"mirrorless" with optical viewfinder

[hunterrichards]

Keep the form factor of a dslr but instead of a flipping mirror use a beam splitter... That way you can always look through the lens (with out looking at pixels!:sick: )

[Dan Hudgins]

Keep the form factor of a dslr but instead of a flipping mirror use a beam splitter... That way you can always look through the lens (with out looking at pixels!:sick: )

In this thread I said something about that issue,

http://www.reduser.net/forum/showthread.php?p=282349&posted=1#post282349

A cube beam splitter would only work with special lenses, and a Pellicle Beamsplitter is hard to clean (but can work with everyday lenses), so there are issues.

Another option is to look at the sensor face with off-axis optics, that way you do not harm the main image, but the eye-cup must fit tight so that you do not get any light going back from your eye to the sensor.

You could use a rotating shutter single blade or butterfly, a single blade would work with most SLR and reflex movie lenses and does not harm the image much. It would also let you use global reset and not get rolling shutter artifact, maybe put it below the frame like in a Konvas on a stepper motor or small servo motor in sync with the sensor. The mirror could be plastic since it is just for the viewfinder.

[Joseph Ward]

I know this sounds outrageous but is this possible? Have a spinning/vibrating/oscillating sensor/sensors, sort of like 35mm adapters, for unique image? :)

[Mike Prevette]

NO BEAMSPLITTER, NO PELLICLE MIRROR! Please, If you look at history it has never worked out well at all.

Hunter there have been several SLR's and film cameras over the years that have worked like that, and they all had huge drawbacks compared to normal slrs.

[DaveT]

I'm wondering if there is a way to use liquid crystals to turn mirrored effect within an angled glass plate, on and off electronically like in a LCD?

In one state the panel would be transparent and light hits the imaging chip. Flip the state and the panel reflects light up into the view finder???

[Dan Hudgins]

I'm wondering if there is a way to use liquid crystals to turn mirrored effect within an angled glass plate, on and off electronically like in a LCD?

In one state the panel would be transparent and light hits the imaging chip. Flip the state and the panel reflects light up into the view finder???

I have not seen anything that is transparent enough to not have haze that would affect the image. If you find something let me know. PLZT can be used as a shutter, there are some other things but I have not seen a good mirror high speed electronic mirror.

Also anything thicker than about 0.0002" will give a ghost image, since it is at 45 degrees to the sensor face. If you put a dust cover window over the Pellicle Beamsplitters it could be used, but some lenses stick back too far for that to be used with them.

Also an inclined plate will cause aberration in the image since one part is closer than the other, the OLPF does that but since it is Parallel to the sensor the aberation is more even over the image, not more on one side.

Spining morror is used on movie cameras because it gets out of the way for the exposure. You could use a Guillotine but on a large sensor that would make for too much mass to move at 24fps or faster, it goes back and forth like a piston. Guillotine are good for Super8 size sensors, not SLR size...

[Dan Hudgins]

NO BEAMSPLITTER, NO PELLICLE MIRROR! Please, If you look at history it has never worked out well at all.

Hunter there have been several SLR's and film cameras over the years that have worked like that, and they all had huge drawbacks compared to normal slrs.

The problem with the flip up mirror for a movie camera is that you cannot see what is going on while you shoot except for the electronic screen on the back of the camera.

Also the flip mirror will break, we have that problem in our scanner, the camera will break after a few thousand feet of negative scanned, the curtain shutter is even more of a problem, since if the DSLR mirror would lock up and STAY up you could shoot a million or more expsoures, but it will not STAY up (do you know of one that will?).

[DaveT]

OK, this is a bit 'out there' but what if the imaging chip was mounted on some sort of horizontal spindle.
On the back of the chip a mirror is mounted.
Spin the assembly so that the imaging chip, then the mirror is presented to the lens.
????

[NuclearNerd]

DaveT: What advantage does that have over a mirror in front of the sensor? Both solutions make optical viewfinder / sensor operation mutually exclusive. Unless maybe you flip the mirror / sensor continuously, dividing the exposure per frame between them. (Good as long as exposures are less than 16ms, and you'd probably notice a flicker in the viewfinder. Nothing comes without compromises!)

[fde101]

Is it possible to make a transparent image sensor and put the mirror behind the sensor, instead of in front?

[VikramTank]

I don't mind having a video viewfinder as long as it includes some good focusing guides. I've used the Leaf live view software, and it is adequate, but not ideal. Of course I'd like to have something that can be used as a waist level finder, and not just as a eye level finder.

[Shaocaholica]

Is it possible to make a transparent image sensor and put the mirror behind the sensor, instead of in front?

No. Even if you could, it would only mean that the image sensor is not collecting all of the light presented reducing its efficiency compared to a sensor that collects all the light.

[Shaocaholica]

It seems like the simplest solution is to use an EVF but instead of all the horrible EVFs of the past, really go all out like an extension of the new Panasonic G1s EVF which uses a high res(1600x900) LED-illuminated Direct-view LCOS system. I'm sure that could be pushed to 1080p with a higher pricepoint.

[Deanan]

It seems like the simplest solution is to use an EVF but instead of all the horrible EVFs of the past, really go all out like an extension of the new Panasonic G1s EVF which uses a high res(1600x900) LED-illuminated Direct-view LCOS system. I'm sure that could be pushed to 1080p with a higher pricepoint.

It's an 800x600 apparently.

[Kevin Olsen]

I mentioned optical viewfinder originally and I had the vision that there would be a 45 degree transparent mirror sending direct reflection to the sensor and passing through to the view finder:

light --------\-----optical viewfinder
.................|
..............sensor

I don't know if you could get the mirror to have the right amount of light pass through the mirror and provide correct color and at the same time deliver maximum light to the sensor. Maybe you'd loose some stops, I don't know.

KO

[fde101]

Hmm, maybe a mirror that can make itself transparent when current is passed through it -- like a HUD display, but reflective instead of just blocking?

[Alan Fletcher]

I mentioned optical viewfinder originally and I had the vision that there would be a 45 degree transparent mirror sending direct reflection to the sensor and passing through to the view finder:

light --------\-----optical viewfinder
.................|
..............sensor

I don't know if you could get the mirror to have the right amount of light pass through the mirror and provide correct color and at the same time deliver maximum light to the sensor. Maybe you'd loose some stops, I don't know.

KO

http://www.dpreview.com/reviews/olympuse10/
http://a.img-dpreview.com/reviews/olympuse10/images/e10cutaway.gif

Unfortunately you lose out both on the brightness in both the viewfinder and the sensor.

[Júlio Taubkin]

They know their EVFs. Just give us something good and the option of mounting the RED EVF for something great.

[Pietro Impagliazzo]

They know their EVFs. Just give us something good and the option of mounting the RED EVF for something great.

I'm sure DSMC built-in EVF will be 2x better than the current RED EVF.

:ninja:

[Jannard]

Why do you want a mirror? I forgot. Oh, I remember now... so you can't see in low light.

I remember going to CML early in the RED ONE development. The only advice I was given was that if we did not have an optical viewfinder, we wouldn't sell any cameras and wouldn't really have a professional camera. We haven't had one request for an optical finder in the last year and a half.

Jim

[Charles Angus]

Is it possible to make a transparent image sensor and put the mirror behind the sensor, instead of in front?

No. In addition to the previously mentioned reason, the image would be out of focus.

edited to remove redundancy.

[Charles Angus]

All ideas which involve something being between the sensor and the lens are impossible.

A flat piece of glass changes the optical path - Bolex came up against this on their reflex cameras. You need a specially designed lens (they call them RX lenses) to compensate.

The only advantage I see of an optical finder is that you can always see your frame, even if the system is off. I still double-take on EVF's when the AC has helpfully shut down the camera during some downtime.

It sounds stupid, but it is very nice to always be able to see.

[James T Mather]

IMHO - nonsense to all this optical finder twaddle - worry about the what's in front of the camera.

[fde101]

No. In addition to the previously mentioned reason, the image would be out of focus.


Isn't that the job of the lens -- to focus the image?

Why would a transparent sensor have any more of a focusing problem than an opaque one?

Just because the tech may not exist yet doesn't mean it's impossible...

[Alan Fletcher]

We haven't had one request for an optical finder in the last year and a half.
Jim

The only problem is :

RED EVF Price: $2,950.00
Nikon D700 Body : $2,999.99

I hope you can do everything else for $49.99 :holloween:

[Pietro Impagliazzo]

Why do you want a mirror? I forgot. Oh, I remember now... so you can't see in low light.

I remember going to CML early in the RED ONE development. The only advice I was given was that if we did not have an optical viewfinder, we wouldn't sell any cameras and wouldn't really have a professional camera. We haven't had one request for an optical finder in the last year and a half.

Jim

Way to go Mr.

Just shut up those nostalgic pricks with some new kick ass EVF.

:red_bandana:

[Lee Jay]

Why do you want a mirror? I forgot. Oh, I remember now... so you can't see in low light.

I remember going to CML early in the RED ONE development. The only advice I was given was that if we did not have an optical viewfinder, we wouldn't sell any cameras and wouldn't really have a professional camera. We haven't had one request for an optical finder in the last year and a half.

Jim

EVFs seem to work acceptably for video. I haven't found that to be the case with stills.

An SLR TTL viewfinder produces a visual resolution of approximately 1800x1200, has zero latency, uses no power, and costs ~$100.

While an electronic system can't ever reach those specs, it may not have to. I would suggest that a practical EVF for stills would be acceptable with a resolution of 1280x848, a maximum latency of 20ms in EV0 (very dark) conditions, a maximum latency of <5ms in >EV8 conditions, a power usage of ~2W (including sensor and processing) to not impact battery usage too severely, and a cost of ~$500.

As far as I can tell, your current system needs a roughly 5x improvement in latency, 10x improvement in power consumption, and 5x improvement in cost to meet those specs.

As for low-light performance, I can tell you that I can see better with my SLR through an f15 optical train than I can with my EVF through an f4 optical train. This is mostly a function of the sensor and larger sensors can close that gap, but the reason EVFs have the odd reputation of working better in low-light is because they can increase their integration times. The Mallincam is famous for this in the astro community, but it uses integration times of up to 56 seconds. Obviously, this approach is not acceptable for stills, and I don't believe EVFs have any current advantage when low-latency is maintained as mentioned above, which will be required by still shooters used to using OVFs in dark conditions.

Finally, there's the issue of autofocus. A mirror enables phase-detection autofocus, which in my experience is at least two orders of magnitude faster, more sensitive to small subjects, and better at tracking than even the best contrast-detection autofocus. This may be the largest barrier to a mirrorless still camera intended to compete with SLRs, unless CD AF can be improved by a couple of orders of magnitude or a way can be found to do TTL PD AF without a mirror.

Hope that helps, and I wish you the best in your development efforts. I'll be interested to see what you come up with and what practical benefits it might bring to those of us that shoot stills in difficult conditions.

[fde101]

Just shut up those nostalgic pricks with some new *** EVF.


Yeah, I'll likely stick with my Canon DSLRs either way, so don't go by me...

[fde101]

We haven't had one request for an optical finder in the last year and a half.

BTW, I'm curious... how many of your company's primarily-for-professional-still-image-capture cameras have you sold in the last year and a half?

There's a BIG difference between pro still work and pro movie work. Even if the camera handles both very well, that doesn't mean its users will...

There is a LOT of potential in what you are doing here, and I wish you well... it is likely to be a long time before I personally consider replacing my Canons for stills, but if/when I do, I seriously doubt I'll even think about anything without an optical viewfinder.

Again, that's me... others will vary of course!

[Jannard]

BTW, I'm curious... how many of your company's primarily-for-professional-still-image-capture cameras have you sold in the last year and a half?

There's a BIG difference between pro still work and pro movie work. Even if the camera handles both very well, that doesn't mean its users will...

There is a LOT of potential in what you are doing here, and I wish you well... it is likely to be a long time before I personally consider replacing my Canons for stills, but if/when I do, I seriously doubt I'll even think about anything without an optical viewfinder.

Again, that's me... others will vary of course!

Having shot stills for over 30 years (Hasselblad, Nikon, Canon, Olympus, Linhof, Mamiya, Pentax, Fuji... and a whole bunch more) I agree that shooting stills is somewhat different than shooting motion. But having heard that there was no substitute for an optical finder in motion less than 2 years ago... I hope you will be open minded that the possibility exists that there just might be another solution here as well.

As for how many "primarily-for-professional-still-image-capture cameras" we have sold in the last year and change... just as many as we did motion cameras two years ago...

Jim

[Joseph Ward]

Its only first generation for motion in DSLR's? I trust Red will make it a standard!:calm:

[Lee Jay]

I hope you will be open minded that the possibility exists that there just might be another solution here as well.

Sure, but it's going to have to match or exceed the focusing performance of existing phase-detection systems, and at least come close to the latency, resolution, low-power, and low-cost of optical viewfinders, or else it will be non-competitive with existing SLRs.

I hope you can do it, but it's a tall order.

[Jannard]

Sure, but it's going to have to match or exceed the focusing performance of existing phase-detection systems, and at least come close to the latency, resolution, low-power, and low-cost of optical viewfinders, or else it will be non-competitive with existing SLRs.

I hope you can do it, but it's a tall order.

I'll disagree with your wording... it is a bit more complicated/interesting than that but will wait to explain until after we announce.

Jim

[Lee Jay]

I'll disagree with your wording... it is a bit more complicated/interesting than that but will wait to explain until after we announce.

Jim

Fair enough. I'm not sure what you're considering, but one thing I can think of is a fast frame-rate with presampling like the Casio F1 or FH-20. Be aware that that approach doesn't solve the latency issue. As for the rest (AF, power, cost), I guess we'll just have to wait and see what you announce. Looking forward to it!

[Matthew Duclos]

Is it possible to make a transparent image sensor and put the mirror behind the sensor, instead of in front?

nope.

[Ken Chan]

Sure, but it's going to have to match or exceed the focusing performance of existing phase-detection systems, and at least come close to the latency, resolution, low-power, and low-cost of optical viewfinders, or else it will be non-competitive with existing SLRs.

I hope you can do it, but it's a tall order.

I think it can exceed the performance of phase-detection systems because they can create a scene recognition system(Computer Vision) that recognises faces, colour, objects, motion and does tracking. Nikon is moving towards this by using a 1005-pixel RGB sensor as a scene recognition system. AF points are also not restricted to a particular position and can be set to any position or customised by the user.

The lack of a mirror will result in no viewfinder blackout and lower latency. With a mirror, the camera has to allow for the mirror to swing clear before making an image. The camera can continue to perform AF in between making images at a very high frame rate without requiring the mirror to be reset to perform AF.

Performance wont be a problem with a high fps and read-reset sensor coupled with a fast processor. So it'll just need algorithms and high speed processors.

An EVF also means a sophisticated viewfinder is possible. E.g. live RGB histogram, display of clipped highlights in the viewfinder, white balance preview etc.

Jim, please make the industrial design really sexy and classy. :turned:

Ken

[Lee Jay]

I think it can exceed the performance of phase-detection systems because they can create a scene recognition system(Computer Vision) that recognises faces, colour, objects, motion and does tracking. Nikon is moving towards this by using a 1005-pixel RGB sensor as a scene recognition system. AF points are also not restricted to a particular position and can be set to any position or customised by the user.

The problem with CD is that it takes many samples to determine the correct focus point, and those many samples have to be read from the main sensor. With PD, you only need one sample for that purpose (two for determining velocity, three for acceleration), and those can be read with a dedicated sensor with linear arrays, which enables faster reads by far. So, it's many reads at 60fps versus a few at 300fps.



The lack of a mirror will result in no viewfinder blackout and lower latency. With a mirror, the camera has to allow for the mirror to swing clear before making an image. The camera can continue to perform AF in between making images at a very high frame rate without requiring the mirror to be reset to perform AF.


That's not the latency I was referring to. I was referring to the time between when the photons enter the lens and when they appear at the view finder. That's a major issue if you are trying to track fast and erratic objects with tight framing.



Performance wont be a problem with a high fps and read-reset sensor coupled with a fast processor. So it'll just need algorithms and high speed processors.


And a lot of power. My existing cameras have ~10Wh of battery energy available, which is sufficient for 8 hours of shooting and ~1500 shots. With an EVF, this would not be possible.



An EVF also means a sophisticated viewfinder is possible. E.g. live RGB histogram, display of clipped highlights in the viewfinder, white balance preview etc.

I'll be interested to see if those are of any practical use. They haven't been on my consumer EVF camera.

[liquidigital]

I have to say, the other day I wandered into Best Buy and saw a D700. I picked it up and looked through the back and it was for lack of a better word incredible compared to my Canon 40d. Whatever you have in store, I hope it is bigger, brighter and clearer than anything on the market. Looking forward to your product, Jim.

[Ken Chan]

http://www.youtube.com/watch?v=7gfQHRP6r2s

Here's a video of the Lumix G1 for those who wonder what an EVF and contrast detect AF might be like.

I think the sneak peek by Jim with the focal plane mark indicates that a still camera is coming :))

[Jeff Coatney]

Analog optical path for viewfinders on a digital camera? I see only disadvantages to this. While many shooters like the Thru The Lens view for "critical focus", having an optical path that bypasses the sensor is limiting, for focus, composition, unit cost, form factor size and factory quality assurance.

1. Focus. Since the critical relationship of lens to sensor plane is the primary determiner of focus, why introduce a physical mechanism that disrupts the pathway that light needs to travel to hit the picture plane?

You are only introducing opportunities for the lens/sensor relationship to be compromised. Mirrors, spinning, fixed or otherwise moving will introduce vibration, possible loss of registration over time (heavy usage) or cause refractive artifacting that will affect image quality (fixed mirror/ beam splitter).

2. Precision. In manufacturing on a mass scale, maintaining critical tolerances between assemblies is expensive. It can eat up hundreds of man-hours and result in lots of waste. On a camera, the rear element of the lens and the sensor plane is the critical tolerance that must be maintained. On a digital camera, the lens/sensor relationship is measured in thousands of an inch, or in microns.

However, with a digital camera, depending on the design, you typically only have one such critical measurement to maintain from camera to camera during assembly. Unless you add an optical pathway that has to match the frameview of the sensor. Then you have two (or three) critical tolerances to maintain, and that only measuring the one axis: the lightpath. Congratulations, you just doubled the number of critical tolerances that you have to maintain throughout your assembly line. You also just doubled your tooling budget, quadrupled your failure rate, added 20% to you assembly process and tripled your retail price.

What did you gain? You gained a means of focusing your camera that completely bypasses the sensor. How do you know the shot is in focus? You can't know for certain because you just eliminated the instant feedback loop that a digital sensor provides you. It is the sensor that records the image, not your eye. If you use a moving mirror mechanism to measure focus, then over time, as the material that the mechanism is made from begins to wear, you will have microscopic particles loose in your camera and the loss of structural material will affect the tolerances of your optical path to viewfinder to sensor relationship. So the more you use your camera, the less precise your focus will be, if you are using only an optical viewfinder to judge it by.

If we know for certain that an analog optical pathway to viewfinder will ultimately fail, why even bother to design it in? Why add the cost? Why needlessly bulk up the camera? Why would you intentionally shut off the instant feedback loop that the sensor provides you in order to judge focus, when that is the only critical focus information that means anything?

Let's use a film analogy: If you could tell if the image is focused by looking at the emulsion of the film, why would you bother looking through the viewfinder?

I'm not even going to list the enormous amount of information and software enhancements that having an electronic viewfinder gives you, that's the killer app.

[Pawel Achtel]

Good post Jeff. Totally agree.

Here is my 2 South Pacific Peso (a.k.a. Australian cents at current exchange rate):

It is quite common that users, instead of giving Requirements, try to give Solutions. That's not productive because they usually have NFI about technology, physics or inner works of the camera.

An optical viewfiner is a Solution. It's not up to you to provide one. Instead, you should formulate your Problem or Requirement.

Red is not interested in Solutions. They have far better experts and capability than an average film maker.

What they like to hear are Requirements: the actual problems that you are trying to fix. For example:

I would like to be able to focus accurately to a single pixel using naked eye at 1 lux of ilumination.

or

It is mandatory that I be able to focus without power.

Then, of course, there is a matter of prioritising requirements and sorting valid onece from invalid or poorly defined.

[Lee Jay]

1. Focus. Since the critical relationship of lens to sensor plane is the primary determiner of focus, why introduce a physical mechanism that disrupts the pathway that light needs to travel to hit the picture plane?

Because phase-detection autofocus is so vastly superior to contrast-detection autofocus. And PD requires a mirror. That's why.



You are only introducing opportunities for the lens/sensor relationship to be compromised. Mirrors, spinning, fixed or otherwise moving will introduce vibration, possible loss of registration over time (heavy usage) or cause refractive artifacting that will affect image quality (fixed mirror/ beam splitter).

Indeed, that's the price to be paid for autofocus that actually gets the job done.



If we know for certain that an analog optical pathway to viewfinder will ultimately fail, why even bother to design it in? Why add the cost? Why needlessly bulk up the camera? Why would you intentionally shut off the instant feedback loop that the sensor provides you in order to judge focus, when that is the only critical focus information that means anything?


Because it's not needless. The zero latency, zero power usage, high resolution, and enabling of PD AF make it well worthwhile. You're exaggerating the difficulty here too - we now have a zillion *very* inexpensive dSLRs that have a mirror, an OVF, a secondary mirror and a PD AF system included in a camera that also has a pretty big sensor and all processing technology. If that system was so expensive, bulky, and difficult to produce, we would not have $500 camera bodies that include it. But we do.

Lee Jay

[reality]

Mechanical still camera shutters are not acceptable for a motion camera though as they will wear out very quickly.

A spinning shutter would be too bulky and too limiting for framerates/stills. They also have a limited life.

[Dan Hudgins]

Mechanical still camera shutters are not acceptable for a motion camera though as they will wear out very quickly.

A spinning shutter would be too bulky and too limiting for framerates/stills. They also have a limited life.


Not sure if this has been gone into but I had an idea for an optical finder that might work,

You make small holes near each pixel on the sensor that go through the dark parts to a clear base or all the way through the sensor chip, you can then view through the back of the sensor.

Yes there would be some light loss, but you get 100% duty cycle, and because the holes are always in the image plane you compensate for the OLPF focus shift issues and such to get true focus without any shifts in the optical finder. The small size of the holes make the focus more accurate and you can use a field lens to focus the light that goes through the holes into your eye piece.

So the sensor would be like a piece of "peg board" with viewing from its backside. You would not see the grid because at 4K the pitch is too small to see clearly with a full view optical viewfinder, at 2X zoom you might see the grid a little if you have good eyes, but that would be a help to focus the eyepiece on.

[Pawel Achtel]

Because phase-detection autofocus is so vastly superior to contrast-detection autofocus. And PD requires a mirror. That's why.


No, DP does not need a mirror. DP needs to focus accurately. And this is big difference.

[Dan Hudgins]

No, DP does not need a mirror. DP needs to focus accurately. And this is big difference.

No matter what kind of optical finder a camera uses, it can be very hard to focus well with the lens stopped down through the eyepiece with a ground glass.

My Brother says that DP for the most part should not focus through the eyepiece while the camera is running, anyone have thoughts on that subject?

What he means is that the follow focus is used for focus and the optical finder is used for framing while the camera is running, if that is the case then you should not need an optical finder for framing if you have an electronic one.

If the optical finder is just used to mark the follow focus when the camera is stopped, then you can make a rack over camera with optical finder like the Mitchell, that lets you view the image to focus and then use the electronic finder for framing with the lens stopped down since you cannot see much in an optical finder at f/22 anyway.

A rackover has problems since the OLPF on the sensor may change the focus point for fast lenses, so it is better to use a 1:1 electronic finder to find the follow focus marks anyway.

For documentary/sports use a reflex optical finder could be useful, maybe someone could make a PL mount optical finder with relay optics that would let you use any PL mount lens on any PL mount camera, and have a 10% or 50% tap off optical finder as an accessory since you would not need the optical finder all the time?

[Lee Jay]

I'll disagree with your wording... it is a bit more complicated/interesting than that but will wait to explain until after we announce.

Jim

Okay, I'm awaiting an explanation. From the specs, it looks like neither Scarlet nor Epic compete in the dSLR space.