Best camera setup for Fine Art Repro? Scanning backs? High MegaPixel DSLR?

[Ken R]

I asume all you guys have seen and read THIS article by Mark Dubovoy right here on LL.

[ErikKaffehr]

Hi,

Yes I have. It seems that different authors have different experience with multishot techniques. Unfortunately very few posters post actual raw images which may serve the purpose of an objective comparison.

That said, some posters have positive experience with multishot. It is very clear that multishot has the advantage of uninterpolated colour and with 16X multishot also a resolution advantage. On the other hand, single shot is in all probability a simpler solution in most cases.

The good news is that some companies are working on implementing more practical multishot. A multishot device can also be used as a single shot device, but offers the option to use uninterpolated color.

Just to mention, there are also superresolution techniques that can gain resolution from multiple shots.

Personally, I would suggest that the gains from multishot are reduced with technologies increasing the fill factor, like shared transistor designs and gapless microlenses.

Best regards
Erik

I asume all you guys have seen and read THIS article by Mark Dubovoy right here on LL.

[AlterEgo]

I think you should think more before you post "whatever"... Olympus method can't do "true colour" at all and Sony doesn't (officially) have any sensor shift method or patent... it's only a rumour.

it can because they have enough precision to shift sensor by sensel or fraction of sensel... so it is just the matter of them deciding let our firmware write true color ("RGB" per each sensel) raw file...

[Theodoros]

Theo,

 The way MS is done in backs AFAIK is that the sensor is mounted on a slightly mobile base that can be micro-shifted by means of super-accurate piezo actuators in the X and Y directions. Now in the last few years camera companies like Olympus and Sony have started to mount their sensors on mobile bases, although they may be using a different type of actuator that can do fast long runs. The actuators in SLRs have mainly been employed for real-time stabilisation, but with appropriate programming they can doubtless do Multi-shot, and I think we are seeing this with Olympus. Sony and I think Pentax have similar in-body stabilisation system, and barring patent issues I would expect them to follow suit. In the end the stabilisers will probably become an industry staple, just like rapid-return mirrorboxes and vertical-run shutters which once upon a time were specialty items.

Edmund

What I say is different, the implementation that Olympus is currently using (a movement of the same pixel in a perimeter around its initial position at 8 different new ones, in a way that each one has some of the pixel covering space from the original pixel's area) is only to increase resolution... it doesn't prove that the method has the accuracy required as to move the pixel at exactly one or half a pixel pitch... so that it will be able to exactly match the same position for all three colors. This particular method would work even if the movement of the pixel is 80% inaccurate... (it would still cover the space between pixels). Sony is yet to be seen...

[eronald]

What I say is different, the implementation that Olympus is currently using (a movement of the same pixel in a perimeter around its initial position at 8 different new ones, in a way that each one has some of the pixel covering space from the original pixel's area) is only to increase resolution... it doesn't prove that the method has the accuracy required as to move the pixel at exactly one or half a pixel pitch... so that it will be able to exactly match the same position for all three colors. This particular method would work even if the movement of the pixel is 80% inaccurate... (it would still cover the space between pixels). Sony is yet to be seen...

Yes, any new method won't work and cannot be improved

Edmund

[AlterEgo]

it doesn't prove that the method has the accuracy required as to move the pixel at exactly one or half a pixel pitch...

and let me ask how do you know (markering statements aside) that backs do exactly a precise shift ? do you have an ind. test (raw files available not from manufacturer) where you can see that multishot will really get a _1_ sensel wide line from the target projected on its sensor to raw file in the same manner (same sensels locations) as w/ the same setup in single shot mode - so we can compare 2 raw files and see that shift was exactly that precise... that in both raw files only exactly the same sensel locations are received exposure ( of it that is a black line - then not receiving :-) ) from that 1 sensel wide line... do you ?

[Theodoros]

Yes, any new method won't work and cannot be improved

Edmund

That's not what I say either... I say that for the moment it doesn't work (for true color) and remains to be seen if it (ever) will...  I just don't like to make assumptions like "here is an alternative technology...", that refer to technologies which the maker themselves (Olympus in this case) have never claimed that they are after...
Sony OTOH, I believe they have a patent that involves moving the Bayer pattern (only) with respect to the (constant) sensor... I find this a brilliant idea, I could then mount it on a Campo Actus, use my C645 lenses on the Campus and get rid of the (bulky) Fuji.... Imagine that? There will be only the C645 lenses needed for all my Nikon cameras (via the JAS adapter I recently purchased), the Actus and the MF system and there will be true color and movement ability all in a bag! ...The photographer's dream if you ask me.

[ErikKaffehr]

Hi,

check this http://petapixel.com/2014/08/20/hasselblads-new-h5d-200c-multi-shot-spits-massive-200-megapixel-files/

And also the image samples here: http://petapixel.com/assets/uploads/2014/08/multishot1.jpg

Don't forget to click on the image for full size

Best regards
Erik

[eronald]

Theo,

 It's just a patent issue. Modern in-camera stabilisers can do this. And of course one needs to write the software to take the images, the software to assemble them, and and and -

Edmund

That's not what I say either... I say that for the moment it doesn't work (for true color) and remains to be seen if it (ever) will...  I just don't like to make assumptions like "here is an alternative technology...", that refer to technologies which the maker themselves (Olympus in this case) have never claimed that they are after...
Sony OTOH, I believe they have a patent that involves moving the Bayer pattern (only) with respect to the (constant) sensor... I find this a brilliant idea, I could then mount it on a Campo Actus, use my C645 lenses on the Campus and get rid of the (bulky) Fuji.... Imagine that? There will be only the C645 lenses needed for all my Nikon cameras (via the JAS adapter I recently purchased), the Actus and the MF system and there will be true color and movement ability all in a bag! ...The photographer's dream if you ask me.

[ErikKaffehr]

Hi,

This was posted by Quentin Bargate over ta GetDPI.

http://www.getdpi.com/forum/598326-post1.html

Forum software resizes the images, click on the images for full size:

Single shot:


Multishot:


Best regards
Erik

[Theodoros]

Theo,

 It's just a patent issue. Modern in-camera stabilisers can do this. And of course one needs to write the software to take the images, the software to assemble them, and and and -

Edmund

I don't think it's a matter of developing the technology as to achieve accuracy... the method that OLY choose to res up things suggests to me that they couldn't go for "dead on" control of accuracy... Don't forget that piezo crystal is a movement device... while image stabilisation is an oscillation damping device... i.e. the first one measures length, while the other accelerates dumbing around a centre... IMO the principal is different to an extend that one can't replace the other... Sony in the OH uses a stepping device to move the BP with respect to the sensor... not an oscillating one... (although they have one).

[eronald]

I don't think it's a matter of developing the technology as to achieve accuracy... the method that OLY choose to res up things suggests to me that they couldn't go for "dead on" control of accuracy... Don't forget that piezo crystal is a movement device... while image stabilisation is an oscillation damping device... i.e. the first one measures length, while the other accelerates dumbing around a centre... IMO the principal is different to an extend that one can't replace the other... Sony in the OH uses a stepping device to move the BP with respect to the sensor... not an oscillating one... (although they have one).

Theo,

 With the greatest respect, we are here to learn from each other.

 Stabilisation means choosing a reference and holding steady for the duration of the exposure. So to do multi-shot you tell the system to shift its reference by one pixel x, y, or xy and hold steady Of course you need to be able to tell the system how far one pixel absolute shift is, but that is ... engineering
 
 Sony have a stabilizer in the A7II which is doubtless a voice coil-actuator system which can do this; they have a different sensor patent but that is in the future.

Edmund

[AlterEgo]

Hi,

check this http://petapixel.com/2014/08/20/hasselblads-new-h5d-200c-multi-shot-spits-massive-200-megapixel-files/

And also the image samples here: http://petapixel.com/assets/uploads/2014/08/multishot1.jpg

Don't forget to click on the image for full size

Best regards
Erik

I am sorry - where are the raw files provided by a 3rd party :-) with the line in the image that we can judge upon the inspection of the said raw files (from a single shot) as being 1 sensel wide ? no such files... you posted a link to a webpage containing JPGs provided my a H. marketing department... hence the claim that in the actual field situation (not a staged shot by marketing department) multishot back can achieve a one sensel precions in sensor shift is yet to be proved/proven beyond words

[Theodoros]

You obviously haven't ever use a multishot back... They have complete accuracy down to half a pixel movement... Imacon/Hasselblad are self calibrated by taking leading shots and then have a failure massage in the software if something has gone wrong, Sinar provides a calibration method in its software and a specially designed card to check alignment for the user in case he detects misalignment of pixels... But the system is so accurate and stable that most users only rarely use it and only to check that the system performs under spec...

[eronald]

You obviously haven't ever use a multishot back... They have complete accuracy down to half a pixel movement... Imacon/Hasselblad are self calibrated by taking leading shots and then have a failure massage in the software if something has gone wrong, Sinar provides a calibration method in its software and a specially designed card to check alignment for the user in case he detects misalignment of pixels... But the system is so accurate and stable that most users only rarely use it and only to check that the system performs under spec...

Theo -

"A failure massage in the software when something has gone wrong "

I'm laughing so loudly, I'm waking up the rest of my household.



Edmund

[ErikKaffehr]

Hi,

I am aware of all that, but I wanted to find examples. I did recall that Quentin Bargate posted some samples of his own:

http://www.luminous-landscape.com/forum/index.php?topic=97360.msg798077#msg798077

Quentin's samples also come without raw files. Quentin is one of those guys who share good info without the bragging factor.

I certainly advocate raw files and try to post raw files for all my articles. There are couple of leading advocates of multishot on these forums, but I never have seen raw images posted by them.

Anders Torger, who is doing some development work on raw conversion software (Lumariver HDR and RawTherapee) says that it is difficult to get raw samples from MFD equipment. Great pity.

Best regards
Erik

[EricWHiss]

One of the many, many reasons why Multishot is an amazing concept, but a very precarious practical application.

Oh no!  I must be more careful then!   I have used the multishot and micro step with a pair of old Metz 60 CT-4 flash units, with all kinds of stuff really.  I had no idea how dangerous this was.  It just seems to work… I had no idea how I was living one the edge.   

[ErikKaffehr]

Hi Eric,

Doing multishot and microstep with the equipment Doug sells is clearly problematic, so I can see he has some reservations about the technique. Would you present some examples demonstrating the efficiency of technique, it would be nice. I have been able to dig up a vendor sample and a very good sample by Quentin Bargate over at GetDPI.

Best regards
Erik

Oh no!  I must be more careful then!   I have used the multishot and micro step with a pair of old Metz 60 CT-4 flash units, with all kinds of stuff really.  I had no idea how dangerous this was.  It just seems to work… I had no idea how I was living one the edge.   

[Theodoros]

Theo -

"A failure massage in the software when something has gone wrong "

I'm laughing so loudly, I'm waking up the rest of my household.



Edmund

Ι don't see why you laugh... Things can go wrong despite the perfect calibration that there is... If the photographer fails to eliminate vibration or lighting variations, the capture may fail, Hasselblad software can detect the failure and a "vibration detected" message appears on the software..

[AlterEgo]

You obviously haven't ever use a multishot back... They have complete accuracy down to half a pixel movement... Imacon/Hasselblad are self calibrated by taking leading shots and then have a failure massage in the software if something has gone wrong, Sinar provides a calibration method in its software and a specially designed card to check alignment for the user in case he detects misalignment of pixels... But the system is so accurate and stable that most users only rarely use it and only to check that the system performs under spec...

I think you did use and you might be so kind as to put raw files behind your words  ... no ? no raw files ? only blah-blah-blah then... that we all can do