Ad
Ad
Ad
Pages: « 1 2 3 [4] 5 6 ... 8 »   Bottom of Page
Print
Author Topic: Open Source CMOS Medium Format Camera/Back - An initial exploration  (Read 12504 times)
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1683


« Reply #60 on: February 27, 2014, 04:20:00 AM »
ReplyReply

This has been tried by Minolta on the RD-3000 and RD-175 cameras, which used 2 and 3 CCDs at the time to increase resolution. In a nutshell: it was not a good idea. The optics needed in front of the CCD limited the system considerably.
Considering that using lenses with large image circles seems to be a significant goal, and large image sensors seems to be expensive.

What about using a common FF/crop CMOS sensor, and moving it mechanically to scan the image circle of a MF lens? Sort of like scanning backs, only using a 2d sensor instead of a 1d strip.

All kinds of mechanical/precision challenges, and a final product with some significant limitations, but avoiding the problem of finding a good, economical 56x56mm sensor.

-h
Logged
eronald
Sr. Member
****
Offline Offline

Posts: 4121



« Reply #61 on: February 27, 2014, 05:09:23 AM »
ReplyReply

One could also have a vernier move the back on the Z axis within the case under computer control and use the computer to set focus.

Anything like this would be left to the user to explore, mechanical mods, would be possible  as I envision the sensor itself to be on a small easily detachable casing that then mates to the camera mount.The cartridge could equally well be mated to a movable carriage.

I envision the electronics coming with an onboard interface module of some kind, maybe just an Arduino, that can control external devices under program control, simplifying all sorts of software-controlled imaging eg.  motor driven sensor movements , focus stacking, camera panning or translation as well as control of external devices. People would need to rig up mechanics, eg a turntable to move an object in front of the camera, but computer control would be easy to synch in the software.

Also, there should be a way to trigger acquiring a frame very quickly. The general idea is that people would be encouraged to mix image acquisition and device control, although many present users of this forum would probably just use a fixed Hasselblad V mount, and possibly an old 500 series Hassy to take 50MP studio portraits Smiley

Flexibility is the whole point of an open-source project. An interesting question open for debate here is whether we should be open to consider using a smaller sensor initially. I am not sure that something the size of a 35mm full-frame sensor in an MF mount makes practical sense, but there are entertaining possibilities of computer-controlled focus, shift and tilt, maybe even using an APS sensor with 35mm lenses that should be debated; with resolutions creeping up maybe a matchbox-sized computer-controlled view camera taking Zeiss Otus lenses is what the studio photography world really needs Smiley

If one goes to the trouble of using  an open source system, one should at least gain some flexibility in compensation of the inconvenience of using a prototype Smiley

Edmund


Considering that using lenses with large image circles seems to be a significant goal, and large image sensors seems to be expensive.

What about using a common FF/crop CMOS sensor, and moving it mechanically to scan the image circle of a MF lens? Sort of like scanning backs, only using a 2d sensor instead of a 1d strip.

All kinds of mechanical/precision challenges, and a final product with some significant limitations, but avoiding the problem of finding a good, economical 56x56mm sensor.

-h
« Last Edit: February 27, 2014, 06:10:23 AM by eronald » Logged
MrSmith
Sr. Member
****
Online Online

Posts: 908



WWW
« Reply #62 on: February 27, 2014, 05:11:21 AM »
ReplyReply

i have nothing of value to contribute to development. Grin

but thinking out loud after reading some of the comments people seem to want to make an add-on to existing cameras? surely just making a mirror less box with a large bayonet mount but short sensor to lens distance and a decent screen on the back for composing would be easier? usb3 tethering plus cf/sd card. the idea being that with an adapter you could use just about any lens you want with live view focus

basically an overblown A7r without the viewfinder.
Logged
EricWHiss
Sr. Member
****
Offline Offline

Posts: 2427



WWW
« Reply #63 on: February 27, 2014, 11:33:57 AM »
ReplyReply

Am I the only one still not convinced CMOS is better?  Let's just say the IQ 250 samples didn't win me over.   Right now, I'd rather have  CCD.  If CMOS is easier to implement then fine, but I think there are existing CCD sensors with a bigger footprint - like the 6.1cm x 6.1cm chip linked to earlier and no available.

I imagined we'd have a box like that Sinar 54 Torger showed that wirelessly connected to a smartphone for previews or tethered to a computer.  I don't expect much from the first pass.   
Logged

Authorized Rolleiflex Dealer:
Find product information, download user manuals, or purchase online - Rolleiflex USA
eronald
Sr. Member
****
Offline Offline

Posts: 4121



« Reply #64 on: February 27, 2014, 11:53:14 AM »
ReplyReply

Am I the only one still not convinced CMOS is better?  Let's just say the IQ 250 samples didn't win me over.   Right now, I'd rather have  CCD.  If CMOS is easier to implement then fine, but I think there are existing CCD sensors with a bigger footprint - like the 6.1cm x 6.1cm chip linked to earlier and no available.

I imagined we'd have a box like that Sinar 54 Torger showed that wirelessly connected to a smartphone for previews or tethered to a computer.  I don't expect much from the first pass.  


Eric,

Would DHW be interested in supporting an open digital back solution?

I am pursuing the sensor question. Possibly you are right, and availability of large chips trumps ease of implementation.

Edmund
« Last Edit: February 27, 2014, 02:08:45 PM by eronald » Logged
jerome_m
Sr. Member
****
Offline Offline

Posts: 621


« Reply #65 on: February 27, 2014, 01:50:29 PM »
ReplyReply

What about using a common FF/crop CMOS sensor, and moving it mechanically to scan the image circle of a MF lens? Sort of like scanning backs, only using a 2d sensor instead of a 1d strip.

You can already buy something like that to adapt a D800 to a view camera... For example this one.

I suppose that the manufacturer is busy doing the same adapter for the Sony A7r as we speak, since the shorter mount register will make shift easier.
Logged
EricWHiss
Sr. Member
****
Offline Offline

Posts: 2427



WWW
« Reply #66 on: March 01, 2014, 11:57:02 AM »
ReplyReply

Eric,

Would DHW be interested in supporting an open digital back solution?

I am pursuing the sensor question. Possibly you are right, and availability of large chips trumps ease of implementation.

Edmund


I myself as a Rolleiflex dealer and photographer am interested.  I'm sure DHW would be very interested in another digital back solution, particularly one that could take advantage of a larger portion of their camera's and lens's 6x6 capability.   But I haven't spoken to them directly about this yet and can't say anything about if or how they would/could get involved. And actually, I wonder now out loud if there may be any licensing issues to also sort out when looking to fit a back to a specific camera or platforms?

« Last Edit: March 01, 2014, 12:13:22 PM by EricWHiss » Logged

Authorized Rolleiflex Dealer:
Find product information, download user manuals, or purchase online - Rolleiflex USA
peterurban
Newbie
*
Offline Offline

Posts: 32


Be kind.


WWW
« Reply #67 on: March 01, 2014, 03:29:28 PM »
ReplyReply

Hi Edmund,

I think a project like this would be very interesting and worthwhile for many reasons - not the least of which to enable individuals or groups to build specialized solutions of "one" that the industry's main vendors can't go after due to lack of market size etc.

I've been working on a personal camera project that I'll be sharing soon and from that experience I believe that a 135 size "medium format" type back would actually find quite a sizeable market. I actually think that 135 chip size backs will be the new "medium format" for technical view cameras of the future (I am not saying that larger chips won't have the known advantages, but also some disadvantages).

Based on the learnings from my project, I am confident to say that 95% of current manufacturers got it wrong when it comes to technical cameras for next generation digital.

I'll be sharing my project here soon which will explain a lot of the above.

For my project I looked into a us vendor called Illunis ( http://www.illunis.com/ ), they - for instance - create a 70mp 135 chip size "camera" that is pretty much exactly what you are looking for as a sensor "module". It interfaces via usb3 and standardized protocols with whatever processing unit (computer) you want to bring to the table. I've talked to their experts and they said the module would initially go for around $10k but would go down with time and volume. They also offer shutter modules etc as well as larger size sensor "cameras".

If I'd be spearheading this a project, this is what I'de be starting with for a good jumpstart. If you get a big enough community behind it - who knows Illunis might be supporting it.

I think the possibilities would be almost endless and many people will be surprised what is possible when commercial and marketing constraints are removed (i.e. closed vs open protocols, open firmwares, total mechanical freedom...)

Will it be hard, frustrating and sometimes seem impossible - yes for sure. Will it a lot of fun, yes I am sure as well. Can it work? Yes, with the right people at the table.

I hope this takes off and I am willing to help where I can.

Cheers,

Peter
Logged

Peter Urban
Courteous Explorer | Photographer | Dog Lover - Passionate about nature, people, technology and design.
eronald
Sr. Member
****
Offline Offline

Posts: 4121



« Reply #68 on: March 01, 2014, 08:21:34 PM »
ReplyReply

Hi Peter,

 Thank you for sharing.

 (Obviously) I agree that a bunch of interesting new possibilities open up once you go open source. One side effect is that you suddenly gain a quasi-infinite pool of developers.

 As Bill Gates once complained, "Linux Torvalds has more programmers than Microsoft".

 On the other hand, I think one should be able to make small-format cameras cheaply. The reason anything based on the 70MP CMOSIS chip is expensive and difficult is that the outputs are analog. Sony chips should considerably simplify and cheapen the construction, I would be very surprised if even an MF solution costs more than $5K since there are no development costs to recoup outside mechanical engineering of the sensor cartridge.

Edmund

Hi Edmund,

I think a project like this would be very interesting and worthwhile for many reasons - not the least of which to enable individuals or groups to build specialized solutions of "one" that the industry's main vendors can't go after due to lack of market size etc.

I've been working on a personal camera project that I'll be sharing soon and from that experience I believe that a 135 size "medium format" type back would actually find quite a sizeable market. I actually think that 135 chip size backs will be the new "medium format" for technical view cameras of the future (I am not saying that larger chips won't have the known advantages, but also some disadvantages).

Based on the learnings from my project, I am confident to say that 95% of current manufacturers got it wrong when it comes to technical cameras for next generation digital.

I'll be sharing my project here soon which will explain a lot of the above.

For my project I looked into a us vendor called Illunis ( http://www.illunis.com/ ), they - for instance - create a 70mp 135 chip size "camera" that is pretty much exactly what you are looking for as a sensor "module". It interfaces via usb3 and standardized protocols with whatever processing unit (computer) you want to bring to the table. I've talked to their experts and they said the module would initially go for around $10k but would go down with time and volume. They also offer shutter modules etc as well as larger size sensor "cameras".

If I'd be spearheading this a project, this is what I'de be starting with for a good jumpstart. If you get a big enough community behind it - who knows Illunis might be supporting it.

I think the possibilities would be almost endless and many people will be surprised what is possible when commercial and marketing constraints are removed (i.e. closed vs open protocols, open firmwares, total mechanical freedom...)

Will it be hard, frustrating and sometimes seem impossible - yes for sure. Will it a lot of fun, yes I am sure as well. Can it work? Yes, with the right people at the table.

I hope this takes off and I am willing to help where I can.

Cheers,

Peter

« Last Edit: March 01, 2014, 08:30:00 PM by eronald » Logged
EricWHiss
Sr. Member
****
Offline Offline

Posts: 2427



WWW
« Reply #69 on: March 02, 2014, 01:58:08 AM »
ReplyReply

Hi Peter,
Thanks for the link - that's an interesting outfit.  I see they also have the 50mp MF chip in some products which for me is what I'm interested in - the larger sensors, but I'm curious to know about your project and why you think its the new MF for tech cameras. Is that because of the larger DOF smaller sensors provide - all else being equal?

I do have a bunch of people asking me about fitting the A7R to the Rollei X-Act2 technical camera, and one guy Richard Gale has done a really smashing job of it.
Logged

Authorized Rolleiflex Dealer:
Find product information, download user manuals, or purchase online - Rolleiflex USA
peterurban
Newbie
*
Offline Offline

Posts: 32


Be kind.


WWW
« Reply #70 on: March 03, 2014, 06:38:58 PM »
ReplyReply

@Eric,

My experience with my project has led me to believe that most technical cameras are conceptually not very suitable to the kind of benefits you'd want from a digital tech camera setup - especially in the filed.

More on that soon Smiley

Peter
Logged

Peter Urban
Courteous Explorer | Photographer | Dog Lover - Passionate about nature, people, technology and design.
eronald
Sr. Member
****
Offline Offline

Posts: 4121



« Reply #71 on: March 04, 2014, 12:15:33 AM »
ReplyReply

The interesting question about an open source camera is not only what the sensor would be like, but also what sort of environmental controls could be synched to the imaging aspects - eg automated stitching, panning etc.

Edmund
Logged
torger
Sr. Member
****
Offline Offline

Posts: 1588


« Reply #72 on: March 04, 2014, 02:05:57 AM »
ReplyReply

@Eric,

My experience with my project has led me to believe that most technical cameras are conceptually not very suitable to the kind of benefits you'd want from a digital tech camera setup - especially in the filed.

More on that soon Smiley

Intriguing Smiley

I'm really curious about what you think is wrong with the current tech cameras for field use. I'd say that there are two faults - no real live view and poor angular response (ie requires LCC). The first is not-using-a-CMOS problem but the second is actually becoming worse with current CMOS technology. Current pixels are like 5um wide and 7um deep. We need new sensors with photodiodes closer to the surface of the sensor, fortunately such technology already exists but so far only in small sensors.
Logged
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1683


« Reply #73 on: March 04, 2014, 02:17:10 AM »
ReplyReply

Judging by my experience (end-user) of open source software projects:

I think it is really hard to beat the large corporations on things like ergonomy, UI, battery life etc. These are things that require some manpower, and a kind of competence that seems to not be easily drawn into open source.

What open source seems to excel at is nitty-gritty technical stuff (e.g. pixel peeping), and adhering to standards. The inherent openness of both product and the process seems to invite thinking out of the box (e.g. the stuff that MagicLantern are putting into old-ish Canon cameras).

-h
Logged
eronald
Sr. Member
****
Offline Offline

Posts: 4121



« Reply #74 on: March 04, 2014, 08:34:20 AM »
ReplyReply

Judging by my experience (end-user) of open source software projects:

I think it is really hard to beat the large corporations on things like ergonomy, UI, battery life etc. These are things that require some manpower, and a kind of competence that seems to not be easily drawn into open source.

What open source seems to excel at is nitty-gritty technical stuff (e.g. pixel peeping), and adhering to standards. The inherent openness of both product and the process seems to invite thinking out of the box (e.g. the stuff that MagicLantern are putting into old-ish Canon cameras).

-h

I agree with everything you say.

For some reason, community open-source developers create software that is not easy to use.

However over the past few years, corporations have also learnt that they can successfully re-use open-source engineering, provided they work hard and wrap and re-skin it.

The most spectacular RECENT example of gift-wrapped open-source is Android. "Desktop Linux" has been a user-interface failure, but the "Touch Linux" called Android has totally defeated Blackberry, Nokia and Microsoft and is even outselling Apple. It is now the leading smartphone and tablet OS.

Another less well known open source success wrapping story is Mac OS X, which builds on an academic Unix/BSD clone, based on a kernel named Mach. The same is true for iOS of course.

In summary, the whole modern mobile phone and tablet industry, iPhone and Android, is based on re-skinned open source.

Between mobile devices and the cloud, open source has already taken over the world, and corporations are making a lot of money from it, thank you very much. Except Microsoft who failed to notice the dagger as it was pushed into their chest.

Edmund
« Last Edit: March 04, 2014, 09:51:11 AM by eronald » Logged
torger
Sr. Member
****
Offline Offline

Posts: 1588


« Reply #75 on: March 04, 2014, 10:14:29 AM »
ReplyReply

Open-source get contributions from geeks, and geeks like to make geeky things, which generally isn't user interface design. Understanding user interface design is a much different competence from designing algorithms or device drivers. Another aspect is that to make a great user interface you need a "dictator", someone that unifies the whole experience. It's less suitable for the distributed type of development open-source generally is.

It's easy to criticize a user interface for not being good (constructive criticism too), but it's hard to actually make one from the ground up. Just like it's easy to criticize a photograph and harder to shoot one. I know this myself, even if I think I am quite good at detecting what a good user interface is, I must really work hard if I'm going to make one myself that doesn't suck.

I would not worry too much though, with an open design you can always fix later. And actually, for things I use often I generally prefer user interfaces that are a bit cumbersome to get into but powerful once you know it. "Great" commercial user interfaces often suffer from omitting features an advanced user would like to have, in order to make the interface more user-friendly.
« Last Edit: March 04, 2014, 10:16:01 AM by torger » Logged
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1683


« Reply #76 on: March 04, 2014, 02:25:50 PM »
ReplyReply

"Great" commercial user interfaces often suffer from omitting features an advanced user would like to have, in order to make the interface more user-friendly.
I think that the great ones have several layers. There is what is presented "in your eyes", this should ideally match my expectations of what I mainly intend to do with the thing. As you use the software and get comfortable doing the basic things, you probably will want to do them more efficiently (or do more advanced operations). At this stage my perfect UI should offer these features at just the place I start looking for them. Kind of like how great Pixar movies offers a great story for your kids, but also adds a deeper layer of fun for the grown-ups.

I think that it is possible (but hardly simple) to make a really great UI for a software application like Lightroom, where you have an advanced set of operations that a large number of people want access to. The thing is that these people have a somewhat common understanding of what they want to do, meaning that it is possible (to some degree) to make a "one-size-fits-all" software for them.

I wish that the mainstream camera manufacturers catered more for us raw photographers. Decent raw histograms would be a start. Focus peaking with good automatic zooming. (hire Guillermo if needed). I like to have immediate hardware access to the fundamental camera parameters (manual focus, AF point/tracking, zoom, exp time, aperture, ISO, EC,...). By having an open philosophy, anyone should be able to write simple "apps" that either run locally on the camera, or on a connected smart phone that have low-level access to image sensor and camera mechanics. There are many skilled people out there, but not all are able or willing to wrestle low-level embedded development.

-h
« Last Edit: March 04, 2014, 02:29:08 PM by hjulenissen » Logged
eronald
Sr. Member
****
Offline Offline

Posts: 4121



« Reply #77 on: March 04, 2014, 05:11:05 PM »
ReplyReply

To get back on topic, do you guys have a preference for a specific FPGA family for open source? Discuss ...

I am thinking of doing a breadboard tethered Raw minicam design with some friends just to see what the issues are.

Edmund
Logged
EricWHiss
Sr. Member
****
Offline Offline

Posts: 2427



WWW
« Reply #78 on: March 04, 2014, 11:45:47 PM »
ReplyReply

We should all agree on some minimum set of key specifications for a first iteration, I think.

For me:
the sensor ideally would be bigger than the leica/pentax/IQ250 size
CMOS or CCD ok with me, but I still favor CCD.
Needs to have a universal design with possible adapter plates for different camera platforms(ie. If I can't get it to fit my camera(s) its not working for me, and probably others feel the same)
Pixel count not as important as sensitivity and ability to handle shifted lenses or wide angle lenses close to sensor. 



Logged

Authorized Rolleiflex Dealer:
Find product information, download user manuals, or purchase online - Rolleiflex USA
haplo602
Newbie
*
Offline Offline

Posts: 22


« Reply #79 on: March 05, 2014, 09:25:02 AM »
ReplyReply

We should all agree on some minimum set of key specifications for a first iteration, I think.

For me:
the sensor ideally would be bigger than the leica/pentax/IQ250 size
CMOS or CCD ok with me, but I still favor CCD.
Needs to have a universal design with possible adapter plates for different camera platforms(ie. If I can't get it to fit my camera(s) its not working for me, and probably others feel the same)
Pixel count not as important as sensitivity and ability to handle shifted lenses or wide angle lenses close to sensor. 




So in current products you are talking about 36x48mm and larger non-microlensed sensors ...
Logged
Pages: « 1 2 3 [4] 5 6 ... 8 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad