[falconeye]

I wanted to publish the following idea (before it gets patented)


By replacing the mirror box of a conventional single lens reflex camera by a trichroic prism assembly with three attached monochrome cmos image sensors and an electronic viewfinder with contrast autofocus, I achieve the following:

1. A continued use of lenses made for the old lens mount
2. An improved low light capability
   (possibly 500% better -- 80% rather than 40% quantum efficiency times 3x because of not filtering 2/3 of colors).
   This is about the improvement of uncropped MF over APS-C!
3. An effective doubling of resolution compared to a Bayer pattern.
4. A possible quadrupling of resolution when using cross channel subpixel information.
5. 2.5 EV better dynamic range (which can be 16 EV with APS-C sensors).
6. Better color rendition as witnessed by "3 ccd" camcorders.
7. No need for an AA filter just to avoid ugly color moiré.
8. A possibly late justification to use APS-C sensors within an FF mount, depending on the registration distance and possible trichroic prism assembly layouts.
9. A justification to call it reflex even after the mirror is gone
10. Continued use of damn cheap APS-C sensors. This camera can be made cheap...

Below is a diagram illustrating the idea. Enjoy

---

Last edited by falconeye; 06-15-2011 at 05:29 AM.

[falconeye]

Possible disadvantages I didn't want to include im my OP :

• Added weight due to solid glass prisms.
• Added calibration step to perfectly align all three chips (may be replaced by software calibration though).
• No easy way to maintain sensor shift shake reduction (possible but three times as expensive).
• A possible loss of EVIL style camera miniaturization.
• Full frame may not be possible with a full frame mount.
• Possibly reflections across additional glass surfaces.
• I never heard nobody else ever talking about this kind of SLR. So, I may be missing something important

I have the feeling that a digital MF camera could be the most likely contender for this type of camera.

[16N64]

I wanted to publish the following idea (before it gets patented)


By replacing the mirror box of a conventional single lens reflex camera by a trichroic prism assembly with three attached monochrome cmos image sensors and an electronic viewfinder with contrast autofocus, I achieve the following:

1. A continued use of lenses made for the old lens mount
2. An improved low light capability
   (possibly 500% better -- 80% rather than 40% quantum efficiency times 3x because of not filtering 2/3 of colors).
   This is about the improvement of uncropped MF over APS-C!
3. An effective doubling of resolution compared to a Bayer pattern.
4. A possible quadrupling of resolution when using cross channel subpixel information.
5. 2.5 EV better dynamic range (which can be 16 EV with APS-C sensors).
6. No need for an AA filter just to avoid ugly color moiré.
7. A possibly late justification to use APS-C sensors within an FF mount, depending on the registration distance and possible trichroic prism assembly layouts.
8. A justification to call it reflex even after the mirror is gone
9. Continued use of damn cheap APS-C sensors. This camera can be made cheap...

Below is a diagram illustrating the idea. Enjoy

Similar technology is already used in 3CCD cameras, but here the sensors are usually much smaller. Note also that the amount of glass needed for the prism would rise approximately with 3rd power of sensor size, so this solution might be expensive.

Stefan

[falconeye]

Similar technology is already used in 3CCD cameras, but here the sensors are usually much smaller. Note also that the amount of glass needed for the prism would rise approximately with 3rd power of sensor size, so this solution might be expensive.

Yeah, I noticed that.

I agree with the 3rd power remark but noticed that the same applies to the pentaprism which doesn't seem to be excessive in cost or weight. I guess the volume would be about 3x the volume of a 100% VF pentaprism (minus the obsolete pentaprism itself) meaning adding the weight and cost of 2 pentaprisms.

The weight of the Pentax LX FA-1 35mm pentaprism viewfinder is 95g. This includes the prism housing and eye piece. So I guess, my idea is adding considerably less than 200g to a camera body, probably less than 100g.

---

Last edited by falconeye; 03-03-2010 at 08:03 AM.

[16N64]

Yeah, I noticed that.

I agree with the 3rd power remark but noticed that the same applies to the pentaprism which doesn't seem to be excessive in cost or weight. I guess the volume would be about 3x the volume of a 100% VF pentaprism (minus the obsolete pentaprism itself) meaning adding the weight and cost of 2 pentaprisms.

The weight of the Pentax LX FA-1 35mm pentaprism viewfinder is 95g. This includes the prism housing and eye piece. So I guess, my idea is adding considerably less than 200g to a camera body, probably less than 100g.

I would guess the optical precision needs to be much higher here as for the viewfinder pentaprism, as it is in the main optical path to the sensor. I can imagine that for current DSLR pentaprism you can get away with cheap but accurate plastic optics, for the dichroic prism this might not be an option.

Stefan

[falconeye]

I can imagine that for current DSLR pentaprism you can get away with cheap but accurate plastic optics, for the dichroic prism this might not be an option.

Hi Stefan,
einem Bonner kann man nix vormachen

You urge me to dig deeper again I found this specification from a Chinese pentaprism manufacturer:

- Material: Glass BK7
- Surface planarity: λ /4

I guess that plastic prisms are not used because penta mirror view finders are used to save money instead. I found other sources saying that SLR use glass prisms.

[Marc Sabatella]

Fascinating idea. Not sure if this is the right forum for it, but then, since you're talking about something new, it's also not clear where else would be...

Moved to General Talk

[falconeye]

Fascinating idea. Not sure if this is the right forum for it, but then, since you're talking about something new, it's also not clear where else would be...

It fits "General Photographic Talk". We don't have this sub forum. As it is photography-related, "General Talk" is the wrong sub forum.

[Ratmagiclady]

Neat concept. The prisms might be heavy, but the weight'd be in a good place. You could get pretty wild with the shape, too. (I'm thinking, maybe, oversized spotmeter with a handle (containing battery and lots of the works and controls) that rotates.

It fits "General Photographic Talk". We don't have this sub forum. As it is photography-related, "General Talk" is the wrong sub forum.

Fair enough, a new design of camera is definitely not a photographic style or technique and should not have been where it was originally started.

Let's settle for the Off Brand forum shall we?

[falconeye]

Let's settle for the Off Brand forum shall we?

Leave it where it is. It doesn't matter really.
AFAIK, Lumolabs isn't another brand (yet)



On another note ...

I am learning that Foveon studied the idea in 2000 with their Foveon studio camera
-> http://dicklyon.com/tech/Photography/PICS2000-Lyon.pdf

It wasn't mount compatible with anything. But in 2000, they concluded: "We expect that due to its inherent advantages of one-shot alias-free color-accurate sensing, the prism-based color separation approach will become the dominant approach for high-end professional photography".

<UPDATE BEGIN>
In another post kxr4trids provided an interesting link with images to the Foveon studio camera:

<UPDATE END>


Yet another detail. I found dimensions for the Philips type trichroic prism assembly (TPA) as invented in Europe by Philips in 1960 (cf. attachments below).
[source: Seperation Prism Technical Data]
If I take the entire width of the exit planes in the diagram (about 1 inch each), the entire length of the assembly would be 1.75x the width.

With a chip height of 16mm, this would require a registration distance of 28mm.
With a chip height of 24mm, this would require a registration distance of 42mm.

The Pentax K mount has a registration distance of 45.46mm, so both sensor sizes should be feasible, although the FF type is on the edge of things.

If I only take the width of the exit planes as drawn in the diagram (denoted "20"), then the entire length of the assembly would be 3.9x the width.

Even with a chip height of 16mm, then this would require a registration distance larger than K mount. But I believe the area denoted "20" serves illustration purposes only -- as I don't see what is special with the inclination angle of the initial ray.



EDIT: I just found their spec sheet Separation Prism Configuration

They specify for a 35mm TPA suitable for f/2.8: 115mm length and 2000g weight. Clearly too heavy and too large for my idea.

Closest comes their version B9901 for a chip 15mm high: 52mm length and 160g weight.


They specify an optical air path-length, which is considerably less: E.g., 29.5mm rather than 52mm for the B9901. It may well be that the prism may be larger than the mirror box as long as the optical air path-length doesn't exceed the registration distance. This would be due to the diffraction index of glass within the box. If so (I have to think about it -- and if this doesn't induce optical aberrations...), then the TPA fits an enlarged mirror box for APS-C sized sensors, even when supporting f/1.4.

[falconeye]

This image is their 3-port RGB with IR filter & Canon mount. So, it has already been done in an experimental way ...

They even have a product sheet for Nikon lenses:
http://www.optec.eu/eng/multichannel/THREE%20CHANNEL%20PRISM%20RGB9901.pdf
They report 15x20mm as max. sensor size. I got the feeling however, that they didn't try to cover full APS-C. This design should cover 16x24 as well (when made wider).

[er1kksen]

This idea has been getting bounced around on the Oly forums for a long time (Olympus acquired some related patents a few years ago and the 4/3 "telecentric lens" specification is particularly well-suited to this setup) and many still expect it to show up on the next 4/3 flagship. Who knows.

One remark about the advantages you list: you note up to 500% of the low-light performance due to the lack of color filters on the sensor. Is there not any light loss in the prism? The light is still passing through SOME kind of filters (or selectively reflective surfaces, as that may be) in order to separate the spectrum.

[pingflood]

Prokudin-Gorskii would approve.