Full Frame Fallacy?

[mpb]

In response to previous posts about larger sensors always being better, hasn't digital technology over the years shown us that things always get better and "smaller?" Certainly, fabrication and large-scale integration has no limit with regard to size at all. Look at microprocessors and chips in general. There should be no reason why a 4/3 sensor can't outperform an FF.

It also amazes me that everyone has fallen for the size and megapixel marketing ploy so easily, while ignoring the things that 'really' have to do with image quality...such as color accuracy, processing algorithms and lens characteristics. I believe the only reason FF exists is to support 35mm lenses, many of which are still around. Anyone else share these views?

 

[Day Hiker]

35mm FF is a format with unique characteristics. The same applies to the various medium format frames, large format, 4/3, and P&S sensors. It is here to stay and serve those who want/need its unique characteristics. I share the belief that anything that can be done to improve a 4/3 sensor can also be done to improve any other sensor, so 4/3 will never outperform 35mm FF. It may, for all intents and purposes, someday match 35mm FF, but never outperform it.

--
Cheers,

Jim Pilcher
Colorado, USA

'Begin each day as if it were on purpose.' -- Mary Anne Radmacher

 

[knoblock]

It is like asking for improvements in film technology to make 35mm equal to medium format film. If the same improvements are applied to larger pieces film (and there is nothing stopping this...it is all film) medium format will always have more resolution, more image quality. 35mm is a sweet spot for a variety of reasons, portability, depth of field, field of view, lens availability, etc. 4/3s is a sweet spot and may equal today's 35mm format sensors some day, but they will use the same technology to surpass 4/3.

It doesn't matter that 35mm sensors are better if you don't need them. If 4/3 can deliver the same quality up to 11x14 prints, you don't need 35mm format. If everyone abandons printing for sharing on the web, all this is pretty meaningless other than for low light capability and other factors. It might be that 35mm suffers the fate of the marketplace and not 4/3.

The images Louis posts show the E3 and D3 fairly close in producing web sized images of good quality. I think the D3 has some nice qualities and might favor it over the E3 were money no object. I like the look that seems to come from increased photosite signal to noise, so either a D3, E3 or even E500 with lower MP and larger photosites produces attractive images (except for the noise in some shots) with the same smooth tonalities. When I look at the 510 images of mine that have the E3 look, they are narrow DR images with the histogram not reaching the ends.

--
Steve

http://www.flickr.com/photos/knoblock/
http://picasaweb.google.com/steve.knoblock

Film will only become art when its materials are as inexpensive as pencil and paper. -- Jean Cocteau

 

[marf]

huge full frame lenses sometimes match 4/3 lens quality,but will never be better [nt]

 

[SphericalAberration]

Even though it may not affect every-day photography or the usefulness of the camera to 99.9% of photographers, compromises have been made.

The mighty Nikon D3 seems to be cheating more than a little applying NR to RAW files.

There are some quite shocking revelations, pertinent to astro-photography here:

http://www.astrosurf.com/buil/nikon_test/test.htm

"The readout noise of the Nikon and Canon CMOS detectors is very similar. The high ISO sensitivity displayed by Nikon is for the marketing:"

"the displayed sensitivity of 6400 ISO for the Nikon D3 #1 corresponds to a sensitivity of 600 ISO for the Canon 40D"

"For Nikon, the hot pixels are eliminated by a sophisticated digital processing external to the sensor. During this digital processing, the signal of the neighboring pixels is also affected. The damage of such processing is well-known in astronomy: the weak stars are also eliminated and the image loose photometric qualities on stellar like objects."

"For Canon, the thermal signal is reduced for each pixel by a differential reading method. The thermal signal level measured at the output of the sensor is very low. The residual can efficiently be removed during the image processing (a simple substraction of a reference dark signal map)."

"It is tragic to see that Nikon solved the problem of thermal signal by a digital processing of the RAW files (i.e. NEF files do not contain true raw data)."

 

[Victor S]

35mm FF is a format with unique characteristics.

Really? Can you tell us what they are?

In response to previous posts about larger sensors always being
better, hasn't digital technology over the years shown us that things
always get better and "smaller?" Certainly, fabrication and
large-scale integration has no limit with regard to size at all. Look
at microprocessors and chips in general. There should be no reason
why a 4/3 sensor can't outperform an FF.

The idea that "fabrication and large-scale integration has no limit with regard to size at all" is simply not correct, we just haven't reached the limit yet. How can you possibly build circuits that are smaller than the molecules you need as parts?

Beyond that, as sensor pixels get more tightly packed they generate more noise. A 4/3 sensor has about four times the density of a 35mm FF sensor. Improved sensor designs can only reduce the effect; less dense sensors will still generate less noise.

Lenses have limits too. Why else would the largest telescopes have mirrors meters across (they use mirrors because if you build a lens that large it will sag under its own weight).

My previous camera, a Konica Minolta A200, has a bunch of features that are better than my e510, but its smaller sensor just can't match the e510's image quality. My 4/3rds system e510 provides everything I need, but might not be acceptable for someone else's applications.

 

[Day Hiker]

...resolves over 125 lp/mm. That outclasses many, if not all, 4/3 lenses.

It all depends upon how much $$$ the mfgr is willing to invest in production. Leica doesn't care that much and has sold at least 17 of those 75mm jewels in the past 3 years.

Also, Sigma shows us how poor 4/3 lenses can be. I use the 55-200mm as my example. A large majority of 35mm lenses will outperform that dog.

--
Cheers,

Jim Pilcher
Colorado, USA

'Begin each day as if it were on purpose.' -- Mary Anne Radmacher

 

[peepingtom]

35mm FF is a format with unique characteristics. The same applies to
the various medium format frames, large format, 4/3, and P&S sensors.
It is here to stay and serve those who want/need its unique
characteristics. I share the belief that anything that can be done to
improve a 4/3 sensor can also be done to improve any other sensor, so
4/3 will never outperform 35mm FF. It may, for all intents and
purposes, someday match 35mm FF, but never outperform it.

I'm sure that's true enough - there will always be those who hanker after the past and believe that their FF lenses will perform optimally on a FF sensor. Unfortunately that's not near the truth. There are some fundamental problems with FF sensors, and the manufacturers know this (even if the users don't). You'll see more and more 'digital ready' lenses appear over time to try and correct this (at least from Nikon, Canon still have their head in the sand). Olympus have it easy - a true digital system where the lenses perform optimally for the sensor.

 

[mpb]

In response to previous posts about larger sensors always being
better, hasn't digital technology over the years shown us that things
always get better and "smaller?" Certainly, fabrication and
large-scale integration has no limit with regard to size at all. Look
at microprocessors and chips in general. There should be no reason
why a 4/3 sensor can't outperform an FF.

The idea that "fabrication and large-scale integration has no limit
with regard to size at all" is simply not correct, we just haven't
reached the limit yet. How can you possibly build circuits that are
smaller than the molecules you need as parts?

Apologies. What I should have said was, "practically" no limit with regards to size. You're right, but I don't think I was talking about going "that" small. The point I was arriving at: at some stage there should be no image quality difference based on size.

 

[Day Hiker]

I wrote:
35mm FF is a format with unique characteristics.
You wrote:
Really?

C'mon, are you that naive? Substitue "35mm FF" with "6x9", "6x7", "6x6", "6x4.5", "35MM half frame", "8mm sub miniature", "4x5 large format", "4/3", "APS-C", etc. and my statement is still correct.

Each format has its unique blend of characteristics that lend themselves to particular styles of photography. If I want more control over shallow DOF, I move to a larger image format (regardless of where I am now). If I want a 2:3 aspect ratio, I will choose 35mm FF or 6x9 MF. If I want the ultimate in grainless B&W landscape photographs, I will choose 8x10 (or larger) large format photography. Panoramas? The 6x17 format is excellent.

And the accessories surrounding a format make it even more unique: 35mm has it hands-down over larger formats if you want zoom lenses. It's also the place to go for prime lenses (at this time) if you shoot in a smaller format. 4x5 gives great control with tilts and swivels (the 35mm T/S lenses cannot compare). The bellows and macro options for 35mm FF far exceed any other format's offerings.

As a 4/3 system user, acquiring a 35mm FF digital camera will give me more control over DOF than I have now. I find that alluring. There are times that the 3:2 aspect ratio is attractive, which I cannot achieve with 4/3 unless I crop. In the digital era, 35mm FF offers different sensors that exhibit different color pallates, ISO sensitivity, and dynamic range than 4/3 has.

Each format is unique.
--
Cheers,

Jim Pilcher
Colorado, USA

'Begin each day as if it were on purpose.' -- Mary Anne Radmacher

 

[leandrod]

I share the belief that anything that can be done to
improve a 4/3 sensor can also be done to improve any other sensor, so
4/3 will never outperform 35mm FF. It may, for all intents and
purposes, someday match 35mm FF, but never outperform it.

Agreed — but if it matches 35mm (which by the way should be either called double frame in reference to the original vertical-oriented-film cinema frames; or just 24×36, without any reference to a frame which was never a formal standard anyway), then compactness and standardisation will tend to make it a hit, unless one needs unique stuff such as fast, cheap wide angulars; and even so the ‘cheap’ part of it may fade away if market changes enough.

 

[Clement Noise]

All manufacturers are now re-designing and introducing digital FF lenses.
The Olympus advantage is being eroded.

You can say that anything you can do to a 4/3 lens can be applied to a FF lens just like you can with sensors.

While Olympus may have the digital lens advantage now....Nikon, Canon, Pentax, Sony, Tamron, Tokina, etc... are all introducing digital compatible lenses and this will eliminate the designed for digital argument that Olympus has been touting.

The other advantage that Olympus DOES NOT HAVE is that the competition allows the use of their FF digital lenses on small and FF sensors. Olympus is painted into a corner with their smaller sensor.

I like 4/3 for now. I got a couple of cheap olympus cameras and lenses because Olympus is always forced to drop prices when their sales stall and they are unable to create demand because they have no money for advertising and they have no discernable marketing department. Who can beat that value! No one in the market can give you a 10MP, 2 lens kit with IS at those prices. WOW!!!

Now, when there is a FF camera at a price I can justify, I will dump 4/3 except for my 410 because it is perfect for hiking and gives me better IQ than a P&S camera.

I keep an open mind and not get let myself get fixated on any fanboy propaganda.

The other thing to keep in mind about your post is physics, wavelength, diffraction, etc.... A smaller sensor will have more problems with this than a larger sensor with larger pixels. And I do like higher resolution because you get better tonal gradations, you can crop a 3:2 sensor to 4:3 if you want to and still get high resolution. And I see more detail at a given enlargement with higher resolution.

the 410/510 are sweet spots in the industry right now. I can see no reason for someone not to pick up this great value, just know going in that you are making some compromises and need to work within the limits of your camera.

Be happy shoot pictures!

 

[joe mama]

In response to previous posts about larger sensors always being
better, hasn't digital technology over the years shown us that things
always get better and "smaller?"

Bigger sensors are not always better, of course, as it depends on their design. But they always have the potential to be better. So, yes, things do tend to "always get better", but not always because they are smaller.

Certainly, fabrication and large-scale integration has no limit with regard to
size at all. Look at microprocessors and chips in general. There should be no
reason why a 4/3 sensor can't outperform an FF.

In fact, there is. Are you suggesting that a microprocessor that is four times the size as another microprocessor will not outperform the smaller chip if both are using the same generation of technology?

So, yes, a 4/3 sensor can outperform a FF sensor, but that's only if the FF sensor has a vastly inferior technology.

It also amazes me that everyone has fallen for the size and megapixel
marketing ploy so easily, while ignoring the things that 'really'
have to do with image quality...such as color accuracy, processing
algorithms and lens characteristics. I believe the only reason FF
exists is to support 35mm lenses, many of which are still around.
Anyone else share these views?

Many people share your views, but the facts don't bare them out. It's true that sharper lenses are more easy to manufacture for smaller sensors than larger sensors, and that modern 4/3 lenses outperform older FF lenses. However, the 4/3 sensors need to outperform the FF lenses since they have a much higher pixel density. In fact, we can be quite specific: a 4/3 lens needs to be twice as sharp as a FF lens for the same number of pixels, since the pixels will be half as long and high.

Now, as FF sensors get higher and higher pixel densities, eventually equalling those of smaller sensors, such as 4/3, then they, too, will need sharper glass to take advantage of the extra pixels, and that's what's being done with the new FF lenses (e.g. Nikon 14-24 / 2.8 and Canon 70-200 / 4L IS).

However, even if some older FF glass is unable to resolve more than, say, 16 MP on a FF sensor, all that means is that more than 16 MP cannot be taken advantage of, not that the IQ is less.

The reality is, of course, that the sharpness of the glass is not even across the frame, so what we see is that FF glass will still be able to resolve the central portion of the image for even much higher MP counts, but the edges will not. Conversely, as the pixel density increases, the FF sensor will be able to simply crop out the central portion of the frame and still have the same number of pixel as 4/3 on the composition, but from the sharper image area of the lens.

Nonetheless, as sensors of all sizes continue to have higher IQ, the higher IQ of FF becomes unnecessary as people still print at the same size. Under those cricumstances, the only advantage of FF will be shallow DOF and ultra low light pics. In fact, we've already reached that point, in my opinion. That is, at 8x12 and smaller, with the exception of shallow DOF and high ISO, the differences between FF and smaller formats is small, if not trivial. Of course, it depends on a person's QT (quality threshold). But for the vast majority of pics I see posted in these forums, it's difficult for me to imagine that the technology matters in terms of IQ; it's the operation of the equipment that matters more, by far.

--
--joe

http://www.josephjamesphotography.com
http://www.pbase.com/joemama/

 

[marf]

1. leica has no full frame digital camera so that lens is no benefit to the full frame fetishists ( which are usually canon marketing victims ).

2. the sigma- 4/3 lenses are APS lenses with a 4/3 mount

--
regards
Martin

-----------------------------
Typing errors are intended to provide a basis for global amusement.
I am a part time wedding photographer (50% of my income)

 

[Franka T.L.]

with the same technology in sensor design. any sensor mfr can had a sensor with larger real estate to had

1. - more photosite
2. - larger photosite

or both ... and the reason behind the FF is not just for the lens compatibility although a few years back that was the primary concern. This format had pretty much allow the capability to stay portable, but yet allow very decent image control ( aspect of lens / image / sensor / ) where the balance is just right for the top end user.

If people just blindly believe that FF is just for the sake of the old film format, then Medium format and even large format mde even less sense as per OP's logic. Truth though is they all got their place in the digital photography world, just that they might nit be for everyone and that FF is prime to go mainstream ... In any case the Canon 5D had shown how a ff can be in term of size and package.

I had to say on this 4/3 forum, there are too many self induced romance towards the 4/3 format. I for one being multiple format / multiple system user hold no such. All format had their good and bad, and its no use keep trashing the othee mfr for their choosen format. Instead let's see oly and 4/3 prove their own right in real product , in the field , and in the market ...

--

• Franka -

 

[Jonathan Demarais]

All manufacturers are now re-designing and introducing digital FF
lenses.
The Olympus advantage is being eroded.

I don't think it will be, completely. Lenses designed to support a FF sensor by using the telecentric design will likely (if their quality is to be high) cost a lot more than film lenses of the same focal length and speed. Even with cheap molded aspherics and inexpensive ED glass. Likely the wide angle stuff will be cheaper from Olympus at least and probably overall higher quality. Everytime I try a kit lens or most wide angle lenses on another brand with a 1.5 or FF sensor, the results disappointing compared to Olympus. It's hard to go from an essentially aberrationless image to having to tolerate colour error and visible edge blurring.



'I cried because I had no E-3. Then I met a man with no E-510'

 

[Day Hiker]

1. leica has no full frame digital camera so that lens is no benefit
to the full frame fetishists ( which are usually canon marketing
victims ).

Hi Martin,

That lens resolves over 125 lp/mm regardless of the sensor/film behind it. It was designed for 35mm FF though. A 35mm FF lens can be damn good, that's the only point I was making. And, BTW, that lens is a benefit to FF fetishists if they are shooting an M film camera.

2. the sigma- 4/3 lenses are APS lenses with a 4/3 mount

I suspect the 55-200mm Sigma is a dog in the APS-C format too.

--
Cheers,

Jim Pilcher
Colorado, USA

'Begin each day as if it were on purpose.' -- Mary Anne Radmacher

 

[marf]

1. Imagine how good a Fourthirds Prime lens for $2500.- can be.

2. The Leica System is good but no SLR or DSLR system. You cannot compare the DSLR world with the Non-DSLR world.

3. Ok, maybe long prime lenses - there an advantage for 35mm may exist. But most people do not shoot long prime lenses, and if someone prefers long prime lenses then medium format would be even better.
In the more relevant areas

• Zoom lenses
• Wide to medium prime lenses

Fourthirds has the advantage

4. THe lenses for a system are more important than the sensor for a system

5. Film is not digital. We are here speaking of digital. I know that in the film world 35mm is better than 4/3 formats. But in the digital world its the other way, if you count the performance of the system lenses.

--
regards
Martin

-----------------------------
Typing errors are intended to provide a basis for global amusement.
I am a part time wedding photographer (50% of my income)

 

[Big Ga]

There are some quite shocking revelations, pertinent to
astro-photography here:

"It is tragic to see that Nikon solved the problem of thermal signal
by a digital processing of the RAW files

Well ... to be fair (and I know you did say this), tragic to astrophotographers, a real boon to the other 99.99% of photographers who then have less PP work to do.

I don't really understand the 'shocking' part of the title. Canon have excellent sensors and have chosen to perform some things one way, Nikon have chosen to do another. It doesn't surprise me if one is better than the other in certain areas and vice versa.

 

[Guy Parsons]

In response to previous posts about larger sensors always being
better, hasn't digital technology over the years shown us that things
always get better and "smaller?" Certainly, fabrication and
large-scale integration has no limit with regard to size at all. Look
at microprocessors and chips in general. There should be no reason
why a 4/3 sensor can't outperform an FF.

The push to make things smaller comes from trying to make CPUs etc perform faster and also use less power. This does not apply to sensors because the fact is that it is the pixel window size and the electron well size under it is the main problem.

As the megapixel count rises and the chip gets smaller, as in 4/3", 1/1.8", 1.1/2.5" etc, those pixel windows and microlenses are getting very small indeed and correspondingly the electron wells are very tiny.

With a large pixel and a large electron well then you can capture many photons and generate and store many electrons and then there is a large difference between a fully exposed pixel and a no-light black one. That means the maximum output signal is way above the noise down in the black region.

A smaller pixel and well means that the maximum bright cells has less output signal above the black (lower maximum photon capture and lower electron store size), hence more random electron noise seen and a lower dynamic range. Attempts are now being made to lessen the random electron noise by using one A/D converter for every column in the sensor (new Sony sensor in upcoming Casio EX-F1 digicam for instance), thus the individual A/D converters can be made to work at lower speed and not create as much stray noise. But unless the chip is at or near absolute zero temperature, then there will always be stray electron noise.

So a 24x26mm sensor will always be better than 4/3" employing the same technology because those pixels can be so much bigger for the same overall pixel count.

The downside to the 24x36mm sensor is the size and how the fabrication machines have to make the whole chip in 4 passes (I think) thus bring in stitching problems etc so the chance of problems is higher and the yield will be lower per die. So cost per chip will be always higher.

If suddenly they could manage 100% yield from a die (maybe they use 12" dies now for that?), just how many 24x26mm sensors could you get from a die compared to 4/3" sensors? Maybe up to 4 times as many? So even with 100% yield the 4/3" sensor will always be cheaper to make.

But the simple matter of physics remains that with whatever the technology of the day is, the individual pixel size and electron well will be always smaller in a smaller chip for the same pixel count.

You can never expect a 4/3" sensor to be equal to the 24x36mm sensor for the same megapixel count.

But the 4/3" works so well for most occasions that it is not a problem, and I'm speaking for my ancient 8 megapixel E-300 here.

It also amazes me that everyone has fallen for the size and megapixel
marketing ploy so easily, while ignoring the things that 'really'
have to do with image quality...such as color accuracy, processing
algorithms and lens characteristics. I believe the only reason FF
exists is to support 35mm lenses, many of which are still around.
Anyone else share these views?

Yes the market has been pushed by megapixels and not by quality, except for some of us who have abandoned pixel peeping and rely on printed results to tell us what is a decent image.

The most critical situations are the pocket cameras with 1/2.5" sensors and I'm happy to report that my 5 megapixel Ricoh pocket camera delivers a less noisy and nicer quality image than any of the later Ricoh models that now go to 8 megapixels, if compared at the same ISO, 64 in my case.

The need for more megapixels really comes from things like landscape to stop the leaves turning to mush at times, and things like school and group photography where people will always get close to the print to try and recognise a face. The more megapixels the better in those cases, but we then have to live with the high ISO noise and lack of dynamic range.

So there's case for 4/3" cameras for most normal photography, and a case for 24x36mm sensors (and even much larger) for occasions that really demand that higher resolution that can be packed in before the noise and dynamic range suffer too much.

The fact remains that though the major companies may be producing 24x36mm sensor cameras for bragging rights, they still are a valid sensor size for those who need it, but of course the lenses need to be upgraded to cope with the more stringent needs of digital sensors. The old film lenses may work OK enough for most occasions, but the ever present pixel peepers will demand better performance.

We see many people floating though who just cannot understand that other sensor sizes and lens focal length ranges do work, they are stuck in a 35mm world of values, so let's sell the really expensive stuff to them and use the cash for doing something useful and interesting.

Regards............. Guy