Mu-43 teleconverters with adapted lenses

[barry]

I've long known that my Oly MC-14 teleconverter fits on most lens adapters for 'legacy' lenses, but I never really got around to trying it until @Brownie asked me what the Minolta Rokkor 58mm lenses (I have both the f/1.4 and f/1.2 models) would look like with a 1.4 teleconverter.
One reason I hadn't tried it is that I had tried a Rokkor 135/2.8 (a fairly sharp lens) with a legacy, third-party, teleconverter and the results were terrible (and worse than cropping).

However, I was pleasantly surprised with the 58/1.4 and the MC-14 (which makes for 82mm f/2.0)...
Wide-open, it's 'glowy' in bright light (as is the bare lens), but probably still usable for portraits (which I don't shoot much).
Stopped down it's very good.

Note these images are basically SooC unless otherwise noted; I would normally add more contrast and make other adjustments, but I didn't want to skew the results...

wide open:

Subscribe to see EXIF info for this image (if available)

f/2.0 (on the lens, not net):

Subscribe to see EXIF info for this image (if available)

wide-open:

Subscribe to see EXIF info for this image (if available)

f/2.0:

Subscribe to see EXIF info for this image (if available)

Although Lincoln is a bit soft, especially in the wide-open one, I can read the text on both plaques without trouble in both the RAW files.

Various, still basically SooC:

(definitely missed the focus on this one, but I like the composition)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

I increased the contrast and blacks on this one:

Subscribe to see EXIF info for this image (if available)

Conclusion:
It's quite usable, and didn't cost me anything extra as I already had the MC-14. I've never come across a Rokkor 85mm (1.8 or 2.0), so this is a great alternative.
I'll have to try again with the 58/1.2, which has better bokeh and is a bit sharper.

Of course, don't expect it to compare in sharpness to the 75/1.8.


Please share your own shots with the Mu-43 TC's and legacy lenses!

 

[Petrochemist]

I've got multiple legacy teleconverters, but very rarely use any of them. Results here look good enough to consider an MC-14 but I don't have any native lenses it will work with

I'll see if I can have a play with a friends MC-14, and compare it with the better legacy ones. - Another thing added to the try this list!

 

[Gerard]

Subscribe to see EXIF info for this image (if available)

My standpoint was 1500 meters from the bridge. The barge in the center of the foto was another 500 meters further. The ship is called "Anna". This is even visible without zooming in.
EM1 OM 300/4.5 @F4.5; 1/50 sec ISO 200; MC-14; Tripod
Not the most engaging photo I ever made, but just like @barry13 and @Petrochemist I was curious about the outcome of this exercize.

 

[Gerard]

At 1/30 seconds not sharp, but still pleasing to my eye View attachment 789099

Same combo as previous post, this time handheld

 

[Brownie]

@barry13 Some of those shots are surprisingly good!

 

[Harvey Melvin Richards]

I played around a little with my MC-14 about 4 years ago. This was taken with the MC-14 and a Phoenix 100 mm f/3.5.

Subscribe to see EXIF info for this image (if available)

M1017427-1 by Harvey Richards, on Flickr

 

[connloyalist]

Interesting.

I am wondering.... Is the image quality you end up with limited by the weakest link in the chain, or do all the components stack?

In the first case ("weakest link"), if the optical performance of the MC-14 is higher than that of an adapted lens attached to it, that would mean that adding the MC-14 has no impact on image quality.

In the second case ("components stack"), every piece of glass the light passes through reduces image quality somewhat, and the good optical performance of the (modern) MC-14 only means that the image quality is reduced less than if a vintage teleconverter had been used.

Regards, C

 

[archaeopteryx]

I am wondering.... Is the image quality you end up with limited by the weakest link in the chain, or do all the components stack?

Yes to both. But not quite as you've described.

For a basic conceptual model I would suggest multiplication. For example, if one component has a relatively low MTF it will tend to control the overall MTF as it drags down the product more than anything else. However, net improvement can result when aberrations cancel each other. For example, combining a bit of barrel distortion (1.0 something) with some pincushion (0.9 something) will probably give something closer to undistorted (1.00).

Actual optical analysis commonly uses methods such as ray casting or wavefront propagation in tools like Zemax. WinLens Basic can be useful for simple investigations, though.

 

[barry]

Also note the TC is magnifying only the center of the lens’ image, which is often the best part.

 

[Bytesmiths]

Is the image quality you end up with limited by the weakest link in the chain, or do all the components stack?

In theory, the latter case. In practice, the former.

Gaussian† bandwidth combines with the inverse of the square root of the sum of the inverses of the squares of the individual components' bandwidths — that's the theory.

That's a big bunch of words! Here is an expression that might be easier to understand, from the electronics realm:

Subscribe to see EXIF info for this image (if available)

You can add any number of terms below the radical, for example, sensor bandwidth, lens bandwidth, teleconverter bandwidth, filter bandwidth, etc.

But this creates a hyperbolic curve, such that if any component's bandwidth is significantly below the bandwidth of other components, the response of that lower bandwidth component predominates — that's the practice.

So, this means a kilobuck teleconverter is not going to improve a $10 lens, and vice-versa.

If both components have a closely matched bandwidth, the the result may be nearly the same as the individual components.

Optical performance is more complicated than simple bandwidth. It is certainly possible to design components such that one will make up for the deficiencies of others. This is a claim that focal reducer designers make.

†A "gaussian" or "natural" or "nominal" bandwidth is one in which response falls off at the rate of 6 decibels per octave, which basically means that line pairs that have twice the pitch will have 1/4th the sharpness. Most spherical lens designs have a gaussian bandwidth. Lenses with aspherical or ED glass may or may not have a gaussian bandwidth.

 

[connloyalist]

Wow, I guess I asked a complicated question Thank you for your attempts at explaining it to me .

Please excuse me being a bit slow of comprehension here, but what I take away from the above explanations is that while in theory a vintage lens (assuming inferior optics*) with a modern teleconverter (assuming superior optics*) will be worse than just the vintage lens by itself, in practice the difference is negligible. Would that be a correct summary?

* The quality of vintage against modern optics is probably a debate unto itself but not relevant to this topic. For the record, far and away most of my lenses are of the "vintage" category.

Regards, C.

 

[barry]

@connloyalist, yes I'd say it's negligible enough that it can be very usable.
The worst case (for pretty much all lenses) is wide-open, and I haven't shot enough wide-open portrait samples to make a firm decision there, but portraits aren't very important to me anyways.

 

[Bytesmiths]

The quality of vintage against modern optics is probably a debate unto itself but not relevant to this topic. For the record, far and away most of my lenses are of the "vintage" category.

In my experience, the best "vintage" lenses are the equal of most of today's lenses.

I shoot mostly "vintage" lenses as well, and all of them that I own appear to be sharper than the "new" 12-200 at the long end.

 

[archaeopteryx]

This is a claim that focal reducer designers make.

Metabones' Speed Booster whitepaper is supported by third party measurement (Cicala 2013), though I'm unsure of the extent to which continuing measurement of subsequent models maintains this. I also haven't looked at measurements of other boosters as they're not something I'm likely to purchase in the foreseeable future.

Also note the TC is magnifying only the center of the lens’ image, which is often the best part.

Yes, though modern matched TCs can still give poorer results than not using a TC and cropping. The 1.4x bundled with the Fuji 100-400 is substantially astigmatic on every lens Fuji specifies it with, for example, and performs worst at 400mm where it's most likely to be used. Though, in fairness to Fuji, I suppose I should note both Olympus and Panasonic duck out of specifying MTF for their TCs. Even in similar cases like the Panasonic-Leica 200mm where the 1.4x TC is bundled.

Lenses with aspherical or ED glass may or may not have a gaussian bandwidth.

I can't think of data suggesting a Gaussian bandwidth model is particularly useful (e.g. Cicala 2019). As a purely technical detail, though, lenses with ED glass can still have only spherical elements.

 

[Bytesmiths]

I can't think of data suggesting a Gaussian bandwidth model is particularly useful

Well, they tend to be less expensive, for one!

 

[Carbonman]

response falls off at the rate of 6 decibels per octave

You're describing a first order crossover!

 

[archaeopteryx]

You're describing a first order crossover!

Well, only the lowpass half of it, and only by proxy in the spatial domain. Any other first order lowpass will have the same general properties, rather by definition.

 

[Bytesmiths]

You're describing a first order crossover!

Yes, that's what a "gaussian" filter is.

Now, if we're talking bessel, butterworth, or other filter characteristics, you've got some higher math to do.

 

[Carbonman]

Now, if we're talking bessel, butterworth, or other filter characteristics, you've got some higher math to do.

What about an Infinite Slope filter? Now it's getting into RF filter theory.

 

[Bytesmiths]

What about an Infinite Slope filter?

I don't have time for measuring or verifying that.