Canon Patent Application: Small Primes for Compact APS-C and Full Frame

Wow, those super aspheric rear elements. I guess these designs are exploring the "simple, low aberration" space of "simple, low aberration, no exotic elements, pick any two"

Sharlin said:

Wow, those super aspheric rear elements. I guess these designs are exploring the "simple, low aberration" space of "simple, low aberration, no exotic elements, pick any two"

Click to expand...

I suspect those will the PMO, like the RF28mm.

Canon please pump out more PMO cheap lenses

Isn't the first formula the one of the RF28 STM? It looks at least pretty similar, doesn't it?
(Sorry, didn't have the time for detailed searching and comparing)

But I'd take a similar 24mm pancake, too.
And a friend surely would like to pair that 33mm F1.8 APS-C with his R7

Maximilian said:

Isn't the first formula the one of the RF28 STM? It looks at least pretty similar, doesn't it?
(Sorry, didn't have the time for detailed searching and comparing)

But I'd take a similar 24mm pancake, too.
And a friend surely would like to pair that 33mm F1.8 APS-C with his R7

Click to expand...

Good catch, and yes it certainly appears to be. Post on CR:



RF 28/2.8 block diagram:


Sorry, Richard...looks like there might be something more you need to take into account...

neuroanatomist said:

Good catch, and yes it certainly appears to be. Post on CR: ...

Click to expand...

Thanks, @neuroanatomist, for doing the work.
I was about to do it later, but you came first

Of course, it directly pops out that the shape of the fourth element seems to be more complex in the RF 28/2.8 block diagram.
Generally, the second group seems to be a bit more complex in the final design.
But I suppose that this was a change in development during the final design of the RF28/2.8

koenkooi said:

I suspect those will the PMO, like the RF28mm.

Click to expand...

What's PMO?

SwissFrank said:

What's PMO?

Click to expand...

Precision MOulded aspherical lens, plastic/resin by any other name.

neuroanatomist said:

Good catch, and yes it certainly appears to be. Post on CR:

Click to expand...

Just to be clear, you're saying the sample embodiment is actually the current 28/2.8?

I think the current-production lens cannot be that shown as an example embodiment of the lens formula patent:

1) You cannot patent something that's already public, even if you make it. It has to be secret. I imagine there's some special cases if it's public due to someone leaking it, etc., but a current product would be absolutely impossible.

2) Both diagrams are strictly informational and need not match the actual patent and product's elements' shapes and dimensions religiously. That said, the diagrams are similar... but differ. Counting from the front, elements 4 and 5 are clearly far smaller in the application while 1 and 2 are bigger. It's not just a question of pixels of difference here, for instance plausibly due to the printout resolution, but substantially different.

SwissFrank said:

Just to be clear, you're saying the sample embodiment is actually the current 28/2.8?

Click to expand...

In essence, yes. The actual product need not exactly match the patent examples (but it must fit within the claims). It’s also possible the patent application was modified and/or amended during review.

SwissFrank said:

I think the current-production lens cannot be that shown as an example embodiment of the lens formula patent:

1) You cannot patent something that's already public, even if you make it. It has to be secret. I imagine there's some special cases if it's public due to someone leaking it, etc., but a current product would be absolutely impossible.

Click to expand...

You need to read up on the patent process. What we’re looking at is a published patent application that was filed 18 months ago. It published on June 19, 2024 meaning it was filed on December 19, 2022. That’s the date before which prior art is actually prior. Was the RF 28/2.8 a launched product on Dec 19, 2022? No.

SwissFrank said:

2) Both diagrams are strictly informational and need not match the actual patent and product's elements' shapes and dimensions religiously. That said, the diagrams are similar... but differ. Counting from the front, elements 4 and 5 are clearly far smaller in the application while 1 and 2 are bigger. It's not just a question of pixels of difference here, for instance plausibly due to the printout resolution, but substantially different.

Click to expand...

Yes, and the first part explains the second. Diagrams in a patent are examples, as long as the product satisfies the claims in the patent, it’s covered by the patent. It’s normal for slight design changes to be made as a patented invention moves into mass production.

neuroanatomist said:

Was the RF 28/2.8 a launched product on Dec 19, 2022? No.

Click to expand...

OK, I didn't note that the RF28/2.8 was released July 2023. I thought it was older. I don't know why you feel a need to phrase your comment as a put-down. I've seen you put down others on multiple occasions.

neuroanatomist said:

The actual product need not exactly match the patent examples (but it must fit within the claims).

Click to expand...

Sure. What's patented here is the mathematical model of the lens formula, not the example embodiments. I used to remind the group admin of that every six months or so.

neuroanatomist said:

You need to read up on the patent process.

Click to expand...

I worked on a half-dozen Japanese patents in the 1990s at Omron, before I switched to finance. I can't think of why I'd need to learn more about it.

That 33 mm suggests that we may see an RF-s replacement for the EF-M 32mm that is lighter and possibly cheaper with excellent performance if the RF 28mm is to be taken as an example.

Dragon said:

That 33 mm suggests that we may see an RF-s replacement for the EF-M 32mm that is lighter and possibly cheaper with excellent performance if the RF 28mm is to be taken as an example.

Click to expand...

That doesn't make a huge amount of sense to me. Lens designs for about the 20-50mm range (in terms of 24x36mm sensors) are much simpler when they don't have to leave a lot of room for the flipping mirror. The natural way to design these lenses results in rear elements closer to the sensor than the mirror would allow. (Past 20mm the lens design starts getting complicated again no matter what you do.)

But the EF-M lenses ARE mirrorless designs, with very short film-to-flange distances. They can't be used on cameras with mirrors, unlike EF-S, right? So an EF-M has basically all the design freedom an RF-S does.

If it is legitimately lighter and cheaper and not suffering in some other way (bigger, less sturdy, slower AF, more aberrations, worse OOF bokeh, etc.), that'd be real news. The only things I can think of are that RF lenses no longer have to care about vignetting or distortion, since these can be undetectably*** fixed digitally. It's possible that new lenses are letting these two metrics slide into the wastecan, and using the resulting freedom to improve every other aspect of the lens (e.g., making it lighter AND cheaper AND higher resolution AND lower aberration AND better OOF highlights etc.).

---

*** I know others say differently but I've asked on this forum politely for counterexamples 50 times and the only counterexample to my statement that I've seen was Milky Way astrophotography where vignetting correction caused a bit more visual noise in the far corners. I totally recognize that as an exception to my statement above, but I would also say that it's a very very unusual and specific case where ISO is already very high, corners EXCEEDINGLY dark, aperture already fully wide, and not much freedom to lengthen exposure, on top of using an ultra-wide angle lens which are typically more prone to cos^4 vignetting on top of other vignetting. Remove any one of those factors, and the digital correction of vignetting would have no visible effect. EG if ISO were lower, no noise. If corners were zone III+, no noise. If aperture not fully open, you could open up, reduce ISO, and no noise. If the subject weren't in very slow motion, you can use a longer exposure, and lower ISO, so no noise.

And as for distortion correction I've had lots of people proudly say that the distortion correction lowered image quality too far for their exacting requirements, but never have seen an actual example of something where you can even tell it was in use.

SwissFrank said:

That doesn't make a huge amount of sense to me. Lens designs for about the 20-50mm range (in terms of 24x36mm sensors) are much simpler when they don't have to leave a lot of room for the flipping mirror. The natural way to design these lenses results in rear elements closer to the sensor than the mirror would allow. (Past 20mm the lens design starts getting complicated again no matter what you do.)

But the EF-M lenses ARE mirrorless designs, with very short film-to-flange distances. They can't be used on cameras with mirrors, unlike EF-S, right? So an EF-M has basically all the design freedom an RF-S does.

If it is legitimately lighter and cheaper and not suffering in some other way (bigger, less sturdy, slower AF, more aberrations, worse OOF bokeh, etc.), that'd be real news. The only things I can think of are that RF lenses no longer have to care about vignetting or distortion, since these can be undetectably*** fixed digitally. It's possible that new lenses are letting these two metrics slide into the wastecan, and using the resulting freedom to improve every other aspect of the lens (e.g., making it lighter AND cheaper AND higher resolution AND lower aberration AND better OOF highlights etc.).

Click to expand...

The EF-M lens has 14 elements (including one Glass Molded Aspheric). The design in the patent has only seven elements and appears to have at least two and possibly three molded elements that may well be some kind of optical plastic. The back two are much more aggressively aspheric than traditional glass molded aspherics. That change in technology would account for my initial comment, "lighter and possibly cheaper, with excellent performance". The RF 28mm makes that case very effectively. Are electronic corrections being thrown into the mix as well? Possibly, but the 28 is pretty decent without that. Molded lenses are the way phones make cheap and quite good (for the size of the imager) cameras. Molded lenses are very cheap to manufacture once you have invested in the outrageously expensive mold (and Canon may well have developed a way to make the molds less expensively).