Diy Digital Rf - Is It Possible?

[Pickett Wilson]

I dunno, Dave. I'm not a luddite. But I can't understand why any Leica user would have the least interest in a digital back that didn't produce top quality output to match the Leica Body and Leica lenses. Why hang a compromise off the back of a no compromise camera?

 

[johannielscom]

It's still only a proof of concept (and an interesting one), not a valid business model. Start with the basics and extend later on, that's the idea.

 

[reklats]

Frankie, so you are saying that the plan is to just rely on the latitude of the sensor and fix the exposure in post? That doesn't seem like the path to optimum image quality. And why would you want to compromise image quality just to get some kind of digital image from an old Leica? A $500 Canon XTi has far fewer compromises.

I own and shoot Leicas, and I just can't get my head around why someone would want something with these kind of minimal specs.

Assuming you have a good enough A/D converter so you don't lose the precision there for low photon counts, then the result is the same. Nothing within the sensor itself changes with changed ISO in any digital camera, it's all after the capture. Of course, if you do have a setting for ISO then you can optimise the readout process.

The IR filter blocks half of the spectral response of a Silicon Sensor. When you remove it, you pick up a full-stop. You are getting all the light out to 11,000A.

(Guessing you mean 11000 Ångström?) Well, of course it blocks quite a bit up there, that's what it's for. That's why said visible range. If IR-photography is your thing, then it's certainly different, but within the visible range it's not so much blocked. Normal black and white film doesn't seem to go that high either, more like ~650 nm. The filter on the Kodak sensor I checked just quickly cuts around there as well. So if you have a scene with a lot of light between 650-1100 nm then it cuts quite a bit, but then do you really want that light all the time? (Of course, options. Can always skip the IR filter and when doing photography where IR is unwanted use an IR-cut filter on the lens. Best of both worlds?)


Depending on sensor type the vignetting can get quite severe. Perhaps not rule out a way to set focal length used so compensation can be encoded in the DNG?
This isn't really my field to be able to help with development (I'm CS, not EE) but I'll sure keep an eye on this thread.

 

[reklats]

The Shutter mechanism controls the exposure of the CCD. The trick is when to start the integration times. "IF" I were doing this -and I am not- an interface controller using a photodiode at the film gate would be able to sense and measure the shutter speed in realtime. Probably an embedded microcontroller such as a PIC16F716 could handle the photodiode, realtime shutter-speed measurement, and signal the main CPU for start/stop time of the Sensor integration.

http://www.microchip.com/wwwproducts/Devices.aspx?dDocName=en010216

Not that i know anything about this stuff.

I'm not sure, but I don't thing it's necessary with a separate controller for that part. Unless the sensor is very different from the ones I've seen then it's not really about starting integration as the shutter fires, more about before the shutter fires clearing out junk from the sensor (noise) and then as soon as the shutter closes (to avoid more noise) reading out the final data. That can just be done with the main cpu (or fpga?). The fewer components the better.
Not that I know anything either, but it's an interesting subject.

 

[djonesii]

I spend a bit of time on the software side of things ..... helping to write geophysical data processing systems, and then to some extent marketing geophysical data processing services, as such, in my professional life, I know little about digital camera hardware development, but

There is a OPEN software development platform out there based around the android phone, while I'm not suggesting that a phone be built into the camera, all the needed control electronics and software to take photos are already in the public domain .... I would suggest that a path to move from alpha on through to product could be built based around some key components from the android, and their SDK.

The more knowledgeable sensor folks can correct me if I'm wrong, but it seems if you have the mechanics/software of taking a photo with a small sensor and cheap plastic lens out there on the web for free, it's a much simpler path to package that FIRST, then "upgrade" the sensor and lens to what you really want.



See here for SDK info:

http://groups.google.com/group/android-developers/browse_thread/thread/922afdd7ecc0146f

 

[djonesii]

A comment about screens .... It would seem to me that a screen is necessary in this type of product. The analogue controls need to be left alone, but it would be handy to get some type of information about the functions of the RF-DB "Range Finder Digital Back" As I recall from looking a medium format digital backs, the early ones just had a simple LCD with basic information.

Some kind of confirmation that the photo is on the card is essential in a commercial product .... even a simple capacity counter would do this job. It would also be nice to know a few of the basic settings, if they can actually be built into the output engine.



As a boot note, I recall reading that the analogue gauge cluster on the RD-1 was the most expensive part of the camera ...

 

[Pickett Wilson]

reklats, so the idea would be to provide some kind of custom imaging editing software with the back that would seamlessly do all the processing on the buyers computer that is normally done in camera? Exposure correction, noise reduction, vignetting correction, etc?

 

[reklats]

reklats, so the idea would be to provide some kind of custom imaging editing software with the back that would seamlessly do all the processing on the buyers computer that is normally done in camera? Exposure correction, noise reduction, vignetting correction, etc?

No, the idea (as I see it) is to dump the raw data in a DNG-file, encode everything needed about noise and vignetting and let any available DNG-capable software (adobe or other) handle the actual processing, as encoded in the file. I just checked the DNG-spec quickly and vignetting correction can be done easily with an opcode.

 

[gandalfk7]

Extremely interesting,
I would help if I could but I almost forgot all the things learned in electronics
highschool..
I work in the IT, I could help in any way?
also financing, when it's time.

 

[Sonnar Brian]

(Guessing you mean 11000 Ångström?) Well, of course it blocks quite a bit up there, that's what it's for. That's why said visible range. If IR-photography is your thing, then it's certainly different, but within the visible range it's not so much blocked. Normal black and white film doesn't seem to go that high either, more like ~650 nm. The filter on the Kodak sensor I checked just quickly cuts around there as well. So if you have a scene with a lot of light between 650-1100 nm then it cuts quite a bit, but then do you really want that light all the time? (Of course, options. Can always skip the IR filter and when doing photography where IR is unwanted use an IR-cut filter on the lens. Best of both worlds?)

This isn't really my field to be able to help with development (I'm CS, not EE) but I'll sure keep an eye on this thread.

I had Kodak leave the IR filter off of my camera to allow visible+Near Infrared. I can use a hot mirror filter for "just visible" and an R60 or deeper for the Infrared. A lot of people would like the Infrared option and pay to get cameras modified to do it.

 

[Sonnar Brian]

I'm not sure, but I don't thing it's necessary with a separate controller for that part. Unless the sensor is very different from the ones I've seen then it's not really about starting integration as the shutter fires, more about before the shutter fires clearing out junk from the sensor (noise) and then as soon as the shutter closes (to avoid more noise) reading out the final data. That can just be done with the main cpu (or fpga?). The fewer components the better.
Not that I know anything either, but it's an interesting subject.

The main CPU (or main circuit) could easily do this. The PIC chip drawa minimal power. It is also a very small and inexpensive part. "IF" i was doing this, using the PIC to start the processor when the shutter is fired would be for power savings.

 

[rxmd]

I had Kodak leave the IR filter off of my camera to allow visible+Near Infrared. I can use a hot mirror filter for "just visible" and an R60 or deeper for the Infrared. A lot of people would like the Infrared option and pay to get cameras modified to do it.

I agree, but I'd also assume that a lot of people who aren't all that interested in IR and don't want to put IR filters on all of their lenses just because the camera is sensitive down to 1100 nm. In fact I'd assume this to be the bigger group. I'd prefer an easily-removable IR filter on the sensor.

 

[reklats]

The main CPU (or main circuit) could easily do this. The PIC chip drawa minimal power. It is also a very small and inexpensive part. "IF" i was doing this, using the PIC to start the processor when the shutter is fired would be for power savings.

Don't get me wrong, I love PIC for building small simple things. I just don't think it's necessary in this case. Either the main CPU needs to be started before to do some preparations, or it's only needed after. In both cases the same signal monitored by the PIC can be used to wake up the main chip from the same deep sleep that the PIC would wake it from.

And for the IR I think you misunderstood me. I wasn't saying it doesn't cut light. I was just saying that for visible light photography, which I guess is what most would be interested in, you don't get a one stop advantage by removing the IR filter. The extra light you get is only up there in the near infrared (with a bit in the upper visible red, depending on filter). Using a sensor that is available both with and without the filter, it might be possible to offer a choice to the buyer perhaps.

 

[Frankie]

The ISO rating of a film is only a manufacturer's suggestion. Some are more optimistic than others.

We can fine tune it the way we want the image to look...overexposure/underdevelopment to lower contrast; underexposure/overdevelopment to increase contrast... Special and hotter soup if you must push [essentially underexposure but adding back some chemical fog in prolonged or hotter processing].

Colour film processing is more restrictive, both in time and temperature, we cannot do much.

Before there was light meter, we use the sunny 16 rule [only applicable in daylight] to start...but for any geographic location and any date, ever wonder why?

Later, we have light meters. We set them according to what the film box says or what we had experience in...whether the meter is hand held, built into the camera, or automatic [in exposure cut off...in time or aperture].

The light meter does not and cannot know if you lied.

Digital is no different. There is a native ISO. Many camera makers mark it as a somewhat pessimistic ISO 200, so that users have less chance to blow out the high light...and let the shadows fall as they may [within the bit depth in the tail end, just like transparency films]. Post-processing can boost the shadow details if you choose, but not much beyond 3 or 4 stops or electronic noise will become visible.

Auto-levelling is the generic term. You can of course print from the DNG files straight...if your exposures were correct.

In using this back-to-basics digital back, first learn the basics…didn’t all Leica aficionados?

New inputs from yesterday...

It's still only a proof of concept (and an interesting one), not a valid business model. Start with the basics and extend later on, that's the idea.

Thanks CLE, you got the whole point.

Frankie, so you are saying that the plan is to just rely on the latitude of the sensor and fix the exposure in post? That doesn't seem like the path to optimum image quality. And why would you want to compromise image quality just to get some kind of digital image from an old Leica? A $500 Canon XTi has far fewer compromises.

I own and shoot Leicas, and I just can't get my head around why someone would want something with these kind of minimal specs.

Before light meters, we all must believe the film manufacturers’ rating and rely on the inherent latitude. After light meters, we still must first believe then set the device.

In digital imaging, the same essentially applies, albeit now opinions of camera makers. [Wasn’t the M8 originally rated a slightly lower ISO…mine is lower than yours?]

In film, optimum image quality is obtained by selecting the correct emulsion; set exposure correctly; process according to manufacturers’ recommendation [before getting creative]; tweak the print somewhat later [within reason]…

In digital imaging, the same protocols apply, albeit exposure can now be aperture-, shutter- or ISO-priority, or a programmed optimal combination of two, or all three variables…so long ISO is selected within reason [Leica says 200~3200; Nikon says 200~12,800…mine is higher than yours].

No enforced compromise of IQ is ever entailed by anyone; and no cradle-to-grave unspecified support is ever possible…it is always up to you. Isn’t that the original Leica way?

……Unless the sensor is very different from the ones I've seen then it's not really about starting integration as the shutter fires, more about before the shutter fires clearing out junk from the sensor (noise) and then as soon as the shutter closes (to avoid more noise) reading out the final data……

No difference…and same concept as keeping film in the dark, refrigerate if necessary, not to load or unload the camera is bright sunlight…according to Kodak.

How best to achieve that goal digitally is engineering.

……I can't understand why any Leica user would have the least interest in a digital back that didn't produce top quality output to match the Leica Body and Leica lenses. Why hang a compromise off the back of a no compromise camera?

M8/9 collectors won’t be interested at all. I am surprised they have not already launched attacks…a side effect of Stockholm syndrome.

M8/9 users might be interested in adding a low-cost [not cheap] back to their legacy M’s, assuming they have some.

M2/3/4/6/7 users would. Their camera body [deemed obsolete in the new Leica paradigm] and beloved lenses are still top quality [in their eyes] and unaltered in any way. My idea of a dM-back simply gives them a choice…as Leica did not.

Assuming you have a good enough A/D converter so you don't lose the precision there for low photon counts, then the result is the same. Nothing within the sensor itself changes with changed ISO in any digital camera, it's all after the capture. Of course, if you do have a setting for ISO then you can optimise the readout process.

Or accept the raw data…ISO is ideally pegged at native with underexposure biasing. A simple membrane switch in the package works.

[At this point, I will not be bogged down by such trivial mechanics. I do have some experience using high MTBF mil.spec. switching, diode confirmation lights and other devices.]

I guess many wished for auto-ISO, among other wishes…from 3-stop stabilization to sensor self cleaning, and other alphabet soup features that might come into mind.

Depending on sensor type the vignetting can get quite severe. Perhaps not rule out a way to set focal length used so compensation can be encoded in the DNG?

Post-processing R-D1 style works well enough. Other post-processing free/share-ware is also available.

Leica patented the M8/9 6-bit encoding for in-camera processing hopefully to garner also all future lens sales. The vignetting correction need not only be done in-camera…6-bit or no-bit is irrelevant. If super-wide only aficionados won’t buy, then they can wait for Leica to retrofit their beloved legacy M’s.

A digital anti- pixel vignetting [not lens] filter is contemplated, pending research into the range and/or classes of lens rear-nodal points likely to be encountered.

reklats, so the idea would be to provide some kind of custom imaging editing software with the back that would seamlessly do all the processing on the buyers computer that is normally done in camera? Exposure correction, noise reduction, vignetting correction, etc?

I wouldn’t hoard that business.

The open DNG data format can be read by all. Interested parties may compete freely.

I agree, but I'd also assume that a lot of people who aren't all that interested in IR and don't want to put IR filters on all of their lenses just because the camera is sensitive down to 1100 nm. In fact I'd assume this to be the bigger group. I'd prefer an easily-removable IR filter on the sensor.

IR imaging is perhaps Brian Sweeney’s professional work. Many photographers played with IR films a bit…I never did and have no interest.

*****​

Thanks everyone for your input.​

I maintain my original intent of simply replicating film for legacy Leica M’s, not because of anything but a large ready-made market abandoned by Leica. I didn’t have to sell customers cameras first; AND, they are stuck...

[If Zeiss would announce a ZMd soon, I would simply buy it and for use with my beloved CV 40/1.4. My film ZM can retire or function as a finger toy. My new focus will be my Nikon F, F2, FM…Nikon won’t retrofit them either. ]

My old stereo-plotter retrofit was a successful business [and earned us some capital] because of the exact same reason. Several years later, I killed it, giving way to our all digital system that requires no precision mechanicals and far surpasses the best instruments Leica and Zeiss had ever achieved…my stepping stone to my final game.

Leica-Heerbrugg was then caught unprepared and had to acquire the marketing rights of a competing system to sell as its own…but eventually got fired. Zeiss had withdrawn from this market entirely…too little too late.

Today, no self-respecting photogrammetrists will speak of their legacy M equivalent of a stereo plotter, including ones we had retrofitted. Scrap metal dealers won’t even take it for free…

This dM-back project will be one of several possible early-retirement project under consideration and hopefully soon to start. It will not be pushed into a beat everything-Leica attempt, rather a project to keep me sharp and make a little fun money. I do have other vices.

 

[Sonnar Brian]

>This dM-back project will be one of several possible early-retirement project under
>consideration and hopefully soon to start. It will not be pushed into a beat
>everything-Leica attempt, rather a project to keep me sharp and make a little fun money.
>I do have other vices.

I applaud you for this. That's the attitude that I want to have when I go down that path.

 

[Jim-st]

This is a fantastic project, and this thread is the most exciting I've read anywhere. Can't wait to order mine!

Just one question: if no LCD (and I kind of hope there will be an LCD) how will we format our cards?

 

[aniMal]

Great that this thread does not drift too much, and I hope the original, barebones idea is the one that prevails. The target is photographers who are used to working on film, so the lacking screen and menus will not be perceived as a problem once it is for actual sale!

Now, this talk of getting BW and IR models... I would buy one colour and one BW if that comes true!!!

Also, if this would give me the same image quality as on my M8 after doing the post processing I do manually on the M8 anyway, then I would sell the M8 and get an M2/4/6 for these backs. Instantly.

Frankie, what you mentioned about mil.spec switches and diodes, keep that as a major feature. If it is milled in metal, and provided with just a few diodes/switches that look and feel serious, then it would fit so good in with a solid old M. How about a green diode lighting for a few seconds every time a file has been saved? Blinking when the card is about to be filled up?

Things like that... The only thing I miss from my sold RD-1 myself, is the fancy gauges

On the note of backs: the M2/3/4 backs should be the same AFAIK, but with some minor differences in feel. I have switched between M2 and M3, and they were the same dimensions - but felt different when closing. Small differences in dimension of the locking pins I believe. Not too sure about the M4, but I think it should be fine... The M6/7 on the other hand, have backs that are linked to the electronics of course. Would need to have something detachable retrofitted, but should be possible!

I agree that there are too many naysayers and experts, the ones that do not see the point of such a KISS design will not be the first buyers anyway... What about setting up a closed group somewhere? Then use RFF to take in the ones that are really interested?

I really hope this will become reality, and that it does not stop on getting too small or too big... What is being discussed now is exactly what I would have wanted to buy instead of the M8, which I am am quite happy with as it is. Only thing is the menus, good as though they are - I could well do without all of them & the screen!

 

[Sonnar Brian]

My Nikon E3 does not have an LCD screen. Nor does the Kodak DCS200ir. The Nikon has two buttons that must be pushed and held simultaneously. It formats the PCMCIA card to FAT16. This is readable with any Laptop or computer PCMCIA card reader. The SD card could be handled the same way for the Digital back, or could be formatted by the computer's card reader.

The Kodak must be plugged into the PC running the TWAIN driver. A command is issued via the PC Twain driver to format the fixed 80MByte SCSI drive built into the camera. Anything is easier than that!

 

[aniMal]

Exactly the format buttons I would prefer - easier than going into a menu, and very hard to do by mistake. Or perhaps a push & slide button...

If this gets built, it would also be possible to write/adapt some software later on that deals with converting the DNGs... Or perhaps just some scripting into existing converters???

 

[dan_a]

UFRaw is a free and open source program which could be used to convert DNGs.

I would have thought for an initial proof of concept that things like formatting a flash card would be best done on the user's PC, not directly on the camera