I know I am going to come across as smug/superior and offend some with this thread. that is not my intent. Getting the science straight is my intent.
For those that think megapixel resolution and how much you pay for it is the only thing that matters.....Sell all your equipment and buy the latest high MP point and shoot. You will be very happy. At least the specs say you will.....
OK - I was being smug and just trying to get your attention.
Do some research on human vision. You will find that the typical human can not resolve detail beyond 0.005 inches (if you put two dots or lines side by side they will appear to be the same if closer than 0.005 inches). Interestingly enough that works out to 200 dpi.
Those that say they can see diferences between a 200dpi print and a 300dpi print are correct. The print made at 200dpi is made with imperfect "dots" (I am not talking about the inkjet dithering etc. but the imperfections in the digital file itself) By shrinking the dots from 200-300dpi you are shrinking the imperfections to the point that the human eye can not resolve them and creating the illusion of increased sharpness.
If you were to make two digital immages, one at 4MP and one at 8MP, and each pixel was absolutely perfect and you printed them you would not be able to tell any difference at all up to a print size of just over 8X12 (8.25X12.75 at 200dpi equals 4MP). At that point going bigger and bigger the 4MP would start to loose sharpness.
Now think of a 4MP with fantastic color and tone rendition, contrast, and etc. compared to a medium quality 8MP file. Up to 8X12 the 4MP camera would far outshine the 8MP giving pictures the sublte differences that we can not visually resolve as detail but call "pop" or "wow". As picture size increases and increases the 8MP file will eventually win out.
Now consider the interpolation programs like Genuine fractiles or Pixel Genious. they do a great job - Especially with a file of very high original quality. They are more than capable of taking a 4MP file, increasing it's size by 400% allowing you to print at 16X24 with almost no reduction in quality. since your 4MP file was higher in quality to begin with you are still better than the 8MP folks or at least as good.
Of course my argument only works if:
the D2Hs gives superior sharpness, tonality, noise, and etc. I think it will since the D2H when used correctly did already.
8X12-11x16 is the typical max size prints you need for the majority of your work and you only need to use file interpolation for a small %.
I know there are plenty that disagree, but I think the Nikon engineers who understand all this stuff way better than me got this one right.
akreager wrote:
Now think of a 4MP with fantastic color and tone rendition, contrast, and etc. compared to a medium quality 8MP file. Up to 8X12 the 4MP camera would far outshine the 8MP giving pictures the sublte differences that we can not visually resolve as detail but call "pop" or "wow". As picture size increases and increases the 8MP file will eventually win out.
Canon has that camera.. it is called the 1DMkII... you can order your cake and eat it.. nuff said
1. I'd be curious to see your sources for the 200 DPI figure. I have no trouble at all resolving detail to finer than 400 DPI. Now, I'm nearsighted, so I have better-than-average closeup vision, BUT I've also done extensive experiments in inkjet-printing output, especially in black-and-white printing. One of the problems in using pigment-based (i.e., archival) inks for BW printing is metamerism, and one way around that is to use black-ink-only printing. But even at 1440 or 2880 DPI, which I've tested on dozens of people, about half the people report noticable texturing in midtones on printed test wedges.
So you should go back and look at the studies you cited: What was the question? Did they ask "can you tell these 2 dots apart?" Or did they show people two images, one printed at 200 DP and the other printed at 400 DPI and ask if they look the same?
2. A bigger problem with your thesis is that it assumes no cropping. I shoot with a D100, which is 6 MP, and it's OK as long as I've composed perfectly. But if I have to crop at all it's not enough. Magazine photo editors love to crop.
3.. Another big problem with your thesis is assuming that 4 MP represents the RESOLUTION of the camera, i.e., its ability to resolve detail. 4 MP is the number of photosensitive elements on the CCD. But they are divided into R, G, and B channels in a bayer matrix. So the camera doesn't actually have 4 MP of luminance resolution. If you shoot a black-and-white test chart (i.e., so the same luminance information is available in all 3 channels), you will find that it has much higher resolution than if you shoot a test chart that's mostly green or mostly red. (I've done this - it's quite interesting). Lots of photographers wonder why they always feel like they have to add sharpening (e.g., via USM) to pictures taken with digicams. Well, this is one of the reasons. Current digicams just can't record as much detail as their megapixel count implies.
Here are the numbers:
for 4x6/8x12 there is no crop and at 8x12 you get 205 pixels per inch for printing ( adequate )
crop to 5x7 = (2285x1632 max) = 326 ppi ( very good)
crop to 8x10 = (2040x1632 max) = 204 ppi ( adequate )
crop to 16x20 = (2040x1632 max) = 102 ppi ( forget about it - time for some interpolation and raw files)
my conclusion for the way I work is that it's livable with but 6-8 megapixels would cure the D2h's achiles heel.
A lot of discussions about whether 4.1MP is "enough" also ingore the impact of lower resolution files on the post-production process and perceived detail. Two common post-production operations - sharpening and noise reduction - both benefit immensely from more pixels.
Noise reduction in particular benefits from more MP. Noise reduction works by blurring/interpolating the noisy pixels using non-noisy pixels (oversimplification - but the details would take pages). Take a 2,400 pixel wide image and apply noise reduction - do the same for a 3,500 pixel image and you will find that you can resolve much more detail in the larger image afterwards. D2H images regularly make trips through the noise reduction process just to get usable images. The result is the actual perceivable detail of a 2,400 pixel noise-reduced image is going to be closer to say an 1,800 pixel image. So now you have an "effective" 1,800 pixels...then you crop to 8x10, you're down to 1,500. Enlarge that crop...ugh. See how it works?
Another thing that amazes me is how people are willing to drop $3,500 for a "well, it's good enough" camera. For that kind of money, I'd really like something better than "good enough". It would be one thing if "good enough" was all that was available due to the state of the technology, but it's not. A few hundred dollars more and you can have a best-of-class camera.
Just a comment to add, not disagreeing with anyone so take my post for what it's worth. As you notice from my signature I own a D2h, and am very satisfied with it.
I print up to 13x19 on a 2200 printer and the results are very satisfactory. Would I like more pixels? Only if the end result justified the additional expense. Put me in the column of people who think 4.1 mp is enough. If nothing horrendous happens to the camera I expect to be using it for many years.
Dave .. that is a good point. I had noticed that with the D2h files I put through Noise Ninja the detail was getting clobered compared to the files from a 10D. I didn't know that it worked that way and it's good to learn. The D2h is definately a camera with compromises ...but for some it's still the best in class. The canon is the best in class for image quality right now .. no argument. The ergonomics while a personal thing and hard to quantify ... is something that is eaqual in importance to image quality if image quality is "good enough". Tough choice but it's not a landslide for canon. I am leaning personally to sacrificing image quality on a nikon rather than sacrificing intuitive operation of the machine on Canon. Don't know which way I will ultimately sacrifice ... but just mean to add that there is logic in going with the less obviously logical. Image quality is easy to discuss and it's canons strenght. Nikons strength is someting that has to be experienced and it has to fit the user. When it does ... it's the best in class.
OK - I knew a bunch of folks would disagree. Does anyone agree or am floating in a big boat by myself?
How about just up to 8X12 size. That is the vast majority of what I do. I also do not want to spend much more than $3000 at this time.
Based on what I have seen, the old D2H blows away the D70 re image quality at least up to 8X12. Every reason to believe that the S will represent an improvement.
I am not concerned with what myself or another photographer can percieve under CLOSE and PROLONGED scrutiny, but what 99.9% of my viewers will think. I will not pull the trigger until the product reviews are out and I have had a chance to test myself, but here is my bet:
At any print size up to 11X14 (16.5 full frame), the new S will be superior to the D70 and 20D.
At any print size up to 8X12, 99.9% of viewers will not be able to tell the difference between a picture taken with the 1D MarkII and the new S.
I will be able to buy a new S for $3000 by this time next year. I would have to spend $3750 plus all the cost to convert if I went with the 1DMarkII.
Two or three years from now I can buy a D2X as a discontinued model and I will have two excellent digital bodies.
PS - Tonality, saturation, contrast, texture, etc. - All important to image quality are factors that tend to be significantly seperated from sharpness/ability to resolve detail. The average person when viewing a photo (say an 8X10 in an alblum on their lap in normal lighting will not be able to tell the diference between 200dpi and 400dpi
Image quality too is a Subjective thing. IF the holy grail of it all is One area namely : Noise profile , then Yes you are right in your assertions. But images Rarely are judged solely on this criterior buy the "consumer" of those images. Rather the topic is bantied about on borads like this one because the numbers are quantifiable realities but really speak only to a select few. I remember when this phenemonon happened in HiFi audio. THe specs from all the mfg published their "number" just like our "numbers" except their acronmys were THD and wow and flutter, and the like. But the older tube amps with Horrible "specs" are still preferred to solid state components. Are the users of the solid state components the "winners" here , it is obvious by the "numbers" that they are , but I was never one of them either. Rather I let my ears be the judge. SOmewhat the same holds true for me now. I simply do not like the "look" of the canon files, do I want a bit less noise in that last critical stop in my Nikon files , heck Yes I do. But it is not a deal breaker for me at all. I cannot tell you how amazing it is to have a Raw convertor like Nikon Capture 4.2, basically I only use CS if I need to clone something and to print the files. NC4 is a godsend. Where is Canon in regard to that issue. NC is at version 4.2 now.......
Do I wish Nikon would release products in a faster rate? hmmm only if the products are a bigger jump and not just incremental improvements.
well from what my eyes have seen ... the image quality is splitting hairs in all attributes besides the noise. Most untrained eyes can't tell that one image from a nikon is shot by brand "a" while a shot from a canon is shot by brand "b". There it is a case of us forum junkie debating with a miopic level of detail. (but fun up to a point) The place where image quality differences scream out is in high iso files. At 1600 even done right the best one can hope for out of the D2h after a good clean up in the noise program of one's choice is that it gets somewhere close to the look of a 20D file before anything is done to it. I've seen that much with my eyes. And then it still won't have the detail that the 20D gives with it's resolution advantage but maybe the noise will not be intrusive. That is the area where if you process files eaqualy ... even a newbie can instantly spot the look differences. So there we have a difference, a limit and that is why we discuss this. Will this limit creap into ones work style and how much. What are techniques to minimise it's negative impact and most fundamentaly .... ARE there techniques that minimise the disadvantage. You have indicated that you believe that there are. I look at your website and and see gorgeous pictures that have no hint of intrusive noise and so I look to you as someone who know of what they are talking about. I pick up a D2h and I, in my newbie state with it can't get anywhere near the results you seem to get with it. That leaves me with the question of what you are doing to get the look you get and can I shoot the same way to get what are satisfactory results for me. Perhaps you don't shoot iso 1600. I can live with that if I had to do it but would rather not have to. perhaps you know a thing or two about setting up your camera and post process magic that I don't. It leaves me with the belief that the canon I can get results quality wise. I've seen the proof coming out of my camera ... in fact nothing to it. The Nikon is not this way. It's a mistress that you have to know how to massage quality out of aparently ... I've seen this much with my "mixed results" shooting it. I am not a beginer photographer. I don't mind having to learn a camera and being constrained by it's ways ... I just want to be sure the rainbow has the pot of gold. The portraits you have on your site btw ... look almost like they were shot with a canon ... how'd you do that? Peace.
Pavel wrote:
well from what my eyes have seen ... the image quality is splitting hairs in all attributes besides the noise. Most untrained eyes can't tell that one image from a nikon is shot by brand "a" while a shot from a canon is shot by brand "b". There it is a case of us forum junkie debating with a miopic level of detail. (but fun up to a point) The place where image quality differences scream out is in high iso files. At 1600 even done right the best one can hope for out of the D2h after a good clean up in the noise program of one's choice is that it gets somewhere close to the look of a 20D file before anything is done to it. I've seen that much with my eyes. And then it still won't have the detail that the 20D gives with it's resolution advantage but maybe the noise will not be intrusive. That is the area where if you process files eaqualy ... even a newbie can instantly spot the look differences. So there we have a difference, a limit and that is why we discuss this. Will this limit creap into ones work style and how much. What are techniques to minimise it's negative impact and most fundamentaly .... ARE there techniques that minimise the disadvantage. You have indicated that you believe that there are. I look at your website and and see gorgeous pictures that have no hint of intrusive noise and so I look to you as someone who know of what they are talking about. I pick up a D2h and I, in my newbie state with it can't get anywhere near the results you seem to get with it. That leaves me with the question of what you are doing to get the look you get and can I shoot the same way to get what are satisfactory results for me. Perhaps you don't shoot iso 1600. I can live with that if I had to do it but would rather not have to. perhaps you know a thing or two about setting up your camera and post process magic that I don't. It leaves me with the belief that the canon I can get results quality wise. I've seen the proof coming out of my camera ... in fact nothing to it. The Nikon is not this way. It's a mistress that you have to know how to massage quality out of aparently ... I've seen this much with my "mixed results" shooting it. I am not a beginer photographer. I don't mind having to learn a camera and being constrained by it's ways ... I just want to be sure the rainbow has the pot of gold. The portraits you have on your site btw ... look almost like they were shot with a canon ... how'd you do that? Peace....Show more →
Thanks Pavel you are too kind
The portraits I have done are more about the Light than the camera. I just have learned to see light in a fundamental way. Having studied and assisted one of the Very top shooters these last 4 years 2 years of which I was not shooting at All, have taught me that it is all about the light and I do not mean it in the cliche way that so many spout . But in a profound way. If my camera is noisy in the last stop what are my options ? Really what are they? Fill that stop better, clip the stop, reduce the noise later ( I hate that). So when I evaluate the picture Before I press the shutter , I see the ratios , and make the decisions Then and there. If you know Anything about me from all these rambles , I have Never Ever blamed a camera for anything, I just adapt to the tool in my hand, period , end of story. This pixel peeping at 100% serves nothing, for me the prints are king and I will do anything to try and make them better, noise that might look objectionable on screen can be insignificant in a print. The new D2x will have USER adjustable Noise Reduction IN camera, this leaves the decisions in the hands of the photographer , where IMO it should be in the first place.
I have Never stated that Nikon is a better system, camera or lens. But I do maintain that there are many compelling reasons to shoot these cameras , use these lenses and so forth. I do tell everyone that NC4.2 is the best editor I have ever used and I have used most all of them. And I gladly pay the 100bux for that application.
I do not do much high iso work but in the times I have shot at 800~1000 and the occasional 1600 , I have been very successful in dealing with it simply because I push the crap out of the EV at the time of capture and pretty much live with the shot as is, I do not attempt "life saving" methods of saving them in post. IMO it is the way to shoot Any camera, and esp the D2h.
My issue with all of this is the purpose. The D2H was designed as a sports camera. It is made with the photojournalist in mind. For that purpose, I would put it head to head against any other phojo camera out there. When the D1H came out I never heard one wedding photographer or landscape artist trying to get their hands on it-- it wasn't designed for them.
That D2H is a fine piece of machinery..but it is a Lamborghini designed for speed not comfort or a family of four or going up the back hills. If you are working for a newspaper, I wouldn't pic up anything else with its accuracy for critical focus, speed and ergonomics. But no matter how many people tell me how much they love it for other things--- I would NEVER again use it to shoot a wedding.
Dragonfly - Why with how accurate it is and it's critical focus etc. would you not want it for a wedding - at least the PJ type pics and especially the HS if it does indeed have great noise characteristics.
Well..because I don't have a D2Hs. I had a D2H. And the noise WAS an issue. If I was in the studio, I would say that there would be no excuse for not perfectly exposing my photos. But weddings are fluid and often in adverse lighting conditions or with restrictions on lighting (time, equipment, power, the rules of the church, etc). Why would you use a camera that was problematic with noise when you could use a different camera (D100, D70, D1X) that gives you much better exposure latitude for safety.
Call me a horrible photographer for not making a Helmut Newton every frame, or just call it job insurance to buy equipment that I can rely on when things don't go perfectly.
The nice thing about sports is that things are pretty consitent for lighting. It might change game to game, but once you are at the job, hockey doesn't dim the lights in the third quarter.
And I don't care what anyone says... if you have to use ninja noise to give your clients something useable then I think that is insanity. Not when you could have used a camera that wouldn't require it. I mean does the client care that your camera goes "clickclickclickclickclickclick" instead of "click......click..... click?" Nope... they want pictures.
Edited by DragonflyDM on Feb 16, 2005 at 06:49 PM GMT
What it all comes down to is...why limit yourself? Why limit yourself to 4.1MP with noise when 8MP with no noise is available? Can you get your gig done with the 4.1MP noisy camera? At a certain level, sure. I shot with a pair of D2H's for about a year. I got my assignments done, but my shooting style changed as I learned that the camera just couldn't handle low light/high ISO situations. I would strobe more - even if it didn't look too hot - just to get a usable subject. If I saw shots in the shadows I wouldn't bother taking them because I knew that there was no way I could pull them out.
Here's a perfect example of a shot that is impossible to get with a D2H:
[url]http://www.daveharpe.com/public/ups_deice.jpg[/url]
Here's the out-of-the-camera histogram:
[url]http://www.daveharpe.com/public/ups_histo.jpg[/url]
This is an ISO 1600 image from a 1DMkII, handheld, during a sleet storm, through a fence, one and one half stops under from about a half a mile away. The lighting was pure sodium vapor - it's white here because I used a sodium vapor custom white balance. This copy is downsized for the web. The original copy of this shot was actually cropped down to 2,500 pixels from 3,500. No noise reduction has been applied - just levels correction using Adobe RAW.
Just take a walk around this image and see the difference a good lowlight camera makes. You can read the UPS logo on the tail. You can actually make out a worker with a hose in the bucket truck by the tail. You can even resolve the pipes along the boom arm of the bucket truck.
For multiple reasons, this image would have been impossible to get from a D2H. First is noise - the image was shot under pure sodium vapor light. That means a huge amount of gain must be applied to the blue channel to even out the color balance. The D2H shows it's "Bleu Cheese" noise chunks when you jack up the blue channel gain. Noise reduce it out and there's no way you could see the detail shown in this sample. The pipes on the boom truck would be fuzz...the UPS logo would be barely resolvable, etc. For this image there's a little bit of red/blue speckle noise...visible mostly on the tail...easily removable, but there's probably no reason to because it's very minor.
The other reason it's a D2H-impossible shot is cropping. I couldn't afford to use a TC in this situation, and I didn't have anything longer than the 70-200 that I used for this shot. So I shot it as tight as I could get and cropped it later. After cropping almost 30% I still had an image that meets the resolution requirements of the paper I shoot for...they want 2,000 pixels on the widest side minimum. A D2H crop of a similar percentage would take it down to around 1,780. Close, but no cigar.
This particular sample isn't exactly award winning, but it got the job done. I couldn't have gotten the job done with a D2H.
Can anyone tell me if there is a law that says how big or what shape a pixel is?? If all pixels were the same size and shape wouldn't that make it easier for a person to determine what size camera they are buying?? I don't understand how you can compare one camera to another unless it is a film based body. I shot Fuji from when they first came out with the S1 Pro and got 58 files from a 1GB Microdrive and I don't see any better looking shots from either my S1 or the S2 that followed it both with large size pixel counts than from my D2H.The 8X10 and 11 X 14 prints comming from the D2H are beautiful and for me this is where it ends for my printing. Anything larger I would send to a Lab and let them figure out how to juggle pixels. BTW as most of us know the Fuji system used pixels that look like a stop sign (8-sided) so I guess this allowed them to get more into an inch. So what size and shape is your pixel?? Hell I don't know I just take the pictures!!!!
Lou
