Yakim Peled wrote:
As I understand it, the opposite argument is very simple. If you need to compare a vast number of cameras with different sensor sizes and MP count then you need to have a baseline in which all will be judged. That baseline is 100% crops.
100% crops are a staggered baseline. Unfairness is built in. A real baseline is the highest resolving file. It's very easy to compare 10 cameras all with different numbers of pixels. Just uprez them all to the pixel size of the largest one. Then you can look at your 100% crops all you want, and see how the cameras actually compare.
>And as these are internet sites, printing is not relevant.
Even when people buy digital cameras to enjoy their photos on a screen, they enjoy the photos, not 100% crops. So what matters is not the per pixel noise, but the amount of noise in the whole image. I am in the habit of calling this 'per image' noise, and it is the correct metric by which to assess a camera's high iso noise performance.
By and large, the ill-informed P&S buying public is right: more megapixels are better. And the online afficionados are wrong because we are comparing 100% crops before resizing them to equal magnification.
vkalia wrote:
That is poor logic. Just because a method is convenient doesnt make it accurate. Sticking to a standard just for the sake of standardization is called become a slave to process, as opposed to results. It is intellectual bureaucracy.
Brainiac is 100% correct in his assessment, and also re. the long-term implication of such misguided practices on Canon's R&D.
There is absolutely no defense for sloppy methodology, and no reason for us to accept it either.
Another poor argument, b/c as far as Internet sites go, a 4MP 1D will do pretty much everything for web resolutions. People buy 21MP cameras for large prints, and any site that is not prepared to adjust their reviewing methodology to account for this is not worth listening to.
Nobody forces you to accept it but, at least currently, it's a fact of life. Have a better way (remember, you need to make a baseline for thousands of cameras with tens of sensor formants and a variable number of MP)? By all means, go ahead and promote it.
As I see it, Richard's method has one fault. When comparing several cameras, which will be picked as a baseline and why? The one with the higher MP count? The one with the lower MP count? The one with the bigger sensor? The one with the smallest sensor? The newest? The oldest? Other?
And let's say that you made a series of tests, comparing A, B C and D. What happens when E, F and G are introduced? How will they be compared? To themselves? To the first series?
That's why I say that people claiming that the 1D3 is superior to the 1Ds3 in some way are almost certainly saying so because they have been making an enormous error in their assessment - comparing 100% crops at radically different magnifications. I am quite sure that so many people are making this error, reviewers too, that the 1Ds3 has had half as many buyers as it deserves. It is the queen of high iso and the king of resolution and picture quality. By quite a margin.
Edited by brainiac on Sep 10, 2008 at 03:26 PM GMT
Yakim Peled wrote:
Nobody forces you to accept it but, at least currently, it's a fact of life. Have a better way (remember, you need to make a baseline for thousands of cameras with tens of sensor formants and a variable number of MP)? By all means, go ahead and promote it.
It's not a fact of life, it is a contingent and avoidable mass delusion. Just about any way of comparing is better than one which has an explicable and demonstrable bias towards lower resolving cameras. For example, a cruddy 6x4 inch print from each camera at a high street minilab would be a fairer analysis.
> As I see it, Richard's method has one fault. When comparing several cameras, which will be picked as a baseline and why? The one with the higher MP count? The one with the lower MP count? The one with the bigger sensor? The one with the smallest sensor? The newest? The oldest? Other?
How about none? Pick a file size 50% bigger than the largest file, and uprez them all. Or when comparing two cameras, compare two sets of crops: one pair at the higher resolution, and one pair at the lower. But never higher to lower.
> And let's say that you made a series of tests, comparing A, B C and D. What happens when E, F and G are introduced?
There's a function in Photoshop called 'Image size'. Uprez all original files to the new higher rez (or beyond it) if you want to be able to compare. presumably a review site is capable of keeping files on record. This isn't an intelligent objection.
> Can I make myself clearer?
What you've made clear is that you still don't understand the seriousness of the error when comparing a 12 megapixel and 21 megapixel camera. As has been said, there's no excuse for sloppy methodology, especially if you wield as much commercial influence as a site like DPReview.
I accept what you say when you compare 2 particular models but not when you need to have a baseline for thousands of very different cameras in the past, present and future.
That's why I say that people claiming that the 1D3 is superior to the 1Ds3 in some way are almost certainly saying so because they have been making an enormous error in their assessment - comparing 100% crops at radically different magnifications. I am quite sure that so many people are making this error, reviewers too, that the 1Ds3 has had half as many buyers as it deserves. It is the queen of high iso and the king of resolution and picture quality. By quite a margin.
Edited by brainiac on Sep 10, 2008 at 03:26 PM GMT...Show more →
I have not read all the posts in this thread, but this one caught my eye; can you explain to me, simply, what these two shots are intended to illustrate? I cannot discern ANY difference between them on calibrated iMac 20" or Sony Vaio 17" monitors - both images look very poor (and I find the subject matter a bit disturbing!).
Colin Key wrote:
...can you explain to me, simply, what these two shots are intended to illustrate? I cannot discern ANY difference between them on calibrated iMac 20" or Sony Vaio 17" monitors
...is the right thing to say! They are taken with the 1D3 and 1Ds3. They are intended to show that there is practically no difference between the cameras at ISO 3200, except that one is capable of recording quite a bit more detail, which isn't visible at this resolution. This is slightly controversial because many review sites examine per pixel noise instead of per image noise, and then declare that their crops show that the 1Ds3 is a less good performer at high ISOs.
>...both images look very poor (and I find the subject matter a bit disturbing!).
Well, I'm not responsible for the subject matter, but they are both at ISO 3200, and are very competently and kindly taken by Chris Miller, for the sole purpose of examining the differences between the 1D3 and 1Ds3. At ISO 3200 both cameras are rightly considered to be in the top 4 performers money can buy.
Yakim Peled wrote:
I accept what you say when you compare 2 particular models but not when you need to have a baseline for thousands of very different cameras in the past, present and future.
So heavily biased tests are unfair between two cameras, but fair when comparing a handful or more of cameras. How many cameras are needed before heavily biased comparison is OK?
Yakim Peled wrote:
Nobody forces you to accept it but, at least currently, it's a fact of life. Have a better way (remember, you need to make a baseline for thousands of cameras with tens of sensor formants and a variable number of MP)? By all means, go ahead and promote it.
It is not a fact of life. It is accepted because people let it slide, b/c people are willing to accept simplicity instead of accuracy. What Brainiac is doing is trying to correct that.
As I see it, Richard's method has one fault. When comparing several cameras, which will be picked as a baseline and why? The one with the higher MP count? The one with the lower MP count? The one with the bigger sensor? The one with the smallest sensor? The newest? The oldest? Other?
The highest sensor size, obviously. Each time a new camera is introduced, up-res to the highest native resolution.
And let's say that you made a series of tests, comparing A, B C and D. What happens when E, F and G are introduced? How will they be compared? To themselves? To the first series?
Well, you up-res them to the highest resolution, obviously. Yes, it is more work but so what? Is the goal to do a fair comparison or is the goal to simply have a neat little process?
Can I make myself clearer?
You are making yourself clear enough - no one is denying that it is harder to test different outputs at the same output size.
But what you are ignoring is: easier/simpler != accurate. Just because testing methodology A is EASIER to perform or standardize doesnt make it accurate, better or desirable.
Pointing out the complexities involved in doing it properly doesnt make the invalid method better. It may be simpler to compare 100% crops but it is inaccurate. What is the point of a standardized methodology when it is wrong?
Option 1 -- more complex, but gives correct, meaningful results
Option 2 -- simpler but gives inaccurate results
For the life of me, I dont understand how you can defend Option 2. Simple + wrong is still wrong.
If I lose my car keys in the garden at night, does it make sense for me to search for them in my living room b/c there is more light and it is easier to search indoors?
I understand the points that Richard is making. They sound pretty reasonable to me.
I would like an equally clear explanation of why the alternate methodology (100% crops) is valid.
With such information I could make up my own mind as to what I think is the better perspective. Without such information on the rational of the other method (100% crop) I guess I would have no choise but to agree with Richard.
EOSDNG wrote:
My Mark III functions perfectly. If some shots are OOF, that always my fault. I love this pro body and hope, that it will serve me for 4 years or so. Just buy a Mark III, do not waste your time with waiting for a Mark IV.
I wouldn't mind using a MK3, but moving up 2MP from a Mk2n is not enough
for me to upgrade. My Mk2n is fast and works perfectly everytime. But if Canon comes up with a 16MP Mk4 then its a big jump from an 8.2MP. and i'm buying one.
lpazxxsh wrote:
[I wouldn't mind using a MK3, but moving up 2MP from a Mk2n is not enough
for me to upgrade. My Mk2n is fast and works perfectly everytime. But if Canon comes up with a 16MP Mk4 then its a big jump from an 8.2MP. and i'm buying one.
I just can't see why 16MP are needed for what amounts to PJ/Sports camera. I shot T&I today with a 1D2N and a 5D. The N was used for the individual shots and there was simply no need for more than 8MP. The 5D did the team photo but 8, 10 or 12 MPs would have been fine.
When I shoot action 10 to 12MP Max might be OK. I just don't have a need for a 16 to 24 MP body for this use. For wildlife and more detailed portraits the alleged specs for the 5D replacement sound pretty good.
It seems to me that folks are getting too megapixel freaky lately. Give me IQ and high ISO purity. Save the excessive megapixels for the fanboys and techies.
lpazxxsh wrote:
I wouldn't mind using a MK3, but moving up 2MP from a Mk2n is not enough
for me to upgrade. My Mk2n is fast and works perfectly everytime. But if Canon comes up with a 16MP Mk4 then its a big jump from an 8.2MP. and i'm buying one.
I agree, I wish it would have at least 12MP. 16MP is perfect for me. If 1Ds3 would have Medium Raw option just like sRaw, I would never sell it. Because it eats memory fast, and files are huge! Most of the time I don't really need 21MP images and sRaw files are too small... I wish they would made M1-Raw, M2-Raw, S-Raw!!!
JDSA wrote:
I just can't see why 16MP are needed for what amounts to PJ/Sports camera. I shot T&I today with a 1D2N and a 5D. The N was used for the individual shots and there was simply no need for more than 8MP. The 5D did the team photo but 8, 10 or 12 MPs would have been fine.
When I shoot action 10 to 12MP Max might be OK. I just don't have a need for a 16 to 24 MP body for this use. For wildlife and more detailed portraits the alleged specs for the 5D replacement sound pretty good.
It seems to me that folks are getting too megapixel freaky lately. Give me IQ and high ISO purity. Save the excessive megapixels for the fanboys and techies....Show more →
It is good that you don't need anymore than 8MP's. A lot of pro-shooters really don't need anymore than that too. Because the pics they shoot generally are published in newspapers and magazins where 8MP's is more than sufficient. However, to call the need for excessive (what is excessive anyway?) MP's is for fanboys and techies only, doesn't make sense at all. There are other fields of photography where even 21MP's are too little: landscape, commercial and fashion. Wildlife and (certain forms of) portraiture are others, just like you already mentioned before. Even more reason why your last statement doesn't make any sense at all except maybe to blow off some frustration.
More MP's means more choice and more room to crop. You can always down-size. Canon seems to have choosen to go with the high MP's route. Look at the 50D - 15MP's. If this doesn't mean a degradation in performance or IQ I welcome this very much. I like choice and versatility
GeorgeK-NJ wrote:
I'm waiting for the next generation of anything right now. I started off with the 20D when it first came out, and just sold it a few months ago when I picked up my first two 1-series, a classic 1D and a 1DmkII. I have a 5D that I picked up in the Fall of 2006 when they had the Double Rebates, and an XTi that I just have for the size.
I feel in love with the 1D's so my next 1-series will be the successor to the 1DmkIII, not because I feel that the MkIII has issues any more, just because I can wait 6 month or whenever the next one comes out. ...Show more →
same for me (in a smaller way).
started with 30D, now I have 40D but instead of reaching the 50D benefits (aprpeciable for my needs) I would like to own the M3 (or M4 if comes in less than 12 months)
I tried it... .another world
JDSA wrote:
I just can't see why 16MP are needed for what amounts to PJ/Sports camera. I shot T&I today with a 1D2N and a 5D. The N was used for the individual shots and there was simply no need for more than 8MP. The 5D did the team photo but 8, 10 or 12 MPs would have been fine.
When I shoot action 10 to 12MP Max might be OK. I just don't have a need for a 16 to 24 MP body for this use. For wildlife and more detailed portraits the alleged specs for the 5D replacement sound pretty good.
It seems to me that folks are getting too megapixel freaky lately. Give me IQ and high ISO purity. Save the excessive megapixels for the fanboys and techies....Show more →
I don't know what kinds or levels of sports you shoot but my clients are now asking for posters, wall clings and life-size prints from team and individuals and action in daylight, indoors with strobes and available light.
I'll take all the megapixels I can get...plus image quality and high ISO performance. I currently use Mark II's and Mark III's for team photo days and action and they do a great job but am strongly considering the 1Ds Mark III's or 50D's starting with the winter sports season.
Also...the higher megapixel cameras give you much more latitude if a client wants a smaller crop out of an image. If the fan boys and techies are the ones pushing for more resolution and the camera makers listen, that's fine by me.
mill4570 wrote:
Sounds interesting, but I would guess a more modest mp increase....more like 12 to 13 and stay a 1.3 crop. Naturally the better lcd screen, and ISO performance that would surpass the D3 by a full stop. Better in camera noise reduction for the jpeg event/pj/sports shooter. I would guess little or no improvement in frame rate, but a larger buffer and faster clearing. IQ will be better, maybe a true 16 bit. There will be a few toys included like horizon leveling, etc. The AF will be so good the Armed Forces will use it instead of laser guided bombs. Not much better than the MKIII you say? Canon only needs to stop some of the bleeding, they aren't on life support yet.
JimboCin wrote:
I understand the points that Richard is making. They sound pretty reasonable to me.
I would like an equally clear explanation of why the alternate methodology (100% crops) is valid.
My guess as to the argument would be: Pixels are the lowest common denominator and a 100% pixel crop is a realistic representation of the image from the camera. It does provide a realistic representation of the image from a camera, true... but is not comparable across different cameras for the reasons listed above.
and i'm buying one.
I wish they would made M1-Raw, M2-Raw, S-Raw!!!
Not much better than the MKIII you say? Canon only needs to stop some of the bleeding, they aren't on life support yet.