'll get best results by grabbing the highest megapixels sensor you can afford, be that a D800 or a Hasselblad or the virtual sensor provided by stack-and-stitch, then optically enlarging the image as needed to fill that many-megapixels sensor.
For smaller subjects, it's much more challenging for a single optical image to out-resolve the sensor. I can do it on APS-C by pushing down a Nikon MRL00102 to 5X NA 0.25 (f/10 effective), but that's a special case. More generally the trick with smaller subjects is to stack-and-stitch using a higher NA lens.
Of course none of this has anything to do with "Reduced NA gives more detail", so we've gone seriously off-topic. I second ChrisR's suggestion of starting a new thread to address this new topic.
--Rik
-
- Posts: 3439
- Joined: Sat Nov 20, 2010 10:40 am
- Location: Santa Clara, CA, USA
- Contact:
Yes, yes, terribly sorry about that, I was inspired by Chris R's last post regarding use of teleconverters and took it a step further. I'll try to keep to topic in future...Rayrjlittlefield wrote:
Of course none of this has anything to do with "Reduced NA gives more detail", so we've gone seriously off-topic. I second ChrisR's suggestion of starting a new thread to address this new topic.
--Rik
Longitudinal chromatic aberration. This aberration is linearly proportional to the NA and independent of the width of the field of view.ChrisR wrote:Um, no, I haven't!
Why should reducing the NA lead to CA?
I can see it happening due to the iris being in the "wrong" place, perhaps, but not as a result of swapping objectives.
It could mean hand-retouching rather than letting the stacker combine images.
The focal point varies depending upon the color, so the aberration appears as a dispersion of color regardless of whether the point is on or off the axis.
Reference:
http://www.nikon.com/products/instrumen ... /index.htm
The numerical aperture of an objective is also dependent, to a certain degree, upon the amount of correction for optical aberration. Highly corrected objectives tend to have much larger numerical apertures.
Reference:
http://www.olympusmicro.com/primer/anat ... rture.html
I think this reference also have some explanations to the original topic of this thread.
Regards,
Conny
I'm struggling to see exactly what you mean.
In a high-pixel output image, you'd only use the reduced NA parts where it's beneficial.
If you used a lower magnification lower NA objective for a web image, and if LoCA reared its head, it might be possible to use PS Blending Modes to combine images to help clear it.
Or did you mean something else?
In a high-pixel output image, you'd only use the reduced NA parts where it's beneficial.
If you used a lower magnification lower NA objective for a web image, and if LoCA reared its head, it might be possible to use PS Blending Modes to combine images to help clear it.
Or did you mean something else?
I probably did, it was references to what happends when it happends.ChrisR wrote:I'm struggling to see exactly what you mean.
In a high-pixel output image, you'd only use the reduced NA parts where it's beneficial.
If you used a lower magnification lower NA objective for a web image, and if LoCA reared its head, it might be possible to use PS Blending Modes to combine images to help clear it.
Or did you mean something else?
I don't have access to my studio setup where I am now, but I will try your tips later and make a stacked image and combining the SL 20x na 0.28 with the 20x na 0.42. I think a mosquito will be a good subject for that test.
Regards,
Conny
If I may add, this explains very well why I instantly recognized a significant increase in IQ when, using my DFI60 10X/0.25 at 5X stopped down to about f/22, contrast improved dramatically. Subtle textures became more evident and DMap artifacts were no longer so frustrating to deal with as they had previouly been.
Having a bit of optics background I dislike diffraction as much as anyone, but all too often the full aperture of this objective when used at 5X brought problems which quite overwhelmed the decrease in diffraction afforded by the effective f/10 aperture.
That said, there are certainly instances when the full aperture @5X does indeed result in the best IQ, at least on sensors with tighter pixels ... I'd suggest that it has to do with the amount of OOF light fogging the image in a given scene/stack.
I do of course always use full aperture on this objective when it's configured for 10X (f/20) -- at it's intended magnification it's far more well-behaved.
Having a bit of optics background I dislike diffraction as much as anyone, but all too often the full aperture of this objective when used at 5X brought problems which quite overwhelmed the decrease in diffraction afforded by the effective f/10 aperture.
That said, there are certainly instances when the full aperture @5X does indeed result in the best IQ, at least on sensors with tighter pixels ... I'd suggest that it has to do with the amount of OOF light fogging the image in a given scene/stack.
I do of course always use full aperture on this objective when it's configured for 10X (f/20) -- at it's intended magnification it's far more well-behaved.