The Omen -anamophic bokeh

[Carl Looper]

It's an interesting question why the larger the bokeh, the more horizontally squeezed it is. I haven't yet figured out how to model this.

 

But given this as a visible effect, would not an oval shaped iris be useful behind an anamorphic lens? You'd make the iris wider than it is taller. The in-focus bokeh would then come out wider than it is tall, but being in-focus it would be too small to see (so no problem), while the out-of-focus bokeh (what we normally mean by bokeh) would come out circular - assuming this being a more desirable outcome than squeezed bokeh.

 

C

[Satsuki Murashige]

I don't know either. If I were to guess, I'd say the largest circles of confusion usually represent point sources at the far distance point from the plane of focus. So since it's physically further away, it would be squeezed more. Not sure if that's logical, but it's the best I can come up with.

 

Hey, people pay a lot of money for the squeezed bokeh! It's a feature, not a bug. If these folks wanted 'normal' bokeh, they would probably just shoot spherical.

[Carl Looper]

Hey, people pay a lot of money for the squeezed bokeh! It's a feature, not a bug. If these folks wanted 'normal' bokeh, they would probably just shoot spherical.

 

Ah okay. So the oval iris it's just a way of faking an anamorphic lens bokeh with an otherwise spherical lens. And Panavision can take from that there's no need to implement an oval iris (of the opposite orientation) in their anamorphic lenses.

 

C

Edited April 10, 2016 by Carl Looper

[Satsuki Murashige]

 

Ah okay. So the oval iris it's just a way of faking an anamorphic lens bokeh with an otherwise spherical lens. And Panavision can take from that there's no need to implement an oval iris (of the opposite orientation) in their anamorphic lenses.

 

C

 

Yup. Now, Richard Gale has an interesting idea to take the 1.5x Iscorama and add a very mild 1.33 oval iris in order to get a true 2x Panavision-style bokeh. But it's probably more trouble than it's worth.

[Satsuki Murashige]

Btw, very interesting anamorphic lens design discussion on Andrew Reid's site EOSHD. Someone found the US Patent for the Cooke Anamorphic lenses online:

 

http://www.eoshd.com/comments/topic/19578-where-did-all-the-single-focus-solutions-come-from/

[Carl Looper]

I don't know either. If I were to guess, I'd say the largest circles of confusion usually represent point sources at the far distance point from the plane of focus. So since it's physically further away, it would be squeezed more. Not sure if that's logical, but it's the best I can come up with.

 

I've got a rough conceptual model going on but it's not yet very clear. It goes something like this:

 

So an anamorphic lens system can be regarded as an otherwise normal lens but with two focal lengths. A vertical one and a horizontal one. The horizontal one gives a wider angle of view than the vertical one. Or to put it another way the horizontal focal length is shorter than the vertical focal length.

 

Now we know that for the same iris, the shorter a lens the smaller the bokeh. Or the longer the lens the larger the bokeh. And so using our conceptual lens with two focal lengths, the bokeh will be smaller in the horizontal than it is in the vertical. An oval bokeh. And this will be on top of the image squeezing and stretching that otherwise goes on in the same system.

 

Since bokeh (or circle of confusion) is also related to focus distance there might very well be an asymmetrical variation between the horizontal and vertical as a function of distance from the focus 'plane'.

 

Just need to test this with a mathematical model for depth of field and bokeh, ie. plugging in two different focal lengths and getting the corresponding bokeh size, for different focus distances. And seeing if this tallies with the observables.

 

C

Edited April 10, 2016 by Carl Looper

[Carl Looper]

I should just add that by "same iris" I mean the same f-number (eg. 5.6). Insofar as the effective diameter of the iris (for calculating DOF or calculating bokeh) is a function of focal length divided by the f-number (eg. f/5.6), our conceptual lens with two focal lengths would then have a smaller diameter in the horizontal than it does in the vertical, ie. an oval shaped iris. This shape is just the "effective" shape rather than a literal one (for purposes of calculation).

 

Another way of saying this is that the depth of field is larger horizontally than it is vertically. Or the image blurs more vertically than it does horizontally (the further away a point is from the focal plane on one side and the focus plane on the other side).

 

C

Edited April 10, 2016 by Carl Looper

[Carl Looper]

Yep - so this model of a conceptual lens with two focal lengths is working. As a focal point moves away from the film/sensor plane, the difference in bokeh size, between the vertical and horizontal will diverge, with the vertical getting taller at a faster rate than the horizontal gets wider. Assuming a 2X anamorphic (one focal length is twice the size of the other), a focal point at infinity would see the vertical bokeh twice the size of the horizontal bokeh.

 

Edited April 10, 2016 by Carl Looper

[Carl Looper]

In case there is any confusion in the above, I'm using the term "focal plane" in the diagram somewhat incorrectly (and using the term "focal point" in the text somewhat incorrectly). The indicated plane (far left of the diagram) is just where a point in world space (and otherwise "out of focus" there) corresponds to a point in camera space (and likewise "out of focus" there). The focal plane, properly speaking, is the film/sensor plane, and the image in this plane corresponds to the image occupying the focus plane in world space (at a distance from the lens corresponding to the lens focus setting).

 

C

Edited April 10, 2016 by Carl Looper

[Robin R Probyn]

Just as an aside..or if you are talking to a bunch of Japanese from the Osaka area.. Boke can also be used as an insult.. to mean someone is stupid !..

[David Mullen ASC]

It's more like "baka" for stupid in Japanese instead of "bokeh"...

[Robin R Probyn]

Not in Osaka/Kansai area.. Boke is often used instead of Baka.. and Boke is also used to describe dementia .. and someone who is forgetful/ not that bright..

Edited April 11, 2016 by Robin R Probyn

[Satsuki Murashige]

Robin, are you based in Japan? If so, I'll have to look you up next time I'm there. I was in Osaka/Kansai for a week this time last year. As I recall, they had their own words for a lot of things. 'Metcha' for 'hella' being one of the most memorable.

[Robin R Probyn]

Hi Satsuki.. Yes but Im based in Tokyo.. although often down in Osaka area.. yes they do have alot of their own words in Kansai.. instead of the usual "genki desk ka" are you healthy.. they say "Mo karimakka" meaning hows your business/are you making money. tend to be much more direct than Tokyo speak.. !

 

Yes if your ever in Tokyo let me know and hopefully meet up sir..

[Satsuki Murashige]

Sounds good, will do! I had heard from friends and relatives down in Yamaguchi/Hiroshima that Osaka natives tend to be much more direct and less patient than Japanese elsewhere. More American-style, basically. I didn't really find that to be true, but again I was only there a week. Have not spent any time in Tokyo other than flying in and out of Narita, will have to rectify that soon.

[Robin R Probyn]

Hiroshima is one of my favorite cities here.. yes please let me know if your Tokyo bound sir.. San fran has a pretty strong Japanese link doesnt it..

[Satsuki Murashige]

Thanks! Yup, most of the west coast cities do. Vancouver, Seattle, SF, and LA in particular.

[Carl Looper]

The model I was using for anamorphic bokeh is a bit too clumsy.

 

Its really quite a peculiar thing this anamorphic bokeh.

 

I've managed to put together a more robust model. The cone of light emerging from an iris behind an anamorphic lens in camera space is more of a four dimensional hyper-cone than a regular three dimensional cone. The tip of this cone (the focus 'point' of the wave) is stretched out into a line along the cone's axis. Bokeh becomes a 2D cross-section of this 4D hypercone. Bokeh in the near foreground I'm finding should be horizontally stretched. At the beginning of the cone's tip the bokeh becomes a small line. Half way along the tip it becomes a small circle and then at the end of it's tip it becomes a small line again, before stretching out into a vertically stretched oval.

 

I've yet to test the model against actual images.

 

C

Edited April 12, 2016 by Carl Looper