Pages: 1 [2] 3   Go Down

Author Topic: How many pixels actually?  (Read 26968 times)

Plekto

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 551
How many pixels actually?
« Reply #20 on: April 30, 2008, 01:15:55 pm »

Quote
I disagree with practically every assertion you have made in this thread.  My D300 has a 12 MP DX sized sensor and I would stack it up against 135 format film without any worries.  The resolution is excellent and the DR is a comfortable 8.5 stops.  The biggest problem with Foveon sensors is that they get noisy at relatively modest ISOs; which is another way that my D300 kicks butt on both Sigma DSLRs and film.
[a href=\"index.php?act=findpost&pid=192589\"][{POST_SNAPBACK}][/a]

You are, of course, entitled to that opinion.  I was merely talking about theoretical limits and what is required to truly replace film at all levels.   Like a typical DB, most people look at the new 40MP models and shrug their shoulders and consider it good enough.  12MP with a large sensor is very close to 16MP, so the 5D could blur the line from time to time.   But as others have said, it's not *quite* a total replacement for film at 12MP.  

I consider 2400dpi scanned to be the lower limit to replace film, since every scanner on the market does at least this with 35mm film, and most of the D-Labs also are about this resolution as well.  That means a lot of data, since that 2400dpi doesn't have micro lenses, moires, anti-aliasing, or other factors affecting it.  It's 2400dpi with full color at every pixel.  That pushes the bar up quite a bit for digital due to the way Bayer patterning works.  

Now, whether we call it 60% or 70% or 75% aside, the fact is that a Bayer sensor has to be higher resolution to replace film because of these inherent design flaws.  There is some loss of information that software can't really get back fully due to the lower percentage of red and blue sensors.   So you have to shoot at higher resolution and then drop it back down with software to get that smooth and clean look of good film.
Logged

dalethorn

  • Guest
How many pixels actually?
« Reply #21 on: April 30, 2008, 02:34:55 pm »

Actually I don't have a problem with "more weight" as you say, but when "more weight" is ten times more, I think it's disingenuous to call that "more weight".
Logged

Tony Beach

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 452
    • http://imageevent.com/tonybeach/twelveimages
How many pixels actually?
« Reply #22 on: April 30, 2008, 06:28:51 pm »

Quote
You are, of course, entitled to that opinion.  I was merely talking about theoretical limits and what is required to truly replace film at all levels.   Like a typical DB, most people look at the new 40MP models and shrug their shoulders and consider it good enough.  12MP with a large sensor is very close to 16MP, so the 5D could blur the line from time to time.   But as others have said, it's not *quite* a total replacement for film at 12MP. 

I consider 2400dpi scanned to be the lower limit to replace film, since every scanner on the market does at least this with 35mm film, and most of the D-Labs also are about this resolution as well.  That means a lot of data, since that 2400dpi doesn't have micro lenses, moires, anti-aliasing, or other factors affecting it.  It's 2400dpi with full color at every pixel.  That pushes the bar up quite a bit for digital due to the way Bayer patterning works. 

Now, whether we call it 60% or 70% or 75% aside, the fact is that a Bayer sensor has to be higher resolution to replace film because of these inherent design flaws.  There is some loss of information that software can't really get back fully due to the lower percentage of red and blue sensors.   So you have to shoot at higher resolution and then drop it back down with software to get that smooth and clean look of good film.
[{POST_SNAPBACK}][/a]

It is not my intention to make this a "film versus digital" thread, and you appear to have backed off on claiming that Foveon sensors are delivering three times the resolution per photosite that BFA sensors deliver.  Yes, we are all entitled to our opinions and I wish I had phrased my objections to your earlier posts more diplomatically, but what I'm referring to is the experience and results that those opinions are premised upon.  Here's [a href=\"http://www.bythom.com/thom.htm]someone[/url] who has a different opinion than yours about this film versus digital debate and the format and megapixels comparisons.  If you ask him (send him an email), he'll tell you that 10-12 MP DX sensors do a better job than 135 format color film does -- if Michael Reichmann wants to get involved here, I'm sure he'll weigh in with a different opinion than yours about the larger format film versus digital comparison.

There's an apt expression that addresses the broader issue of pixel equivalence being discussed here:  The proof of the pudding is in the eating.  Now if the pudding you want to eat is black and white resolution charts (or black and white anything for that matter), then black and white film is the better choice.  If you want to measure black on a 100% saturated red or blue background then I am sure that Sigma's Foveon sensors would excel at that as long as the ISO wasn't pushed too high.  However, for small format color photography in the real world 10-12 MP BFA sensors (even smaller DX and APS-C sensors) are every bit as good as 135 format color film assuming you know what your doing with the files they produce.
Logged

dalethorn

  • Guest
How many pixels actually?
« Reply #23 on: April 30, 2008, 06:56:39 pm »

I think a more precise description of this thread than "pixel equivalence" is Pixel Count -vs- Pixel Quality. And regardless of the opinion of the person who deposits facts here, just raising those facts or issues in this context has provided some really useful pointers for further investigation. Now we should all chant together "Praise the mfr's for the true good that they've done, and don't spare the criticism when they weasel out some new design that panders to the quality-unconscious crowd."
Logged

Tony Beach

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 452
    • http://imageevent.com/tonybeach/twelveimages
How many pixels actually?
« Reply #24 on: April 30, 2008, 07:16:19 pm »

Quote
Actually I don't have a problem with "more weight" as you say, but when "more weight" is ten times more, I think it's disingenuous to call that "more weight".
[a href=\"index.php?act=findpost&pid=192715\"][{POST_SNAPBACK}][/a]

Let's be clear about some things here.  When you say "500mm", you are really referring to some "equivalent" focal length based on 135 film format.  The actual focal length is more like 100mm, and to be fair (which marketing hype never is) the "equivalence" should extend to aperture as well as focal length.  My rough estimate is that a DX sensor enjoys a 5 stop advantage in DOF isolation over a 2/3 sensor; so the question is, how much would a 100mm f/1.4 zoom lens weigh compared to a 267mm f/11 lens?

As for the overall weight.  Even ignoring that the 80-400 mounted on a D60 would give you 50% more reach than a 400mm "equivalent" P&S, off the charts more DOF isolation and high ISO performance; the actual difference in weight between the Nikon DSLR/tele-zoom combo and the Fujifilm Finepix S100FS is double (around 1882 grams compared to 968 grams; now if you can find a 200 gram P&S with a 500mm "equivalent" focal length, I would like to hear about it).

Therefore, I do not think I'm being "disingenuous" writing "more weight".
Logged

dalethorn

  • Guest
How many pixels actually?
« Reply #25 on: April 30, 2008, 10:37:41 pm »

If you don't mind I'll put that back on you with the "proof of the pudding" analogy. I have two years worth of shots with a Nikon 8800 and a Panasonic FZ50, and I've been through enough exhibitions to note that my 420mm equivalent zoom works exactly as advertised, and while I don't doubt that your big heavy 500mm zoom has more dynamic range, you're just not getting anywhere near the variety of shots when you're locked down on that  tripod. Of course if you're doing mostly landscapes, you can ignore the foregoing, unless you'd want to reconsider and use a MF camera for those landscapes (to gain the extra edge you're selling so passionately).
Logged

Tony Beach

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 452
    • http://imageevent.com/tonybeach/twelveimages
How many pixels actually?
« Reply #26 on: May 01, 2008, 12:52:40 am »

Quote
If you don't mind I'll put that back on you with the "proof of the pudding" analogy. I have two years worth of shots with a Nikon 8800 and a Panasonic FZ50, and I've been through enough exhibitions to note that my 420mm equivalent zoom works exactly as advertised, and while I don't doubt that your big heavy 500mm zoom has more dynamic range, you're just not getting anywhere near the variety of shots when you're locked down on that  tripod. Of course if you're doing mostly landscapes, you can ignore the foregoing, unless you'd want to reconsider and use a MF camera for those landscapes (to gain the extra edge you're selling so passionately).
[{POST_SNAPBACK}][/a]

I don't have a 500mm lens, but I do have a manual focus 300mm f/2.8 lens that I rarely use and I can testify to it's weight.  On the other hand, there is just no comparing the results I get with that 450mm "equivalent" (I use it on a DX camera) to what you are getting from your 420mm "equivalent" lens on a P&S camera.  It's just a fact of life that you get what you pay for, and some of that payment is weight and size.  There is no way faster, longer focal length glass can be replicated with slower, shorter focal lengths used on a tiny sensor -- the creative potential of the latter is much more limited than the former, although that doesn't mean you can't get perfectly fine photographs with a smaller format sensor.

I have nothing to prove to anyone, but this discussion does become meaningless without some examples; so here's a handheld shot with the aforementioned Tokina 300/2.8 shot at f/5.6 with a D200:



I doubt the background would be as blurred using a P&S digicam.

What is notable about [a href=\"http://photos.imageevent.com/tonybeach/mypicturesfolder/april2008//2008-04-19_Montage-web.jpg]this[/url] sequence is that the images are cropped and still hold excellent detail despite being cropped.  It is also useful to note that pixels are not the only thing that matters in an image and that a camera's AF and fps capabilities also contribute to telling some stories that would otherwise not be told.
Logged

Plekto

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 551
How many pixels actually?
« Reply #27 on: May 01, 2008, 02:57:49 pm »

Quote
...and you appear to have backed off on claiming that Foveon sensors are delivering three times the resolution per photosite that BFA sensors deliver.

Actually, I have never once in any post that I've made online, anywhere, claimed that the Foveon sensor wasn't flat out lying about it being 3x the sensors.  It's 4.6MP, because according to the definition of a "pixel" by every other source in printing and optical history other than digital camera manufacturers marketing claims, a "pixel" is:

(Dictionary.com)
The most basic unit of an image displayed on a computer or television screen or on a printer. Pixels are generally arranged in rows and columns; a given combination among the pixels of various brightness and color values forms an image. ◇ A subpixel is one of three components of a pixel used in the representation of a color image. Each subpixel represents the contribution of a single color—red, green, or blue—to the overall color and brightness of the pixel.
*****

Labeling sub-pixels as the real thing has always been a *major* pet peeve of mine, since it's flat out lying.  12,000 sub-pixels arranged in a Bayer pattern isn't close to what you really need to fully replace film.  

Since it's not quite as easy to convert a Bayer pattern into exact true pixels, due to the interpolation that's required, a conversion factor of about 60-65% has proven to be fairly accurate.  I'm being generous at 65%, though, since a smaller sensor with tiny sub-pixels is going to take a massive hit to its ability to capture vibrant colors and deal with contrast.  And few DSLRs have 1:1 ratio sensors.(do any?  I honestly might have missed a model or two)  Most cameras seem to be making them *smaller* lately, when they really need to get larger, if anything.(or move to a non-Bayer pattern)

See, my point wasn't about absolute resolution, but about what is required by a typical Bayer patten sensor to overcome these technical limitations and be equivalent to a 2400DPI scan of film in terms of how it looks.(which most Pros consider to be a *minimum* acceptable limit - most say it's closer to 3000dpi with good equipment and low speed film)  No jaggies, no moires, no fringing... Because film doesn't have those digital artifacts to begin with.  Claiming 12MP is equal to film when defects in the image are present that have to be blurred/anti-aliased out at that resolution is a bit disingenuous.

P.S. I'm not attacking you, but rather the marketing lies by the camera makers.  Current technologies have serious flaws in their designs that need to be solved and their marketing hype and misinformation isn't making it any better.
« Last Edit: May 01, 2008, 03:09:29 pm by Plekto »
Logged

dalethorn

  • Guest
How many pixels actually?
« Reply #28 on: May 01, 2008, 10:48:00 pm »

Quote
I don't have a 500mm lens, but I do have a manual focus 300mm f/2.8 lens that I rarely use and I can testify to it's weight.  On the other hand, there is just no comparing the results I get with that 450mm "equivalent" (I use it on a DX camera) to what you are getting from your 420mm "equivalent" lens on a P&S camera.  It's just a fact of life that you get what you pay for, and some of that payment is weight and size.  There is no way faster, longer focal length glass can be replicated with slower, shorter focal lengths used on a tiny sensor -- the creative potential of the latter is much more limited than the former, although that doesn't mean you can't get perfectly fine photographs with a smaller format sensor.

I have nothing to prove to anyone, but this discussion does become meaningless without some examples; so here's a handheld shot with the aforementioned Tokina 300/2.8 shot at f/5.6 with a D200:

I doubt the background would be as blurred using a P&S digicam.

[a href=\"index.php?act=findpost&pid=192804\"][{POST_SNAPBACK}][/a]


I'm trusting I did the snip OK on the quote -- anyway here's two from Bolsa Chica - wild birds at the wetlands, using Panasonic FZ50 at 420mm "equivalent". I'm sure you can see all kinds of problems with these non-pro images, being expert as you are, but seriously, do I really need a 50-lb camera to do better than this? All I would ask for is the mfr. to do the best they can in the package size I'm using now, and they can do a *lot* better.  BTW this is a two-year old camera, which apparently has been replaced with "superzoom" junk using a 1/2.33 sensor. That's progress in the 21st century.
Logged

Tony Beach

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 452
    • http://imageevent.com/tonybeach/twelveimages
How many pixels actually?
« Reply #29 on: May 02, 2008, 12:56:12 am »

Quote
Claiming 12MP is equal to film when defects in the image are present that have to be blurred/anti-aliased out at that resolution is a bit disingenuous.
[{POST_SNAPBACK}][/a]

It isn't a marketing claim, it's what photographers experience in real world prints.  Again I cite Thom Hogan (this time directly), "But let me tell you something: all this deep ending into detail isn't particularly useful. The D2x, 5D and 1DsMarkII all outperform 35mm film...", at:  [a href=\"http://www.bythom.com/d2xreview.htm]http://www.bythom.com/d2xreview.htm[/url]
Logged

BernardLanguillier

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 13983
    • http://www.flickr.com/photos/bernardlanguillier/sets/
How many pixels actually?
« Reply #30 on: May 02, 2008, 01:08:33 am »

Quote
I use point & shoots, which I imagine few of you do. Since their pixel specs are usually way overrated, I wonder if there's a better way to rate them. For example, if I were to take a series of landscape pics under ideal lighting conditions, do minimal or no post processing, then make several copies of each at different sizes (say, 3072x2304, 2560x1920, 2048x1536 etc.), and view all copies at full size on screen, could I say that the True Effective Pixels is the smallest size at which I can plainly see all of the detail of the original, ignoring edge artifacts etc. added by the in-camera processing? I ask this because my highly-promoted 9mp camera appears not to have any real detail beyond about 3mp.
[a href=\"index.php?act=findpost&pid=192231\"][{POST_SNAPBACK}][/a]

My view is that more than a few compact cameras used at base ISO and in the middle of their zoom range do resolve close to as much as a DSLR of the same pixel count (I am basing this on an extended usage of my Ricoh GX100).

Since they are typically designed for users with very limited post processing skills, they tend to offer a pretty strong AA filter to prevent moire etc... which limits a little bit their pixel sharpness, plus some noise is present even at noise ISO which tends to hide some detail also.

But overall, it remains pretty close.

Regards,
Bernard

Tony Beach

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 452
    • http://imageevent.com/tonybeach/twelveimages
How many pixels actually?
« Reply #31 on: May 02, 2008, 01:23:39 am »

Quote
...anyway here's two from Bolsa Chica - wild birds at the wetlands, using Panasonic FZ50 at 420mm "equivalent". I'm sure you can see all kinds of problems with these non-pro images, being expert as you are, but seriously, do I really need a 50-lb camera to do better than this?[a href=\"index.php?act=findpost&pid=193043\"][{POST_SNAPBACK}][/a]

They are fine images, but the background in the first one is cluttered and would have benefited from a shallower DOF.  Now please Dale, stop exaggerating about the weight of improving upon that.  I posted the weight of gear that would deliver better results, the total weight is one tenth of what you are claiming.  While being half as heavy is not inconsiderable (remember that in my previous post on this that the comparisons were 1882 grams versus 968 grams), it is not even remotely ten times as much.

Here's what I routinely tell my son (and anyone else who cares to take my advice) on the topic of photography:  the more effort you put into an image, the more it will distinguish itself to the viewer.  In this discussion that means if you want to carry a lighter camera with a small sensor and slow lens and shoot handheld then (all other things being equal) you will have less to show for your efforts than someone who is willing to haul more gear and work harder at getting the shot.

I am sorry you are disappointed with camera makers not making better P&S cameras.  Many photographers share your frustration.  However, many photographers want bigger sensors and faster wide angle, fixed focal length lenses with a simple UI and the ability to shoot RAW; while many consumers prefer a more telephoto image (they want the camera to get them closer to the subject)  in a small package and they want it to come out of the camera ready to view, which means they end up with a "superzoom" lens and a less capable smaller sensor in a camera that doesn't shoot RAW.  Since the camera makers are all chasing after consumers we end up with the latter and not the former in the P&S digicams they are marketing.
Logged

dalethorn

  • Guest
How many pixels actually?
« Reply #32 on: May 02, 2008, 07:54:40 am »

I can't even quote Tony because of the endless repetition of "justifications" for the way things are. Well, Tony, I and my cameras are the way things are, and no offense intended, but I'm not concerned about your gear and how you use it. I've been at this since the 1960's so I'm well aware of the way things are for you. And LL shows me all the other formats as well. What I want is forward progress in my interests, and the best way you could help is to stop excusing the mfr's and just tell it like it is - they're pandering not progressing in most cases. And not only are those good images I posted, but thanks to my lightweight kit, I was able to share hundreds of those with other photogs at the wetlands using the Libretto computer I also carried. And not one other photog ever showed me anything - they were all bogged down with "camera gear". So what's the point of this "hobby" if you can't share? I know - this forum, away from the action. Well, it's a poor substitute for that purpose.
Logged

dalethorn

  • Guest
How many pixels actually?
« Reply #33 on: May 02, 2008, 08:03:15 am »

One last point on the subject of "having less to show because of smaller less capable gear":  I totally disagree - I have far more good images to show than the Big Gear folks from Bolsa, for what should be obvious reasons - I get anound the site much better, get into smaller spaces, get closer to the animals (scare them less), have no issues of tiring out, and on it goes. You probably can't appreciate that if it's outside of your experience.
Logged

Tony Beach

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 452
    • http://imageevent.com/tonybeach/twelveimages
How many pixels actually?
« Reply #34 on: May 02, 2008, 10:55:16 am »

Quote
One last point on the subject of "having less to show because of smaller less capable gear":  I totally disagree - I have far more good images to show than the Big Gear folks from Bolsa, for what should be obvious reasons - I get anound the site much better, get into smaller spaces, get closer to the animals (scare them less), have no issues of tiring out, and on it goes. You probably can't appreciate that if it's outside of your experience.
[a href=\"index.php?act=findpost&pid=193096\"][{POST_SNAPBACK}][/a]

First, on "justification"; I am not justifying anything -- it's just a fact that camera manufacturers are more interested in "pandering" to consumers than making a camera that you want.  I don't like it anymore than you do and I'm sorry that you are upset by this state of affairs.

Second, if you are going to quote me, then do me the courtesy of accurately quoting me.  I said "...(all other things being equal) you will have less to show for your efforts than someone who is willing to haul more gear and work harder at getting the shot."  That means they have to work harder; now if you worked harder then you got the better images, and I acknowledged that I thought you got fine images.
« Last Edit: May 02, 2008, 08:56:17 pm by Tony Beach »
Logged

dalethorn

  • Guest
How many pixels actually?
« Reply #35 on: May 02, 2008, 01:26:10 pm »

Two points: We're here to provide useful info - not to state the obvious, or say "that's the way it is" etc. When there's no other reason to state those things -- it defaults to justification. Point two: I didn't work harder to get my pics, I worked *less* because I had better (i.e. more appropriate gear).
Logged

Plekto

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 551
How many pixels actually?
« Reply #36 on: May 02, 2008, 08:47:26 pm »

Quote
It isn't a marketing claim, it's what photographers experience in real world prints.  Again I cite Thom Hogan (this time directly), "But let me tell you something: all this deep ending into detail isn't particularly useful. The D2x, 5D and 1DsMarkII all outperform 35mm film...", at:  http://www.bythom.com/d2xreview.htm
[a href=\"index.php?act=findpost&pid=193059\"][{POST_SNAPBACK}][/a]

That's one sentence in the whole review, and is quite obviously his opinion.  I don't see him giving hard facts about it.   12MP to compete with a 35mm scan at 2400dpi would require about a 75% ratio for interpolation and that's just not physically possible beased upon how the technology works.(66% is the most you can get, actually, out of the raw, which is why 65% is being very generous.  As is 2400DPI.  If you bump it to 3000dpi,  it jumps to nearly 13 million actual full pixels, which the Bayer patterned sensor isn't possibly going to do - that's over 100% radio.  Some desktop scanners go to 3600 DPI even.(though I think 3000 is about the practical limit - and is what the Minolta scanners do)
Logged

Tony Beach

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 452
    • http://imageevent.com/tonybeach/twelveimages
How many pixels actually?
« Reply #37 on: May 02, 2008, 09:14:33 pm »

Quote
That's one sentence in the whole review, and is quite obviously his opinion.  I don't see him giving hard facts about it.

Some desktop scanners go to 3600 DPI even.(though I think 3000 is about the practical limit - and is what the Minolta scanners do)
[a href=\"index.php?act=findpost&pid=193225\"][{POST_SNAPBACK}][/a]

In theory I could devise a scanner that could deliver 50 or a 100 MPs from a 35mm negative or slide; but that doesn't mean there is that much meaningful detail in that negative or slide.  The question that needs to be answered (and may have an actual connection to this thread) is how many pixels can you really extract from a 35mm negative or slide?  Not addressing that question and the lack of side by side examples makes your arguments unsubstantiated speculation.

You can choose to disbelieve Thom Hogan on this issue; but then you are saying he is either wrong or lying.  It is specious of you to try to claim simultaneously that Thom may not have really meant what I quoted (because it is only one sentence in a longer review) and to then argue that is just his opinion.  Thom's "opinion" is based on his experience, and I trust his experience and mine over your speculations.
Logged

Plekto

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 551
How many pixels actually?
« Reply #38 on: May 02, 2008, 09:21:08 pm »

Most experts agree that the effective maximum for film is between 2400 and 3000DPI.  Certainly not less than 2400DPI for good film. Even the cheapest DLabs try to keep it at or near 2400DPI.   That raises the bar to slightly higher than 12MP with a Bayer sensor.

Again, if we were talking about 16MP, I'd say it was a wash, or close to it, though with AA filters and other nonsense being so common, you might have to go to 20MP or so or find a camera without such a filter.
Logged

Tony Beach

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 452
    • http://imageevent.com/tonybeach/twelveimages
How many pixels actually?
« Reply #39 on: May 02, 2008, 09:39:24 pm »

Quote
Most experts agree that the effective maximum for film is between 2400 and 3000DPI.  Certainly not less than 2400DPI for good film. Even the cheapest DLabs try to keep it at or near 2400DPI.   That raises the bar to slightly higher than 12MP with a Bayer sensor.
[a href=\"index.php?act=findpost&pid=193227\"][{POST_SNAPBACK}][/a]

My D300 has 12.2 MPs.  Anyway, just show a comparison of a scanned slide side by side with a 12 MP DSLR file -- if the shots are taken under identical conditions then that would put the debate to rest.
Logged
Pages: 1 [2] 3   Go Up