Peacock feathers

Images made through a microscope. All subject types.

Moderators: Pau, rjlittlefield, ChrisR, Chris S.

hkv
Posts: 469
Joined: Wed Jan 02, 2013 3:58 pm
Location: Sweden

Re: Peacock feathers

Post by hkv »

Sorry for the late response, but I have now had time to sort out my calculations. Still a bit of a mystery how we can get to so different conclusions so I will try to be as clear as possible and perhaps you can spot if I am making some error. I may be blind to my own miscalculations ;-)

Here is how I calculate. Assumptions (which has been modified to be aligned to 550 nm wavelength from your example):

Wavelength of light: 550 nm
Photo tube magnification: 2.5X (Olympus PE 2.5X)
I am using the Reyleigh Criteria: Resolution = 0.61*Wavelength/NA

Example Objective:

40X NA 0.95

1. Maximum resolution at specimen plane: 0.61*550/0.95 = 353.158 nm (this is the resolving power of the objective and the pixel pitch needed if the sensor was just above the specimen)
2. Applying Nyquist theorem meaning that we need to sample at least twice that to achieve good results: 353.158/2 -> We need to resolve/sample 176.579 nm
3. The light is moving on through the objective magnifying 40X. This is the resolution we need to meet at the objective backplane. 176.579 * 40 = 7063.16 nm = 7.063 um.
3. The light is moving on through the photo eyepiece and has a magnification factor of 2.5: 7.063 * 2.5 =17.65 um. This is the resolution we need to meet at the sensor plane.
4. The full frame sensor has a size of 35.6 x 23.8 mm.
5. 17.65 um = 0.01765 mm: Required X-resolution: 35.6/0.01765 = 2017 pixels
6. 17.65 um = 0.01765 mm: Required Y-resolution: 23.8/0.01765 = 1348 pixels

Sensor resolution: 2017 x 1348 = 2.72 MP.


60X NA 1.20

1. Maximum resolution at specimen plane: 0.61*550/1.20 = 279.58 nm (this is the resolving power of the objective)
2. Applying Nyquist theorem meaning that we need to sample at least twice that to achieve good results: 279.58/2 -> We need to resolve/sample 139.79 nm
3. The light is moving on through the objective magnifying 60X. This is the resolution we need to meet at the objective backplane. 139.79 * 60 = 8387.5 nm = 8.387 um.
3. The light is moving on through the photo eyepiece and has a magnification factor of 2.5: 8.387 * 2.5 = 20.969 um. This is the resolution we need to meet at the sensor plane.
4. The full frame sensor has a size of 35.6 x 23,8 mm.
5. 20.969 um = 0.02969 mm: Required X-resolution: 35.6/0.020969 = 1698 pixels
6. 20.969 um = 0.02969 mm: Required Y-resolution: 23,8/0.020969 = 1135 pixels

Sensor resolution: 1698 x 1135 = 1.93 MP.


Using The Nikons webpage and filling in data for the 40X NA0.95 but with a 1-inch sensor [12.8 x 9.6 mm] (biggest selectable) and a 1X tube magnification I get a pixel size of 5.8 um.
Looking at the three formulas at the Nikon page, it appears they are using the first formula: 0.5*wavelength/NA. I am using formula 2: 0.61X*wavelength/NA, but change it below to compare

Link to Nikon's calculations:

https://www.microscopyu.com/tutorials/m ... resolution

Using my formula above applied and calculating It manually I get:

1. Maximum resolution at specimen plane: 0.61*550/0.95 = 353.158 nm (this is the resolving power of the objective and the pixel pitch needed if the sensor was just above the specimen)
2. Applying Nyquist theorem meaning that we need to sample at least twice that to achieve good results: 353.158/2 -> We need to resolve/sample 176.579 nm
3. The light is moving on through the objective magnifying 40X. This is the resolution we need to meet at the objective back plane. 176.579 * 40 = 7063.16 nm = 7.063 um.
3. The light is moving on through the photo eyepiece and has a magnification factor of 1: 7.063 * 1 = 7.063 um. This is the resolution we need to meet at the sensor plane.
4. The 1 inch sensor has a size of 9.6 x 12.8 mm.
5. 7.063 um = 0.007063 mm: Required X-resolution: 12.8/0.007063 = 1812 pixels
6. 7.063 um = 0.007063 mm: Required Y-resolution: 9.6/0.007063 = 1359 pixels

Sensor resolution size: 1812 x 1359 = 2.46 MP.

Using Nikons slightly more conservative formula with 0.5*wavelength/NA we get:

1. Maximum resolution at specimen plane: 0.5*550/0.95 = 289.47 nm (this is the resolving power of the objective and the pixel pitch needed if the sensor was just above the specimen)
2. Applying Nyquist theorem meaning that we need to sample at least twice that to achieve good results: 289.47/2 -> We need to resolve/sample 144.74 nm
3. The light is moving on through the objective magnifying 40X. This is the resolution we need to meet at the objective backplane. 176.579 * 40 = 5789.47 nm = 5.79 um.
3. The light is moving on through the photo eyepiece and has a magnification factor of 1: 5.79 * 1 = 5.79 um. This is the resolution we need to meet at the sensor plane.
4. The 1 inch sensor has a size of 9.6 x 12.8 mm.
5. 5.79 um = 0.00579 mm: Required X-resolution: 12.8/0.00579 = 2211 pixels
6. 5.79 um = 0.00579 mm: Required Y-resolution: 9.6/0.00579 = 1658 pixels

Sensor resolution: 2211 x 1658 = 3.66 MP. (Slightly more than 2.46 due to the more conservative formula of 0.5 instead of 0.61)

If I recalculate my resolution table I previously presented and change 0.61 to 0.5 (same as Nikon) and also increase the Nyquist theorem requirement from 2 to 3 (to be even further on the safe side), I get the following:

Code: Select all

Magnification	Megapixel

4X (NA 0.16)	25,81
10X (NA 0.4)	25,81
20X (NA 0.75)	22,69
40X (NA 0.95)	9,10
60X (NA 1.35)	8,17
60X (NA 1.20)	6,45
100X (NA 1.4)	3,16
The conclusion is that I get the same results as Nikon's little app if I adapt my photo eyepiece mag and sensor size. Perhaps you are projecting the image without a phototube magnification? That could explain the difference, but then you are not using the full area of the sensor to capture an image, and buying a full-frame camera would be a waste. Many of the pixels would be black and wasted, like 30% of the sensor area.

rjlittlefield
Site Admin
Posts: 21036
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Re: Peacock feathers

Post by rjlittlefield »

I'm glad to hear that you're now able to match Nikon's app. That certainly makes it simpler to spot where differences are due to the assumptions.
hkv wrote:
Sun Aug 16, 2020 4:20 pm
Perhaps you are projecting the image without a phototube magnification? That could explain the difference, but then you are not using the full area of the sensor to capture an image, and buying a full-frame camera would be a waste.
On the other hand, using a a 2.5X phototube with a 35.6 x 23.8 mm sensor means that you're using an area of the usual intermediate image that is only 14.24 mm x 9.52 mm.

Do your objectives only cover such a small area? If they cover more, then capturing such a small area is wasting some of their capability.

But if that's what you're going to use, then I agree with your calculations.

--Rik

hkv
Posts: 469
Joined: Wed Jan 02, 2013 3:58 pm
Location: Sweden

Re: Peacock feathers

Post by hkv »

rjlittlefield wrote:
Sun Aug 16, 2020 5:52 pm
I'm glad to hear that you're now able to match Nikon's app. That certainly makes it simpler to spot where differences are due to the assumptions.
hkv wrote:
Sun Aug 16, 2020 4:20 pm
Perhaps you are projecting the image without a phototube magnification? That could explain the difference, but then you are not using the full area of the sensor to capture an image, and buying a full-frame camera would be a waste.
On the other hand, using a a 2.5X phototube with a 35.6 x 23.8 mm sensor means that you're using an area of the usual intermediate image that is only 14.24 mm x 9.52 mm.

Do your objectives only cover such a small area? If they cover more, then capturing such a small area is wasting some of their capability.

But if that's what you're going to use, then I agree with your calculations.

--Rik
Totally agree with you Rik. I am wasting the capabilities of the objectives instead. I will try to get hold of a Sony A7R Mark IV and make a comparison. I am quite happy with my current setup, but would be interesting to see the increased field of view as well as compare the effect of the low pass filter compared to the A7R that don't have such filter. Maybe rent a camera for a day or two. I also have the 2X photo eyepiece. It produces a slight vignetting around the edges, but I use it if I have an object that I *just* cannot get into the field of view and need a few additional pixels.

Macro_Cosmos
Posts: 948
Joined: Mon Jan 15, 2018 9:23 pm
Location: Sydney

Re: Peacock feathers

Post by Macro_Cosmos »

Alright, that clears stuff out. We're on the same page!
Embarrassingly, I didn't really know about photo eyepieces when I made that post. I was always puzzled by that U-SPT tube, "it's 101mm for U-TLU to be focused at infinity, why is this weird thing so long, the dovetails are weird and unworkable, I'm not using it!" I have a PE3.3x on the way now and I'll design an adaptor for it. Still not gonna use the U-SPT.

Direct projection does yield the best image quality. Therefore for specimens that are large enough, not using a photo eyepiece is worth it. I have some diatom samples that cover FF at 100x. I also have some arrangements that require a stitch of 4 40x exposures. My 100x at 1.35NA ~8.2709 mono ~24MP colour, but the 40x oil at NA1 ~28.634MP mono ~60+MP colour. This is why I want a higher resolution camera than the Z6 now. I will still be using the Z6 for video however, so it's staying.

Oh, and could you please briefly explain which trinocular head is better for phtoography purposes, viewing as a supplement experience? The U-TR30-2 or the U-SWTR-2. The U-SWTR-2 uses the probably discontinued oly U-SWTLU-SP, and I read that no photo eyepiece is able to "capture what I see", however I was also told that the U-SWTR-2 has better image quality since it uses a far better tube lens.
I appreciate the comments!

If you pixel peep, the a7r4 for sure will yield better images. It's in the different realm in terms of resolution. However since a PE is used, at screen level, I'm guessing the difference wouldn't be visible.
Effects of the OLPF would be interesting to examine. I know for a fact that my naughty Z6's engine tries to "correct" for moire, it just ends up spewing nonsense and ruining images. Luckily I don't shoot resolution charts for fun.
viewtopic.php?f=25&t=41908&p=263998&hilit=moire#p263998

Never seen it in real world shooting. Considering that Sony's processing is no match to Nikon's, it might be worse. Either way, the camera can do naughty stuff in the engine all it likes, just don't show up in my actual images.

hkv
Posts: 469
Joined: Wed Jan 02, 2013 3:58 pm
Location: Sweden

Re: Peacock feathers

Post by hkv »

Agree with you that direct projection is preferred for image quality. A lens is a lens and they are never perfect of course. Where have you read that the u-swtr is of higher quality than the u-tr30? I am interested to understand more about this. I am using the u-tr30.

What is you setup to mount the camera? What tubes and connectors are you using?

Lou Jost
Posts: 4589
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Re: Peacock feathers

Post by Lou Jost »

Agree with you that direct projection is preferred for image quality.
Be careful, I see you are an Olympus and Zeiss owner. The older Olymous objectives leave some CA uncorrected, amd their proprietary correcting projection lenses or eyepieces will usually be needed for maximum quality. So in these cases, direct projection is worse than projection lenses. Many Zeiss and Lomo and Leitz objectives are in the same class (but each needs a different correction, except that Lomo = Zeiss, and Olympus is said to be close to Leitz)..

hkv
Posts: 469
Joined: Wed Jan 02, 2013 3:58 pm
Location: Sweden

Re: Peacock feathers

Post by hkv »

Lou Jost wrote:
Mon Aug 17, 2020 3:04 pm
Agree with you that direct projection is preferred for image quality.
Be careful, I see you are an Olympus and Zeiss owner. The older Olymous objectives leave some CA uncorrected, amd their proprietary correcting projection lenses or eyepieces will usually be needed for maximum quality. So in these cases, direct projection is worse than projection lenses. Many Zeiss and Lomo and Leitz objectives are in the same class (but each needs a different correction, except that Lomo = Zeiss, and Olympus is said to be close to Leitz)..
Thanks Lou for pointing that out. I use Olympus infinity objectives which I believe are fully corrected from start. At least that is what I have read and in most cases, e.g. on smaller sensors, there is no tube lens in the adapters. Even from Olympus themselves.

Lou Jost
Posts: 4589
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Re: Peacock feathers

Post by Lou Jost »

Yes, the UIS and UIS-2 series, and possibly others, don't need correction.

"no tube lens in the adapters"

That's got me confused. If your setup is an infinity set-up, there has to be a tube lens somewhere.

hkv
Posts: 469
Joined: Wed Jan 02, 2013 3:58 pm
Location: Sweden

Re: Peacock feathers

Post by hkv »

Lou Jost wrote:
Mon Aug 17, 2020 3:17 pm
Yes, the UIS and UIS-2 series, and possibly others, don't need correction.

"no tube lens in the adapters"

That's got me confused. If your setup is an infinity set-up, there has to be a tube lens somewhere.
I was referring to the photo eyepieces that goes into U-SPT. Or many of the other Olympus video adapters U-TV0.5X and so on that has a lens, but there are also video adpaters with no lens, U-TV1X I believe it is called. Just a hollow piece of tubing with a c-mount on top.

Lou Jost
Posts: 4589
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Re: Peacock feathers

Post by Lou Jost »

I'm afraid I still don't understand. There has to be a tube lens somewhere if the device is made for infinity-corrected objectives....I don't know what the U-SPT is, but maybe that has a tube lens?

hkv
Posts: 469
Joined: Wed Jan 02, 2013 3:58 pm
Location: Sweden

Re: Peacock feathers

Post by hkv »

Lou Jost wrote:
Mon Aug 17, 2020 3:48 pm
I'm afraid I still don't understand. There has to be a tube lens somewhere if the device is made for infinity-corrected objectives....I don't know what the U-SPT is, but maybe that has a tube lens?
Well, yes. I believe the trinocular head (e.g. u-tr30) contains that lens. That is the last device before the camera adapters are mounted.

The u-spt goes directly on top of the trinoc head and is used to hold a photo eye piece. They come with different magnifications 2X, 2.5X, 3.3X, 4X and 5X. The 2.5X is meant to be used with a full frame sensor camera (36x24 mm). On top of the U-SPT you mount the Olympus Photoadpater L and then a camera-specific adapter ring, eg. OM to Sony/Canon/Nikon.

Macro_Cosmos
Posts: 948
Joined: Mon Jan 15, 2018 9:23 pm
Location: Sydney

Re: Peacock feathers

Post by Macro_Cosmos »

hkv wrote:
Mon Aug 17, 2020 2:37 pm
Agree with you that direct projection is preferred for image quality. A lens is a lens and they are never perfect of course. Where have you read that the u-swtr is of higher quality than the u-tr30? I am interested to understand more about this. I am using the u-tr30.

What is you setup to mount the camera? What tubes and connectors are you using?
It's according to a friend and contact of mine, he's a scientist and has access to a lot of stuff we don't really get to see, he's helped me with many things, including the custom 200mm tube lens optimised for FF, that came from him too; therefore he has my unconditional trust. That said, everyone makes mistakes!

I will be getting the u-swtr-2 most likely. I can then compare it to a U-TLU, the TL that's embedded in a U-TR30 line.

I'm using a U-TLU tube lens, it's adapted to my Z6 via a custom adaptor, via the FTZ, since F-mount is my standard. It will change as I recently got some Z-mount adaptors of extremely high quality, zero wobble unlike F-mount adaptors. My system is based around Thorlabs' SM2 tubes.

Lou Jost wrote:
Mon Aug 17, 2020 3:04 pm
Agree with you that direct projection is preferred for image quality.
Be careful, I see you are an Olympus and Zeiss owner. The older Olymous objectives leave some CA uncorrected, amd their proprietary correcting projection lenses or eyepieces will usually be needed for maximum quality. So in these cases, direct projection is worse than projection lenses. Many Zeiss and Lomo and Leitz objectives are in the same class (but each needs a different correction, except that Lomo = Zeiss, and Olympus is said to be close to Leitz)..
The UIS and UIS2 system is fully corrected for CA at an objective level. Can't speak about their finite ones, I have zero experience with those.

hkv is likely referring to Olympus' own adaptors, such as the U-SPT for photo eyepieces, which is just a hallow tube that holds the eyepiece. The U-CMAD and other garbled messy part numbers are simply hallow tubes.
The U-TR30 type has a build-in U-TLU. The U-SWTR types feature their super wide tube lens, likely the U-SWTLU-SP. There's also IR ones that feature the U-TLUIR.
I'm pretty sure these super wide TLs found in trinoculars (and the part I listed above) are different to Oly's current offerings, the U-SWTLU-C is $399 for Edmund Optics, cheaper than a U-TLU.

Lou Jost
Posts: 4589
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Re: Peacock feathers

Post by Lou Jost »

The UIS and UIS2 system is fully corrected for CA at an objective level. Can't speak about their finite ones, I have zero experience with those.
As I said above, the UIS and UIS-2 series are fully corrected, and the finite ones are not.

Post Reply Previous topicNext topic