[Qwntm]

A quick comparison of the 8mm Samyang for Fuji against the 8mm Rokinon for Pentax.

You Tube

You Tube

[ramseybuckeye]

It's interesting because so many of the mirrorless lenses are not made that small, and why not?

[stevebrot]

They are two different lens designs, both made for APS-C , and Samyang makes both in native Fuji-X. The larger one (I have one on the shelf behind me in K-mount) is twice as heavy as the smaller and has one less element (10/7) vs 11/8. I believe the two provide different projections. The larger lens is well-known for being a stereographic fisheye. The smaller lens probably uses the more common equisolid projection since Samyang does not call the projection out in the specs. That difference alone is probably enough to explain the size difference. Edit: Some reviewers seem to feel that the f/2.8 lens is also stereographic. The difference is pretty easy to spot (I have fisheyes of both types), but demonstrating on comparison photos requires same camera and identical camera position.

FWIW, both are probably retrofocus. The difference can be easily told by comparing the apparent diameter of entry and exit pupils (LINK)


Steve

[clackers]

It's interesting because so many of the mirrorless lenses are not made that small, and why not?

Nothing about having a mirror or not in the body alters the size needed to make a high performance lens.

You can make any lens small - look at the M series for Pentax film cameras - but it's always a compromise. Those Sony G Master lens are enormous by comparison.

You can make lenses smaller if they don't cover the full sensor, so APS-C Fuji and the micro Four Thirds can do that. They can even make them smaller than strictly necessary by deliberately not even correcting for the smaller sensor. Software can increase the sharpness in the corners, stretch the image to compensate for distortion, and increase the brightness at the edges - in RAW, even! You *think* you're getting a good lens, but you're not.

In theory, because of the shorter registration distance, a wide angle lens can be made without a retrofocal group, so that can save size and weight. But in practice, especially on digital, with the rays hitting the sensor at an awkward angle and not registering, wide angle lenses for mirrorless often put in a retrofocal group anyway, AFAIK, so there's no advantage.

[Tony Belding]

In theory, because of the shorter registration distance, a wide angle lens can be made without a retrofocal group, so that can save size and weight. But in practice, especially on digital, with the rays hitting the sensor at an awkward angle and not registering, wide angle lenses for mirrorless often put in a retrofocal group anyway, AFAIK, so there's no advantage.

This seems like a logical explanation, and yet it doesn't jibe too well with my experience.

Back in the film days, it was unusual to see a rectilinear lens for SLR cameras wider than 20mm, and even those tended to be expensive. Wider lenses did exist for rangefinders, though. When the Sony A7 hit the market, people were very soon adapting those rangefinder lenses to it. They found with very wide lenses there was a problem of vignetting and color shift due to the incident angle, as you alluded. However, newer models have improved microprisms to mitigate that effect, and companies like Voigtländer have easily adapted their rangefinder lens formulas for mirrorless. We’ve seen rectilinear lenses from Voigtländer for Sony in 15mm, 12mm and even 10mm with full frame coverage, and the lenses are quite compact. Maybe I'm just not looking in the right places, but I haven't noticed anything like that for SLRs.

[stevebrot]

In theory, because of the shorter registration distance, a wide angle lens can be made without a retrofocal group, so that can save size and weight.

I own such a lens, though it is a bit of a pain. The Zeiss Biogon-derived Jupiter-12 35/2.8 for both Leica thread mount and Contax/Kiev rangefinder is quite tiny and is not retro focus. In fact, it is mildly telephoto...go figure. The reason it is a pain is that the rear element protrudes so deeply into the camera that the face of the rear element is less than 5mm from the shutter on my FED-2 and quite a bit closer on the R3M and Canon P. I have read reviews for the J-12 on the Sony A7 and the edge/corner performance suffered quite badly away from center due to severe angle of incidence to the sensor.


Steve

[clackers]

This seems like a logical explanation, and yet it doesn't jibe too well with my experience.

Back in the film days, it was unusual to see a rectilinear lens for SLR cameras wider than 20mm, and even those tended to be expensive. Wider lenses did exist for rangefinders, though. When the Sony A7 hit the market, people were very soon adapting those rangefinder lenses to it. They found with very wide lenses there was a problem of vignetting and color shift due to the incident angle, as you alluded. However, newer models have improved microprisms to mitigate that effect, and companies like Voigtländer have easily adapted their rangefinder lens formulas for mirrorless. We’ve seen rectilinear lenses from Voigtländer for Sony in 15mm, 12mm and even 10mm with full frame coverage, and the lenses are quite compact. Maybe I'm just not looking in the right places, but I haven't noticed anything like that for SLRs.

Of course you can make them that size for the SLRs, I have a Voigtlander 20mm f3.5 for my K-1, but why, Tony?

It's light, but the pictures aren't sharp except in the centre, lots of distortion, plenty of CA. It focuses quite closely so it's almost pseudo-macro.

A prime is meant to be superior optically to a zoom, but that 12mm Voigtlander is f5.6, like your kit 18-55.

Here's how bad it is: Review: Voigtlander 12mm 5.6 Aspherical Ultra Wide Heliar - phillipreeve.net

Check out the vignetting at f8, f11. Its poor sharpness mirrors my 20mm. You can't cheat physics.

My Irix 15mm f2.4 is big and beautiful, making big and beautiful images.

And microprisms are a kludge, with an SLR you don't need them at all, this extra obstacle for the light, they just degrade the image further and add artefacts, there's no upside to them, they put them in because the mirrorless registration distance is unsatisfactory - light is not coming in parallel from the rear element.

---

Last edited by clackers; 11-24-2020 at 10:32 PM.

[Rondec]

My impression is that size of lenses has to do with registration distance of the mount, size of the sensor needing to be covered, and how corrected the optics are.

So, you can make an M40 f2.8 that is tiny for an SLR and covers a full frame sensor, because it is close in focal length to the registration distance and the optical folks didn't do a lot of corrections in the design. Lenses like the DFA *50 and DFA *85 are very large, not because they are SLR designs, but because Pentax threw the kitchen sink at them to try to make them sharp wide open and with minimal aberrations. I think in many cases Sony is content to use software manipulation to take care of flaws in their lenses. That may be fine, but it is part of the reason that some of their lenses are smaller than SLR counterparts.

[zapp]

A mirror makes a difference in lens design. Particularly for any FL shorter than the flange focal distance, here retrofocus designs are necessary. Modern high Performance Glass also applies retrofocus design to mirrorless mounts for better quality and incident light ray angle towards the edges of a digital sensor.
So, no Universal answer here that hold for film, digital, slr, mlc...

[stevebrot]

I own two 180° fisheyes, both in Pentax K mount. One covers the 24x36mm FF format and the other APS-C. Conventional wisdom would hold that the smaller of the two should be the APS-C lens (Rokinon 8/3.5 Fisheye on right), but that is not the case.


Note that the Rokinon is the same lens in the comparison for this thread.

Below is a photo lifted from an ephotozine article. On the left is a FF fisheye and on the right an APS-C fisheye. Note that both are of similar size.


From: Nikon 10.5mm Zenitar 16mm Fisheye Lens Comparison Interchangeable Lens Review | ePHOTOzine

FWIW, if I had a comparison photo of the FF Sigma 15/2.8 EX DG Fisheye, it would be about the same size as the Zenitar 16/2.8. For the most part the Samyang 8/3.5 fisheye is quite a bit larger in length, girth, and front element size than other fisheye lenses across brands.

Question? What do either of those photo have to do with the subject of this thread, other than both involving fisheyes and comparison? The common element is the Samyang 8/3.5 Fisheye and the question might be whether Samyang might create an SLR version of the new compact design. If there were money to be made, they might.

What I am thinking is the video is a comparison of old vs. new Samyang lens generations and that the size difference has little to do flange focal distance.

Note: It is frequently asserted on this site that APS-C allows for more compact lens design and that short flange focal distance has a similar benefit. The latter is backed by the larger body of evidence, but even then there are exceptions to prove the "rule", say for lenses over 80mm.


Steve

[house]

Its interesting to see how this short registration distance plays out. We know that it simplified wide angle lenses but that the rays come at to steep angles for digital sensors.

Recent mirrorless wides seem very good. That sony 20mm looks excellent. By the way sony forces less software corrections than canikon right?

So the question becomes are there still over all advantages even when you have to use retrofocal designs. Perhaps its still easier to correct some aberrations.

[Alex645]

A quick comparison of the 8mm Samyang for Fuji against the 8mm Rokinon for Pentax.

The YouTube video is somewhat misleading in that both lenses are called 'fisheye' but the 8mm Samyang f/2.8 on a mirrorless APS-C has a corner to corner to image, whereas the 8mm Rokinon f/3.5 can be used with a FF DSLR showing the image circle and surrounding black corners and edges.

Personally I'd rather have the option of a circular image (or cropping it for the image only) on a FF DSLR than not. Half a stop difference on an ultrawide is also not a big deal as it can be handheld at much slower shutter speeds than a longer focal length lens.

Yes, if you want the compactness and lightweight of a mirrorless system, the video makes its point. But in the same way he fumbled one lens, there is more to the comparison that is not explained.

[Fogel70]

Its interesting to see how this short registration distance plays out. We know that it simplified wide angle lenses but that the rays come at to steep angles for digital sensors.

Recent mirrorless wides seem very good. That sony 20mm looks excellent. By the way sony forces less software corrections than canikon right?

So the question becomes are there still over all advantages even when you have to use retrofocal designs. Perhaps its still easier to correct some aberrations.

A shorter register distance give the designer more freedom in designing the lens even if using retrofocal design. Moving the lens further away from the sensor has no advantage for optical design of the lens.

An intresting case are the two retrofocal Zeiss 18/2.8, one for mirrorless and one for DSLR. The size does not differ so much, but optical design and weight differs alot (and price).
Batis for mirrorless: ZEISS Batis 2.8/18 | Fullframe autofocus lens for Sony ? series
Milvus for DSLR: ZEISS Milvus 2.8/18 | Manual focus lens for Canon & Nikon SLR cameras

[stevebrot]

A shorter register distance give the designer more freedom in designing the lens even if using retrofocal design. Moving the lens further away from the sensor has no advantage for optical design of the lens.

The first point is true, though the combination of mount diameter and short register can put a crimp on things. The second point is often false. Angle of incidence is important.


Steve

[house]

The first point is true, though the combination of mount diameter and short register can put a crimp on things. The second point is often false. Angle of incidence is important.


Steve

I know nothing about optical design but I though you used the retro focus design to make the rays more parallel. If the rays are parallel the short register is not much of an issue? Assuming the rays aren't parallel the milc lens sitting close to the sensor can have a larger rear element with the same angle of incident as a slr lens where the rear element will shrink with the distance from the sensor.

I can't make much sense out of the Batis Milvus comparison. Both distagons buth the former have newer more fancy glass in addition to a simpler formula. What is due to the new glass elements and what is milc vs slr is hard to know?

---------- Post added 11-25-20 at 01:05 PM ----------

That sony 20mm looks excellent.

Just saw that dpreview agrees with me on that one... chose it for some best 2020 award. Funny because is saw samples and though this looks spectacular then checked some reviews which were meh.