Luminous Landscape Forum
Equipment & Techniques => Medium Format / Film / Digital Backs – and Large Sensor Photography => Topic started by: Willow Photography on August 21, 2014, 11:04:56 am
-
Some month ago I bought a Hasselblad H5D40 (great offer from Hasselblad , 45% of new ).
I love the skin tones and the viewfinder and the D800E is not getting so much use these days.
A couple days ago, I bought a Hasselblad Ixpress 528C on a HD1 used for under $2000.
Its 1-shot,4-shot or 16-shot.
I tested it against my H5D40 and the 4-shot is sooooo much better.
The resolution, colours and the lack of artefacts are out of this world.
Its a 22M back, but even when I resize up to 40M, its way much better than the H5D40.
Shooting products and interiors will not be shot with the H5D from now on, thats for sure.
And I haven't even tested the 16 shoot yet.
That will give a 88M back!!
Willow
Are there any other multishot users here?
Any tips? What to do and what not to do?
-
Hi DougPeterson
You sent me a personal message on this topic.
I hope it is ok that I show it here, because I think it is a valid discussion.
You wrote : One of the reasons your 4-shot looks so good is because Hassy and LightRoom's algorithms are way behind Capture One for detail on 1-shot captures. If you compared to the quality of an IQ140 processed in Capture One v7 you'd be surprised how much of that advantage goes away (or wins if there is ANY vibration or imprecision in the multishot). I've done this test many times, as our Cultural Heritage division has been competing against multishot for many years (http://www.dtdch.com/).
I so not agree with you on this.
I have had a lot of P1 backs ( Lightphase, P30, P65+ and IQ140. )
And I have had some Hasselblads.
I used C1 for a long time, both on P1 backs and Canon and Nikon.
Don't use it anymore for several reasons.
I like the skin tones from the Hasselblads much more than the P1s, and I don't find them any
less sharper than the P1s.
But of course C1 have some tricks. Even if you turn the sharpening down to zero, it adds some sharpening.
Just for fun I found some of my IQ140 shots from 2 years back and compared them to my H5D40 files.
Resolution wise there is no difference IMO.
-
I am a long time multi-shot user tho' less so these days as I shoot mostly food which moves. The difference is indeed dramatic! 4-shot mode is actually tolerant of sight movement but a decent tripod and mirror up is good practice. I always grab a single shot as well just in case. 16 shot is much more demanding, a really solid support and a delay between shots to allow the camera to settle but the results are pretty amazing.
HTH
Nick-T
-
And YI there is now a 50C multishot that has a six-shot mode that produces 200MP files:
http://www.pdnonline.com/gear/Hasselblad-H5D-200c--11438.shtml
Hopefully Doug will chime in to tell us how terrible they are :)
-
Some month ago I bought a Hasselblad H5D40 (great offer from Hasselblad , 45% of new ).
I love the skin tones and the viewfinder and the D800E is not getting so much use these days.
A couple days ago, I bought a Hasselblad Ixpress 528C on a HD1 used for under $2000.
Its 1-shot,4-shot or 16-shot.
I tested it against my H5D40 and the 4-shot is sooooo much better.
The resolution, colours and the lack of artefacts are out of this world.
Its a 22M back, but even when I resize up to 40M, its way much better than the H5D40.
Shooting products and interiors will not be shot with the H5D from now on, thats for sure.
And I haven't even tested the 16 shoot yet.
That will give a 88M back!!
Willow
Are there any other multishot users here?
Any tips? What to do and what not to do?
Well... I hate to say this, but you'll be sorry you ever sold your Contax system if you ever try the 528c with Contax glass! :) Trust me! I own the combo and I've tried it on a friend's H2 for comparison (bought the back in H-mount and still have the adapter).... Especially with the 120mm Apo in 16x microstep the result is simply ...shocking! :o Welcome to the club! 8)
-
Hi DougPeterson
You sent me a personal message on this topic.
I hope it is ok that I show it here, because I think it is a valid discussion.
You wrote : One of the reasons your 4-shot looks so good is because Hassy and LightRoom's algorithms are way behind Capture One for detail on 1-shot captures. If you compared to the quality of an IQ140 processed in Capture One v7 you'd be surprised how much of that advantage goes away (or wins if there is ANY vibration or imprecision in the multishot). I've done this test many times, as our Cultural Heritage division has been competing against multishot for many years (http://www.dtdch.com/).
I so not agree with you on this.
I have had a lot of P1 backs ( Lightphase, P30, P65+ and IQ140. )
And I have had some Hasselblads.
I used C1 for a long time, both on P1 backs and Canon and Nikon.
Don't use it anymore for several reasons.
I like the skin tones from the Hasselblads much more than the P1s, and I don't find them any
less sharper than the P1s.
But of course C1 have some tricks. Even if you turn the sharpening down to zero, it adds some sharpening.
Just for fun I found some of my IQ140 shots from 2 years back and compared them to my H5D40 files.
Resolution wise there is no difference IMO.
The reason that 4x shots of only 22mp look absolutely stunning (and 16x look extraterrestrial) is that no colour interpolation (and the inevitable errors of it) exist... As a result, the human eye understands the real colour that is recorded on each different pixel as resolution (which it is) and all that without the presence of any artefacts (or hint of artefact existence like moire) what so ever! The additional 2 stops of DR only helps things further for the difference between neighbouring pixels to be more evident!
It's simply the time when one realises that 22MP are more than plenty to have for stunning prints of more than 2m^2 of size... (if interpolation of colour was absent) :'(
Additionally, the superb image (when compared with other modern single shot backs) of all Kodak sensor22mp backs (the now "famous" "fat pixel magic") when combined with the fact that single shots (common DSLR use) only very rarely needs higher resolution and the additional benefit that the sensors that where used on 528c where "hand selected" for optimum performance, as well as that the "fat pixels" make the back particularly "friendly" with view and tech or other cameras that offer lens movements, ...all contribute to what one may call "the perfect MFDB"! ;D
-
I used a Hassey CF39-MS for years. The 4-shot files are extraordinary. I don't care what anyone says, I've seen files from just about any camera ever made for fine art repro. That back mated with the 72mm Digitar (electronic shutter instead of mechanical) on a pancake camera produced better files than any single shot system I've seen. And for as much grief as people vent towards Phocus, it makes for a fine RAW developer. With repro work I shot in "repro" mode which rendered a flat response curve--exactly appropriate for copy work. Prior to using the CF39-MS, I worked with the 528C's little brother, the 384C. It had 1, 4, 16 shot capabilities too. I rarely used 16-shot. If I had to use that back today, I'd work in a darkened studio and shoot use the 16-shot capability in "pinhole" mode to bypass using a shutter. The studio was in a virtual zero traffic zone, and the floor 2 feet thick concrete. Rock steady!!
-
What Doug keeps proving over and over again is that good advocacy and salesmanship will improve a user's perception and experience of a product but some of us realize that this doesn't improve the product itself. Adding a concierge to a third rate hotel doesn't magically fix the airconditioning. Phase makes good single shot backs, but multishot is still better than single shot, especially for textiles and jewelry, trucks still carry more weight than motorbikes, and and pigs still need help to fly (http://abcnews.go.com/US/story?id=94861).
Frankly, I think that the institutional market would be better served with a mono sensor and filterwheel setup, but that's a different story.
I think that it was Kayne West who said that the recipe for having better music in his repertory was to make better music - Doug might tell this to Phase.
Edmund
Hi DougPeterson
You sent me a personal message on this topic.
I hope it is ok that I show it here, because I think it is a valid discussion.
You wrote : One of the reasons your 4-shot looks so good is because Hassy and LightRoom's algorithms are way behind Capture One for detail on 1-shot captures. If you compared to the quality of an IQ140 processed in Capture One v7 you'd be surprised how much of that advantage goes away (or wins if there is ANY vibration or imprecision in the multishot). I've done this test many times, as our Cultural Heritage division has been competing against multishot for many years (http://www.dtdch.com/).
I so not agree with you on this.
I have had a lot of P1 backs ( Lightphase, P30, P65+ and IQ140. )
And I have had some Hasselblads.
I used C1 for a long time, both on P1 backs and Canon and Nikon.
Don't use it anymore for several reasons.
I like the skin tones from the Hasselblads much more than the P1s, and I don't find them any
less sharper than the P1s.
But of course C1 have some tricks. Even if you turn the sharpening down to zero, it adds some sharpening.
Just for fun I found some of my IQ140 shots from 2 years back and compared them to my H5D40 files.
Resolution wise there is no difference IMO.
-
I do use the multishot backs still. Nick's advice to take a single shot is good. Occasionally there will be some fine cross hatching in specular highlight areas, boundaries or other areas and if you have the single shot you can overlay this with a black mask and paint in over the the funky areas.
With the 528c and I've just notice the Sinar 54H, the physical orientation of the back may affect performance - i.e. - if you point the camera down instead (like on a copy stand) of horizontal the micro step images may be affected more. Some fluorescent lighting will not work for all shutter speeds.
Multishot gives you 1.5 - 2 stops more DR than single shots - I've measured it. I don't know how you measure tonality, but that is also greatly improved. Also you can stop the aperture down more with multishot than single shot without seeing diffraction - somehow assembling the different frames the software is recovering something or perhaps better worded the bayer interpolation is loosing something that true RGB color for each pixel doesn't.
If you like to compare the new 80mp backs to the 528, you'll find a few things:
The most obvious - The high MP single shot backs like the Phase and Leaf 80mp backs really win in workflow.
For images like maps - where the contrast in the printed lines is high - the 80mp backs will come very very close - perhaps even look better
For images like of textiles or fruit or paintings - the multishot will really shine
I think I was shooting a $20 bill taped to a test chart with the two and I didn't notice much difference until I realized with the multishot I could see the fingerprints on the tape, but with the 80mp back not. It's subtle, but the differences are there.
I shot an art reproduction job with both the CF 528 and the AFi-ii 12 where I was able to use the same lens on both cameras. I ended up using the AFi-ii 12 files for output to the client because it was faster and it was easier for me to get to matching color with C1. Maybe I know C1 better than flex color? (to get micro step you still have to use Flex Color since Phocus does not support 16 shot). The software does make a difference, as does a user's knowledge of how to get the best out of it. To get close to the same micro detail as the CF 528 in that job, I had to output the files out of C1 with low small radius sharpening and then run them through LR sharpening again on a second pass with a larger radius and the detail slider way up. No doubt a wizard with C1 like Doug could do better, but that's what I found worked for me. In my observation, the multishot files require less sharpening overall and gain less from 2nd pass tricks and deconvolution software.
But yeah, I'm a fan of multishot. Tedious and you can only use it on some subjects, but when it comes out right it - it's amazing.
-
One boring thing with multishot is that software choice is limited... it would be cool if the backs could in a self-contained way generate a raw file that then could be processed in various software. Then one could even shoot multi-shot in the field when conditions are good :). Well, for us tech cam users there's a need of electronic shutters too of course, which is a mess in the field...
-
Thx for your input guys.
I have now tested the 16-shot.
No problem with vibration and the result is stunning!!!
The mirror goes up and 16 shots and a few minutes later, it has done its magic.
All automatic thru Flexcolor.
4-shot tested in Phocus and its faster than Flexcolor.
I am amazed that I got this for next to nothing.
Now I only have to find a used H2F or a H4X so I can use my 28 on interiors and architecture.
-
I agree. but phocus only serve 4shot in 39ms back.
22mpx back serve 4shot and microstep 16shot.
I hope phocus serve 16shot to 39mpx back.
it needs software develop.not hardware.
Only hasselblad can do.
but they want to sell newer 50mpx based multishot with expensive price.
-
I agree. but phocus only serve 4shot in 39ms back.
22mpx back serve 4shot and microstep 16shot.
I hope phocus serve 16shot to 39mpx back.
it needs software develop.not hardware.
Only hasselblad can do.
but they want to sell newer 50mpx based multishot with expensive price.
I am afraid that this is not the case... 16X mode is a matter of mechanical capability not of software! It seems that the piezzo electric crystal that moves the sensor can retain accuracy when moved at 4.5μm but 3.4μm (half of 6.8μm) that is needed for the 39mp sensor is too small to be accurately controlled! I believe that this is why the 200ms is using a new technique to create a 4x50mp=200mp file out of a 50mp sensor in which it is moving the sensor only 6 times rather than 16 times that the 528c is using...
Now, I am not sure how exactly that process works, (perhaps Steven Hendrix could enlighten us here) but it seems that the 200ms does the same with 50MS for the first four shots (i.e. moves the sensor by one pixel thus aligning all the Bayer pattern colours on all pixels) and then it adds two more movements of 1+1/2 pixel length in magnitude (obviously to increase the size of the movement) once horizontally and once more vertically so that it then interpolates the presence of 200 pixels....
I haven't tried the 200MS but I think that the result may be impressive but may also be not up to the "real" 16x mode that the 528c can achieve, since there is no interpolation involved there for the 88mp file to be created... That said, I have to say that one must be very careful when comparing his 16x microstep results from his 528c back with anything else. He has to make sure that he has taken the maximum out of his 528c which is not easy to judge at all since one may consider that "because flexcolor detected no fault - the method was done to its perfection" which is rarely the case... My experience out of every day use of the 528's 16X microstep mode has shown to me that the software has some latitude in forgiving mistakes and that if one seeks for perfection there is simply nothing that compare to the result and that a modern 80mp back will pale in detail comparison even if the file is BW converted!
-
It may be You're right.
but I guess piezo microstep motor drive by software.
I heard hasselblad cf-22ms and cf-39ms multishot hardware is same. just only operated by software.
Anyone knows this informations?
-
I think the new Sinar eXact backs are 50mp and can do 16 shots mode. The sinar micro step backs seem to have a piezo calibration routine - just learning about this as I have picked up a sinar 54H which is the equivalent to the 528c (sort of). If I can get the sinar happy on the Rolleiflex 6008AF then I'll be doing a comparison between these two forgotten imaging giants. I had the CF-ii 39MS and its a great back, but only does 4 shot mode not 16 shot.
-
I am looking forward to the Sinar and Imacon comparison. One being a Dalsa multishot the other a Kodak multishot AFAIK.
The new Sinar back 48MP does 16 shot? I thought it only did 4. Indeed it does do 16shot! It is the only back that uses a 48x36 Dalsa sensor with 48Mp (AFAIK).
I am rally curious how that back performs but I have never met anyone that has one. I also like the thought Sinar put behind these backs, eg. No. Screen, to get the cleanest file possible. These things are completely designed as studio and multishot backs. No hybrid stuff like being able to use it on a body in the field with a screen.
-
i think the Sinar backs are made by Jenoptik who i believe just happen to be the originators of the piezo multishot tech.
Edmund
I am looking forward to the Sinar and Imacon comparison. One being a Dalsa multishot the other a Kodak multishot AFAIK.
The new Sinar back 48MP does 16 shot? I thought it only did 4. Indeed it does do 16shot! It is the only back that uses a 48x36 Dalsa sensor with 48Mp (AFAIK).
I am rally curious how that back performs but I have never met anyone that has one. I also like the thought Sinar put behind these backs, eg. No. Screen, to get the cleanest file possible. These things are completely designed as studio and multishot backs. No hybrid stuff like being able to use it on a body in the field with a screen.
-
I am looking forward to the Sinar and Imacon comparison. One being a Dalsa multishot the other a Kodak multishot AFAIK.
The new Sinar back 48MP does 16 shot? I thought it only did 4. Indeed it does do 16shot! It is the only back that uses a 48x36 Dalsa sensor with 48Mp (AFAIK).
I am rally curious how that back performs but I have never met anyone that has one. I also like the thought Sinar put behind these backs, eg. No. Screen, to get the cleanest file possible. These things are completely designed as studio and multishot backs. No hybrid stuff like being able to use it on a body in the field with a screen.
The Sinarback 54h and the Imacon/Hasselblad 528c/22ms both share the same sensor, the Kodak 22mp 37x49mm... The difference between the two (if any) should be only in software (since both backs record true colour images and there is no processor involved to produce colour).
The Sinarback 86h and E-xact again, both share the same sensor, the Dalsa 48mp 36x48mm which essentially is the same as the Dalsa 60mp sensor of FF size used by other makers, but cropped down to 36x48mm, in this version the sensor also lacks microlenses. The difference between them is that 86H is only able for up to 4x multishot, while E-xact can do up to 16x and offers 192mp true colour analysis.
-
i think the Sinar backs are made by Jenoptik who i believe just happen to be the originators of the piezo multishot tech.
Edmund
True, at one point Jenoptik used to own the majority of Sinar shares... After the partnership has ended, Jenoptik is obliged to continue the provision of backs to Sinar. Jenoptik makes them, Sinar owns the rights, distributes them and supports them.
-
True, at one point Jenoptik used to own the majority of Sinar shares... After the partnership has ended, Jenoptik is obliged to continue the provision of backs to Sinar. Jenoptik makes them, Sinar owns the rights, distributes them and supports them.
So basically the backs are sold by zeiss to leica (add time warp).
e.
-
So basically the backs are sold by zeiss to leica (add time warp).
e.
I don't believe that you can put a Zeiss back onto a Leica Ed... ::) But you can put a Sinarback on many different MF bodies that accept an MFDB via an interchangeable adapter plate! ;D Practically, if one buys new, there is no other "open system" MS back that one may have... The backs all offer an interchangeable adapter, which is the same for all three backs. However, Sinar decided not to support via an adapter the Rolleis (or the Contax which I use), so Eric won't ever be able to experience an E-xact unless he invests on another system ;). Lucky me however, I have a friend who has the latest 54m adapter for Contax (which he doesn't need), which is fully compatible with the new series of backs, so if I invest on an E-xact, I'll make Eric jealous. :-*
-
I don't believe that you can put a Zeiss back onto a Leica
onto, maybe not until now. inside, certainly.
jenoptik is zeiss JENA. M8, 9, maybe S2 are jenoptik.
Edmund
-
Yeah, I wish I had a new eXact to play with because I'd really love to see what it can do in 16 shot mode. But I only have the 54H which is about the same vintage as the CF 528 and if I'm not mistaken uses the same sensor.
So far the competition between these two venerable old backs may not be fair because I can't figure out how to keep the mirror up between frames on the Sinar, while the CF 528 does this automatically. It's entirely possible I have the wrong cable or I'm not using the software correctly. If any one has sinar multishot backs with Rollei 6000 series experience I would love to hear from you.
Also the seller who sold me the 54H seems to have misplaced the reference file CD which I understand may improve the IQ. I think a new reference file can be made for it, but a SinarCam2 is needed which I don't have.
It's hard to believe that a screen could add so much heat as to affect the noise in an image, but I do believe a memory card could. If you are shooting fast, those do get hot. So in theory the Sinar backs could produce better IQ with the active cooling and the heat sources removed.
I think the newest Sinar backs have peltier cooling as well.
-
Yeah, I wish I had a new eXact to play with because I'd really love to see what it can do in 16 shot mode. But I only have the 54H which is about the same vintage as the CF 528 and if I'm not mistaken uses the same sensor.
So far the competition between these two venerable old backs may not be fair because I can't figure out how to keep the mirror up between frames on the Sinar, while the CF 528 does this automatically. It's entirely possible I have the wrong cable or I'm not using the software correctly. If any one has sinar multishot backs with Rollei 6000 series experience I would love to hear from you.
Also the seller who sold me the 54H seems to have misplaced the reference file CD which I understand may improve the IQ. I think a new reference file can be made for it, but a SinarCam2 is needed which I don't have.
It's hard to believe that a screen could add so much heat as to affect the noise in an image, but I do believe a memory card could. If you are shooting fast, those do get hot. So in theory the Sinar backs could produce better IQ with the active cooling and the heat sources removed.
I think the newest Sinar backs have peltier cooling as well.
Eric, the recorded colour in a real colour mode is 48bits depth (3x16), the output of 16bits is made later by the software and it is compressed from the original 48bit one. You can't have 48 bits of colour recorded on a card in a 3fr or fff file! The PITA with 54h (as with all sinarbacks)... is the software, ....also, Imacon has a much better (more reliable - better mechanical implementation) piezoelectric mechanism which is totally separate from the rest of the back and has its own heat sinks. On the mirror lock problem you have, I suggest you look at the following...
1. Check if the cable is for Rollei 6008 or for Fuji GX680 (they look like one another) bear in mind that Fuji GX680 can't lock its mirror up permanently which is the behaviour you get on the Rollei. (if it is for Fuji you may sent it to me, I need it!)
2. Check the amount of the delay you've set...
-
Also the seller who sold me the 54H seems to have misplaced the reference file CD which I understand may improve the IQ. I think a new reference file can be made for it, but a SinarCam2 is needed which I don't have.
I bought my 54h secondhand too and without the ref file. I wrote a letter to Sinar and they uploaded the Ref File on their FTP server for me. Maybe they keep the ref files in the archive.
-
Anyone got any samples shot with 16x? So far I have never been able to find any true multishot photos anywhere.
-
Eric, the recorded colour in a real colour mode is 48bits depth (3x16), the output of 16bits is made later by the software and it is compressed from the original 48bit one. You can't have 48 bits of colour recorded on a card in a 3fr or fff file! The PITA with 54h (as with all sinarbacks)... is the software, ....also, Imacon has a much better (more reliable - better mechanical implementation) piezoelectric mechanism which is totally separate from the rest of the back and has its own heat sinks. On the mirror lock problem you have, I suggest you look at the following...
1. Check if the cable is for Rollei 6008 or for Fuji GX680 (they look like one another) bear in mind that Fuji GX680 can't lock its mirror up permanently which is the behaviour you get on the Rollei. (if it is for Fuji you may sent it to me, I need it!)
2. Check the amount of the delay you've set...
Theodoros, I have no problems with the CF 528 on the Rollei - works perfect. The Rollei does lock the mirror up between shots with the CF 528. It's the Sinar 54H that is not keeping the mirror up during multishot and its not the camera, it's either the Sinar cable I have is incorrect or the Sinar software isn't smart enough.
Probably you are right about the Sinar software, but Hasselblad software isn't perfect either. I find it odd that Phocus can only do the 4 shot and if I want to do 16 shot I have to use their older software Flexcolor. But flex color in many ways is more powerful than Phocus.
Lastly, are you saying that's data depth is the reason why they can't write the multishot sequence to the memory card and let uses take multishot to the card and process later?
-
Theodoros, I have no problems with the CF 528 on the Rollei - works perfect. The Rollei does lock the mirror up between shots with the CF 528. It's the Sinar 54H that is not keeping the mirror up during multishot and its not the camera, it's either the Sinar cable I have is incorrect or the Sinar software isn't smart enough.
Probably you are right about the Sinar software, but Hasselblad software isn't perfect either. I find it odd that Phocus can only do the 4 shot and if I want to do 16 shot I have to use their older software Flexcolor. But flex color in many ways is more powerful than Phocus.
Lastly, are you saying that's data depth is the reason why they can't write the multishot sequence to the memory card and let uses take multishot to the card and process later?
Yes, fff (or 3fr) is a 3x16bit file per shot, which can't be recorded as ONE file on a card... The card should have the ability to record three different files (either in 4x or in 16x mode) and "translate" them into one by itself. To do that, the card should have Flexcolour or Phocus (for 4x only) build in and thus should be able to combine the files... Also, the card should be able to detect possible movement and do all the 16 shots adding, after the 19 (in +mode) or the 21 shots have been shot... This means that the card should be able to do what a computer does. Don't forget that with the original 528c, there was no image recorded in the image bank before the computer would finish the process, this means that the image bank was recording the image after the software has finished the process. :)
-
I don't see the problem in writing the entire sequence of files to the card, then doing the processing when it can be put in a computer. If you have motion then too bad, but I don't normally have this issue.
-
I don't see the problem in writing the entire sequence of files to the card, then doing the processing when it can be put in a computer. If you have motion then too bad, but I don't normally have this issue.
What I am saying is different. I suggest to read it back once more.
EDIT: There is NO sequence of shots (as in Raw material).... there is ONE MS file... in true colour! It's the file that is controlled by the software, not a mathematical "translation" of the sequence... that's why you have the pre-shots. That's why you get (or not) a "motion detection" signal for ONE picture, that's why you have the single shot exposure applying to ALL the sequence, it's ONE file...
-
deleted.
Edmund
-
Theodoros,
Yes, I know well what you are saying. What I am saying is each frame collects a new piece of the data and could also be written to the card instead of being assembled on the fly. It's only data. Yes, it might require a special format, but that is not to say it isn't possible.
And in any case, its not so important to this thread. The 528 is a great back regardless if the multishot is tether only.
-
I believe this is correct. I use a 200MS and the biggest gain is 4 shot mode which looks superb. 6 shot is a compromise and works as you indicate. It does recover more detail than 4 shot, but there is some interpolation going on to get to 200mp. Either way, its a brilliant back capable of astonishing quality.
I am afraid that this is not the case... 16X mode is a matter of mechanical capability not of software! It seems that the piezzo electric crystal that moves the sensor can retain accuracy when moved at 4.5μm but 3.4μm (half of 6.8μm) that is needed for the 39mp sensor is too small to be accurately controlled! I believe that this is why the 200ms is using a new technique to create a 4x50mp=200mp file out of a 50mp sensor in which it is moving the sensor only 6 times rather than 16 times that the 528c is using...
Now, I am not sure how exactly that process works, (perhaps Steven Hendrix could enlighten us here) but it seems that the 200ms does the same with 50MS for the first four shots (i.e. moves the sensor by one pixel thus aligning all the Bayer pattern colours on all pixels) and then it adds two more movements of 1+1/2 pixel length in magnitude (obviously to increase the size of the movement) once horizontally and once more vertically so that it then interpolates the presence of 200 pixels....
I haven't tried the 200MS but I think that the result may be impressive but may also be not up to the "real" 16x mode that the 528c can achieve, since there is no interpolation involved there for the 88mp file to be created... That said, I have to say that one must be very careful when comparing his 16x microstep results from his 528c back with anything else. He has to make sure that he has taken the maximum out of his 528c which is not easy to judge at all since one may consider that "because flexcolor detected no fault - the method was done to its perfection" which is rarely the case... My experience out of every day use of the 528's 16X microstep mode has shown to me that the software has some latitude in forgiving mistakes and that if one seeks for perfection there is simply nothing that compare to the result and that a modern 80mp back will pale in detail comparison even if the file is BW converted!
-
On another site, I posted this quick comparison between a crop from a single shot 50mp file from my H5D-200MS, rezzed up to the equivalent of 200mp, and a similar crop from a 200mp file. Please forgive the dull subject matter.
First, the crop from the single shot capture that had been interpolated to 200mp
(https://dl.dropboxusercontent.com/u/9806585/July_2014_onwards/H5D-200MS%20single%20shot.jpg)
Next, the crop from the 200mp file
(https://dl.dropboxusercontent.com/u/9806585/July_2014_onwards/H5D-200MS%20multi%20shot.jpg)
-
I believe this is correct. I use a 200MS and the biggest gain is 4 shot mode which looks superb. 6 shot is a compromise and works as you indicate. It does recover more detail than 4 shot, but there is some interpolation going on to get to 200mp. Either way, its a brilliant back capable of astonishing quality.
Hi Quentin. The 2 more movements happening with 200MS do cover the positions between pixels so that there is real 200mp resolution recorded. The thing is that on those two extra positions, there is no aligning of all the different RGB information happening, so (I believe) the colour is interpolated by the information of the overlapped pixels that has been the result from the 4x action that has pre occurred. I don't see any other way that this could work...
-
On another site, I posted this quick comparison between a crop from a single shot 50mp file from my H5D-200MS, rezzed up to the equivalent of 200mp, and a similar crop from a 200mp file. Please forgive the dull subject matter.
First, the crop from the single shot capture that had been interpolated to 200mp
(https://dl.dropboxusercontent.com/u/9806585/July_2014_onwards/H5D-200MS%20single%20shot.jpg)
Next, the crop from the 200mp file
(https://dl.dropboxusercontent.com/u/9806585/July_2014_onwards/H5D-200MS%20multi%20shot.jpg)
I believe you mean up-sampled (instead of interpolated) to 200mps for the single shot 50mp file...
MO is that there should be another shot at 4x for the comparison and that the 200mp should be down sampled to 50mp instead of the others been up sampled... In fact I would really appreciate it if you may post such a comparison... Never the less, since the (huge) resolution difference is to be expected, what I find most impressive is the clear and undeniable difference on the DR recorded on the two files (which may be of 2 stops - same as I notice with my 528c)... I would expect the DR to be the same with the 4x file as with the 6x file... but I would like to see the resolution comparison with the 200mp file down sampled to the size of the others... If you have the 120 macro lens, you may decide to use that... Regards, Theodoros.
-
I believe you mean up-sampled (instead of interpolated) to 200mps for the single shot 50mp file...
MO is that there should be another shot at 4x for the comparison and that the 200mp should be down sampled to 50mp instead of the others been up sampled... Regards, Theodoros.
Hi Thoedorus
No, I meant interpolated. Also, given the whole point of a larger file is more detail / ability to crop, it makes no sense to me to downsize a full resolution 200MS shot. In my view, the point of the exercise is to see if we really need those big files, or can we get away with interpolating the data in the smaller file to the larger file size?
So at your request, here is the test you asked for ;D
I have taken three identical shots using single shot, 4 shot and full 6 shot capture, and beloew I will post the links to the files for samples taken from each shot.
The lens used was the HC-50 II, tripod mounted (of course). There are some lighting differences because these images were taken using natural light. A 0.6 ND grad was used right side side to balance the light, and some shadow / highlight correction used in Photoshop to bring up the shadows slightly.
The single shot capture and the 4 shot multi-shot were each interpolated to the same size as the 6-shot, 200mp image, using Photozoom pro. No sharpening was applied. I have only linked to the samples, because of image file size restrictions on this forum
Here is the main shot
(https://dl.dropboxusercontent.com/u/9806585/July_2014_onwards/H5D-200MS%20Outbuilding.jpg)
Next, a link to the Single Shot Capture sample:
https://dl.dropboxusercontent.com/u/9806585/July_2014_onwards/H5D-200MS%20single%20shot%201.jpg (https://dl.dropboxusercontent.com/u/9806585/July_2014_onwards/H5D-200MS%20single%20shot%201.jpg)
Then the 4-Shot capture:
https://dl.dropboxusercontent.com/u/9806585/July_2014_onwards/H5D-200MS%204%20shot%201.jpg (https://dl.dropboxusercontent.com/u/9806585/July_2014_onwards/H5D-200MS%204%20shot%201.jpg)
Finally, the big banana, the 6 shot capture
https://dl.dropboxusercontent.com/u/9806585/July_2014_onwards/H5D-200MS%20Six%20shot.jpg (https://dl.dropboxusercontent.com/u/9806585/July_2014_onwards/H5D-200MS%20Six%20shot.jpg)
-
Hi Thoedorus
No, I meant interpolated. Also, given the whole point of a larger file is more detail / ability to crop, it makes no sense to me to downsize a full resolution 200MS shot. In my view, the point of the exercise is to see if we really need those big files, or can we get away with interpolating the data in the smaller file to the larger file size?
So at your request, here is the test you asked for ;D
I have taken three identical shots using single shot, 4 shot and full 6 shot capture, and beloew I will post the links to the files for samples taken from each shot.
The lens used was the HC-50 II, tripod mounted (of course). There are some lighting differences because these images were taken using natural light. A 0.6 ND grad was used right side side to balance the light, and some shadow / highlight correction used in Photoshop to bring up the shadows slightly.
The single shot capture and the 4 shot multi-shot were each interpolated to the same size as the 6-shot, 200mp image, using Photozoom pro. No sharpening was applied. I have only linked to the samples, because of image file size restrictions on this forum
Here is the main shot
(https://dl.dropboxusercontent.com/u/9806585/July_2014_onwards/H5D-200MS%20Outbuilding.jpg)
Next, a link to the Single Shot Capture sample:
https://dl.dropboxusercontent.com/u/9806585/July_2014_onwards/H5D-200MS%20single%20shot%201.jpg (https://dl.dropboxusercontent.com/u/9806585/July_2014_onwards/H5D-200MS%20single%20shot%201.jpg)
Then the 4-Shot capture:
https://dl.dropboxusercontent.com/u/9806585/July_2014_onwards/H5D-200MS%204%20shot%201.jpg (https://dl.dropboxusercontent.com/u/9806585/July_2014_onwards/H5D-200MS%204%20shot%201.jpg)
Finally, the big banana, the 6 shot capture
https://dl.dropboxusercontent.com/u/9806585/July_2014_onwards/H5D-200MS%20Six%20shot.jpg (https://dl.dropboxusercontent.com/u/9806585/July_2014_onwards/H5D-200MS%20Six%20shot.jpg)
Thanks for the post Quentin! Still... sampling a file involves no interpolation! Never the less, if one is to compare two files of different resolution, it's always best to down sample the larger file than up sample the lower rez file, since there is no additional information to be added with the lower resolution.
EDIT:Up-sampling is a tricky process during which it is easy for artefacts to be induced (since information that hasn't been captured is asked to be added in the file).
-
By definition interpolation involves guesswork. The definition of interpolation includes guessing or estimating intermediate values, which is what is happening in software when an image is uspsized, with varying degrees of accuracy, because data is being added to based upon software assumptions. I used Photozoom Pro S-Spline Max with sharpening turned off the for the 50mp shots, but other algorithms such as Lanczos sometimes work better.
I really do not see any point at all in reducing the size of an image to compare it with a lower resolution version. The point of a bigger file is... its bigger! The question I want answered is how well the smaller files compare when their sized is increased to match the bigger file. Some do better than others. For example, the single shot 50mp capture is riddled with moire once you look closely, as we have become used to with mosaic or bayer type sensors. Frankly if this was not photography, but something (even) more important, we'd be howling in outrage at such obvious flaws. This becomes more obvious when its size is scaled up to match the larger file, quite apart from other differences. Fact is, a 40-80mp single shot file will do very nicely for most purposes - right up to the point you put it side by side with a multishot file ::)
And this, of course, is why Foveon sensors have a cult following, despite their limitations.
-
By definition interpolation involves guesswork. The definition of interpolation includes guessing or estimating intermediate values, which is what is happening in software when an image is uspsized, with varying degrees of accuracy, because data is being added to based upon software assumptions. I used Photozoom Pro S-Spline Max with sharpening turned off the for the 50mp shots, but other algorithms such as Lanczos sometimes work better.
I really do not see any point at all in reducing the size of an image to compare it with a lower resolution version. The point of a bigger file is... its bigger! The question I want answered is how well the smaller files compare when their sized is increased to match the bigger file. Some do better than others. For example, the single shot 50mp capture is riddled with moire once you look closely, as we have become used to with mosaic or bayer type sensors. Frankly if this was not photography, but something (even) more important, we'd be howling in outrage at such obvious flaws. This becomes more obvious when its size is scaled up to match the larger file, quite apart from other differences. Fact is, a 40-80mp single shot file will do very nicely for most purposes - right up to the point you put it side by side with a multishot file ::)
And this, of course, is why Foveon sensors have a cult following, despite their limitations.
Yes Quentin, interpolation involves guesswork, but it is a process that the guesswork is based on some origin specifically designed for the process to work (like with interpolated colour and bayer pattern). Sampling is different, the maker of the sensor never specified the pixels to be up sampled... so there is no interpolation involved there.
OTOH, a 50mp file can be printed (if one knows what he is doing) as large as 6x8feet which is huge by any standard, so there is no need to up sample it for comparison reasons on a monitor. The artefacts you mention on the single shot file are obviously because the file's colour is interpolated which of course involves mistakes, these mistakes are obviously multiplied when up sampled. The true colour file in contradiction up samples particularly well (its surprising the little difference between the up-sampled 4x file with respect to the 6x one) because there is no interpolation involved in the original. I believe that if pixels where bigger, both the 50mp files would up-sample even better. It leads me to suspect that the 16x file of the 528c which is 88mp, but is created from 9μm size pixels (which is more than twice the size of the pixel "entrance" area) would end up being better than the 6x result, even if up sampled!