Ad
Ad
Ad
Pages: « 1 [2] 3 4 ... 8 »   Bottom of Page
Print
Author Topic: Open Source CMOS Medium Format Camera/Back - An initial exploration  (Read 13638 times)
jerome_m
Sr. Member
****
Offline Offline

Posts: 666


« Reply #20 on: February 25, 2014, 01:45:56 PM »
ReplyReply

There are quite a few alternative firmwares for existing cameras. For example, for Canon: CHDK.
Logged
torger
Sr. Member
****
Offline Offline

Posts: 1590


« Reply #21 on: February 25, 2014, 01:49:41 PM »
ReplyReply

A large sensor could be key for a success, that's a way to be truly different. If the sensor is 56x56 everything else could be pretty lame and it would still be attractive Smiley

A monochrome and a bayer color edition of a basic 56x56 back. It would be perfect for Hassy V, Hy6, and every legacy 6x7 camera out there.

As far as I understand the dynamic range of the astronomy chips is a disaster though, unless you have like liquid cooling.

Here's data sheets of two large CCDs, neither is very suitable for this type of back:

80x80mm sensor: http://www.fairchildimaging.com/files/data_sheet_ccd_595.pdf
61x61mm sensor: http://www.fairchildimaging.com/files/ccd6161_data_sheet_january_2013.pdf
« Last Edit: February 25, 2014, 02:02:29 PM by torger » Logged
ErikKaffehr
Sr. Member
****
Online Online

Posts: 7910


WWW
« Reply #22 on: February 25, 2014, 03:37:34 PM »
ReplyReply

Hi,

Possible? Yes! Probable? No!

Why?

Significant initial investment. We would need a few samples of a sensor and evaluation board.

Can we develop a digital interface for sensor, or do we need an ASIC?

The sensor needs to be mounted with decent precision, can we do that?

Just some questions…

Best regards
Erik

Hi -

There are a variety of reasons why an open medium format camera platform makes sense. In this topic we could discuss whether there would be people interested in investing time and money or contributing such skills in such a product, and what is reasonably attainable in a reasonable amount of time.

In order to fix expectations, I would like to remind people that Linux Torvalds was not very skilled technically when he started Linux, but he got the project off the ground and the pre-existing GNU software suite made the system viable.

Also, the Linux project was  initially a big success for scientists and servers run by professional sysadmins, but  failed to convince the users of conventional desktop PCs. Of course things changed, and now a substantial fraction of the world's phones and tablets run Linux.

In the same way, I believe it would be feasible to do an open source tethered setup in limited time; external processing of the files could then take over: Many on this forum are skilled software engineers. I am not so sure that a mobile camera is an attainable short time goal. The Magic Lantern project has proved very useful to the film-making community, but what we discuss here starts from the sensor, not from an existing working and integrated platform.

Let's discuss -

- what you'd like to see as useful features
- what platform should be a base for the mechanics/optics
- what skills you could bring to the table, and how much time you could contribute


Edmund
Logged

eronald
Sr. Member
****
Offline Offline

Posts: 4208



« Reply #23 on: February 25, 2014, 05:11:17 PM »
ReplyReply

It seems that DIY cameras have been quite well received in the astro community, because the cost (BOM) of components is less than $1K over the sensor cost.  Engineering (dud) sensors are usually cheap, so I guess any of us could afford to do this as a project. Production sensors for MF can be expected to cost, although 35mm chips might be cheaply available with fairly compatible architectures.

Maybe people here could indicate what they would like to do with a tethered liveview camera - that could help us choose an initial camera platform. In other words, it would be interesting to know whether people need to do AF or motor-driven focus stacking, tilt/shift, panorama stitching etc. or whether a back for an Alpa style device would attract interest.

Edmund
« Last Edit: February 25, 2014, 05:17:09 PM by eronald » Logged
haplo602
Newbie
*
Offline Offline

Posts: 22


« Reply #24 on: February 26, 2014, 01:11:19 AM »
ReplyReply

I have an old Megavision S4 that I don't have the cables or interfaces to. So I took it apart :-) It contains a KAF-16801-CE sensor. The spec sheet and design documents do not look that complicated. The interface is quite easy, main problem is power delivery for me (not good in electronics). The ASIC can be any arm controller chip (cheapo STM Cortex-M4 ?) with proper cache. Anyway I'll try to hook it up to a devel board and see what I get.

Anyway, the basic mount should be Hasselblad V, since it is adaptable to about any camera mount there is. Not self contained, the processing/interface done by a smartphone (iPhone, Android) via USB. This makes the back lighter and cheaper since almost anybody has a smartphone these days. Also enables use of already available smartphone apps for processing images and smartphone CPUs are impressive these days (quad cores and such). DNG format RAW.

Sensor size is not deciding, the platform could be made sensor agnostic based on the chip family ...

However the above is just wishes so far :-))) I have little programing skills to contribute.
« Last Edit: February 26, 2014, 04:10:06 AM by haplo602 » Logged
bpepz
Full Member
***
Offline Offline

Posts: 203


« Reply #25 on: February 26, 2014, 07:34:53 AM »
ReplyReply

I have an old Megavision S4 that I don't have the cables or interfaces to. So I took it apart :-) It contains a KAF-16801-CE sensor. The spec sheet and design documents do not look that complicated. The interface is quite easy, main problem is power delivery for me (not good in electronics). The ASIC can be any arm controller chip (cheapo STM Cortex-M4 ?) with proper cache. Anyway I'll try to hook it up to a devel board and see what I get.

Anyway, the basic mount should be Hasselblad V, since it is adaptable to about any camera mount there is. Not self contained, the processing/interface done by a smartphone (iPhone, Android) via USB. This makes the back lighter and cheaper since almost anybody has a smartphone these days. Also enables use of already available smartphone apps for processing images and smartphone CPUs are impressive these days (quad cores and such). DNG format RAW.

Sensor size is not deciding, the platform could be made sensor agnostic based on the chip family ...

However the above is just wishes so far :-))) I have little programing skills to contribute.

That is really cool. I really like the idea of offloading the processing to a smart phone. I wonder if down the road with increased computing power, and decent firmware, you could actually get higher quality files out of existing sensors. At the very least you would have faster processing.

This actually bring ups a whole new idea for me, taking old DBs and switching out the electronics in them. Take some  ancient H25s and put brand new processing in them, usb 3.0 connectivity, LCD via smartphone, stuff like that. If such a service existed where I could send an old back in for an "upgrade" like this I would happily throw down some big bucks on something like that.
Logged
haplo602
Newbie
*
Offline Offline

Posts: 22


« Reply #26 on: February 26, 2014, 08:42:08 AM »
ReplyReply

That is really cool. I really like the idea of offloading the processing to a smart phone. I wonder if down the road with increased computing power, and decent firmware, you could actually get higher quality files out of existing sensors. At the very least you would have faster processing.

This actually bring ups a whole new idea for me, taking old DBs and switching out the electronics in them. Take some  ancient H25s and put brand new processing in them, usb 3.0 connectivity, LCD via smartphone, stuff like that. If such a service existed where I could send an old back in for an "upgrade" like this I would happily throw down some big bucks on something like that.

hold your horses :-)

CCD chips require some analog to digital circuitry around them and decent power delivery. Also there are some limits to what the CCD sensor itself can deliver. It's a nice idea, but the RoI would be very small if any ...
Logged
lance_schad
Sr. Member
****
Offline Offline

Posts: 257


« Reply #27 on: February 26, 2014, 09:00:19 AM »
ReplyReply

Eronald,
I was at the International Center for Photography yesterday and I walked past the gift store and thought that this may be a good way to for you to possibly jumpstart the project.


All kidding aside it sounds like an interesting project and wish you the best of luck with it.


Lance


Logged

LANCE SCHAD - DIGITAL TRANSITIONS - Phase One,Leaf/Mamiya,Arca-Swiss,Cambo value added reseller
direct/cell:610-496-5586  office:877-367-8537 x 224
http://www.digitaltransitions.com email:lns@digitaltransitions.com
torger
Sr. Member
****
Offline Offline

Posts: 1590


« Reply #28 on: February 26, 2014, 09:13:36 AM »
ReplyReply

I'd suspect it will be quite difficult to get low noise out of the CCD sensor, ie you can't just put the components in any order you'd like and hope you'll get maximum image quality out of it, but you'd need to do some tricks to avoid thermal noise etc. To satisfy the customer I think one would have to be pretty competitive on the DR property.

Making the back tether only and with mobile phone interface (wireless etc) is interesting, but I think it would be very valuable to have a stand-alone back so one should look into how difficult that part is. I would imagine that today there would exist kits with display and processing to do the GUI part.

Looking how much issues there are with legacy backs and Compact Flash I get the impression that making a compact flash reader that works with all cards in all conditions is quite complicated.
Logged
eronald
Sr. Member
****
Offline Offline

Posts: 4208



« Reply #29 on: February 26, 2014, 09:26:00 AM »
ReplyReply

Torger,

I would be thankful of you discussed my proposal, not somebody else's.

Let le stress the point again: CMOS IS A BREAKTHROUGH FOR SENSOR HACKERS BECAUSE IT IS DIGITAL, EASY TO INTERFACE, AND HAS LIVEVIEW SO ONE CAN COMPOSE ON THE COMPUTER.

I agree that redoing CCD stuff is a pain. Not an impossibly large pain though, because the astro community have been doing it - the circuitry is already out there. I think that an open source CCD back would be a hack of last resort because you can now buy older backs cheaply, so the only real purpose would be to get eg. 80MP performance at a reasonable price. BUT YOU DON'T GET LIVEVIEW FOR FREE.

However CMOS sensors are digital, so there are no noise issues except thermal dissipation, and by tethering and doing the heavy lifting on an attached device, you bypass all the painful issues with writing files and preparing previews.

Edmund

I'd suspect it will be quite difficult to get low noise out of the CCD sensor, ie you can't just put the components in any order you'd like and hope you'll get maximum image quality out of it, but you'd need to do some tricks to avoid thermal noise etc. To satisfy the customer I think one would have to be pretty competitive on the DR property.

Making the back tether only and with mobile phone interface (wireless etc) is interesting, but I think it would be very valuable to have a stand-alone back so one should look into how difficult that part is. I would imagine that today there would exist kits with display and processing to do the GUI part.

Looking how much issues there are with legacy backs and Compact Flash I get the impression that making a compact flash reader that works with all cards in all conditions is quite complicated.
« Last Edit: February 26, 2014, 09:30:55 AM by eronald » Logged
torger
Sr. Member
****
Offline Offline

Posts: 1590


« Reply #30 on: February 26, 2014, 09:46:16 AM »
ReplyReply

I would be thankful of you discussed my proposal, not somebody else's.

Ah okay, did not understand your proposal was locked to CMOS, sorry.

With CMOS I'd say it's not a suitable time to make an open source digital back Undecided. There is only one option which is the Sony 44x33mm sensor, and I think the Pentax offer will be cheap enough, and for those that want a detachable back the sensor is a bit smallish (for V mount or RZ), and for technical cameras it has too large issues with crosstalk to work well with technical wides. I've made quite thorough analysis of the crosstalk issue on tech wides over at getdpi.

I guess some would still like to have it, but I would personally not be too interested in the back. I'd like to hear those that would be interested in such a back what they would use it for. I'd guess for some sort of table-top technical photography. But for that application you could get an Arca-Swiss MF-two and a Sony A7r, a 36x24mm sensor is not that much smaller, and soon the 50+ megapixel is coming to that.

I think what people mainly desire in the 'MF geek space' is larger sensors for the legacy MF SLR systems (hassy H and mamiya RZ) and the Hy6, the larger sensor is needed to get the 'look' from the the format's lenses, ie large sensor area is more important than resolution, and live view is of little interest. High ISO would be nice for those working in natural light though. When it comes to tech users I think people are quite satisfied with sensor sizes but lower prices would be nice, and good wide angle compatibility is a must, which the current CMOS doesn't provide. Live view would be great of course, but without wide angle compatibility it's a no go.
« Last Edit: February 26, 2014, 09:51:16 AM by torger » Logged
eronald
Sr. Member
****
Offline Offline

Posts: 4208



« Reply #31 on: February 26, 2014, 09:51:11 AM »
ReplyReply

Yes, Hasselblad V mount sounds like a VERY SENSIBLE idea, because there are plenty of camera test mules out there, put out to pasture Smiley and there are adapters out there for lots of interesting systems like Alpa and Mamiya.

Also, I think quite a few pros would be happy to have a tethered V in the studio for portraits.

I think family chip interfaces are fairly similar, so if we get one chip from a family done, all others will probably fall. In this sense, it might be worth doing prototype work with a smaller Sony sensor.

Edmund

Quote
Anyway, the basic mount should be Hasselblad V, since it is adaptable to about any camera mount there is. Not self contained, the processing/interface done by a smartphone (iPhone, Android) via USB. This makes the back lighter and cheaper since almost anybody has a smartphone these days. Also enables use of already available smartphone apps for processing images and smartphone CPUs are impressive these days (quad cores and such). DNG format RAW.

Sensor size is not deciding, the platform could be made sensor agnostic based on the chip family ...

However the above is just wishes so far :-))) I have little programing skills to contribute.
Logged
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1712


« Reply #32 on: February 26, 2014, 09:55:47 AM »
ReplyReply

I have seen references to open source cinema cameras. Although technically different, perhaps these projects can give some insight into practical dos and don'ts?

https://www.apertus.org/axiom
Logged
G*
Newbie
*
Offline Offline

Posts: 48


« Reply #33 on: February 26, 2014, 09:58:55 AM »
ReplyReply

Hi, for me the larger would be the better, and I think that would be a comfortable niche above P1/Leaf/Hassy/Pentax. What are the problems with large sensors? Production environment? Maybe nobody has tried to build a machine that produces large sensors because nobody sees a market there. One should just try, maybe. Or increasing number of errors on one chip and therefore increasing costs until you have produced an error-free sensor? Maybe one should try harder to overcome errors (blind sensels, whatever) with software. When I have really wild dreams I’m seeing a digital 4x5.  :-)
Logged
eronald
Sr. Member
****
Offline Offline

Posts: 4208



« Reply #34 on: February 26, 2014, 10:04:34 AM »
ReplyReply

Torger,

 These are interesting thoughts.

 I believe the large Sony sensor will exist in other versions, among which mono and multispectral, so there might be versions which don't suffer from color crosstalk, and audiences which don't come here - eg. repro for paintings.

 Regarding the A7/A7R,  as these chips are from the same family and CMOS too, we can probably use the smaller ones to prototype the larger ones; an offshoot of this work might be a very cheap 35mm back, and an offshoot from this might be someone like Arca Swiss making a miniature tech camera that can work well with 35mm sensors - just think of how nice that would be!

 Frankly, I think that a miniature Alpa-style 35mm camera with interchangeable backs and iPhone display will appear sooner or later, even if people have to saw Sony cameras in two to make the backs. Taking a hammer to an existing product to make a new one is a well-known design technique Smiley However one reason to start with MF is the cameras already exist and larger devices with larger tolerances are easier to deal with mechanically.

 I don't know what people really "want". This is the reason why we are having this discussion ...

Edmund

 
Ah okay, did not understand your proposal was locked to CMOS, sorry.

With CMOS I'd say it's not a suitable time to make an open source digital back Undecided. There is only one option which is the Sony 44x33mm sensor, and I think the Pentax offer will be cheap enough, and for those that want a detachable back the sensor is a bit smallish (for V mount or RZ), and for technical cameras it has too large issues with crosstalk to work well with technical wides. I've made quite thorough analysis of the crosstalk issue on tech wides over at getdpi.

I guess some would still like to have it, but I would personally not be too interested in the back. I'd like to hear those that would be interested in such a back what they would use it for. I'd guess for some sort of table-top technical photography. But for that application you could get an Arca-Swiss MF-two and a Sony A7r, a 36x24mm sensor is not that much smaller, and soon the 50+ megapixel is coming to that.

I think what people mainly desire in the 'MF geek space' is larger sensors for the legacy MF SLR systems (hassy H and mamiya RZ) and the Hy6, the larger sensor is needed to get the 'look' from the the format's lenses, ie large sensor area is more important than resolution, and live view is of little interest. High ISO would be nice for those working in natural light though. When it comes to tech users I think people are quite satisfied with sensor sizes but lower prices would be nice, and good wide angle compatibility is a must, which the current CMOS doesn't provide. Live view would be great of course, but without wide angle compatibility it's a no go.
« Last Edit: February 26, 2014, 10:15:52 AM by eronald » Logged
eronald
Sr. Member
****
Offline Offline

Posts: 4208



« Reply #35 on: February 26, 2014, 10:05:54 AM »
ReplyReply

I have seen references to open source cinema cameras. Although technically different, perhaps these projects can give some insight into practical dos and don'ts?

https://www.apertus.org/axiom

Absolutely. This is a very serious modular project. Of course, it is aimed at a smaller sensor, and is electronically and computationally much more difficult than what we are discussing, due to the huge data rates of moving images.  

Edmund
« Last Edit: February 26, 2014, 10:17:01 AM by eronald » Logged
eronald
Sr. Member
****
Offline Offline

Posts: 4208



« Reply #36 on: February 26, 2014, 10:08:47 AM »
ReplyReply

Hi, for me the larger would be the better, and I think that would be a comfortable niche above P1/Leaf/Hassy/Pentax. What are the problems with large sensors? Production environment? Maybe nobody has tried to build a machine that produces large sensors because nobody sees a market there. One should just try, maybe. Or increasing number of errors on one chip and therefore increasing costs until you have produced an error-free sensor? Maybe one should try harder to overcome errors (blind sensels, whatever) with software. When I have really wild dreams I’m seeing a digital 4x5.  :-)

This is an interesting set of questions. I believe there are some sensors out there but they are CCD-based. However if you would do some research and report back to us ...

Edmund
Logged
Fine_Art
Sr. Member
****
Offline Offline

Posts: 1152


« Reply #37 on: February 26, 2014, 10:37:29 AM »
ReplyReply

Hi -

There are a variety of reasons why an open medium format camera platform makes sense. In this topic we could discuss whether there would be people interested in investing time and money or contributing such skills in such a product, and what is reasonably attainable in a reasonable amount of time.

In order to fix expectations, I would like to remind people that Linux Torvalds was not very skilled technically when he started Linux, but he got the project off the ground and the pre-existing GNU software suite made the system viable.

Also, the Linux project was  initially a big success for scientists and servers run by professional sysadmins, but  failed to convince the users of conventional desktop PCs. Of course things changed, and now a substantial fraction of the world's phones and tablets run Linux.

In the same way, I believe it would be feasible to do an open source tethered setup in limited time; external processing of the files could then take over: Many on this forum are skilled software engineers. I am not so sure that a mobile camera is an attainable short time goal. The Magic Lantern project has proved very useful to the film-making community, but what we discuss here starts from the sensor, not from an existing working and integrated platform.

Let's discuss -

- what you'd like to see as useful features
- what platform should be a base for the mechanics/optics
- what skills you could bring to the table, and how much time you could contribute


Edmund

Just to be clear, you are talking about a camera OS, not a camera.

Once you start making a real device there are all sorts of quality standards that come up. CE mark. Safety standards based on destructive testing of many units showing it will not start a fire. It will not shock anyone. Industry standards of warranty service.

The software industry has always got to weasel out based on the idea the user can delete something that does not work. The system pretends software can not damage anything. They then wrap their product in EULA to indemnify themselves. This is all something that real companies could not do with real products.

It might be better to make an open source 135 camera format OS. You would get far more coders interested.
Logged
Fine_Art
Sr. Member
****
Offline Offline

Posts: 1152


« Reply #38 on: February 26, 2014, 10:43:09 AM »
ReplyReply

Off topic:

If the software industry was held to the same standards as other industries, you would not have weekly "security patches" of dysfunctional crap. You would not have hackers botting your PC or taking your credit card data.

The current state of software is a disgrace. It is so because everyone lets them avoid all accountability.
Logged
EricWHiss
Sr. Member
****
Offline Offline

Posts: 2440



WWW
« Reply #39 on: February 26, 2014, 10:57:01 AM »
ReplyReply

I'm definitely interested in this project …   

Larger sensors seem more appealing - and I wonder if the 6.1x6.1 sensor someone linked to earlier could be made to work with the Rollei's, RZ's, Mamiya 7's, Fuji GX680, Tech cameras, etc?   Going bigger and square would be my interest.   

To me a back with the sensor standing proud of the rest with some kind of adapter plate with pass through electronics is going to be the most useful to all. You can't exactly start with a Hasselblad V and make it fit to everything.   Plus I know that tech camera users are always looking for a few more millimeters to get the widest lens possible.

I think some of the astronomy sensor packages might offer some ideas - built in pixel binning, active coolers, ability to control the read out speed (which affects noise).  Overall more user control would be awesome! 

I'm okay with a tethered first solution without on back display screen,  but like the ideas of using smart phones for control or previews.
Logged

Authorized Rolleiflex Dealer:
Find product information, download user manuals, or purchase online - Rolleiflex USA
Pages: « 1 [2] 3 4 ... 8 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad