Patent: Canon RF APS-C pancake prime lens optical formulas

[John Wilde]

Strictly speaking, Canon's strategy document shows that they should release about eight new lenses this year.

 

[Jasonmc89]

There's really nothing there to hold on to.

So you’ve never used a point and shoot? Or a camera phone?

Pancakes really aren’t designed for long, demanding photo shoots. They’re supposed to be small.

 

[neuroanatomist]

Okay, but how do you hold the camera? I usually hold the weight of the camera in my left hand by putting my left hand under the CG of the camera-lens combination, which is always somewhere along the length of the lens. With a small prime, the CG is in the body and there's no place for me to put my left hand. That leads to trying to hold the body with my left hand on the left edge of the body where there's no grip.

Really not sure what the issue is here. You just hold the camera from the bottom. Most cameras have one.

Personally, I use an E1 hand strap on my bodies, and that lets me support the weight on camera + pancake with my right hand. I just need light touch of my left hand for stability.

 

[Lee Jay]

So you’ve never used a point and shoot? Or a camera phone?

Pancakes really aren’t designed for long, demanding photo shoots. They’re supposed to be small.

Yeah, and they're just as hard to hold. Camera phones are much worse because they're both too small and have no viewfinders.

For me, one of the big attractions of an ILC is that it's just so much easier to hold than a compact or a phone.

 

[Lee Jay]

Really not sure what the issue is here. You just hold the camera from the bottom. Most cameras have one.

Personally, I use an E1 hand strap on my bodies, and that lets me support the weight on camera + pancake with my right hand. I just need light r
touch of my left hand for stability.

That's just poor technique. Holding under the CG is more stable both because forces don't translate into rotations and because your left arm can be braced against your torso. The right hand should be holding no weight, just operating controls and helping with pointing.

 

[entoman]

You know, you don’t have to use a pancake lens..?

Slap that massive 50mm 1.2 on it if you want.

What's better, a pancake or a beer barrel? I'll take the beer!

 

[Franklyok]

What's better, a pancake or a beer barrel? I'll take the beer!

Planty of space for sensor ibis movement… looking forward to 25 stop ibis

 

[neuroanatomist]

That's just poor technique. Holding under the CG is more stable both because forces don't translate into rotations and because your left arm can be braced against your torso. The right hand should be holding no weight, just operating controls and helping with pointing.

My technique is fine, thanks. I get that you can only handle doing things one way, but that doesn’t mean your way is the only way. Would you have no weight on your right hand when handholding a 1-series body with a 600/4 II lens?

Have you ever used a hand strap on a gripped body? The weight is entirely on the back of your hand, your fingers aren’t supporting anything.

With a pancake lens, the camera body itself is the center of gravity. Put your hand under the camera. This just isn’t that hard.

Sounds like you speak from a lack of experience in this particular area.

 

[Dragon]

As you point out, the patent application makes no mention of "APS-C" or "RF-S".

The "specs" are, as you say, an informed guess - by a highly experienced commentator, Keith Cooper of Northlight.

I'd be extremely surprised if his guess was wrong.

If the larger image circle was purely to accommodate IBIS, I would have expected Canon to design *all* of its RF-S lenses to have this image circle, not just the pancakes under examination here.

I think it's more likely that the larger image circle has been chosen to minimise corner softness and vignetting, both of which are a bigger problem with wide-angle pancakes than with longer focal lengths.

But like Keith, I'm only guessing

These lenses are pancakes w/o IS, so more IBIS travel room is needed for good stabilization. With an IS lens, the IBIS needs to do less (if anything all), so the image circle can be smaller.

 

[entoman]

Just as an aside:

Can anyone explain why the angle of view of lenses (in every Canon patent application that I've seen) is given as a "half angle of view"?

 

[AJ]

The specs state "RF-S" so they must be for APS-C models.

Maybe a misprint, but perhaps Canon has deliberately elected to have a larger than normal image circle with these wide-angle optics, in order to minimise vignetting and softness in the corners?

Or am I being daft?

If so, that's opposite to what Canon did with full-frame. RF 16/2.8 and 24-105 kit lens and 24-240 need distortion corrections to be applied to avoid black corners. I would expect Canon would take the same approach with RF-S. As such I don't quite know what to make of these patents.

 

[neuroanatomist]

Just as an aside:

Can anyone explain why the angle of view of lenses (in every Canon patent application that I've seen) is given as a "half angle of view"?

The angle given is half of the total angle of view. In other words it’s the angle measured from the edge of the lens to the center/optical axis (as opposed to edge-to-edge, which would be the full angle of view).

It’s consistent with the value stated for image height, which is the radius (not the diameter) of the image circle.

 

[entoman]

If so, that's opposite to what Canon did with full-frame. RF 16/2.8 and 24-105 kit lens and 24-240 need distortion corrections to be applied to avoid black corners. I would expect Canon would take the same approach with RF-S. As such I don't quite know what to make of these patents.

Distortion correction seems to be applied to virtually all modern lenses, as it is easier and cheaper to produce a mediocre lens and correct it via firmware trickery, than to produce an optically better lens. But correcting distortion via firmware involves shrinking and/or stretching the edges/corners, which reduces sharpness. So it's better to start with a fairly decent lens design and thus reduce the amount of firmware correction needed.

Similarly you can "correct" vignetting via firmware (or in post), but this involves amplifying the signal in the corners, with a corresponding increase in luminance noise. So again, it's better to produce a lens design that does not have pronounced vigneting in the first place.

As others have pointed out, it would seem that the larger image circle is primarily to enable more efficient IBIS, (as these lenses don't have OIS), but it also should result in sharper corners and less obvious vignetting.

 

[entoman]

The angle given is half of the total angle of view. In other words it’s the angle measured from the edge of the lens to the center/optical axis (as opposed to edge-to-edge, which would be the full angle of view).

It’s consistent with the value stated for image height, which is the radius (not the diameter) of the image circle.

Thanks, I understood what the figures represented, it just struck me as strange that radius and half-angle were adopted as a convention, as opposed to diameter and full angle.

 

[Jasonmc89]

What's better, a pancake or a beer barrel? I'll take the beer!

Beer and pancakes..? Sounds like a decent lunch to me!

 

[AJ]

Distortion correction seems to be applied to virtually all modern lenses, as it is easier and cheaper to produce a mediocre lens and correct it via firmware trickery, than to produce an optically better lens. But correcting distortion via firmware involves shrinking and/or stretching the edges/corners, which reduces sharpness. So it's better to start with a fairly decent lens design and thus reduce the amount of firmware correction needed.

Similarly you can "correct" vignetting via firmware (or in post), but this involves amplifying the signal in the corners, with a corresponding increase in luminance noise. So again, it's better to produce a lens design that does not have pronounced vigneting in the first place.

As others have pointed out, it would seem that the larger image circle is primarily to enable more efficient IBIS, (as these lenses don't have OIS), but it also should result in sharper corners and less obvious vignetting.

Or else there is something else going on. Maybe a misprint on the image height, maybe something else. Yes it'd be an advantage with an ibis-equipped body, but it seems silly to me to have a larger image height than needed, especially for wideangle. That would be wasted optics.

 

[sanj]

Expected! Good news.

 

[f119a]

That's not APS-C. Closest would be Super35 or (gasp) APS-H.

The third example is the acutal design of RF 16mm f/2.8 STM.
The image height in patent is uncorrected data. And there's 20%+ distortion on all 4 designs - multiply 18.2 by 1.2 you got 21.8, which will cover full frame sensors.

 

[f119a]

Don’t get us excited.

These are all Full frame lenses. Image height is low due to their hugh amount of distortion. Just add the 20% distortion back and you get 21.8mm coverage (equiv.)
Even though the RF 16mm is capable of covering the whole sensor at infinity, those extreme corners are stretched and thrown away during correction anyway.

 

[f119a]

Continue reading...

There are not APS-C lenses.
Example 3 is the RF 16mm f/2.8 STM available on the street and I don't expect Canon to put the others into production at this time.