the D800 is cool and all, but let's talk about a real mfd killer!

[simonstucki]

Hi I haven't posted much on (in?) this forum so far, but I have been reading for some time and I like it a lot here. very interesting threads and I've learned a lot so far.

anyway I had an idea for a nice camera I'd like to share here (im not a native English speaker so I hope you understand what I mean).

the basic idea: a modular, mirrorless, sensorshift, sensorfocusing, interchangable lens mount camera.

the sensor: about fx size, however if I could decide I'd go with a 4:5 aspect ratio (but still ok to use with fx lenses), pixel density very high (probably higher than d800 (and more importantly an even higher dynamic range and even better colour), because it will of course take some time to finish the design of such a camera, thus at that time better sensors will be available . no aa filter (because of the multishot feature)

sensorshift (the full shift range would use the image circle for 6x7 lenses):
1. for regular shifting and rising and so on
2. image stabilization
3. multishot (like the hasselblad mfd backs)
4. and of course for sensor shift stitching

mirrorless:
well no mirror but a very nice evf with peaking

sensorfocusing:
in combination with regular "lens focusing" for fine focus and as well as in combination with mf (tripod use you focus by hand with live view and then you use the touchscreen to tell the camera where you want the focus to be perfect) with short focal lenghts you will only have to focus with the lens when you want to use some sort of close range correction of the lens.
and of course for fast automatic focus stacking

interchangable lens mount: ef, F, leica m, mamiya 7? (do those lenses work well with digital sensors?, large format (for those nice digital lenses) and whatever you like plus a new lens mount optimized for this camera (if the mamiya 7 lenses would work well, and if it would be at all possible and usable I would make the new lens mount to be compatible with the mamiya 7 lenses)

extra feature if possible: some lense mounts (those for large image circle lenses) could be mounted in two orientations, the normal orientation would place the center of the image circle at the center of the sensor (no shift or rise) the other orientation turned by 180° would place the center of the image circle of the lens at the center of the sensor when shifted to the lower right corner (when looking at the camera from the front) for more shift and rise and fall.

the lens mounts would also feature controls for aperture, exposure time

modular:
1: sensor modules: square!
1a well the sensor with the shift and focusing capabilities the attachment points for the lens mount and arca swiss style dovetails on top, bottom, left and right (for rock solid tripod mount) electronic contacts for the accessories. on all 6 sides, small battery.
1b high speed (video and still) lower resolution no shift sensor module


2: top, left, right or bottom mount accessories (slides over dove tail, very firmly attached, can be attached on all four sides)
2a viewfinder module: nex-7 style evf (same position, better resolution and faster, peaking, 3d), and basic control dials, wheels and stuff, shutter release, top lcd
2b right hand hand grip with shutter release, some controls, extra battery, you can even combine two of those (right, and bottom or left and right if your not sure whether you are right or left handed
2c battery, different sizes
2d non functional rubber protector lets say you want to use the camera hand held and with only one handgrip and the evf module attached you can put two of those on the to other sides. for better handling

3: backside accessory:
3a small lcd back (high resoltion "retina"), in combination with two handgrips and the evf the camera will look sort of like a pro dslr lots of physical buttons (except for the topsilde (lack of the prism...)
3b large lcd back (very very high resolution 4k or better) a few physical buttons but mainly touchscreen based operation. optional would be a lcd loupe (like the hoodman hoodloupe, very very high quality)
3c no lcd back, just some control dials (if you only want to use the evf), two versions one with battery and a very slim and light one without
3d wireless back (wifi and 3g, and what ever might make sense) the large lcd back can be used as a remote


everything (except some lens mounts) would be weather sealed and the touchscreen would be usable even if wet.

for one shot 3d just combine two sensor modules.

price: in three years: sensormodule, one lens mount, evf, one hand grip, one extra battery: less than $10k

new lenses: if the mamiya 7 lenses work, only some wide angle lenses would be necessary lets say a 35/4, 28/4, 24/5.6, 20/5.6 that cover maybe a little less than 6x7. very high quality and unfortunately probably quite expensive.

what do you think? I'm looking forward to your comments. am I completely crazy (do not hesitate to point that out, if you think that is the case), hopelessly idealistic, just plain stupid, very unrealistic or maybe not so stupid at all?

[NancyP]

Interesting!  I am not a mechanical engineer, and the sensor movement would seem to be a real challenge.

[ErikKaffehr]

Hi,

This is what I call thinking out of the box. Nice convergence of existing and forthcoming technologies.

Just some reflections:

I think that some of your suggestions may be in conflict. I have some doubts about the feasibility of sensor based image stabilization, I guess that a fixed sensor may be easier to hold in required alignment with lens, for instance.


Best regards
Erik

[shadowblade]

I don't see the sensor shift as being a problem - after all, Pentax (I think - one of the smaller manufacturers anyway) already uses a rotating sensor to correct for uneven horizons, and digital scanning backs move the sensor along a plane.

I've always been interested in a tilting sensor, actually, to give every lens the capability of focal plane adjustment, in order to get everything in focus (for a typical landscape with a ground plane, anyway) from zero to infinity.

[Fine_Art]

I don't see the sensor shift as being a problem - after all, Pentax (I think - one of the smaller manufacturers anyway) already uses a rotating sensor to correct for uneven horizons, and digital scanning backs move the sensor along a plane.

I've always been interested in a tilting sensor, actually, to give every lens the capability of focal plane adjustment, in order to get everything in focus (for a typical landscape with a ground plane, anyway) from zero to infinity.

A great idea.

[BernardLanguillier]

A great idea.

Well.... how do you get it back in perfect position when you don't want tilting?...

Cheers,
Bernard

[shadowblade]

Well.... how do you get it back in perfect position when you don't want tilting?...

Cheers,
Bernard

Press a button and have it return to a centred position?

[Rob C]

I don't see the sensor shift as being a problem - after all, Pentax (I think - one of the smaller manufacturers anyway) already uses a rotating sensor to correct for uneven horizons, and digital scanning backs move the sensor along a plane.

I've always been interested in a tilting sensor, actually, to give every lens the capability of focal plane adjustment, in order to get everything in focus (for a typical landscape with a ground plane, anyway) from zero to infinity.

Have you used a proper 'movements' camera such as a monorail?

Moving a back around, on its own - as would be with a sensor - isn't any kind of answer to what you seek - you'll simply eff up the shape of the shot.

Rob C

[shadowblade]

Have you used a proper 'movements' camera such as a monorail?

Moving a back around, on its own - as would be with a sensor - isn't any kind of answer to what you seek - you'll simply eff up the shape of the shot.

Rob C

You'll get some keystoning, but, with certain landscapes, that would be acceptable.

[BernardLanguillier]

Press a button and have it return to a centred position?

I am not talking about the user interface. 

I am talking about the technical difficulty of bringing a sensor back in perfect alignement.

Considering the resolutions at stake, I believe that this would be a major technical challenge to maintain this ability over the long run.

Cheers,
Bernard

[Fine_Art]

You'll get some keystoning, but, with certain landscapes, that would be acceptable.

A tilting lens mount would be good.

[Eddy M]

I am not talking about the user interface. 

I am talking about the technical difficulty of bringing a sensor back in perfect alignement.

Considering the resolutions at stake, I believe that this would be a major technical challenge to maintain this ability over the long run.

Cheers,
Bernard

I think it's doable. IBIS moves sensor, yet the sensor is able to get back to its original position. Practical or not is another story. But the biggest story is - are manufacturers ready to offer something new rather than selling film camera with digital sensor?

[BernardLanguillier]

I think it's doable. IBIS moves sensor, yet the sensor is able to get back to its original position. Practical or not is another story. But the biggest story is - are manufacturers ready to offer something new rather than selling film camera with digital sensor?

Such a decision is typically not taken in terms of degree of innovation. It is taken in terms of measured benefits vs cost/downsides.

The fact is that even many 4x5 cameras do not offer back tilt.

There is some value, in particular if made assymetrical, but for most applications it is probably more practical to tilt the lens, isn't it?

Cheers,
Bernard

[simonstucki]

thanks for the comments and ideas.

i think the sensor of the olympus om-d em-5 does all movements (and I assume also is able to get back to a 0 position precisely):
http://www.youtube.com/watch?feature=player_embedded&v=KuIoTPeISWI

if the range of the movements could be expanded easily then even a tilting sensor would apparently be possible (and not probably not too expensive), and in combination with sensor shift and tilting the camera, it would have the same effect as tilting the lens right? but an additional tilting lens mount would be nice too.

the reason why I was thinking about the movement of the sensor rather than the camera or lens ist that the sensor is lighter and therefore it would probably be easier to achieve the necessary high precision. also motorized movement seems to be more precise and durable than manual movement (at least that is what hartblei says about its motorized sliding back).

however I must admit, that I have no real idea if such a camera would be possible at an affordable price. What I did is just think about what I know already exists (e.g. high res evfs, sensorshift, high pixelpitch sensors with good dr and colour and so on, interchangeable lens mount (red)) and how this could be improved and combined. I also put in all the cool features I would like to have but i would already be very happy with a few of these features (most importantly the 4:5 sensor, the shift, and the interchangeable lensmount and the lack of mirror)


to get back to the tilting (or depth of field) business: I think in the future with even higher resolution sensors this is going to become more important, already now with todays high resolution sensors it is getting harder and harder to achieve a great depth of field at maximum resolution. one solution is focus stacking, the other is tilting the lens, but both come with limitations that are in some situations quite serious. so I'm wondering what else could be done?
the best I come up with so far is a very fast moving sensor that moves back or forth and captures several images (or just very very very fast automatic focus stacking). the problem is of course speed, if you need 1/250s to get a sharp picture and you need three pictures to get the depth of field you woud have to take all three images during this 1/250s so the shutter time for each pictures would be more than tree times faster and the sensor would have to move very fast. the good news is that if you can't achieve high depth of field at max resolution it is because you can't close the aperture more due to diffraction, so that means you will be shooting at a large aperture what means it is easier to get a fast exposure time. another approach might be software, what if you move the sensor slower and use something like that announced deblurring feature from adobe: http://www.youtube.com/watch?v=xxjiQoTp864& to make up for differences between the images.
why not? that would be really something new, less like "painting with light" more like "using measurements of light to make an image".
I'm looking forward to your comments.

[MatthewCromer]

Such a decision is typically not taken in terms of degree of innovation. It is taken in terms of measured benefits vs cost/downsides.

The fact is that even many 4x5 cameras do not offer back tilt.

I almost always used back tilt with my Shen Hao 4x5, as it made little geometric difference for the landscapes I was shooting, avoided coverage issues with lens tilt, and was far easier, and faster to perform accurately in the field.  At least on that particular body.

There is some value, in particular if made assymetrical, but for most applications it is probably more practical to tilt the lens, isn't it?

Cheers,
Bernard

I think you face significant issues with coverage and internal lens baffles when tilting the lens mount for non tilt/shift lenses.  Alternately, there are very few purpose-built t/s lenses, and they are expensive and "fiddly" with significant limitations (no AF, no zoom).

[MatthewCromer]

Have you used a proper 'movements' camera such as a monorail?

Moving a back around, on its own - as would be with a sensor - isn't any kind of answer to what you seek - you'll simply eff up the shape of the shot.

Rob C

That's what photoshop is for.

[uaiomex]

Tilting the image plane would introduce distortion altering the related size between near and far objects. In many instances, it could be desirable but not in most. Lens tilting introduces a lot less of this issue. If carefully applied is not legible. Probably not a bad thing in general cause it could turn your entire system into a tilt system. Lenses with movements are always bigger, heavier and expensive. I guess I could live with it.
Software could return the sensor to "Neutral" position. How accurate? That would depend on the whole mechanism and firmware own tolerances. Every camera manufacturer would have their own standards. Probably only very expensive cameras and backs could achieve such stunt, at least in the first years.
Eduardo  

[MatthewCromer]

Tilting the image plane would introduce distortion altering the related size between near and far objects.

Isn't this what all wide angle lenses do?  Create wildly distorted perspective between objects near and far from the lens?

The main point of a tilting sensor would be to allow the rotation of the plane of focus through the image.

That's the effect I wanted when I used rear tilt with my Shen Hao, and any distortions were pretty much unnoticable with my subject matter (landscapes).

I can see why architectural photographers would prefer T/S lenses, but I'd prefer to get tilt with the lenses I already own.

[Fine_Art]

I almost always used back tilt with my Shen Hao 4x5, as it made little geometric difference for the landscapes I was shooting, avoided coverage issues with lens tilt, and was far easier, and faster to perform accurately in the field.  At least on that particular body.
I think you face significant issues with coverage and internal lens baffles when tilting the lens mount for non tilt/shift lenses.  Alternately, there are very few purpose-built t/s lenses, and they are expensive and "fiddly" with significant limitations (no AF, no zoom).

All my lenses are full frame 35mm. My digitals are APS-C. Coverage is not a problem for any cropped sensor camera. You don't need baffles if it is a rack and pinion arc. you turn the dial, the dovetail blocks any light. You can argue the tolerance has to be very high at extreme tilt to keep the lens in the perfect position with reduced rack contact. If Arax and Schneider can do it I'm sure the big manufacturers can too.

[Stefan.Steib]

You are close ! I am sure someone will build this soon. Parts of this are already happening in smaller formats. look at the GoProHero 2.
There is even much more which would make sense, like direct raw output for Video and Photo, large batteries, WLAN, electronic lenscalibration (like in the D800)

Thanks for sharing your thoughts, I have read this with large interest !

Greetings from Germany
Stefan