Patent: More Curved Sensor Information

[Loswr]

I too am curious about the potential use for a curve-able sensor. Would it help with lens design perhaps? to my knowledge, all current lenses were designed with a flat imaging sensor in mind..

That's the point. Current lenses often require additional elements to project a flat field, so a curved sensor would mean fewer elements, meaning smaller, lighter lenses and potentially better optics.

I do not believe, that it will be an Advantage to relinguish correcting a lens in regard to a flat projection plane. Canon is a Lens-Company - why should they give up their expertise in making good lenses?

In the case of wide-angle-zooms such an approach may result in an amount of changing curvature of the projection plan, that you have to adust the bending of the sensor in respect to multiple combinations of focal length (in case of a zoom-lens), distance and aperture. And it may be different in every single copy of an lense - remember that many extrem wide-angle-lenses are not adjusted exactly orthogonal to the plane of the bajonet. Nobody will do all those adjustments, its far too difficult.

Perhaps it will be interesting to have a curved sensor in the case of a Camera with a fixed ultra-wide prime. Something like a digital Version of the Hasselblad SWC.

I disagree with your first point. How would designing lenses for a curved sensor mean 'giving up their expertise in making good lenses'? It would merely allow them to make different – and quite likely better – designs. Any correction (flat field, CA, etc.) comes with trade offs, some of which then require additional corrections. Not having to flatten the field is one less compromise to make. From a business standpoint, it means new lenses for people to buy. Also, it means the potential to raise a price for a 'better' lens, while fewer elements means less cost to produce...and that means more profit.

As for the need to adjust the sensor curvature based on focal length, etc., Canon recently had a patent published for a sensor with electronically-controlled dynamically adjustable curvature, so they are clearly thinking beyond just a fixed lens, wide prime application.

 

[Berowne]

.... As for the need to adjust the sensor curvature based on focal length, etc., Canon recently had a patent published for a sensor with electronically-controlled dynamically adjustable curvature, so they are clearly thinking beyond just a fixed lens, wide prime application.

Neuro, this is really interesting, I didnt know. Sounds like an analogy to adaptive optics. I guess, it will be technically and computationally somewhat demanding to apply this principle to camera-sensors, remember in Astronomy you Need a Laser guide star to adjust the mirror of a telescope in real time.

I do not doubt, that Canon will be able to develop a somewhat similar System for DSLR's, but I think it will not be available for the consumer-market soon.

BTW: Happy New Year to you and your Family - Greetings from Germany - Andy

 

[Loswr]

.... As for the need to adjust the sensor curvature based on focal length, etc., Canon recently had a patent published for a sensor with electronically-controlled dynamically adjustable curvature, so they are clearly thinking beyond just a fixed lens, wide prime application.

Neuro, this is really interesting, I didnt know. Sounds like an analogy to adaptive optics. I guess, it will be technically and computationally somewhat demanding to apply this principle to camera-sensors, remember in Astronomy you Need a Laser guide star to adjust the mirror of a telescope in real time.

I do not doubt, that Canon will be able to develop a somewhat similar System for DSLR's, but I think it will not be available for the consumer-market soon.

BTW: Happy New Year to you and your Family - Greetings from Germany - Andy

I do think it's reasonable that the first application would be a niche fixed wide angle camera.

Happy New Year to you and your family, as well. —John

 

[Mikehit]

I can see a two-stage release - the first sensor would be on a fixed curve. This would still enable simpler lenses to be made (less flat-field correction needed), maybe with an adaptible sensor in mind for the future and eventually they go full-bore with an adaptible curved sensor.
Canon DPP software already has a lens correction function which can be quite impressive so this could be updated to account for a lens to be fitted to a flat sensor, fixed curved sensor and adaptible curved sensor. So someone buying a current lens could fit it onto a fixed-curve sensor with lens correction software available. Someone buying future lens manufactured for curved sensor would be able to put it on flat-sensor camera or a curved sensor camera.


I recall reading somewhere that up to 2/3 of the elements in a lens are there to correct image artefacts so the opportunities for reducing lens size/weight are significant.

 

[ISO64]

Curved sensor is still straight top to bottom, so image is affected/improved in one direction only. But, DPAF will work much better as the incoming ray angle at the sensor's extremes now comes closer to normal.

ISO64

 

[Loswr]

Curved sensor is still straight top to bottom, so image is affected/improved in one direction only.

Is that stated in the patent, or are you making an assumption based on the diagrams posted here?

 

[zim]

Curved sensor is still straight top to bottom, so image is affected/improved in one direction only.

Is that stated in the patent, or are you making an assumption based on the diagrams posted here?

That's exactly what I asked in (in a deferent way)
http://www.canonrumors.com/forum/index.php?topic=31461.0
with no answer, maybe it's obvious but not to be.

This would seem to me to be for very specialised applications.

 

[wockawocka]

I really wish Canon would stop playing around, and actually do something innovative. :

One of the reason Kodak had so much money after the death of film was the sheer number of patents they had.

 

[Berowne]

Curved sensor is still straight top to bottom, so image is affected/improved in one direction only.

Is that stated in the patent, or are you making an assumption based on the diagrams posted here?

That's exactly what I asked in (in a deferent way)
http://www.canonrumors.com/forum/index.php?topic=31461.0
with no answer, maybe it's obvious but not to be.

This would seem to me to be for very specialised applications.

For anamorphic lenses?

 

[ISO64]

Curved sensor is still straight top to bottom, so image is affected/improved in one direction only.

Is that stated in the patent, or are you making an assumption based on the diagrams posted here?

That's exactly what I asked in (in a deferent way)
http://www.canonrumors.com/forum/index.php?topic=31461.0
with no answer, maybe it's obvious but not to be.

This would seem to me to be for very specialised applications.

For anamorphic lenses?

Assumption based on the property of piezo elements that change length (extend or shrink) along a single axis only; you can disregard how much applying the voltage will result in change in total volume. So, if you wish the sensor to follow a spherical shape, at least two separately driven deforming layers would be required. Mechanical coupling between them would be a tug-of-war.

Image imperfections are getting progressively worse moving away from the axis so applying corrections in that area between image height square format and full 3:2 size is where the effort will be. The illustration on the patent application shows it clearly.