One (biased) Opinion for MORE MORE MORE IQ

[Doug Peterson]

First let me frank and clear up front: I work for a dealer so sure read this with a grain of salt. However, those who know me I do a pretty good job of not drinking the kool-aid. Cynics might see this post as entirely self serving, but I promise you that I'm just using the only outlet I know of to discuss something that has been bubbling up in my mind for the last year.

I started off shooting point and shoot digitals. Then 35mm film. Then MF film. Then MF digital.

This forum has been exhausted with plenty of good arguments against the need for more resolution or DR
1) Most compelling images aren't compelling because of resolution but rather lighting, subject, emotion, composition etc etc etc etc. e.g. A Holga t-max shot becomes less beautiful if you add sharpness.
2) Most magazine use can be accomplished with far less than current max resolution. Depending on all the mitigating factors (lens quality, crop, lighting) and who you ask a 2-page spread might require 4, 12, or 22 megapixels for images that benefit from sharpenss, but certainly not 60 megapixels (the current ceiling on single-shot resolution).
3) For most other commercial applications the current high-end of resolution (32-60 megapixels) is also considered adequate.

But in my job I can pull any back/body/lens off the shelf to shoot (provided it's not being otherwise used. Recently I took a large case full of gear to Moab for GetDPI.com's landscape workshop. In the kit were technical bodies, MF bodies, the best lenses of each system, and P45+s and P65+s (39 and 60 megapixels respectively) and it solidified what I've begun to feel for the last few months: if you remove cost I will always use the maximum resolution possible. In fact 60 megapixels was not enough for me; every shot I liked I shot two or four image stitches on the tech camera (moving the back within the static image circle rather than rotating the camera/lens) to achieve 100 megapixels or more.

Large Prints:
To get a print which is twice as big visually (2x each dimension) you need a file which is four times as large. So resolution shows up in prints much earlier than most photographers think. There is a strong trend away from printing photographers (in favor of sharing albums online or on devices like iPhones) and I'm a solid part of this trend. The only prints I make anymore are at least 20x30 and, for images where sharpness is part of the intention, I want them to be wicked sharp.

Secondary Use:
It's impossible to know where/how your best few images each year will be used. You may have shot them for a magazine or for the web, but if the file is huge and clean you can use it later for anything.

Raised Expectations:
I'm young so I need to look at the next several decades. At some point in the future point-and-shoots will produce the same IQ as the P65+. At that point any file in my library which was shot on a mid-range dSLR will be limited from a re-use point of view (again, for images that benefit from resolution, DR etc).

I guess the bottom line of this rant is that I'm 100% against the arguments that say we've reached a plateau above which photographers will only want more IQ for very specialized applications. I think that is short sighted. Give me a 300 megapixel camera and, for subject matter which allows it, I will still ask if I can stitch on the back of a tech camera. Give me 20 stops of dynamic range. Heck give me 100 stops of dynamic range so that I can shoot any scene at any aperture and shutter speed. Give me a raw format aimed specifically for still image quality that breaks the 1/100th sec or 1 hour long exposure into a millions slices in time so I can select in post which fraction of a moment in time to create my 10 foot print from. While I'm dreaming, virtualize the entire world so that I can virtually tour the earth and and place my virtual camera anywhere at any place. Lower the barrier of entry to nadda and let the art battles begin!

Did I veer of course there at the end?

*going to get coffee*

Doug Peterson,  Head of Technical Services
Capture Integration, Phase One & Canon Dealer  |  Personal Portfolio

[Guy Mancuso]

Ah hell your a pig like the rest of us. Big enough will never happen but in the meantime we can be happy with what we have or maybe better said there is always that little devil whispering in your ear , are you sure you have enough. LOL

Heck I am pretty happy but I will fully admit , I am a pig too. I would love to have more, but right now money seems to be my target. Great to dream though and if you stop doing that than you know there is 6 feet of dirt sitting on top of you.

[russell a]

[quote name='dougpetersonci' date='Feb 19 2009, 05:12 PM' post='261342'

*going to get coffee*

[/quote]

Doug:  Doesn't sound to me as if you have a caffeine deficiency.  

[bradleygibson]

Doug,

I think you are right on the mark here...  I feel exactly the same way.

After 4 years of contemplating the switch to medium format, and now approaching 2 years of having been here, I agree, the 39 and 60-megapixel offerings are a "nice start".  I am very excited to see if RED can ship their 261 megapixel 617--that is a super-appealing product from my standpoint.

I also favour large prints--I have work hanging at 4 feet by 12 feet, and while it does look great, the 160 megapixels (it is a stitch) are at their max.

I'm not saying I have a need to print even larger, but I would love to see what the same print would look like at 1/2-3/4 of a gigapixel.  I'm sure it would improve noticably as well.

Of course there are at least two practial limits to all of this (cost, which we've removed for the sake of discusson) and technology.  Photoshop CS3 really labours under 160 megapixels (particularly with a non-destructive editing workflow--smart objects, smart filters and the like).  So when it takes me 45 minutes to run a filter on my image, it becomes too expensive time-wise to work very many of these images.

Digital photography pioneer Steven Johnson has also run into this limitation -- he would often take photographs knowing he wouldn't even be able to open them for years until the technology caught up (back in the Photoshop 5-ish days, if memory serves).  I feel like ultra-high resolution capture is in this same boat to some degree.  (64-bit Photoshop, and i7 Xeon's with the capability to handle 128GB RAM are very welcome innovations in my book).

In the end all the tools are about delivering emotional impact.  And in addition to the artistic content, the sheer presence and clarity of these prints helps to deliver that impact.

So I'm in agreement--folks who don't need more resolution are probably right.  And those that do are as well.  

-Brad

[Ben Rubinstein]

Seems like the OP makes (possible) sense to rich amatuers but absolutely no sense to any business. How can any business plan justify more than is needed or can possibly be charged to clients just to fulfil some inner need for the very best possible however unnecessary?

[BernardLanguillier]

But in my job I can pull any back/body/lens off the shelf to shoot (provided it's not being otherwise used. Recently I took a large case full of gear to Moab for GetDPI.com's landscape workshop. In the kit were technical bodies, MF bodies, the best lenses of each system, and P45+s and P65+s (39 and 60 megapixels respectively) and it solidified what I've begun to feel for the last few months: if you remove cost I will always use the maximum resolution possible. In fact 60 megapixels was not enough for me; every shot I liked I shot two or four image stitches on the tech camera (moving the back within the static image circle rather than rotating the camera/lens) to achieve 100 megapixels or more.

What prevents you from stitching many more frames with a high resolution DSLR?

The following image is a 250MP pano with infinite DoF...



Cheers,
Bernard

[Joe Behar]

What prevents you from stitching many more frames with a high resolution DSLR?

The following image is a 250MP pano with infinite DoF...


Cheers,
Bernard

Time...

When you're shooting a sunset at its last stages you may have as little as  30 seconds before the scene changes completely. I'm with Doug on this, I'd love to have 1 frame per second of 200 megapixels each with a dynamic range of 50 stops.

You can always dumb it down if you really want to, but I suspect no one will.

WARNING FOLLOWS: I'm in the same situation as Doug, I have my choice of virtually any toy I care to play with if its not being used by our rental department or at a sales demo.

[hubell]

To me, the constant quest for extra resolution is misguided. I need more depth of field than I can get shooting at f/11 or even f/16, not more resolution. If parts of my image are not in good focus because of insufficient depth of field, all the resolution in the world will not overcome it. I wish that Phase would devote its R&D to that. (Rebadging Hartblei stuff and jacking the price by 4x doesn't count.) If the Hasselblad Tilt/Shift device works really well, and the proof will soon be in the pudding, that will be functionally way more valuable to me with 39mp than moving to 60mp, or 80mp or.....

[Guy Mancuso]

Nothing is going to help your DOF except a T/S OR and i have been using it with great success Helicon Focus. http://forum.getdpi.com/forum/showthread.php?t=5306

For the money a darn nice solution for a lot of work

[hubell]

Nothing is going to help your DOF except a T/S OR and i have been using it with great success Helicon Focus. http://forum.getdpi.com/forum/showthread.php?t=5306

For the money a darn nice solution for a lot of work

Yes, I know. I bought Helicon Focus over a year ago and I have gotten some very good results with it, but it has its limitations, like images with foliage, which I shoot regularly. The focus blending in CS4 is a bad joke. A view camera with tilt movements is not an option for me. The only other option that's on the table is the forthcoming Hasselblad T/S device. We will see how it performs in the real world in the hands of photographers with no axe to grind. (Is that an oxymoron today?) I am particularly curious about how hard it is to do critical focus with it through the viewfinder. I wonder if Hasselblad has given any thought to the need for some sort of supplemental magnifying device to use for focusing.

[Guy Mancuso]

It could be tough to focus but I agree like to see how it really works out for the Hassy. Just no real wide here is the bummer

[DesW]

Doug,

After 4 years of contemplating the switch to medium format, and now approaching 2 years of having been here, I agree, the 39 and 60-megapixel offerings are a "nice start".  I am very excited to see if RED can ship their 261 megapixel 617--that is a super-appealing product from my standpoint.

-Brad

Hi,
 Yes I  believe the Mega RED 617  is well on the way to be a viable product we have a slight in to it's production and have our order in at the top of the queue !

Frankly  in RED's case it is absolute pleasure to deal with a company that refrains from  the  MFDB Dealer/ Importer/Retailer/sales shacking/etc/etc that has taken hold of these forums lately.

Good Shooting

DesW

[BernardLanguillier]

Time...

When you're shooting a sunset at its last stages you may have as little as  30 seconds before the scene changes completely. I'm with Doug on this, I'd love to have 1 frame per second of 200 megapixels each with a dynamic range of 50 stops.

I am denying the fact that higher resolution sensors have value, but considering the huge price difference vs the quality gap once stitching is part of the equation, the value proposition of backs is even worse than when a single frame only is considered.

About sunsets or low light shooting in general... my view is that this is in fact where the DSLR has the largest advantage over the backs if stitching is involved on both sides...  :-)

Here is why (this is coming from extended real work experience with both solutions...). What matters then is how long an exposue you need... and DSLR have a 3 fold advantage on this:

1. Their high ISO is much cleaner, you can shoot at ISO 400 with a D3x and pretty much get away with it,
2. They don't need dark frame substration up to 8 sec (for the D3x at least) and image quality is not affected at all up to 8 sec (I seriously doubt this is the case with a P65+),
3. You don't need to stop down as much to get the same depth of field... a 2 sec frame at f11 with the D3x will have more DoF than a 4 sec f16 frame with a P65+ (and that one becomes 8 sec on the P65+ because of dark frame substraction - see point 2 above),

With all of time combined, I could shoot at ISO400, 1 sec, f11 on the D3x, vs ISO200, 4 sec * 2 = 8 sec on the P65+, all other things being mostly equal.

Staying conservative (taking into accout pano head operation adding one second between frames), all in all, you can shoot at least 4 times more frames with a D3x than with a P65+ in the same amount of time, all other things being mostly equal, but you will have a lot less time gap between successive frames in the sky with the DSLR... and that is what really matters.

Both the 6 frames DSLR and the 2 frames MFDB pano will be shootable in 15 seconds or less, but there will be only 2 sec gap between each segment of the sky with the DSLR, while there will be 8 sec gap between left and right with the P65+...

So what we have in fact is the very opposite of what you are writing above, if stitching is done with the back of course. It is true that a single P65+ frame is easier to handle... but you probably still have problems with lower image quality at longer exposures.

You can of course say that a 200MP sensor would be great, but considering the current curves, the image quality of DSLR's sensor is improving much faster than that of MFDB (think nvidia vs Silicon Graphics in the high end graphic card market). So the value of stitching with a DSLR vs single captures/Stitching with a back is going to only increase in the coming years.

Cheers,
Bernard

[Joe Behar]

I am denying the fact that higher resolution sensors have value, but considering the huge price difference vs the quality gap once stitching is part of the equation, the value proposition of backs is even worse than when a single frame only is considered.

About sunsets or low light shooting in general... my view is that this is in fact where the DSLR has the largest advantage over the backs if stitching is involved on both sides...  :-)

Here is why (this is coming from extended real work experience with both solutions...). What matters then is how long an exposue you need... and DSLR have a 3 fold advantage on this:

1. Their high ISO is much cleaner, you can shoot at ISO 400 with a D3x and pretty much get away with it,
2. They don't need dark frame substration up to 8 sec (for the D3x at least) and image quality is not affected at all up to 8 sec (I seriously doubt this is the case with a P65+),
3. You don't need to stop down as much to get the same depth of field... a 2 sec frame at f11 with the D3x will have more DoF than a 4 sec f16 frame with a P65+ (and that one becomes 8 sec on the P65+ because of dark frame substraction - see point 2 above),

With all of time combined, I could shoot at ISO400, 1 sec, f11 on the D3x, vs ISO200, 4 sec * 2 = 8 sec on the P65+, all other things being mostly equal.

Staying conservative (taking into accout pano head operation adding one second between frames), all in all, you can shoot at least 4 times more frames with a D3x than with a P65+ in the same amount of time, all other things being mostly equal, but you will have a lot less time gap between successive frames in the sky with the DSLR... and that is what really matters.

Both the 6 frames DSLR and the 2 frames MFDB pano will be shootable in 15 seconds or less, but there will be only 2 sec gap between each segment of the sky with the DSLR, while there will be 8 sec gap between left and right with the P65+...

So what we have in fact is the very opposite of what you are writing above, if stitching is done with the back of course. It is true that a single P65+ frame is easier to handle... but you probably still have problems with lower image quality at longer exposures.

You can of course say that a 200MP sensor would be great, but considering the current curves, the image quality of DSLR's sensor is improving much faster than that of MFDB (think nvidia vs Silicon Graphics in the high end graphic card market). So the value of stitching with a DSLR vs single captures/Stitching with a back is going to only increase in the coming years.

Cheers,
Bernard

Bernard, You are correct in your math. I believe the original intent of the post was to remind people that if we say we have enough resolution, dynamic range, etc we won't be encouraging progress. When the Phase One H20 came out we though we had all we would ever need. The H25 was better, the P25 even better, the P45 even more so and now the P65+ is beyond all those. At every step of the way, a lot of people said it was too much, we don't need all this. Today most people won't consider less than 12-16 megapixels in a camera, and the trend is to 21+ MP. The reality is we are still far away from the ideal picture making machine. We've surpassed all previous quality marks but we're still very far from what the eye sees, especially in dynamic range.

I know, printers need to catch up as well, but the fact still remains that if don't at least verbalize our wish lists, how will we get there.

Wouldn't you like to have your 200 megapixels, 20 stop dynamic range and ISO 1,000,000 all in one package and one exposure?

I don't think we're talking about value for money here, just a wish list.

[BJL]

Let us look at what it really takes to double linear resolution of the main subject.
First, doubling sensor resolution.
Second, doubling lens resolution: hard to say how hard that will be when over 100MP is desired.
Third, keeping diffraction spot size small enough: the minimum acceptable aperture diameter is doubled. This sounds like accepting half the maximum DOF, but in fact it is worse due to the greater degree of enlargement making OOF effects more noticeable.
Fourth, to keep the same DOF, meaning keeping all the same stuff looking in focus despite doubling of print size, requires halving of the maximum acceptable aperture diameter.

So the range of aperture choices between "too large, causing things I want in focus to be visibly OOF" and "too small, leading to a visible loss of sharpness due to diffraction" is reduced by two stops for each doubling of linear resolution. In other words, by one stop for each doubling of pixel count, if you wish to get the full resolution benefit of the extra pixels. At some point very few subjects will benefit.

[Guy Mancuso]

Seriously after shooting the P65 several times now with tests on the Mamiya 80 , 150 D glass and even the 28d we still have yet in my mind surpassed the lens with that back. I think we still have some room to go before we do. I would seriously like to move up to the P45 plus from the 25 plus and I will when things settle out some with business but just being in Mf even on the lower end side with a 22 mpx back is such a improvement to me. I think we all would like to improve what we have be it mpx, features , ergonomics and glass and than some. Not sure those desires or needs will ever go away.

[yaya]

I am denying the fact that higher resolution sensors have value, but considering the huge price difference vs the quality gap once stitching is part of the equation, the value proposition of backs is even worse than when a single frame only is considered.

About sunsets or low light shooting in general... my view is that this is in fact where the DSLR has the largest advantage over the backs if stitching is involved on both sides...  :-)

Here is why (this is coming from extended real work experience with both solutions...). What matters then is how long an exposue you need... and DSLR have a 3 fold advantage on this:

1. Their high ISO is much cleaner, you can shoot at ISO 400 with a D3x and pretty much get away with it,
2. They don't need dark frame substration up to 8 sec (for the D3x at least) and image quality is not affected at all up to 8 sec (I seriously doubt this is the case with a P65+),
3. You don't need to stop down as much to get the same depth of field... a 2 sec frame at f11 with the D3x will have more DoF than a 4 sec f16 frame with a P65+ (and that one becomes 8 sec on the P65+ because of dark frame substraction - see point 2 above),

With all of time combined, I could shoot at ISO400, 1 sec, f11 on the D3x, vs ISO200, 4 sec * 2 = 8 sec on the P65+, all other things being mostly equal.

Staying conservative (taking into accout pano head operation adding one second between frames), all in all, you can shoot at least 4 times more frames with a D3x than with a P65+ in the same amount of time, all other things being mostly equal, but you will have a lot less time gap between successive frames in the sky with the DSLR... and that is what really matters.

Both the 6 frames DSLR and the 2 frames MFDB pano will be shootable in 15 seconds or less, but there will be only 2 sec gap between each segment of the sky with the DSLR, while there will be 8 sec gap between left and right with the P65+...

So what we have in fact is the very opposite of what you are writing above, if stitching is done with the back of course. It is true that a single P65+ frame is easier to handle... but you probably still have problems with lower image quality at longer exposures.

You can of course say that a 200MP sensor would be great, but considering the current curves, the image quality of DSLR's sensor is improving much faster than that of MFDB (think nvidia vs Silicon Graphics in the high end graphic card market). So the value of stitching with a DSLR vs single captures/Stitching with a back is going to only increase in the coming years.

Cheers,
Bernard

Hmmm...

This really calls for a test (which I am going to do soon enough)

6 images Vs 2 images means 7 stitches Vs 1 stitch. Let's ignore the time factor and just look at the possible errors/ scene changes, knocking the tripod/ mishandling the pano head. That's 6 beautiful frames going straight into the dustbin.

Now let's take the DB and mount it on a pancake camera and leave the pano head at home. With a lens such as let's say 35mm Digitar you will be shifting 15mm to the right, Click, 15mm to the Left, Click, done. Composition is done is ONE ACCURATE GO versus "let's guestimate where my actual frame is" and a single-stitch 300MB+ file is 2 minutes away in CS4. Straight lines, no CA, no sharpness falloff and massive amount of detail. Also at f11 with the above lens you will get more or less infinite DOF.

This is not to say that there are no situations where a good DSLR will give the best result, but your scenario above I'm afraid is not one of these.

See if you can get hold of Richard Bryant's book or even better if you're in London before June go and see the exhibition at Somerset House. Those 6 metre Lambda prints will knock your socks off. Looking at the raw files and/ or the 16-bit stitched files will cause a real damage to your bank account.

all IMHO

Yair

[BernardLanguillier]

Wouldn't you like to have your 200 megapixels, 20 stop dynamic range and ISO 1,000,000 all in one package and one exposure?

I don't think we're talking about value for money here, just a wish list.

Joe,

Sure, I far from being against progress!  As I mentioned above, I am not saying that stitching is an answer for all needs, nor that it doesn't have any drawback compared to a single high resolution frame. There is room for P65+ like devices and I would probably buy one myself if I could justify the cost.

I just want to keep things in perspective, there are many things we ca ndo today that were simply impossible just 2 years ago.

Cheers,
Bernard

[BernardLanguillier]

This really calls for a test (which I am going to do soon enough)

6 images Vs 2 images means 7 stitches Vs 1 stitch. Let's ignore the time factor and just look at the possible errors/ scene changes, knocking the tripod/ mishandling the pano head. That's 6 beautiful frames going straight into the dustbin.

Yes, you need reasonnable skills. Shooting 4x5 was not super easy either, was it?

I shoot quite a lot, and mostly shoot only panos. The truth be told, I have a success rate in excess of 99%. I do in fact not remember a single pano I tried to shoot that didn't end up as a totally seamless image in the last 12 months. And I am talking about litterally thousands of them. I don't consider myself as a particularly skilled person.

Granted, there are some scenes I am not trying to address because I know the result will not be good. For these cases a single high resolution frame is clearly superior.

Now let's take the DB and mount it on a pancake camera and leave the pano head at home. With a lens such as let's say 35mm Digitar you will be shifting 15mm to the right, Click, 15mm to the Left, Click, done. Composition is done is ONE ACCURATE GO versus "let's guestimate where my actual frame is" and a single-stitch 300MB+ file is 2 minutes away in CS4. Straight lines, no CA, no sharpness falloff and massive amount of detail. Also at f11 with the above lens you will get more or less infinite DOF.

Yes on all of your points but the infinite DoF. I have shot a Mamiya 28mm for long enough to know that f11 is far from ensuring adequate sharpness along the Z axis if that is what you trying to achieve. Even if it were, the DSLR would still feature more DoF at f8. You cannot wi against smaller sensors when taking DoF into account.

This is not to say that there are no situations where a good DSLR will give the best result, but your scenario above I'm afraid is not one of these.

See if you can get hold of Richard Bryant's book or even better if you're in London before June go and see the exhibition at Somerset House. Those 6 metre Lambda prints will knock your socks off. Looking at the raw files and/ or the 16-bit stitched files will cause a real damage to your bank account.

I will unfortunately probably not have the chance to attend, but I am sure that this is remarkable work.

Cheers,
Bernard

[PeterA]

Yair

Why will looking at the raw files cause real damage to one's bank account? - serious question -
ie what did you mean by this?