Ad
Ad
Ad
Pages: « 1 2 [3] 4 »   Bottom of Page
Print
Author Topic: Why the D800 is not at all a Milestone  (Read 11738 times)
alba63
Jr. Member
**
Offline Offline

Posts: 66


« Reply #40 on: May 05, 2012, 02:43:22 PM »
ReplyReply

But don't get me wrong, I am not in particular bashing Nikon here and I think that the D800 is probably the best camera in the - well 35mmm heritage shooting gear market place.
But a milestone?

Such a long posting for such a weak message?

If d800 is the best 35mm DSLR, and at a very moderate price, this can easily be called a milestone. Virtually every review so far has pointed out that colour, DR and resolution are superb, many have shown that it rivals MF cameras (much more expensive) in more than one or two respects.
It has often been said that putting more pixels on the same surface in not revolutionary etc. But people expect probably a bit too much. Compare the d800 to a 6-8 year old one, and you may start to see that the amount of evolution over the years in also a revolution.

Bayer Sensors may technically not be perfect, because they have to interpolate colour for the surrounding pixels. But as pixels get smaller and denser, the colour information gets more and more precise, fine detail is better rendered, specially if noise and DR are not getting worse (which is not the case for the d800). In other words: The higher the pixel count goes (noise kept low and DR high), the less relevant the Bayer disadvantage will get. Add RAW converters that keep getting better and better and there is less and less to critizise.

I am not a sensor expert, but I have read - several times - that inside a Foveon based camera the amount of processing and calculating (with highly complex algorithms) is higher than in a Bayer sensor. In theory the Bayer filter approach seems not very perfect, but in practice it works pretty well. And it is telling that noone so far has presented an approach that delivers results that are - all counted in - better.

Bottom of the line seems that in order to get better image quality (in a much more limited types of shooting conditions), you have to invest at least 5x as much money, lenses included even more. To make such a camera available for under 3000 bucks can be easily called a milestone. I have seen detailed comparisions between the Leica S2 and the d800 that showed that the Leica struggles to come out as the better camera: Some slight advantages here, some weaker points there...  If you count in the number of situations (low light, fast shooting required, availability of different and specialized lenses) where the DSLR (d800) is superior, it looks like a real winner here. To ask 15000 for a camera and 4-5000 for a lens (that is optically excellent, as in the Leica S2 system) is hardly revolutionary.

So I think - as most reviewers who tested the d800 so far - that the d800 is pretty much an important milestone. Although I do not have one (yet).

Bernie
Logged
Andreas_M
Newbie
*
Offline Offline

Posts: 17


« Reply #41 on: May 05, 2012, 03:34:04 PM »
ReplyReply

Such a long posting for such a weak message?
What is waek aboutr stating that I do not bash a acamera / firm?

Quote
If d800 is the best 35mm DSLR, and at a very moderate price, this can easily be called a milestone.

I don't know if I agree - because this means that we have a milestone every half year or so...well, I guess that it will take probably a bit longer in the case of the D 800, but anyway: As soon as we have something either better in IQ or similar but cheaper - we have a new milestone?

Quote
I am not a sensor expert, but I have read - several times - that inside a Foveon based camera the amount of processing and calculating (with highly complex algorithms) is higher than in a Bayer sensor. In theory the Bayer filter approach seems not very perfect, but in practice it works pretty well. And it is telling that noone so far has presented an approach that delivers results that are - all counted in - better.
Why are so many people try to "nail" my statement to Foveon? I said it clearly and I repeat: I do not believe that Foveon, like we know it today, is an alternative.
Stepping back to your statement about Bayer issues becoming less and less of a real world issue with smaller sensel size (if I got your statement right?), which I agree to, the question has to be asked why no one apart from Fuji has the idea to do something different - the XTrans is still a Bayer CFA - but taken to a different level. I don't know if it works and Fuji will most probably not make any Full Frame sensor (so we have no real comparison), but if this works and avoids some of the major Standard - Bayer CFA shortcomings, then this could well be a milestone. Better resolution at less nominal pixel, simply because the used Bayer pattern allows to not use AA and to have better de-Bayer algorithm seems to be a natural approach to me, as closer we come to sensel sizes taht become harder and harder to manager (also in terms of lens resolution and so on) - don't you think so?


Quote
Bottom of the line seems that in order to get better image quality (in a much more limited types of shooting conditions), you have to invest at least 5x as much money, lenses included even more. To make such a camera available for under 3000 bucks can be easily called a milestone. I have seen detailed comparisions between the Leica S2 and the d800 that showed that the Leica struggles to come out as the better camera: Some slight advantages here, some weaker points there...  If you count in the number of situations (low light, fast shooting required, availability of different and specialized lenses) where the DSLR (d800) is superior, it looks like a real winner here. To ask 15000 for a camera and 4-5000 for a lens (that is optically excellent, as in the Leica S2 system) is hardly revolutionary.

I agree.
I would be interested to read this comparison - do you have a link?

Logged
BJL
Sr. Member
****
Offline Offline

Posts: 5081


« Reply #42 on: May 05, 2012, 03:54:00 PM »
ReplyReply

Why are so many people try to "nail" my statement to Foveon? I said it clearly and I repeat: I do not believe that Foveon, like we know it today, is an alternative.
People are probably using "Foveon" in the general sense of "X3", measuring three colors at each location, as opposed to using the Bayer CFA approach. And they probably think that you are a zealous partisan in favour of Foveon _style_ as superior to the Bayer CFA approach because of your flood of nasty comments about the Bayer CFA approach and suggestions that the X3 approach and not interpolating, as with Foveon, is the way to go. That and your comments dismissing or ignoring the huge technological improvements in sensors made over the last decade, apparently because simply staying with the Bayer CFA means to you that a company like Sony is not innovating.

Subsequent revisions and retractions of your claims in the face of refutations do not entirely release you from the tone set by comments like:

So here we are, it is 2012, everybody is freaking out about a camera that puts 36 million pixels onto a single chip of (roughly) 35 mm. But - and this is my point - nothing has changed:
- We still have a stupid Bayer filter in front of the sensor
- As advanced as it may be - we still do stupid interpolation to get the full color information
- We still live with strange artefacts - or use an AA filter

... Given the advances in sensors, I see no reason that we could not have something more advanced than Bayer. Foveron shows that it is possible - even the Fujis XTrans is at least a bit more modern.

I for one am sick of this blowing up a device that is essentially flawed. ...

... milking the dead cow called Bayer is a guarantee to nake money.

But as a matter of fact the Bayer filter, at least as we know it, is in fact not an ideal solution:
It relies on interpolation, which is in other words an educated guess of what the color and luminace of the pixel has to be.

...

But the effect is very clear: Bayer does spoil the resolution and the color fidelity. Depending on the de - Bayer filter, you will get more or less half of the nominal resolution - sometimes a bit more.

You do so a the cost of
- processing power: Do get good de - Bayer results you need at least a 7x7 kernel and a Laplace filter - in addition to many other tricks the companies do. This is an incedible processing load.
Ok we have the fast Asic's like Digic, Expeed - whatever they are called. But it remains somewhat questionable if you could not use that power for something else - of have smaller (read less power hungry) Asic's for extended battery life.

- producing artefacts: Not much to say here. Everybody knows about the Moiree problem. Some are affected more, other less

- Waisting sensor reel estate. YES, this is what they do: Insted of having somehow 3 colors in one place, you need 3x3 pixel to have all colors. I call this less than clever...

« Last Edit: May 05, 2012, 04:00:19 PM by BJL » Logged
Andreas_M
Newbie
*
Offline Offline

Posts: 17


« Reply #43 on: May 05, 2012, 04:16:50 PM »
ReplyReply

Good point - overlooking one important sentence in the last quote YOU selected:

"But as a matter of fact the Bayer filter, at least as we know it, is in fact not an ideal solution:"

At least as we know it is the key here. Not too hard to understand, I thought.

And by the way - I never said that X3 in the sense of three layer is the way to go. It might or might not be. Fuji had some years ago the idea to have one big and one small sensel under one Micro lens for increased dynamic - and it worked to some extend.
I have to be careful, that is what I learn here, so let me make it clear: I do not say that XTrans or a kind of SuperCCP (would be called SuperCMOS today) is the RIGHT way - who am I to say that.

All I say: I am worried that the big guys (and Canon at the forefront right now) are not looking enough out of the box (or - maybe that is wrong alltogether, maybe they do look out of the box and are simply not ready yet - the point is: Yet another conventional Bayer CFA camera is something I have a hard time to understand to be a milestone - it is for sure a great camera, to make this also - again - clear)
Logged
jeremypayne
Guest
« Reply #44 on: May 05, 2012, 04:17:04 PM »
ReplyReply

What is waek aboutr stating that I do not bash a acamera / firm?

You have been bashing the D800, Nikon, Canon and pretty much every other manufacturer except Fuji and Sigma.

What is weak is your understand of the topics at hand.

Calling the Bayer CFA approach to digital imaging "stupid" is ... well ... stupid.



Logged
BJL
Sr. Member
****
Offline Offline

Posts: 5081


« Reply #45 on: May 05, 2012, 04:47:38 PM »
ReplyReply

"But as a matter of fact the Bayer filter, at least as we know it, is in fact not an ideal solution:"
Not being "ideal" is no basis for any of your criticisms of the main sensor makers, because it is consistent with being the best method available, or even being the best method that is physically possible. The constraints of physical reality do not always allow an "ideal" solution to our problems. One of those physical constraints is that the photo-electric effect is almost color blind: it will detect photons of all colors at roughly the same rate. It is even harder to get the photo-electric effect to detect one color but not another of shorter wavelength. For example, we might dream of a sensor with one layer that detects all red photons but lets all blue and green ones pass through, followed by a layer that detects only green, and so on, but that might not be physically possible.

Do you have any evidence that a better solution is possible, or that camera companies could have a better method now if they had worked hard enough on finding one? It is not enough to point to the disadvantages of one method, because it might be that any other method without those disadvantages has other, greater disadvantages.
Logged
Andreas_M
Newbie
*
Offline Offline

Posts: 17


« Reply #46 on: May 05, 2012, 05:22:01 PM »
ReplyReply

You have been bashing the D800, Nikon, Canon and pretty much every other manufacturer except Fuji and Sigma.

What is weak is your understand of the topics at hand.

Calling the Bayer CFA approach to digital imaging "stupid" is ... well ... stupid.





Thanks :-)
Point taken.
But will all respect - not thinking out of the box is also, well at least not too clever.

But I agree, my wording was not precise enough and this is a valid point: I should have said Bayer CFA as we know it.

The first part of your statement is what I do not at all agree to. I do not call it bashing, if I point out what I -feel- might be wrong. I don't and I think I made the pointseveral times that I am sure that the D800 is a great camera. That's no bashing.
If you want me to I could bash Fuji and in fact I have - in your terms - bashed Sigma.

I can't understand what the problem is if someone says that the D800, being a great camera, is not a milestone, simply because it does not introduce any new technology but instead blowing up an Exmore sensor, much of the kind we know from APS C cameras like the Nex 7, A77 or Nikon D7000 - with great results, for sure.
Why do so many users here take that so - well, nearly personal?

I just wanted to express my surprise that so many people talk about a milestone - and it's not more than a very well executed, very predictable evolution of the Exmore sensor family.
Logged
Andreas_M
Newbie
*
Offline Offline

Posts: 17


« Reply #47 on: May 05, 2012, 05:40:01 PM »
ReplyReply

Not being "ideal" is no basis for any of your criticisms of the main sensor makers, because it is consistent with being the best method available, or even being the best method that is physically possible.
Precisely this is the point I am questioning.
Don't get me (again) wrong - I am not talking about a X3 solution (at least not as we know it today) - this might still be an option in the long term (or a dead end - which is probably more likely today), I am talking about new ways that help us getting rid of some of the standard Bayer shortcomings.

Quote
The constraints of physical reality do not always allow an "ideal" solution to our problems.
Your definition of ideal is a bit different than what I had in mind: You define ideal as a physical ideal, right? And there is no dissagreement betwen us in this aspect: There is no ideal solution to our problems and probably there will never be one.
There is another definition of ideal: Ideal in the sense of the best possible compromise. That's what I meant and I just expect comapnies to look for that kind of solution.
Now, you might still believe that standard BAyer CFA is the best solution. Maybe you are right, but I have my doubts.

Quote
One of those physical constraints is that the photo-electric effect is almost color blind: it will detect photons of all colors at roughly the same rate.
True. But this is not the point. It is all about the question if a standard pattern of G-B-G-B in line one and R-G-R-G in lien two (and so on) is the most effective way. Do not forget: This pattern has been primarily developed to match monitor outputs. We are talking about prints, aren't we? A different pattern might be a good idea.

Quote
Do you have any evidence that a better solution is possible, or that camera companies could have a better method now if they had worked hard enough on finding one? It is not enough to point to the disadvantages of one method, because it might be that any other method without those disadvantages has other, greater disadvantages.

A good and valid point.
To your first rethoric question: No, I have no evidence that something else will work better. But I have two other things: A lot of professional experience with the standard Bayer CFA dilema. This is not in photography, I must admit, but in Machine Vision.
The other thing: I have hope. Hope that things like Xtrans will work, finally.
Logged
BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3012


« Reply #48 on: May 05, 2012, 07:03:26 PM »
ReplyReply

Now, you might still believe that standard BAyer CFA is the best solution. Maybe you are right, but I have my doubts.

Hi Andreas,

Doubts, why??

Quote
It is all about the question if a standard pattern of G-B-G-B in line one and R-G-R-G in lien two (and so on) is the most effective way. Do not forget: This pattern has been primarily developed to match monitor outputs.

Says who?

Quote
To your first rethoric question: No, I have no evidence that something else will work better. But I have two other things: A lot of professional experience with the standard Bayer CFA dilema. This is not in photography, I must admit, but in Machine Vision.

Ah, that explains a lot. Machine vision is all about making relevant differences (more than) visible, not about a rendering that closely models human vision, does it?

Cheers,
Bart
Logged
BJL
Sr. Member
****
Offline Offline

Posts: 5081


« Reply #49 on: May 05, 2012, 07:23:42 PM »
ReplyReply

To your first rethoric question: No, I have no evidence that something else will work better. ... I have hope.
That is all that needs to be said really, and I think we will all agree that it is good to hope for a truly new and better idea than the Bayer CFA, or any CFA.  Which has absolutely nothing to do with any denigration of the progress that is or is not represented by the D800, because there are many other ways that camera technology can be improved that do not inolve getting rid of the Bayer CFA.

I will stick my neck out and guess that the improvement we have seen over the last decade from the passive sensor approach of the CCD to today's best active pixel CMOS sensor technology is greater than any improvement possible by changing the CFA arrangement (as Fujifilm has tried on several occasions) or even by eliminating the need for a CFA entirely.
Logged
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1615


« Reply #50 on: May 06, 2012, 10:42:15 AM »
ReplyReply

If you are shooting B&W landscape at base ISO to jpeg and never adjust the color contribution, Bayer might be a poor choice. Just stripping the CFA and removing or reducing the AA filtering would give you more resolution, and more photon efficiency.

If you are pushing extreme low-light photography to the point where colors does not matter, Bayer might be a poor choice.

If you are shooting flowers in a vase in a studio and want the highest possible spatial resolution, you might want to get rid of the cfa and use a color wheel.

If you use your camera for machine vision, you may want to optimize the camera for that application.


For most of us, Bayer seems to be a rational and economic method to give us widely usable cameras at a given technological level.

-h
Logged
BernardLanguillier
Sr. Member
****
Offline Offline

Posts: 7523



WWW
« Reply #51 on: May 08, 2012, 01:10:54 AM »
ReplyReply

It may not be a milestone, but it helps me capture challenging images more easily than any other cameras before it and that is the only thing that really matters, is it not?



Cheers,
Bernard
« Last Edit: May 08, 2012, 01:15:17 AM by BernardLanguillier » Logged

A few images online here!
Fine_Art
Sr. Member
****
Offline Offline

Posts: 890


« Reply #52 on: May 08, 2012, 03:25:16 AM »
ReplyReply

Looks good to me. In fact it is well off the chart. Look at the measuring wheel with fine lines on the right. Compare to other cameras.

http://www.imaging-resource.com/PRODS/nikon-d800e/FULLRES/D800EhSLI00050.HTM
Logged
AJSJones
Sr. Member
****
Offline Offline

Posts: 353



« Reply #53 on: May 10, 2012, 03:06:54 PM »
ReplyReply

So here we are, it is 2012, everybody is freaking out about a camera that puts 36 million pixels onto a single chip of (roughly) 35 mm. But - and this is my point - nothing has changed:
- We still have a stupid Bayer filter in front of the sensor
- As advanced as it may be - we still do stupid interpolation to get the full color information
- We still live with strange artefacts - or use an AA filter

We get a milestone every mile Cheesy

Even the human eye has not progressed beyond using an array of three different kinds of color sensor cells and it does a massive amount of (stupid?) interpolation, interpretation and even compression (like mpeg) too, before it ever sends the information to the brain Cheesy  The brain just suppresses what would otherwise be called flaws.  Seems like digital is doing quite well in getting the eye the right information - we just have to process it correctly.

Now, what was your question, again?
Logged
jeremypayne
Guest
« Reply #54 on: May 10, 2012, 03:54:34 PM »
ReplyReply

We get a milestone every mile Cheesy

Even the human eye has not progressed beyond using an array of three different kinds of color sensor cells and it does a massive amount of (stupid?) interpolation, interpretation and even compression (like mpeg) too, before it ever sends the information to the brain Cheesy  The brain just suppresses what would otherwise be called flaws.  Seems like digital is doing quite well in getting the eye the right information - we just have to process it correctly.

Now, what was your question, again?

Exactly.
Logged
John Nollendorfs
Sr. Member
****
Offline Offline

Posts: 297


« Reply #55 on: May 11, 2012, 10:44:20 AM »
ReplyReply

You have been bashing the D800, Nikon, Canon and pretty much every other manufacturer except Fuji and Sigma.

What is weak is your understand of the topics at hand.

Calling the Bayer CFA approach to digital imaging "stupid" is ... well ... stupid.





I think what Andreas is meaning to imply is that short of Fuji and Sigma--Canon, Nikon & Sony are not trying new imaging technologies, but only refining existing ones. I hope I have summarized the "crux" of your statements when applied to the D800, Andreas?

Pure research into imaging for companies  such as Canon, Nikon & Sony is incredibly expensive. It is much easier and logical for them to further refine and develop existing technologies. (Their way of thinking) But it is interesting that a European cell phone company would invest and develop a new imaging technology with a 44 MP sensor, and unique processing techniques for that sensor!
Logged
AJSJones
Sr. Member
****
Offline Offline

Posts: 353



« Reply #56 on: May 11, 2012, 01:18:47 PM »
ReplyReply

I think what Andreas is meaning to imply is that short of Fuji and Sigma--Canon, Nikon & Sony are not trying new imaging technologies, but only refining existing ones. I hope I have summarized the "crux" of your statements when applied to the D800, Andreas?

Pure research into imaging for companies  such as Canon, Nikon & Sony is incredibly expensive. It is much easier and logical for them to further refine and develop existing technologies. (Their way of thinking) But it is interesting that a European cell phone company would invest and develop a new imaging technology with a 44 MP sensor, and unique processing techniques for that sensor!

The Nokia 41MP still uses a Bayer array, but they have gone further down the road of processing, as you say.
Quote
Since each pixel in a finished 5MP image on the 808 PureView gets data from eight different pixel locations on the sensor, a value for red, green, and blue can be determined without extra interpolation. Furthermore, noise caused by some pixels returning random or inaccurate values is effectively averaged out by combining data from more than one pixel location. This results in more accurate color and reduced noise overall.
Additionally, oversampling reduces the softening effects of the Bayer filter, the anti-aliasing filter, and diffraction of the tiny lens, resulting in sharper images.

 I think Andreas is just frustrated that no-one has yet come up with something much better than the Bayer array (or CFA in general) and got it to work well.  Until something comes along to make it look stupid (in retrospect, of course), we are stuck with it as the smartest thing around (barring 3 separate exposures with 3 filters etc).  That doesn't mean they're not working on it, though...
Logged
t3hh
Newbie
*
Offline Offline

Posts: 9


Marek T3hh


WWW
« Reply #57 on: May 20, 2012, 08:46:44 AM »
ReplyReply

Looks to me that DRReview guys are starting to trickle into LuLa.
Or, in other words, tire kickers are starting to kick in. Wink
EDIT: Correction: DPReview, of course
what a useful and intelligent comment! really represents how different is LuLa forum from DPR...
Logged
BernardLanguillier
Sr. Member
****
Offline Offline

Posts: 7523



WWW
« Reply #58 on: May 20, 2012, 08:58:23 AM »
ReplyReply

I think what Andreas is meaning to imply is that short of Fuji and Sigma--Canon, Nikon & Sony are not trying new imaging technologies, but only refining existing ones. I hope I have summarized the "crux" of your statements when applied to the D800, Andreas?

The same could be said about most car manufacturers having done little to go beyond our current engines technologies.

The thing though is that those companies are running different streams of activities in parallel. They have to make tough calls in terms of selecting what technology to use in a given product.

The focus is not innovation, it is value for customer and stakeholders.

Sigma tried with Foveon, I believe they have lost a lot of money in this endeavor, perhaps enough to threaten the future of the company. Was their focus on Foveon interesting for those interested in sensor technology? Amazingly interesting. Was it a decision stakeholders are happy about? That's a lot less obvious.

So unless you consider Canon and Nikon to be research entities, I am not sure they can be blamed for the way they conduct their business.

Cheers,
Bernard
Logged

A few images online here!
Slobodan Blagojevic
Sr. Member
****
Offline Offline

Posts: 5000



WWW
« Reply #59 on: May 20, 2012, 11:59:14 AM »
ReplyReply

what a useful and intelligent comment! really represents how different is LuLa forum from DPR...

And you would know it after five anonymous comments?
Logged

Slobodan

Flickr
500px
Pages: « 1 2 [3] 4 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad