Canon 70D Focus Tests

[rnabholz]

Some of you know that I have been struggling a bit getting my new camera, a Canon 70D to perform to my expectations.

I am the first to admit that it is very likely that the issue rests with me. This is my first serious attempt at DSLR Photomicrography, and I am sure that there are aspects of the process that I could improve on.

My largest concern at this time is centered on the apparent lack of sharpness. With that in mind I set up a test to see if I could isolate the issue.

The Platform:

AO One Ten
10x Plan Achro Objective
Subject - Citric Acid Crystals
Canon 70D tethered to a Laptop

The test will be to image the same portion of the slide using four different approaches

#1 AO 1054 Photo Eyepiece
This is my default set up. A photo tube connected to the the trinoc head, a microscope adapter attached to the tube, housing an AO 1054 Photo Eyepiece

#2 AO 180 Eyepiece
The same set uo as #1 with the exception of the use of an AO 180 Eyepiece instead of the 1054.

#3 Eyepiece Adapter
This is an adapter attached to the camera via Tmount which permits mounting the camera directly to the eyepiece tube. It contains no optics.

#4 Afocal using my Nexus 6 phone.
This is a control using my previously preferred setup, a Nexus 6 phone shooting afocally through the AO 180 eyepiece.

Image Capture

My test would be to shoot the same subject, the camera would remain at the same exposure, 1/60th at ISO 400 for all of the camera shots, the phone sets its exposure automatically, so those settings may vary.

I would shoot multiple images, a few at best focus - on the 70D using focus magnification feature to assure best focus, on the phone using the 6" screen to set focus. I would then shoot images on both sides of the "best" focus in the event I missed with my determination.

Then comparing them on my larger desktop computer screen, pick the best image from each of the four methods.

First is a comparison of the center of all four images at 100% size. These are the JPEGs straight out of the camera and phone with no manipulation or enhancement.

Crop 1054 Photo Eyepiece.JPG (203.75 KiB) Viewed 8812 times

Crop 180 Eyepiece.JPG (163.91 KiB) Viewed 8812 times

Crop Eyepiece Adapter.JPG (125.89 KiB) Viewed 8812 times

Crop Phone.jpg (109.55 KiB) Viewed 8812 times

The size differences in the Canon images are due to slightly different scale caused by minor differences in the distance to the sensor between the eyepieces and adapter. The Phone is a much smaller sensor both in physical size and pixels, 13mp vs 20mp in the camera.

My first reaction is that the phone seems to be applying some sharpening enhancement, while the camera does not seem to be doing any obvious enhancements. The second is that the phone's sensor size and pixel density would seem to put it at a significant disadvantage compared to the others.

Among the camera images, the 180 eyepiece would seem to leave the most to be desired. Seems to be much less sharp than the 1054 and Eyepiece Adapter images.

Continued.....

[rnabholz]

Next was to take each of the images full frame and scale it to a width of 1024 to make it easily viewable and apply typical adjustments, in this case, just slight levels tweaks and unsharp mask using the same settings for all images.

1024 1054 Photo Eyepiece.JPG (387.36 KiB) Viewed 8808 times

1024 180 Eyepiece.JPG (419.39 KiB) Viewed 8808 times

1024 Eyepiece Adapter.JPG (341.92 KiB) Viewed 8808 times

1024 Phone.jpg (212.79 KiB) Viewed 8808 times

The first obvious observation is that the phone is at a serious disadvantage with regard to image size, and the only remedy is to crop or effectively crop in camera by "zooming" during capture - which of course is the same thing as cropping after post capture. While this may not be a big consideration for web presentation, it may limit options with regard to printing.

Other observations:

The 180 eyepiece and Eyepiece Adapter present a larger field than the 1054. Any comparison of the sharpness at the edges should consider that difference and compare the same field.

That said, to my eyes the 1054 presents the best image across the entire field. The center is sharper and the edge of field still delivers a sharpness that seems to be diminished with the other two methods.

With the 1054 exhibiting the best performance, is there more I should be doing in post to maximize the potential?

I will readily admit to being a bit conservative in the application of sharpening methods. Nothing looks worse to my eye that an oversharpened image. I tend to "leave something on the table" rather than oversharpen.

With that in mind I revisited the 1054 image and applied a bit more Unsharp Mask and came up with this.

1024 1054 Photo Eyepiece Higher Unsharp Mask Levels.JPG (390.65 KiB) Viewed 8808 times

I am seeking all your opinions on all of this. Let me know what you see, offer tips, suggestions, guidance, recommendations for alternate hobbies.....whatever you think will be helpful :^)

[zzffnn]

Rod,

Thank you for doing this. I learned quite a bit from your posts.

To my not-so-good eyes:

1054 > EP adapter > 180 >> Phone

Sharpening makes difference among the top 3 smaller.

Phone seemed to apply sharpening by itself. While big structures/lines appear sharp, the smallest details are lost in the phone camera, sharpened or not.

I started with a smartphone too, so I am not biased against it as much as others.

Without seeing the eyepiece images, I am guessing I would vote for using slightly less sharpening than your less sharpening versions. Maybe 70%-80% of your less sharpening version . The more sharpening version seems a bit grainny.

[75RR]

A 5th approach, using the camera afocal would have been interesting.

Was wondering how you set up your camera?

This is what I do:
I use Zeiss KPL 12.5/18 eyepiece as a projection lens.
I drop the eyepiece in the trinocular port and adjust the port to obtain parfocality with the eyepieces in the binocular.
I then raise the eyepiece with a spacer. I had several made from PVC tubing in 2.5mm increments.
I have it raised by 15mm at the moment which gives me a 1.8x magnification.
This covers the central part of the field of view which suits most of my subject matter.

Do you set up parfocality between all 3 eyepieces?

[JimT]

Rod, really hard to tell given the subject complexity and the number of alternatives you tested. Maybe a better test would be to compare only two at a time on something less colourful with sharp edges and angles.

Got any diatom slides lying around?

JimT

[KurtM]

Agree, crystals are better test subjects for color rendition than critical focusing, and nothing beats diatoms for the purpose. Another good test is carapace texturing on cladocera or ostracod shells, but there are obvious problems with that route.

[rnabholz]

Rod,

Thank you for doing this. I learned quite a bit from your posts.

To my not-so-good eyes:

1054 > EP adapter > 180 >> Phone

Sharpening makes difference among the top 3 smaller.

Phone seemed to apply sharpening by itself. While big structures/lines appear sharp, the smallest details are lost in the phone camera, sharpened or not.

I started with a smartphone too, so I am not biased against it as much as others.

Without seeing the eyepiece images, I am guessing I would vote for using slightly less sharpening than your less sharpening versions. Maybe 70%-80% of your less sharpening version . The more sharpening version seems a bit grainny.

Thanks zz

We are in agreement on the ranking of the techniques.

The phone is definitely up to some mischief. At least sharpening and possible levels including HDR.

Below is a revised version of the 1054 image with a 20% reduction of the Unsharp Mask parameters. Thoughts? Thanks

1024 1054 Photo Eyepiece Lower Unsharp Mask Levels.JPG (356.55 KiB) Viewed 8778 times

[rnabholz]

A 5th approach, using the camera afocal would have been interesting.

Was wondering how you set up your camera?

This is what I do:
I use Zeiss KPL 12.5/18 eyepiece as a projection lens.
I drop the eyepiece in the trinocular port and adjust the port to obtain parfocality with the eyepieces in the binocular.
I then raise the eyepiece with a spacer. I had several made from PVC tubing in 2.5mm increments.
I have it raised by 15mm at the moment which gives me a 1.8x magnification.
This covers the central part of the field of view which suits most of my subject matter.

Do you set up parfocality between all 3 eyepieces?

Can you describe what you mean regarding afocal using the camera? Do you mean shooting through an attached lens through an eyepiece?

As far as setting up the camera, it was as described above, and depending on the target, I can vary the insertion of the tube into the microscope adapter to change the distance to the sensor. On the wide end, I can vignette the corners of the frame. I generally adjust so that no vignetting takes place in a 4:3 framing.

I have not really pursued parfocality of the eyepieces as of yet. I have been relying on the camera screen with the magnification inset feature to set focus on the image sent to the sensor. That pretty much mirrors the process that I was using with the phone and did not seem to produce any sharpness problems.

Thanks for the interest.

[rnabholz]

Rod, really hard to tell given the subject complexity and the number of alternatives you tested. Maybe a better test would be to compare only two at a time on something less colourful with sharp edges and angles.

Got any diatom slides lying around?

JimT

No diatom slides, sorry.

I was hoping that the fine lines and details in crystals would serve as a decent test subject. I'll see what I can come up with.

Thanks

[zzffnn]

Rod,

The final judgement can only made by yourself, since only you can see the (best) eyepiece view. I suspect those grains are there to start with.

Your original less sharpening version seems slightly better overall than that 20% reduction. Maybe stay with that one or do 10% less. It probably varies among subjects and lighting too.............

[75RR]

Do you mean shooting through an attached lens through an eyepiece?

Yes, just to make sure the camera is working as it should.

That pretty much mirrors the process that I was using with the phone and did not seem to produce any sharpness problems.

Were you taking photos through the trinocular port with your phone? Thought you were using it on a binocular eyepiece.

Have read that parfocality between the binoculars and the trinocular port is important, though I am not sure how that relates to a raised eyepiece.

[rnabholz]

Do you mean shooting through an attached lens through an eyepiece?

Yes, just to make sure the camera is working as it should.

That pretty much mirrors the process that I was using with the phone and did not seem to produce any sharpness problems.

Were you taking photos through the trinocular port with your phone? Thought you were using it on a binocular eyepiece.

Have read that parfocality between the binoculars and the trinocular port is important, though I am not sure how that relates to a raised eyepiece.

I have used the camera for non microscopy uses, and it does seem to be working just fine if that addresses that concern.

I was shooting the phone through the binocular head and eyepiece.

Regarding parfocality, please correct me if I misunderstand. I understood the importance of having the Bino Eyepieces parfocal with the Trinoc was strictly related to the ability to focus with eyes on the eyepieces and then switch the output to the Trinoc port and shoot without refocusing.

If one switches the output to the Trinoc port and then uses a monitor screen that depicts what the camera sensor is seeing and adjusts focus in real time based on the image on that screen, doesn't that alleviate the necessity for absolute parfocality?

Thanks

Rod

[rnabholz]

Rod,

The final judgement can only made by yourself, since only you can see the (best) eyepiece view. I suspect those grains are there to start with.

Your original less sharpening version seems slightly better overall than that 20% reduction. Maybe stay with that one or do 10% less. It probably varies among subjects and lighting too.............

Thanks zz.

I do agree that the lessened sharpening does seem to be soft.

I will keep working on it.

Here are a couple of quick edits on some that I just took.

Euchlanis 4-2-16.JPG (211.41 KiB) Viewed 8760 times

Euchlanis 4-2-16 II.JPG (175 KiB) Viewed 8760 times

[75RR]

If one switches the output to the Trinoc port and then uses a monitor screen that depicts what the camera sensor is seeing and adjusts focus in real time based on the image on that screen, doesn't that alleviate the necessity for absolute parfocality?

"Microscope objectives are designed to very specific parameters including lens-to- subject distance and (for finite systems) a specific tube length. When your microscope head is set up properly and focus is achieved through the eyepieces these parameters are met, and optical quality can be maximized. With the possible exception of a very slight "fine-tuning", the focus should not be changed from proper eyepiece focus in order to provide accurate focus in a camera. While a poorly set-up camera can often be brought into focus by significantly changing the microscope focus (and thus losing proper focus in the eyepieces), this would upset the optical design parameters and can result in serious image degradation."

Quote from: Basic Considerations When Mounting a Camera on a Trinocular Tube by Charles Krebs.

See pdf:
http://www.krebsmicro.com/pdf/trinoc_a3.pdf

[rnabholz]

If one switches the output to the Trinoc port and then uses a monitor screen that depicts what the camera sensor is seeing and adjusts focus in real time based on the image on that screen, doesn't that alleviate the necessity for absolute parfocality?

"Microscope objectives are designed to very specific parameters including lens-to- subject distance and (for finite systems) a specific tube length. When your microscope head is set up properly and focus is achieved through the eyepieces these parameters are met, and optical quality can be maximized. With the possible exception of a very slight "fine-tuning", the focus should not be changed from proper eyepiece focus in order to provide accurate focus in a camera. While a poorly set-up camera can often be brought into focus by significantly changing the microscope focus (and thus losing proper focus in the eyepieces), this would upset the optical design parameters and can result in serious image degradation."

Quote from: Basic Considerations When Mounting a Camera on a Trinocular Tube by Charles Krebs.

See pdf:
http://www.krebsmicro.com/pdf/trinoc_a3.pdf

Thanks 75

I wonder if that still hold true with infinity systems like my AO?

[75RR]

I wonder if that still hold true with infinity systems like my AO?

Worth a try, you only have to set it up once.

[gekko]

If one switches the output to the Trinoc port and then uses a monitor screen that depicts what the camera sensor is seeing and adjusts focus in real time based on the image on that screen, doesn't that alleviate the necessity for absolute parfocality?

"Microscope objectives are designed to very specific parameters including lens-to- subject distance and (for finite systems) a specific tube length. When your microscope head is set up properly and focus is achieved through the eyepieces these parameters are met, and optical quality can be maximized. With the possible exception of a very slight "fine-tuning", the focus should not be changed from proper eyepiece focus in order to provide accurate focus in a camera. While a poorly set-up camera can often be brought into focus by significantly changing the microscope focus (and thus losing proper focus in the eyepieces), this would upset the optical design parameters and can result in serious image degradation."

Quote from: Basic Considerations When Mounting a Camera on a Trinocular Tube by Charles Krebs.

See pdf:
http://www.krebsmicro.com/pdf/trinoc_a3.pdf

Thanks 75

I wonder if that still hold true with infinity systems like my AO?

Yes. I believe that is the first order of business.

[zzffnn]

I could be wrong, though I thought it matters if you change thing after the tube lens (which is probably inside the head of AO infinity scopes, before the eyepieces).

[75RR]

I could be wrong, though I thought it matters if you change thing after the tube lens (which is probably inside the head of AO infinity scopes, before the eyepieces).

The photo tube is an attachment that allows one to conveniently photograph a subject that has been correctly setup through the binocular eyepieces. Parfocality effectively links them together. I think this is equally applicable to both finite and infinite systems.
One must not confuse microscope setup (Köhler, focus, etc ...) with camera setup (White Balance, shutter speed, etc ...) Apples and Oranges.

[rnabholz]

I plan on working to set up parfocality later today, but for perspective, the focus adjustment that I am making now when moving from the eyepiece to the camera on the trinoc is very very slight, just a tweak.

I will see if I can eliminate it so as to take it out of the equation.

[zzffnn]

Sorry, I was too vague regarding "things after tube lens". I should have said distance between (photo/visual) eyepiece and tube lens. As to whether CA/SA/flatness correction is done by tube lens or eyepiece or both, different manufacturers have different practice. I cannot remember what AO does.

[gekko]

My 2 cents' worth: The criterion is that the objective should be so positioned from the object that the image is in focus through the eyepiece, in other words, parfocality between eyepieces and camera sensor with the objective at its specified working distance. A tiny tweak (a few divisions or µm on the fine focus knobw) is ok but then, as in cover glass thickness, the closer to the goal the better, and low NA objectives are more tolerant of a slight lack of parfocality. In fact, the effect of non-parfocality is the same as deviations from the specified cover glass thickness. To get as close as possible to parfocality, it is best (I would say necessary) to use the low power objective (higher power objectives have successively more depth of focus and can tolerate more error in the location of the camera sensor but less error in the objective's actual working distance employed, if that makes any sense.)