The Missing Factor in Sony F828 Reviews

[Andrzej Poniatowski]

I wonder why NOBODY criticizes the biggest flaw in 828 - the quality of the electronic viewfinder. I think that the camera deliveres very good quality images (fantastic value for money) BUT looking throught the viewfinder makes me "sick" and "unwilling" to photograph - though I know that ALL will look much better later.....
The viewfinder made me send the camera back! No more electronic viewfinders! Bought E-1 instead!

[Howard Smith]

Gosh Ray, was that a compliment?  Unless quantum mechanics has changed a great deal since I was in college, I think I can understand and appreciate the subtle difference betwee digital and analog.  But feel free to explain it if you must.

[BJL]

BJL, I would point out that except for Pentax and Fuji, I don't think any camera maker you mentioned is a player in the medium format market.

Howard,

   even ignoring medium format, there is room for Nikon, Fuji, Pentax and Sony to try to move DSLR sensor sizes up towards 35mm frame size, but, contrary to your claim that "(the Canon and Nikon pros for starters) are staying up nights trying to make bigger sensors" I know of no evidence that Nikon or any of these companies are trying to do so. There is also room for all digital camera makers to move towards bigger sensors for digicams than the current 2/3" limit, but there is no sign of movement there either.

    Could you give some evidence for your claim that Canon and Nikon engineers are working on increasing sensor sizes, or of any current effort to increase sensor sizes other than in the medium format back realm?

[Howard Smith]

Perhaps the key word is "amateur  I think with that there comes things like cost and the need ot desire to make really big prints.  When amateur cameras start racking a thousand US dollars or two, the market must drop off pretty fast.  Too expensive for the amateur and not good enough for the pro and very serious amateur.

I have very little knowledge about electronic cameras, but with film there is an advantage to size.  Larger fil can make larger prints with less grain and lower quality lenses.    I have two 4x5 lenses that cost less than one Hasselblas lens, and they work just fine.  Perhaps if film (or electronic Fujichrome) were to get so good that a fine 30x40" print equal to a large format camera could be made from a less than 35mm digital camera, then large format would get pushed to better lenses and bigger electronic Fujichrome.  Of course, here will be some limit beyond which no one cares if a print that big could be made of no one but the government could afford it.

[Ray]

Howard,
I knew you knew that!   I've also known it for some time. The obstacle is in the concept of a 'perfect' film (or sensor). With analogue processes, improvements happen gradually and by small increments. If cameras had not gone 'digital', sort of, (they're not yet truly digital) film might never have progressed to the point where it could be so fine grained and have a sufficiently high MTF response, at say 60 lp/mm, that 35mm could rival the quality of 9x12cm or 8x10 in normal, everyday conditions.

A similar analogue problem now plagues so-called digital sensors, which gives the edge to the larger sensor. The smaller the sensor and photodiodes, the greater the noise and the smaller the dynamic range. The closer one gets to capturing the full resolution of the smaller format lens, the more apparent the noise becomes at equivalent enlargement and the more difficult it becomes to maintain a DR that at best is rarely equal to that of negative film in the current crop of digital cameras, at any price.

By designing sensors that are truly digital, they become virtually noise-free like an audio CD, with terrific dynamic range. Huge enlargements become possible because there is no grain or noise to spoil the show. Problem solved!  

[BJL]

About the resolution of larger format versus smaller format lenses, what testing like that at photodo sems to show is that

a) larger format lenses have somewhat lower resolution in the technical sense of fewer "line pairs per millimetre" in the image formed in the camera

 however, the larger format leads to more detail in the sense of "line pairs per picture height", which translates in to more "print resolution" in the sense of more line pairs per millimetre on prints of a given size.

[BJL]

To further illustrate this point, the Olympus E-1 would be an excellent example because the Zuiko lenses appear to be so much better than the rather noisy and low resolution 5MP sensor.

Ray,

   first, I have asked indirectly before but am still curious: how do you come to your conclusions about the resolution and noise performance of the E-1? By reading lab. test results only, or also by actually looking at properly prepared prints? I have heard mostly favourable comments from peope judging from prints. The letter from Mike Sims mentioned in the What's New section ( http://luminous-landscape.com/essays/cogna....shtml#feedback ) seem apposite.

    Secondly, the one part of your proposal that I think has the most chance of coming to pass is photosite sizes of 4 microns and even below 3 microns in many DSLR sensors, not just 4/3 format. This is because, once those huge numbers of pixels can be read out and processed fast enough, there is almost no image quality downside to having more, smaller pixels in a given sensor size:

a) Printing at the same size and hence with proportionately higher ppi levels effectively downsamples, which recovers roughly the same visible noise levels, dynamic range, tonal gradations and such as if one had had fewer pixels to start with: prints of the same size from a sensor of the same size and technology should show roughly the same image quality as photosite size shrinks, except for increased resolution in regions where noise does not interfere.

 For subjects of normal contrast range with adequate light levels, photosites of around 3 microns have already been shown to give quite good image quality, with competely satisfactory noise levels, so you would often be able to expolit the greater resolution by making larger prints. (The new book "America 24/7" shows what pros can do with the 2.8 micron photosites of the 5MP 1/1.8" sensor in the Olympus C-5050.)

c) Often there would be visible noise in some regions (shadows) but not others (main subject), and then noise reduction post-processing could fix the shadows (at some resolution cost there) while maintaining higher resolution in the better lit parts.

[Ray]

That makes perfect sense to me, and was so obvious that it is hardly worth mentioning in the first place.

There are many things that seem obvious after the thought has occurred, including Darwin's Theory of Evolution.  

The lack of development of large and medium format gear has been apparent for a long time. To the camera buying public, the cost of such gear has always seemed outrageous. The cost of MF digital backs is even more outrageous.

There's a very high price to pay for that additional image quality, so don't be surprised if these dinosaurs become extinct.

A true digital sensor will seal their fate.    

[Ray]

I'll see if I can set one up, but I can assure you that there will be no comparison in depth of field.

Lin,
How about the following comparisons?

(1) the F828 at ISO 100 and F8, and the 10D at ISO 400 and f13.

(2) the F828 at ISO 64 and F8, and the 1Ds at ISO 800 and F32?

Same exposure, same DoF and equivalent focal length in each case. This would represent a true level playing field and would let us all know just what the F828 can do that these other DSLRs can't.  

[BJL]

How about the following comparisons?

(1) the F828 at ISO 100 and F8, and the 10D at ISO 400 and f13.

(2) the F828 at ISO 64 and F8, and the 1Ds at ISO 800 and F32?

Same exposure, same DoF and equivalent focal length in each case.

Where do you get your 10D f-stop value of f/13? The 10D sensor is larger than the 828's by a linear factor of about 2.5, and its photosites are larger by an even larger factor (in case one compares by cropping to equal pixel counts) so the comensurate f-stop for the 10D should be about f/20, and hence ISO speed a bit over 600.

[BJL]

Is there a general rule to get the number of electrons caused by a single photon (I remember a mean efficiency of about 30%)?
Any hints are appreciated.

Georg

Thanks for that number; I have been trying to infer that value from data for particular sensors, but get only about 4,000 electrons for the 6.8µm^2 photosites of the E-1 sensor, and so about 15,000 photons assuming that "Quantum Efficiency" (electrons per photon) of about 30%.

That quantum efficiency figure of about 30% is about right though, according to Kodak's spec sheets, which are a good place to look. It is also close to the maximum possible for colour photography, when you remember that each site should only respond to light in the correct part of the spectrum.

[Howard Smith]

Ray, you say complex organisms evolve very slowly over millions of years.  Let's look at but one thing required for human life - hemoglobin.  Not an optical system, but one component of blood.

Using the formula for the linear arrangement of amino acids to make hemoglobin, there are about 7.4 X 10(654) combinations.  (Not a bad approximation of infinity.)  Keep in mind that ONE misplaced acid is usually fatal.  Sickle cell anemia is an example.  And each time you combination fails and theorganism dies, you have to start over.

Now try to imagine 10(500+) random trialsto evolve just hemoglobin.  The accepted age of the universe is about 10(17) seconds.  So you would have to try about about 10(480) combinations per second.  That's a lot of megahertz!

There ar about 10(66) atoms in the universe.  Not much material to waste evolving hemoglobin.

There doesn't seem to be enough time or material to make hemoglobin by chance.  You need a lot of external help.  A Designer or Creator.

One might argue that a computer could evolve naturally.  But without software, the computer is pretty lame.  Where do evolutionists claim "information" came from?

The gurus of evolution aren't about to admit there precious theory is in real trouble.  It takes a lot of commitment to fallicies and blindness what is around you to be an atheist.  Just as I cannot "prove" there is God, you cannot prove there isn't.  I am not willing to gamble eternity on the arrogant assumption there isn't.

[BJL]

Ray,

    thanks for the link on read noise; yes read noise increases with read rate, and the 5 electron figure is for the extreme case of very slow read out (40kHz, meaning 25 seconds per MP?) from a sensor with huge photosites. Even the 1MHz read rate for the case with 11 electrons is way below what digital cameras need, since I think that 5MP @ 3fps needs 15MHz. I will search for some more up-to-date read noise information; I think Dalsa has a graph of read noie vs read rate at their web site, relating to their 22MP FFT CCD for medium format backs.

[Howard Smith]

So, ae you saying that smaller isn't better?  Maybe bigger could be better?

[georgj]

Hi BJL,

thanks for your answer, it makes me feel a bit more comfortable with my 'homebrewn phisics'.
In 3 µm sensors, even if you leave the read-out-error and thermal noise aside and thus go down to abt. 100 electrons or 300 photons (giving a 10/1 S/N-ratio), you'll have roughly 500 distinguishable steps of intensitiy (by dividing the [18,000;300] interval in classes that are half wide the mean error (e.g. 18,000; mean error 134; next step is 17993, class width decreasing continously). Of course, one might add some more steps at the top end towards maximum so that a 100% reflectivity wouldn't blow out.
What did I learn? I come to about 6 f stops dynamic range; one cannot read information from the sensor without enough photons; I still want the F828 ;-).

Thanks, Georg

I can hardly wait for Michael tp use the new test equipment - DxO I think it is called.  It will be wonderful to have real information.

And the lens correction program - if they make it available for enough lenses -will be worth a great deal.  I expect by next PMA that most top-end prosumer digicams like the 828 will have this built into the software, and problems like CA and PF will be quaint relics of the past...

It should be possible to put it in DSLRs, I guess, since the DSLR has to know which lens it is using anyway, and should be able to apply the correction for that specific lens.  

This whole thing should really finish film for keeps since there IS no software in a film camera and corrections could only be applied after scanning....

Like MR says, everything is about to change....

willie

[Howard Smith]

Ray, there is considerable difference between agnostic and atheist.  I believe there is God, and know I can't prove it.  I guess I may be an agnostic too, accepting God on faith.  If God could be "proven," we would all believe in God.

BJL, why can't a debate of God be rational?  And isn't the "totality of informed assessment of the evidence and arguments" simply the summation of all the opinions of individuals?  I really don't whether Darwin believed his theory or not.  I have examined the evidence and find evolution to quite impossible.  If one takes the Bible seriously, one would also deny evolution.

Probably enough said.  Back to photography.

[Ray]

Ray, you mentioned that the Pope is not an "evolution denier."

No, that was BJL  . The history of persecution of rational, clear thinking individuals (such as Galileo) by the Church is a disgrace and has done great harm to the Christian movement, and that's a great pity.

However, I think it is possible to have a belief in God and accept the Evolutionary process. It just requires a bit of reinterpretation of the Bible. A day of the 'six days of creation' then becomes 4.8 billion divided by 6, and the process of evolution then becomes the method which God has used to create the universe. He's set the laws in motion and allowed things to happen and fall where they will (sort of thing   )

[Ray]

Ray, the brown moth thing is bunk.  It was later discovered that the spotted moth does not rest on tree truncks or branchs.  The moths used to support that rot were dead and glued or pinned to trees to prove the theory.

Can you give me some sources for this information? My source is Richard Dawkins.

I'm also confused about why you care about the "Christian movement."  I am laboring under the opinion you are an atheist

First of all, I make a distinction between "disbelief in God" and "belief that God does not exist". The former is passive and the latter is active. The terms Atheist and Agnostic are fairly rigid categories. To state one's true position, one is up against the limitation of language. Putting myself in the category of 'agnostic' seems a bit wimpish. Putting myself in the category of 'atheist' seems a bit foolish, so what can I say?  

I care about Christianity because it seems clear the true Christian view of the world, with its origins in Judaism, is actually favourable to scientific enquiry. Neither the ancient Greeks nor the Arabs who gave us our numer system, really caught on to the scientific process. There's something about the Judaic view of the world that has allowed the development of science, despite occasional ignorant outbursts of persecution by the Christian church.

[A Scottish soul]

OOps

my mistake  it is a 14 bit counter not a 13 bit.


John