Pages: 1 [2]   Go Down

Author Topic: Waiting for Moore's Law to plateau  (Read 16382 times)

Ellis Vener

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2151
    • http://www.ellisvener.com
Re: Waiting for Moore's Law to plateau
« Reply #20 on: June 12, 2012, 03:15:28 pm »

It seems that the Canon 5DIII is not a major upgrade regarding DR and noise characteristics. It seems that the Canon 5DIII is not a major upgrade regarding DR and noise characteristics. It is a great upgrade regarding shooting performance.

I've never used the 5D Mark II but have shot tens of thousands of frames with the 1Ds Mark III.  assumign the 5D Mark II and 1Ds Mark III are essententially the same in dynamic range and signal to noise characteristics below ISO 800 you are correct; Above that the improvements are significant. And as you point out "It is a great upgrade regarding shooting performance." My evaluation is that the 5D Mark III replaces the 1Ds Mark III as the benchmark in Canon DSLR performance. (The 1D X is primarily designed for shooting sports and journalism and so it has a different set of benchmarks to hit.

 I've also shot a lot with the D3X and D800 and D800E and the differences are significantly in favor of the D800 series.
Logged

Wayne Fox

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4237
    • waynefox.com
Re: Waiting for Moore's Law to plateau
« Reply #21 on: June 13, 2012, 02:52:39 am »

Hi,

It seems that the Canon 5DIII is not a major upgrade regarding DR and noise characteristics. It is a great upgrade regarding shooting performance.  So if image quality is all you care about just keep the 5DII you have and be happy.

Nikon's 800D is a major upgrade over the D800, but perhaps less so over the D3X.

Best regards
Erik


W
The 5d mark 3 is more of an upgrade than people are giving it credit for.  As you pointed out the real difference is the jump from the d700 to the d800 is huge.  I've never understood why NIkon didn't put the sony 24mp sensor in the d700 body a long time ago.  Celebration time for Nikon users ... and I myself now own my first Nikon ever.

You mentioned noise as not being a factor, I guess it depends on your parameters.  At base ISO, the 5D2 is pretty clean.  No increased dynamic range so deep shadow noise is about the same if you try to pull them up, but at higher ISO's the noise characteristics appear to be pretty significant.  I haven't done my own comparisons, just seen some stuff around the web such as this.  (scroll to the bottom to see the side by sides).

We've sold as many 5D Mark 3's as we have d800's, but part of that is how hard the d800's are to come by, and no one is complaining. One of my closest friends bought a 3 early on, he's an outstanding wedding/portrait photographer, and he's told me he feels he's getting much better files, and loves the higher ISO for available light at weddings.  His landscape stuff is pretty nice too .. he's just getting back for a week in southern Utah.  I'll be very interested to play with a few of his files from the slot canyons down there.

Canon isn't finished, but the rumors of a 40+mp sensor aren't super intriguing, since Canon still hasn't figured out how to pull the dynamic range out of their sensors yet.  Sony has leaped well ahead of Canon on this.  To me the increased dynamic range of the d800 is the real story ... it is extremely hard to find good enough glass to put in front of that 36mp sensor, but even with lesser glass, the files are still amazing.

I've always felt that we need to oversample the data to get good results.  I would rather see a large image printed from a camera with a higher resolution sensor, even with a weak lens because it still looks better than one printed from a substantially lower resolution sensor and the uprezzed too far.  the weak data of one to me is still superior to the made up fictitious data of the other.  That's what I"m seeing the my d800e right now, even with files from my 24-70 which isn't up to the task of resolving to that sensor.
Logged

hjulenissen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2051
Re: Waiting for Moore's Law to plateau
« Reply #22 on: June 13, 2012, 05:17:20 am »

...the weak data of one to me is still superior to the made up fictitious data of the other...
Interpolation is not the same as "inventing" data. Any measurement system will obtain measurements that 1)measure what we want, and 2)measure something that we don't want (e.g. internal noise, spatial averaging,...). Interpolation _can_ reduce 2). I.e. the interpolation brings the signal "closer" (mathematically or perceptually) to the phenomena that we wanted to measure.

Rather than object to the (potentially misguided) philosophical sides of the measurement/reproduction, I would focus on the observed total performance. If having to delve into the nitty-gritty stuff, I would claim that the reason that the D800 may (or may not) produce more detailed large prints than the 5Dmk3 is not primarily due to the demosaic/interpolation/scaling one applies to the 5Dmk3 images, but simply the fact that the D800 seems to be better able to measure image detail.

I am sure that the 5Dmk3 have many other qualities that ensure that many of its owners will be able to shot many good images using it.

-h
Logged

Wayne Fox

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4237
    • waynefox.com
Re: Waiting for Moore's Law to plateau
« Reply #23 on: June 13, 2012, 11:28:37 pm »

Interpolation is not the same as "inventing" data.
call it what you want, but at some point it reaches a limit that becomes visually apparent. 

I'm not a typical pixel peeper. Sure I look at files and examine things, but to me it's all in the prints.  At my store we print photographs from all kinds of photographers and all kinds of cameras every day.  I've seen people try to get a 100" print out of a point and shoot, and I have some very talented photographers stuck with smaller resolution sensors that push their size far beyond what they can really handle.   Sure you can also step farther away, but this to me has always been sort of an "excuse" for printing very large images from too small of files.  There are many locations where images are displayed that might even restrict viewers to closer distances.  I know many claim only photographers stick their nose in prints, but I've stood around watching people in galleries that have images worthy of sticking your nose in and there are certainly many that do so ... probably because it actually has something to offer.  this isn't like normal art where most of the time their isn't an expectation of detail.  But even in the art world there are disciplines and styles where intricate and amazing detail is critical, and art critics don't stand back and say isn't that nice, they walk up and say wow.

When printing a file a 200% uprez seems to be where things begin to be noticeable.  This is certainly not an absolute, it depends a lot of the file and subject matter ... some images just don't have any expectation of micro detail, others fall apart even sooner. But generally speaking this means at some point around 200% uprez you will begin to see issues and the further you go the more will become apparent. 

I guess the point I was trying to my make in my post was if I use the same lens on two different sensors, and even if that lens isn't the greatest, the limits still seem to fall in place.  Yes the image will be "softer" than if I had used a higher quality lens, but that softness is far different visually than what I get if I uprez a lower sensor image to the same size.

Over sampling of data is a pretty tried and true technique to achieve what we expect as realistic, and I think the same principal applies to capturing the data.  I fully expect files from my d800 to be extremely good at 40", and I certainly most I can push further than that.  And even if the lens isn't the greatest, I don't think you'll get the same look by under sampling it with a lower rez camera then over rezzing it it back up.

The true limit to the system should be the lens, not the sensor.  If can sample enough data, then I can get as good as it can be.  Many thought we were getting there with the 1DsMark2 all those years ago, perhaps we're getting there now. 
Logged

hjulenissen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2051
Re: Waiting for Moore's Law to plateau
« Reply #24 on: June 14, 2012, 03:59:36 am »

The true limit to the system should be the lens, not the sensor.  If can sample enough data, then I can get as good as it can be.  
I see your point, but I disagree. The true limit to the system should be the photographer. Or the price. Or the weight.

A sensible system should be balanced. Increase the quality of each component until the total system reach the point of limited returns (total cost increases without tangible total quality increases).

Set your aperture to f/32, and your image detail will mostly be lense-limited, and any modern sensor will be wastly oversampling the blurry detail. May not produce any good images, though :-) (depending, of course, on what you are shooting and what you want it to look like)
Quote
Over sampling of data is a pretty tried and true technique
I am quite familiar with the concept of oversampling, I wrote my thesis on multirate dsp.

I think you are falling into a familiar trap of "how many percent can I blow up this image", similar to the "how many dpi do you need?". In my opinion, that line of thought leads to strange conclusions. I think that a more relevant question is "how large can I typically/ideally make a print from the D800/5Dmk3 for viewing at a sensible distance?". Predictably, images from the D800 can be printed larger than the 5Dmk3 under fair (and fairly optimal) conditions, since the D800 is capable of recording more spatial detail than the 5Dmk3.

-h
« Last Edit: June 14, 2012, 04:08:37 am by hjulenissen »
Logged

williamrohr

  • Full Member
  • ***
  • Offline Offline
  • Posts: 105
Re: Waiting for Moore's Law to plateau
« Reply #25 on: June 14, 2012, 01:33:06 pm »

As someone who has not only viewed most of Nick Brandt's collection in person but actually owns one of his photographs ... I think this discussion is missing a point that Michael, Jeff Schewe and others have been stressing for years  .... and Nick's work helped me understand even more clearly ... Nick has a very unique style that is not (as best I can tell) enhanced by a digital MF camera.  He uses film grain in a rather unique and dramatic way. Through very selective focus and the grain effects in out-of-focus areas ... it creates a unique and dramatic style.   Remember also that he spends hours sitting in a very digitally hostile environment .... dust, dust and more dust and as everyone who owns digital equipment knows, it is far more of an issue for us than in the film days.  Now I know that many will argue that many of Nick's effects could be reproduced with digital effects ... why do so when he has it nailed so perfectly with his present equipment.  Again, as Michael, Jeff and others have stressed  ..it's about the image NOT the technology.  P.S.  Go look at his images in person ... and look at the prices they sell for  .... he certainly is doing something right. One man's opinion  :)
Logged

Wayne Fox

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4237
    • waynefox.com
Re: Waiting for Moore's Law to plateau
« Reply #26 on: June 14, 2012, 02:22:32 pm »

I see your point, but I disagree. The true limit to the system should be the photographer. Or the price. Or the weight.

I don't think you can roll all of the factors of exceptional photography into a single point.  My post is simply related to my experience in printing large prints from almost every digital camera that's ever been made (thousands of prints from hundreds of different photographers each year). There are certainly many factors in the entire process, and I'm certainly aware of and have experience in those other things but they are different topics.

Quote
I think you are falling into a familiar trap of "how many percent can I blow up this image", similar to the "how many dpi do you need?". In my opinion, that line of thought leads to strange conclusions. I think that a more relevant question is "how large can I typically/ideally make a print from the D800/5Dmk3 for viewing at a sensible distance?". Predictably, images from the D800 can be printed larger than the 5Dmk3 under fair (and fairly optimal) conditions, since the D800 is capable of recording more spatial detail than the 5Dmk3.

-h
Don't know what to say.   I don't think I'm falling into any trap, I never said anything about 200% being a limit or there being a limit at all.  You can enlarge an image as far as you want.  Your statement about the d800 vs the 5D3 seems to imply the same thing that I'm talking about.  I'm looking at this from the practical experience of printing images, and my observation is interpolated data at about 200% is where things begin to be observable. The further you go the more you may see.  But its certainly image/subject matter dependent. 

I've already stated my opinion of a "sensible" viewing distance.  certainly it is a factor, but that doesn't mean it should be a restriction.
Logged

williamrohr

  • Full Member
  • ***
  • Offline Offline
  • Posts: 105
Re: Waiting for Moore's Law to plateau
« Reply #27 on: June 14, 2012, 02:59:36 pm »

Wayne:  My comment was not in any way directed at your points (and I have always enjoyed your discussion) ... just that Nick Brandt's work (at least for me) drives home the point that his style is intimately tied to his technique and that technique is not easily transferred to digital.  In fact, as I understand it, he owned for a while a Hasselblad H4D60 which he gave up at a loss because it just does not work for him.  I have personally been guilty of chasing the latest and greatest instead of spending the time on creating the right image.  When I moved from the H3DII31 to the H4D50, all I really did was give myself more options for cropping ... which if I had taken more time I could have easily done in camera. (and besides, I've saved a ton of money NOT chasing the technology  :))  Bill
Logged

texshooter

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 575
Re: Waiting for Moore's Law to plateau
« Reply #28 on: June 14, 2012, 07:21:29 pm »

can someone translate what the debate is about. it sounds like wayne depends more on sensor resolution than bill does to achieve maximum large print quality.  we've all heard endless times that you can get away with less resolution if the print is viewed from a reasonable distance. i think this is an outdated arguement because unless you hang your print above the fireplace or on a billboard, i can promise you your print will be scrutinized up close. therefore, the more rez the better, even if your scene is deliberately blurry or grainy.  nobody looks at fine art prints from afar these days. just look at those people at nick brandt's gallery exhibit. they are all practically breathing on the glass.  i just don't know how many pixels is enough for 60" prints--enough for the discerning eye that is.
Logged

williamrohr

  • Full Member
  • ***
  • Offline Offline
  • Posts: 105
Re: Waiting for Moore's Law to plateau
« Reply #29 on: June 14, 2012, 07:51:35 pm »

Its not that Wayne and I are in any disagreement.  I was just stressing (someone used Nick Brandt's work as an example) that when it comes to really powerful images ... no one really cares about the pixel count.  My wife is normally the financially conservative one in our family ... but once we saw Nick Brandt's work in person  ... she was not leaving that gallery without one, regardless of the price (and they are not cheap).  Just once I'd like to have an image that powerful (and trust me ... she couldn't care less about pixels.  In fact when she found out it was done with film ... it made my next MF back purchase that more difficult).  Bill
Logged

Wayne Fox

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4237
    • waynefox.com
Re: Waiting for Moore's Law to plateau
« Reply #30 on: June 15, 2012, 06:15:36 am »

My response was only in answer to the posts that I quoted in the response.  Nothing in that reply was intended towards the comments by Bill, and in fact I really like Nick Brandt's work as well, and have complete respect for any photographer who uses any tools to create work that is creative, well done, and enjoyable to view.

There are varying opinions about how far you can push a file, and certainly there are different expectations by different photographers as to what is acceptable.  Some advocate very strongly the appropriate viewing distance theory, I'm not one of them as I've stated in my posts, one of the reasons is just what was stated ... you can't control where the image will be displayed and in fact there are many cases where viewing the image up close is something that can't be helped .. such as a hallway, entrance ways, stairway, etc.  I don't think it's a black and white issue, and I think both point of views have some validity. Certainly circumstances and the intended market/audience is a big factor. We don't all shoot the same stuff or sell the same stuff. But once an image is purchased there is little control over it's presentation and viewing distances.
Logged

williamrohr

  • Full Member
  • ***
  • Offline Offline
  • Posts: 105
Re: Waiting for Moore's Law to plateau
« Reply #31 on: June 15, 2012, 12:04:34 pm »

As a recovering "pixel peeper" (thanks to Alain Briot) I'd like to share a little experiment going on in our house.  We have a restored antique home that has large open staircase and plenty of wall space for photographs.  We like to collect photographs from well known photographers (some of whom contribute to this website) that are particularly striking (and because I like to really explore their work up close and personal - remember I am a recovering pixel-peeper).  Among the photographs is a 30 X 40 print from a standard DSLR (16 megapixel I believe) and a contact print we bought from Ansel Adams back in the 1960's and lots of pictures of other resolutions and technologies in between.  Ansel's picture is of Bridal Vail Falls and was printed by him from as a contact print from a glass negative ( I believe its the only photograph in which he used a glass negative).  The difference in "resolution" is quite significant.  When people come to our home (many of whom are photographers) they invariably comment on the photographs and their favorites vary widely but they NEVER comment upon the resolution ... even the pixel peepers.  Again the point I am trying to stress is that the images are so striking (and skillfully printed) that no one seems concerned about all the technical issues we often obsess over.  I am beginning to think that if people comment upon the resolution etc. ... my photograph has invariably failed.  The really successful images are those where people never even consider the technical aspects (or really don't care because the image is so striking)  :)  Bill
Logged

lfeagan

  • Full Member
  • ***
  • Offline Offline
  • Posts: 196
Re: Waiting for Moore's Law to plateau
« Reply #32 on: June 15, 2012, 12:43:55 pm »

I need a +1 button for Bill's comment: "The really successful images are those where people never even consider the technical aspects."

Right on mark Bill.
Logged
Lance

Nikon: D700, D800E, PC-E 24mm f/3.5D ED, PC-E 45mm f/2.8D ED, PC-E 85mm f/2.8D, 50mm f/1.4G, 14-24 f/2.8G ED, 24-70 f/2.8G ED, 70-200 f/2.8G ED VR II, 400mm f/2.8G ED VR
Fuji: X-Pro 1, 14mm f/2.8, 18mm f/2.0, 35mm f/1.4

uaiomex

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1211
    • http://www.eduardocervantes.com
Re: Waiting for Moore's Law to plateau
« Reply #33 on: June 15, 2012, 01:53:41 pm »

+2 now.  I see all the time "here" and "there" too many pictures made with tons of mp's that communicate nothing. They are like and endless string of IQ test images.
Eduardo

I need a +1 button for Bill's comment: "The really successful images are those where people never even consider the technical aspects."

Right on mark Bill.
Logged

Wayne Fox

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4237
    • waynefox.com
Re: Waiting for Moore's Law to plateau
« Reply #34 on: June 15, 2012, 08:04:20 pm »

Everyone seems to see everything as black and white, and interpret everything written so narrowly and literally.

There are many aspects to great images, and certainly the most important is the artistic/aesthetic quality of the print.  I've never advocated or stated anything about technical quality being the predominate element of a great image.  But just because you put a great composition and subject on a print, you can certainly miss the boat technically to the point that the image falls far short of it's potential. Ansel Adams was a great photographer, partially because of his outstanding technical expertise and skill.

I'm also just stating my opinion that you can only push digital files so far before the technology can limit the experience, no matter how great the image is.  This really isn't the technical fault of a photographer, usually it's just a budget constraint ... something that has become less and less of a factor over the past 15 years since the first 2 mp dSLR's came out costing over $20,000.  How far you can go depends on many factors, certainly the most important being the perspective of the creator of the work.  In the example by bill, it doesn't sound like we're talking about anything very large, 40" is certainly doable for a good capture from a 16mp camera and is really just a nice decent size for most walls. It sounds like they all were created by a skilled photographer with a great eye as well as great technical expertise. That's why he chose them.  I'm guessing all can handle decently close scrutiny if an observer so chooses to do so.  But that doesn't mean the opposite cannot occur, and indeed I've seem some terrific images print as very large prints destined to be hung in very nice homes that even Bill would probably not be so keen on.

My home is full of my work.  Most of the images are in the 48" range, some a little smaller, some a little larger. Most are shot with MFDB, but some are with various dSLR"s I've used over the years as well. I"ve also observed many people as they come into my home and take some time to look at the work.  And I would agree, no one has ever stated "wow, those are technically fantastic".  But on the other hand, I have seen many who after looking at print, suddenly decide to walk up closer and look at the detail in the field of lavender, or examine the sky to discover what they were seeing from 10 feet away was actually a crescent moon in the evening sky.  Or comment on how real the petals of the bougainvillea look.  As with many things, qualities can exist that might be extremely subtle yet can be part of the experience of seeing, tasting, or hearing ... and usually those experiencing  it can't really describe the factors that are involved in their feelings.

This isn't a black and white, cut and dry issue here.  It is very complex, extremely variable, and the audience goes from people who could care less about photographic images to those that obsess over them.

And my apologies for diverting this thread off topic.   
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: Waiting for Moore's Law to plateau
« Reply #35 on: June 16, 2012, 02:33:15 am »

Wayne,

I don't think you are diverting this topic. The topic is about need of pixels and affordability of pixels. Some folks have a lot of experience with high end equipment, large prints and at least I appreciate very much when that experience is shared.

Regarding technique, I would say it is an important aspect of photography. Not that 16, 32 or even 64 MPixels makes for a better photographer, but execution has always been an important  factor in photography. We have different shooting traditions, a journalistic one where technical quality matters less and an aesthetic one where execution matters a lot.

There is a lot to the viewing distance argument, but I would say that the viewing distance is decided by the viewer, unless he/she is hindered to walk up and view close. Many images may contain different regions of interest, so the viewer may be interested in some detail and look close.

The enclosed picture is one of my old images from a trip to Iceland shot with a 6MP DSLR. It hangs on my wall, printed in A2-size. It's OK but more pixels would not hurt. There is a group of people in the middle. Perhaps the viewer wants to see the faces.

Best regards
Erik

Everyone seems to see everything as black and white, and interpret everything written so narrowly and literally.

There are many aspects to great images, and certainly the most important is the artistic/aesthetic quality of the print.  I've never advocated or stated anything about technical quality being the predominate element of a great image.  But just because you put a great composition and subject on a print, you can certainly miss the boat technically to the point that the image falls far short of it's potential. Ansel Adams was a great photographer, partially because of his outstanding technical expertise and skill.

I'm also just stating my opinion that you can only push digital files so far before the technology can limit the experience, no matter how great the image is.  This really isn't the technical fault of a photographer, usually it's just a budget constraint ... something that has become less and less of a factor over the past 15 years since the first 2 mp dSLR's came out costing over $20,000.  How far you can go depends on many factors, certainly the most important being the perspective of the creator of the work.  In the example by bill, it doesn't sound like we're talking about anything very large, 40" is certainly doable for a good capture from a 16mp camera and is really just a nice decent size for most walls. It sounds like they all were created by a skilled photographer with a great eye as well as great technical expertise. That's why he chose them.  I'm guessing all can handle decently close scrutiny if an observer so chooses to do so.  But that doesn't mean the opposite cannot occur, and indeed I've seem some terrific images print as very large prints destined to be hung in very nice homes that even Bill would probably not be so keen on.

My home is full of my work.  Most of the images are in the 48" range, some a little smaller, some a little larger. Most are shot with MFDB, but some are with various dSLR"s I've used over the years as well. I"ve also observed many people as they come into my home and take some time to look at the work.  And I would agree, no one has ever stated "wow, those are technically fantastic".  But on the other hand, I have seen many who after looking at print, suddenly decide to walk up closer and look at the detail in the field of lavender, or examine the sky to discover what they were seeing from 10 feet away was actually a crescent moon in the evening sky.  Or comment on how real the petals of the bougainvillea look.  As with many things, qualities can exist that might be extremely subtle yet can be part of the experience of seeing, tasting, or hearing ... and usually those experiencing  it can't really describe the factors that are involved in their feelings.

This isn't a black and white, cut and dry issue here.  It is very complex, extremely variable, and the audience goes from people who could care less about photographic images to those that obsess over them.

And my apologies for diverting this thread off topic.  
« Last Edit: June 16, 2012, 01:11:36 pm by ErikKaffehr »
Logged
Erik Kaffehr
 

Pingang

  • Full Member
  • ***
  • Offline Offline
  • Posts: 117
Re: Waiting for Moore's Law to plateau
« Reply #36 on: June 16, 2012, 12:33:03 pm »

I don't agree with that conclusion.

I've shot a lot over the years with 6x7, and still have my RZ67 and my Pentax 67 systems.  I've also been shooting digital for since the Kodak DCS 520, and have owned nearly every dSLR system as well as medium format beginning with Kodak's original DCS back up to know with a PhaseOne IQ180.

I've printed from those files as well as scanned film.  One of my good customers has been shooting landscapes for quite some time, and much of his older stuff is scanned 6x7 transparencies which I've printed.

From that experience I feel I can get as good and perhaps slightly better quality of prints out of a well exposed/processed digital image from a good dSLR in the 20mp range.  As with any of this, final print size is really the determining factor, but the 36mp Nikon or 40mp MFDB's will easily handle getting pushed to larger sizes than the scanned 6x7 film.
I shot many 135 and 6x6, 6X7, 6X9 film before going digital, scanned many of them and wish the digital arrived a lot earlier. It may be true the digital file has higher clarity than film and many still talk about the film look but, a big but, the old film can't reach the clarity of digital while digital can processed the created to have film feel, it is that was desired.

Pingang
Shanghai
Logged
Pages: 1 [2]   Go Up