No "Bad" Olympus digital lenses...

msusic said:

Noooooo....
Did you forget "we" don't have that magic 400/5.6 equivalent?

I agree completely, 43rds is easily the most complete and sensible system on the market with no bad film-era lenses.

Sure Canon FF has huge selection of lenses, but you've got to take into an account that cheapest FF camera costs 2500$ €, and APS systems are not nearly well designed as 43rds.

--
Cheers,
Marin

Click to expand...

I'm disagreeing because I ( and many like me ) own and frequently use "third party" equipment like my ( "full frame" ) film cameras as per last evening (7/8/2010) when I shot my EOS 1n (HS) with my 70-200 f2.8 zoom.

While I can still easily utilize my "full frame" film gear I will, 3 4/3rs DSLRs to the contrary.

( We do have access to third party 300 & 400mm lenses in at least 11 other mounts, to include LEICA "R" lens' if we do not mind the hassle of using adapters

Tyrone Wellhung said:

boggis the cat said:

If you have a set scene available, try it out. A TC that reduces the f-number 1 stop produces an EV shift of -1 stop; a TC that reduces the f-number 2 stops produces an EV shift of -2 stops.

I don't have a TC for my Olympus system, but here is what happens with my Canon film SLR (EOS 10 and 35-135 f/4-5.6, at 35 mm f/4) and 2x TC (Tamron 2X C-AF BBAR MC7):

• No TC 1/15 s
• 2x TC 1/4 s.

It works exactly as it should.
Tyrone Wellhung wrote:

Click to expand...

If you compare two cameras, same lens, one with a TC, one without the TC you will notice that the exposures are different, but then it's become a circular argument, the 'correct' f-number is the one that gives the 'correct' exposure.

Click to expand...

The Canon SLR didn't "know" it had a 2x TC present. It simply metered for an extra 2 EV, proving that the 2x TC introduced a 2 stop reduction in the f-number.

Remember that the f-number is just a convenience used for calculating exposure, and a TC is magnifying the focal length.

Tyrone Wellhung said:

Look at it a different way, maybe you'll see it.

Imagine we have a 12MP camera and take a photo of a grey card such that n photons are recorded at every pixel location. Now we fit a 2x teleconverter in front of the lens. This magnifies the image projected onto the sensor to four times the area, but the number of photons in that image is the same. This means that each pixel location records a quarter of the number of photons, n/4.

Now we take the TC off and mount the lens directly. Since our camera is a modular E-5 we can slip out the 12MP sensor and replace it with a 48MP one. With this new sensor the pixels each has a quarter of the area, so each records n/4 photons. We now crop out the centre 12MP and compare the resulting image with the one taken with the TC. We find that in both cases we have 12MP with n/4 photons recorded at each pixel location.

Click to expand...

I don't see what you are trying to prove here. The hypothetical 48 MPixel sensor would require four times the sensitivity of the 12 MPixel, so it all balances out in exposure terms.

Tyrone Wellhung said:

Tyrone Wellhung said:

Edit:

Just an after-thought -- it is a film camera, so no clever reporting by the TC or other hi-jinks can take place. The camera doesn't know the TC is there, it just meters for the scene.

Click to expand...

A film camera is a completely different kettle of fish. If we use the same speed film, n/4 photons over 1/12Mth of the frame will produce a completely different density from n/4 photons over 1/48Mth of the frame, because film records light non-linearly.

Click to expand...

Only at very low light levels, or for long exposures.

For the purposes of your argument that digital systems somehow "cheat" on the effect of teleconverters on the f-number this isn't relevant.

Tyrone Wellhung said:

A digital sensor, on the other hand, records light linearly. There is no 'density' until we read the photon count recorded in each pixel and use software to map that count to some density.

Click to expand...

Film response can be considered linear, and camera metering assumes so. As an example of this: for very long exposures with film you must take account of reciprocity failure and extend your exposure time accordingly.

If you have a TC available you can check this for yourself. The TC does exactly as it should, adjusting the focal length of the lens mounted to it and thus the f-number.

FYI, after a lot of to and fro it turns out that Rikke's original claims about the change in f-number were with regard to "equivalence theory". An f/x lens on a large sensor is still an f/x lens on a smaller sensor.

boggis the cat said:

Tyrone Wellhung said:

boggis the cat wrote:

If you compare two cameras, same lens, one with a TC, one without the TC you will notice that the exposures are different, but then it's become a circular argument, the 'correct' f-number is the one that gives the 'correct' exposure.

Click to expand...

The Canon SLR didn't "know" it had a 2x TC present. It simply metered for an extra 2 EV, proving that the 2x TC introduced a 2 stop reduction in the f-number.

Click to expand...

This was just a repeat of what I'd said in the earlier post and you repeated. We both agree. the exposures are different. Where we disagree is that you insist that one is 'correct'. What I am asking is how your experiment would demonstrate which one is 'correct'.

boggis the cat said:

Remember that the f-number is just a convenience used for calculating exposure, and a TC is magnifying the focal length.

Click to expand...

Exactly, just a convenience. A convenience with its roots in film practice when an emulsion imposed the same densiometric S-curve whatever area you chose to use. That is not the case with a digital sensor, but more on this later.

boggis the cat said:

boggis the cat said:

Look at it a different way, maybe you'll see it.
...

Click to expand...

I don't see what you are trying to prove here.

Click to expand...

What I'm trying to prove is that the result, in terms of photon counts attributed to pixels, is precisely the same in either case. Since the raw image data consists essentially of photon counts attributed to pixels, it demonstrates that as far as the sensor data is concerned the 2x crop and 2x TC are doing the same thing, as Rikke suggested.

boggis the cat said:

The hypothetical 48 MPixel sensor would require four times the sensitivity of the 12 MPixel,

Click to expand...

Firstly, It has nothing to do with the sensitivity of the sensor at all. The sensitivity of the sensor determines how close it actually gets to counting individual photons and there is no reason for the sensitivity of the sensors to be any different.

boggis the cat said:

so it all balances out in exposure terms.

Click to expand...

I am a loss to understand why 'exposure terms' are so important. I have established that the raw file, in terms of photons counted at each pixel location, is identical. That is a very real thing related to the image produced. Exposure is not, it is simply a value of interest if you're trying to use a particular film emulsion at the best spot on its S-curve. So why are we at all interested in things balancing out in exposure terms?

boggis the cat said:

boggis the cat said:

because film records light non-linearly.

Click to expand...

Only at very low light levels, or for long exposures.

Click to expand...

At every value, it is intrinsic to the nature of film. This, from the Wikipedia article on sensiometry ( http://en.wikipedia.org/wiki/Sensitometry ), explains it nicely:



. A digital sensor does not have a similar s-curve characteristic. It records linearly over a very large range of values. The S-curve apparent in digital images we view is imposed entirely in processing, and is arbitrary.

boggis the cat said:

For the purposes of your argument that digital systems somehow "cheat" on the effect of teleconverters on the f-number this isn't relevant.

Click to expand...

What is relevant is that 'correct' exposure is important for film to ensure that the tonal range of the image is recorded on the right art of the S-curve, as explained in the illustration cited here. That is why people brought up on film have the primacy of 'correct' exposure drummed into them. With it's linear curve, digital is different, within reason, wherever you record the tonal value of the scene on the linear characteristic of the sensor, the results will be similar, without the crushing of highlights or shadows that occurs with film. The S-curve is applied in processing to match whatever values have been recorded by the sensor.'Correct' exposure really doesn't mean very much with the digital medium.

boggis the cat said:

boggis the cat said:

A digital sensor, on the other hand, records light linearly. There is no 'density' until we read the photon count recorded in each pixel and use software to map that count to some density.

Click to expand...

Film response can be considered linear, and camera metering assumes so.

Click to expand...

Incorrect. Metering assumes that it should try to place midtones on the mid part of the S-curve. If what yous aid was true, metering with film would be a simple process, with no need for multi pattern metering or zone systems. In fact, sorting out from a scene which of the different brightnesses need to be matched to different parts of the film characteristic is highly complex.

boggis the cat said:

As an example of this: for very long exposures with film you must take account of reciprocity failure and extend your exposure time accordingly.

Click to expand...

reciprocity failure is a completely different thing from the film's densiometric curve.

boggis the cat said:

If you have a TC available you can check this for yourself. The TC does exactly as it should, adjusting the focal length of the lens mounted to it and thus the f-number.

Click to expand...

No one has argued with that, indeed I suggested it above. The issue is, given that it appears that taking a crop is doing just the same thing as is a TC, that regarding that as adjusting the f-number also is a reasonable thing to do.

boggis the cat said:

FYI, after a lot of to and fro it turns out that Rikke's original claims about the change in f-number were with regard to "equivalence theory". An f/x lens on a large sensor is still an f/x lens on a smaller sensor.

Click to expand...

To quote from you: 'the f-number is just a convenience used for calculating exposure'.

Same mistake as others have made, Newman.

Digital zoom is not "equivalent" to optical zoom, whether enlargement or reduction.

You ought to try actually taking some photographs, you would soon learn the difference this makes in practice...

Funny thing about the English language - " A thin chance " and " A fat chance " mean exactly the same thing ...

-

This obsession with exposure is, as I have discussed, a hangover from film practice, when we used a non-linear medium, which required precise control of exposure, and used the same films between different formats, so that you needed to get the right exposure for Velvia, whether you loaded it into a Pen-F, an OM-1 or a 500CM (actually, Velvia wasn't available for MF - and its worthwhile thinking why that might have been the case). Four Thirds was designed as a fresh start digital system, and Olympus correctly based its design around the idea that it was the number of photons collected that mattered, not the size of the sensor. Thus a designed for digital system might have a smaller sensor than a film system and still produce very high quality results. You and others are trying to argue that exposure is the important thing, which inevitably implies that larger sensors have an intrinsic advantage, and undercuts the whole FT design philosophy.

The ideas called 'equivalence' here are absolutely fundamental to the design precepts of Four Thirds, namely

1. It is total light captured in an image which ultimately determines the image noise, not the exposure or size of the sensor. This means that a small sensor can produce images with the same noise characteristics as a larger sensor so long as it receives the same total light.
2. To provide the same total light for the same shutter speed lenses with smaller sensor systems need the same aperture for the same angle of view, which means lower f-numbers, given that the FL for a given angle of view will be shorter.
3. Provision of lower f-numbers is not a technical problem, since the lens diameter will be no more than that for the larger sensor system, and given that the length will be reduce, there is a net saving in size.
4. Since lenses for a smaller sensor size are required to cover a smaller image circle, they can be made to produce higher resolutions, and therefore maintain the overall resolution of the image.
5. In practice, smaller pixels perform better than large ones in terms of read noise, so a smaller sensor system can maintain parity of resolution with a large sensor system without incurring a noise penalty.
6. So overall, reduction in sensor size provides the opportunity to produce the same image quality with a net reduction in size and cost.

Overall this is a compelling argument (interestilngly, one Nikon seems to be about to follow with it's forthcoming EVILs, which are rumoured to have a sensor smaller than FT). What puzzles me is why you and others wish to argue that exposure is paramount and therefore FT has an inherent disadvantage with respect to larger sensor systems.

John King said:

Digital zoom is not "equivalent" to optical zoom, whether enlargement or reduction.

Click to expand...

I explained precisely how the two situations end up with the same photon counts in corresponding pixel locations. Perhaps you could explain how that is not 'equivalent'.

Tyrone Wellhung said:

1. In practice, smaller pixels perform better than large ones in terms of read noise, so a smaller sensor system can maintain parity of resolution with a large sensor system without incurring a noise penalty.

Click to expand...

In practice a FF camera with the same resolution as a 4/3 does have better noise performance.

--
http://dslr-video.com/blogmag/

PerL said:

Tyrone Wellhung said:

1. In practice, smaller pixels perform better than large ones in terms of read noise, so a smaller sensor system can maintain parity of resolution with a large sensor system without incurring a noise penalty.

Click to expand...

In practice a FF camera with the same resolution as a 4/3 does have better noise performance.

Click to expand...

In read noise? That is certainly not a given.

It is, however, in some circumstances possible to take advantage of an FF camera's ability to collect photons over an area about four times larger than that of a 4/3 sensor. This of course will a mean a two stop better photon shot SNR. Note that this advantage isn't always there, you may be able to collect a similar amount of photons with 4/3 in many circumstances, and is has quite literally nothing to do with the spatial sampling rate (pixel count).

A hypothetical 1 square centimeter sensor can collect (count) the same number of photons regardless of it being divided into a million photo sites or simply consisting of one, large 1 square centimeter photo site. The photon shot SNR will be the same in both cases. The second case will just have been sampled at a rather low spatial frequency and might be considered a tad uninteresting from a photographic point of view ;-)
--
Rikke

PerL said:

Tyrone Wellhung said:

1. In practice, smaller pixels perform better than large ones in terms of read noise, so a smaller sensor system can maintain parity of resolution with a large sensor system without incurring a noise penalty.

Click to expand...

In practice a FF camera with the same resolution as a 4/3 does have better noise performance.

Click to expand...

In practice, there's a very few FF and FT cameras with the same resolution we could compare. Essentially it is the 12MP Olympuses, vs the 5D, D3, D700 and D3S. I'm working on a spreadsheet to extract read noise data from DxO data, but until then we can do a quick check on the DxO 18% noise scores, which with with the same photon count gives 36dB for the E-620 (at 100 ISO) and 35.7dB for the D700 (at 400 ISO). Since with the same photon count the photon shot noise is the same (if the QE is the same), it would seem that the E-620 is producing less read noise than the D700. Against the 5D it's 36dB plays 34. The 5D has very low QE, so it's likely that we're seeing that come into play as well as the read noise difference. The Nikon D3S wins with 37dB to 36, but that camera has exceptional QE. On the whole, though, the evidence supports my contention.

Rikke Rask said:

PerL said:

Tyrone Wellhung said:

1. In practice, smaller pixels perform better than large ones in terms of read noise, so a smaller sensor system can maintain parity of resolution with a large sensor system without incurring a noise penalty.

Click to expand...

In practice a FF camera with the same resolution as a 4/3 does have better noise performance.

Click to expand...

In read noise? That is certainly not a given.

It is, however, in some circumstances possible to take advantage of an FF camera's ability to collect photons over an area about four times larger than that of a 4/3 sensor. This of course will a mean a two stop better photon shot SNR.

Click to expand...

Suppose Olympus had commissioned a sensor for the E-5 with a quadrupled photoelectron saturation capacity (perhaps by increasing the cell capacitance). Such a sensor would collect four times the number of photons. Of course, it would be no more sensitive, so it would take four times the amount of light to fill it up. If the original sensor had a base ISO of 200, this one would have a base ISO of 50. Suppose also that Olympus releases a range of f/0.7 primes to go with the E-5. My submission would be an E-5 fitted with a 25/0.7 could take any photograph that could be taken by a D700 fitted with a 50/1.4. Only the control settings would be different. For example a photo needing 1/1000 sec, 50/1.4, 200 ISO would be pretty much identical to one of the same subject taken with 1/1000 sec, 25/0.7, 50 ISO on the E-5.

Edit: for those of the feline persuasion, time to start thinking why films such as Velvia and K25 were only available in 35mm, not MF.

Tyrone Wellhung said:

PerL said:

Tyrone Wellhung said:

1. In practice, smaller pixels perform better than large ones in terms of read noise, so a smaller sensor system can maintain parity of resolution with a large sensor system without incurring a noise penalty.

Click to expand...

In practice a FF camera with the same resolution as a 4/3 does have better noise performance.

Click to expand...

In practice, there's a very few FF and FT cameras with the same resolution we could compare. Essentially it is the 12MP Olympuses, vs the 5D, D3, D700 and D3S. I'm working on a spreadsheet to extract read noise data from DxO data, but until then we can do a quick check on the DxO 18% noise scores, which with with the same photon count gives 36dB for the E-620 (at 100 ISO) and 35.7dB for the D700 (at 400 ISO). Since with the same photon count the photon shot noise is the same (if the QE is the same), it would seem that the E-620 is producing less read noise than the D700. Against the 5D it's 36dB plays 34. The 5D has very low QE, so it's likely that we're seeing that come into play as well as the read noise difference. The Nikon D3S wins with 37dB to 36, but that camera has exceptional QE. On the whole, though, the evidence supports my contention.

Click to expand...

Go to an indoor sports game armed with a E620 and a D700, both with a 70-200 2.8 lens. Shoot the game at ISO 6400 and 12800 and check the results.

--
http://dslr-video.com/blogmag/

PerL said:

Tyrone Wellhung said:

PerL said:

Tyrone Wellhung said:

1. In practice, smaller pixels perform better than large ones in terms of read noise, so a smaller sensor system can maintain parity of resolution with a large sensor system without incurring a noise penalty.

Click to expand...

In practice a FF camera with the same resolution as a 4/3 does have better noise performance.

Click to expand...

In practice, there's a very few FF and FT cameras with the same resolution we could compare. Essentially it is the 12MP Olympuses, vs the 5D, D3, D700 and D3S. I'm working on a spreadsheet to extract read noise data from DxO data, but until then we can do a quick check on the DxO 18% noise scores, which with with the same photon count gives 36dB for the E-620 (at 100 ISO) and 35.7dB for the D700 (at 400 ISO). Since with the same photon count the photon shot noise is the same (if the QE is the same), it would seem that the E-620 is producing less read noise than the D700. Against the 5D it's 36dB plays 34. The 5D has very low QE, so it's likely that we're seeing that come into play as well as the read noise difference. The Nikon D3S wins with 37dB to 36, but that camera has exceptional QE. On the whole, though, the evidence supports my contention.

Click to expand...

Go to an indoor sports game armed with a E620 and a D700, both with a 70-200 2.8 lens. Shoot the game at ISO 6400 and 12800 and check the results.

Click to expand...

That has no relevance to the discussion at all. The read noise performance of the E620 and D700 will not be revealed by such an experience, and both used with an f/2.8 lens, the total light on the sensor will be two stops different, so the photon noise will be different (in the D700's advantage). Moreover, the lenses give you a very different angle of view, so you'll never take the same photo with both.

A much more revealing test would be to go to your sports day armed with an E620 + 35-100/2 and a 5D + 70-200/4L. Then tell me what difference you see in the noise at 6400 and 12800, shooting wide open with the same shutter speed.

Tyrone Wellhung said:

PerL said:

Tyrone Wellhung said:

PerL said:

Tyrone Wellhung said:

1. In practice, smaller pixels perform better than large ones in terms of read noise, so a smaller sensor system can maintain parity of resolution with a large sensor system without incurring a noise penalty.

Click to expand...

In practice a FF camera with the same resolution as a 4/3 does have better noise performance.

Click to expand...

In practice, there's a very few FF and FT cameras with the same resolution we could compare. Essentially it is the 12MP Olympuses, vs the 5D, D3, D700 and D3S. I'm working on a spreadsheet to extract read noise data from DxO data, but until then we can do a quick check on the DxO 18% noise scores, which with with the same photon count gives 36dB for the E-620 (at 100 ISO) and 35.7dB for the D700 (at 400 ISO). Since with the same photon count the photon shot noise is the same (if the QE is the same), it would seem that the E-620 is producing less read noise than the D700. Against the 5D it's 36dB plays 34. The 5D has very low QE, so it's likely that we're seeing that come into play as well as the read noise difference. The Nikon D3S wins with 37dB to 36, but that camera has exceptional QE. On the whole, though, the evidence supports my contention.

Click to expand...

Go to an indoor sports game armed with a E620 and a D700, both with a 70-200 2.8 lens. Shoot the game at ISO 6400 and 12800 and check the results.

Click to expand...

That has no relevance to the discussion at all. The read noise performance of the E620 and D700 will not be revealed by such an experience, and both used with an f/2.8 lens, the total light on the sensor will be two stops different, so the photon noise will be different (in the D700's advantage). Moreover, the lenses give you a very different angle of view, so you'll never take the same photo with both.

Click to expand...

Tyrone Wellhung said:

A much more revealing test would be to go to your sports day armed with an E620 + 35-100/2 and a 5D + 70-200/4L. Then tell me what difference you see in the noise at 6400 and 12800, shooting wide open with the same shutter speed.

Click to expand...

If the 4/3 camera has a two stops advantage, what is the purpose of claiming that the smaller sensor has the same noise performance. I dont care about read nosie or shot noise - I care about usefulness in a real situation. The one I described is exactly a reason why most pro photographers dont use Olympus.
--
http://dslr-video.com/blogmag/

PerL said:

Tyrone Wellhung said:

PerL said:

Tyrone Wellhung said:

PerL said:

Tyrone Wellhung said:

1. In practice, smaller pixels perform better than large ones in terms of read noise, so a smaller sensor system can maintain parity of resolution with a large sensor system without incurring a noise penalty.

Click to expand...

In practice a FF camera with the same resolution as a 4/3 does have better noise performance.

Click to expand...

In practice, there's a very few FF and FT cameras with the same resolution we could compare. Essentially it is the 12MP Olympuses, vs the 5D, D3, D700 and D3S. I'm working on a spreadsheet to extract read noise data from DxO data, but until then we can do a quick check on the DxO 18% noise scores, which with with the same photon count gives 36dB for the E-620 (at 100 ISO) and 35.7dB for the D700 (at 400 ISO). Since with the same photon count the photon shot noise is the same (if the QE is the same), it would seem that the E-620 is producing less read noise than the D700. Against the 5D it's 36dB plays 34. The 5D has very low QE, so it's likely that we're seeing that come into play as well as the read noise difference. The Nikon D3S wins with 37dB to 36, but that camera has exceptional QE. On the whole, though, the evidence supports my contention.

Click to expand...

Go to an indoor sports game armed with a E620 and a D700, both with a 70-200 2.8 lens. Shoot the game at ISO 6400 and 12800 and check the results.

Click to expand...

That has no relevance to the discussion at all. The read noise performance of the E620 and D700 will not be revealed by such an experience, and both used with an f/2.8 lens, the total light on the sensor will be two stops different, so the photon noise will be different (in the D700's advantage). Moreover, the lenses give you a very different angle of view, so you'll never take the same photo with both.

Click to expand...

PerL said:

A much more revealing test would be to go to your sports day armed with an E620 + 35-100/2 and a 5D + 70-200/4L. Then tell me what difference you see in the noise at 6400 and 12800, shooting wide open with the same shutter speed.

Click to expand...

If the 4/3 camera has a two stops advantage, what is the purpose of claiming that the smaller sensor has the same noise performance. I dont care about read nosie or shot noise - I care about usefulness in a real situation.

Click to expand...

My example was exactly the real situation. The 35-100/2 on the E-620 is the equivalent of the 70-200/4L on the 5DII. If you fitted a 70-200 on a E-620, you'd be looking to fit a 140-400 on a D700 to do the same job, or maybe the same 70-200 with a 2x TC. Do that, you've got a 140-400/5.6 and you'll find its performance pretty much exactly the same as the 70-200/2.8 on the E-620 (although the FT system would probably be optically superior). The point is that lenses and cameras are part of a system, and you need to compare parts of the system that do the same job, not different jobs.

PerL said:

The one I described is exactly a reason why most pro photographers dont use Olympus.

Click to expand...

Maybe, but not a very big one. The major reason is that Olympus has never had a big penetration into the pro markets (didn't even in OM days) and would have had to develop a system of the range and power of Canon's and Nikon's and rely on them both messing up simultaneously to get a foothold. Olympus hasn't fulfilled even the first condition (the second was outside its control) by providing the system lenses that would allow the E-system to perform at the same levels as the Canon and Nikon pro systems. There were no fundamental technical problems doing so. My guess is that they haven't simply because it was not worthwhile for them. A full range of f/0.7 primes and f/1.4 zooms would have been enormously expensive to develop and probably wouldn't have made a big dent into the pro marketplace in any case. They certainly would have made a big dent in Olympus' bottom line. Instead they've adapted FT designs into the SHG range which has given much cheaper development costs and put a place marker at the top end, even though they are deeply compromised with respect to what they could have been if purpose designed.

Tyrone Wellhung said:

PerL said:

Tyrone Wellhung said:

PerL said:

Tyrone Wellhung said:

PerL said:

Tyrone Wellhung said:

1. In practice, smaller pixels perform better than large ones in terms of read noise, so a smaller sensor system can maintain parity of resolution with a large sensor system without incurring a noise penalty.

Click to expand...

In practice a FF camera with the same resolution as a 4/3 does have better noise performance.

Click to expand...

In practice, there's a very few FF and FT cameras with the same resolution we could compare. Essentially it is the 12MP Olympuses, vs the 5D, D3, D700 and D3S. I'm working on a spreadsheet to extract read noise data from DxO data, but until then we can do a quick check on the DxO 18% noise scores, which with with the same photon count gives 36dB for the E-620 (at 100 ISO) and 35.7dB for the D700 (at 400 ISO). Since with the same photon count the photon shot noise is the same (if the QE is the same), it would seem that the E-620 is producing less read noise than the D700. Against the 5D it's 36dB plays 34. The 5D has very low QE, so it's likely that we're seeing that come into play as well as the read noise difference. The Nikon D3S wins with 37dB to 36, but that camera has exceptional QE. On the whole, though, the evidence supports my contention.

Click to expand...

Go to an indoor sports game armed with a E620 and a D700, both with a 70-200 2.8 lens. Shoot the game at ISO 6400 and 12800 and check the results.

Click to expand...

That has no relevance to the discussion at all. The read noise performance of the E620 and D700 will not be revealed by such an experience, and both used with an f/2.8 lens, the total light on the sensor will be two stops different, so the photon noise will be different (in the D700's advantage). Moreover, the lenses give you a very different angle of view, so you'll never take the same photo with both.

Click to expand...

Tyrone Wellhung said:

A much more revealing test would be to go to your sports day armed with an E620 + 35-100/2 and a 5D + 70-200/4L. Then tell me what difference you see in the noise at 6400 and 12800, shooting wide open with the same shutter speed.

Click to expand...

If the 4/3 camera has a two stops advantage, what is the purpose of claiming that the smaller sensor has the same noise performance. I dont care about read nosie or shot noise - I care about usefulness in a real situation.

Click to expand...

My example was exactly the real situation. The 35-100/2 on the E-620 is the equivalent of the 70-200/4L on the 5DII. If you fitted a 70-200 on a E-620, you'd be looming to fit a 140-400 on a D700, or maybe the same 70-200 with a 2x TC. Do that, you've got a 140-400/5.6 and you'll find its performance pretty much exactly the same as the 70-200/2.8 on the E-620 (although the FT system would probably be optically superior). The point is that lenses and cameras are part of a system, and you need to compare parts of the system that do the same job, not different jobs.

Tyrone Wellhung said:

The one I described is exactly a reason why most pro photographers dont use Olympus.

Click to expand...

Maybe, but not a very big one. The major reason is that Olympus has never had a big penetration into the pro markets (didn't even in OM days) and would have had to develop a system of the range and power of Canon's and Nikon's and rely on them both messing up simultaneously to get a foothold. Olympus hasn't fulfilled even the first condition (the second was outside its control) by providing the system lenses that would allow the E-system to perform at the same levels as the Canon and Nikon pro systems. There were no fundamental technical problems doing so. My guess is that they haven't simply because it was not worthwhile for them. A full range of f/0.7 primes and f/1.4 zooms would have been enormously expensive to develop and probably wouldn't have made a big dent into the pro marketplace in any case. They certainly would have made a big dent in Olympus' bottom line. Instead they've adapted FT designs into the SHG range which has given much cheaper development costs and put a place marker at the top end, even though they are deeply compromised with respect to what they could have been if purpose designed.

Click to expand...

I dont think we disagree about the noise performance if we say it is about two stops, give or take. I jumped in and read your statement about the smaller sensor having no noise penalty, but obviously you did not mean at the same F stops.

And yes, Olympus never had any large impact on the pro market even with OM-system (except forcing Nikon to make the compact FM/FE series that were quite popular among pros). But the current lag in low light ability of the Oly system vs FF and 1.3x cameras is makes even more difficult for those that want to use Oly for pro use. No editor is interested in explanations why you cant get the shot the competing photographers can.
--
http://dslr-video.com/blogmag/

PerL said:

I dont think we disagree about the noise performance if we say it is about two stops, give or take. I jumped in and read your statement about the smaller sensor having no noise penalty, but obviously you did not mean at the same F stops.

Click to expand...

If you'd read my point 1:

It is total light captured in an image which ultimately determines the image noise, not the exposure or size of the sensor. This means that a small sensor can produce images with the same noise characteristics as a larger sensor so long as it receives the same total light.
and 2:

To provide the same total light for the same shutter speed lenses with smaller sensor systems need the same aperture for the same angle of view, which means lower f-numbers, given that the FL for a given angle of view will be shorter.
You'd have seen that was exactly what I was saying in the first place.

PerL said:

And yes, Olympus never had any large impact on the pro market even with OM-system (except forcing Nikon to make the compact FM/FE series that were quite popular among pros). But the current lag in low light ability of the Oly system vs FF and 1.3x cameras is makes even more difficult for those that want to use Oly for pro use. No editor is interested in explanations why you cant get the shot the competing photographers can.

Click to expand...

and in point 3:

I said Provision of lower f-numbers is not a technical problem, since the lens diameter will be no more than that for the larger sensor system, and given that the length will be reduce, there is a net saving in size.

As an example, a 35-100/1.4 on a FT will get the same shots as a 70-200/2 on a FF. In fact, the 35-100/2 really is a 35-100/1.4 - only Olympus won't let it open up all the way, maybe because its wide-open optical performance isn't what they'd want. The departed sergeant-major always used to claim there was a trick to make it open to f/1.6 or so, which would get the same shot as an f/3.2 on FF. The problem it is designed by tacking a 2x wide converter onto the back of a 70-200/2.8, so it's actuallly larger, heavier and more expensive than the competing FF lenses, offering the system no advantages in shot gettability. Had it been directly designed as a 35-100/1.4 it would have been as fat as the FF lenses but quite a bit shorter and probably lighter, and would have offered a distinct advantage in shot gettability, since no editor is interested in explanations why a lens was too big and heavy for your baggage. On the other hand, it would have cost Olympus a great deal more to develop, and they may never have recouped that cost, given their agreed low penetration in the digital market. After the rapturous reception that the E-1 got, Olympus could have gone for it and made the system it needed to give it a chance in the PJ market, but they chose caution instead, probably for the best - they could have bust the company if they put that investment in and it hadn't come off.

Barring new information I'm content with that answer, though I imagine many will not be.

Tyrone Wellhung said:

PerL said:

I dont think we disagree about the noise performance if we say it is about two stops, give or take. I jumped in and read your statement about the smaller sensor having no noise penalty, but obviously you did not mean at the same F stops.

Click to expand...

If you'd read my point 1:

It is total light captured in an image which ultimately determines the image noise, not the exposure or size of the sensor. This means that a small sensor can produce images with the same noise characteristics as a larger sensor so long as it receives the same total light.
and 2:

To provide the same total light for the same shutter speed lenses with smaller sensor systems need the same aperture for the same angle of view, which means lower f-numbers, given that the FL for a given angle of view will be shorter.
You'd have seen that was exactly what I was saying in the first place.

Tyrone Wellhung said:

And yes, Olympus never had any large impact on the pro market even with OM-system (except forcing Nikon to make the compact FM/FE series that were quite popular among pros). But the current lag in low light ability of the Oly system vs FF and 1.3x cameras is makes even more difficult for those that want to use Oly for pro use. No editor is interested in explanations why you cant get the shot the competing photographers can.

Click to expand...

and in point 3:

I said Provision of lower f-numbers is not a technical problem, since the lens diameter will be no more than that for the larger sensor system, and given that the length will be reduce, there is a net saving in size.

As an example, a 35-100/1.4 on a FT will get the same shots as a 70-200/2 on a FF. In fact, the 35-100/2 really is a 35-100/1.4 - only Olympus won't let it open up all the way, maybe because its wide-open optical performance isn't what they'd want. The departed sergeant-major always used to claim there was a trick to make it open to f/1.6 or so, which would get the same shot as an f/3.2 on FF. The problem it is designed by tacking a 2x wide converter onto the back of a 70-200/2.8, so it's actuallly larger, heavier and more expensive than the competing FF lenses, offering the system no advantages in shot gettability. Had it been directly designed as a 35-100/1.4 it would have been as fat as the FF lenses but quite a bit shorter and probably lighter, and would have offered a distinct advantage in shot gettability, since no editor is interested in explanations why a lens was too big and heavy for your baggage. On the other hand, it would have cost Olympus a great deal more to develop, and they may never have recouped that cost, given their agreed low penetration in the digital market. After the rapturous reception that the E-1 got, Olympus could have gone for it and made the system it needed to give it a chance in the PJ market, but they chose caution instead, probably for the best - they could have bust the company if they put that investment in and it hadn't come off.

Click to expand...

There seems to be some difficulty in making these ultra fast lenses for smaller formats that would be needed. There is a lack of them even for APS-C, and for instance the 35 1.8 G DX Nikkor is larger than the 50 1.8 for FF. Not to mention the size of the Sigma 30 1.4 and Leica 25 1.4, comapred to a 50 1.4 for FF.
--
http://dslr-video.com/blogmag/

PerL said:

There seems to be some difficulty in making these ultra fast lenses for smaller formats that would be needed. There is a lack of them even for APS-C, and for instance the 35 1.8 G DX Nikkor is larger than the 50 1.8 for FF. Not to mention the size of the Sigma 30 1.4 and Leica 25 1.4, comapred to a 50 1.4 for FF.

Click to expand...

In theory, you take an f/1.8, f/1.4, f/1.0 or f/0.7 lens with one FL/image circle, scale it down by 2, and you've got an lens of the same f-number, half the focal length and half the image circle. In the past, there has been no shortage of spectacularly fast lenses in small formats, such as the f/1.0 3x zoom on the Canon 310XL super 8 camera. Very fast lenses are common for TV applications and one such (f/0.95) has found its way onto mFT as a 'Noktious' or somesuch. Famously, Stanley Kubrick used a Zeiss f/0.7 to film Barry Lyndon (using a frame bigger than FT). The problem is making them fast and retrofocus - the lens gets very big. The crop formats have lens registers designed for FF, so even standard lenses must be very retrofocus. FT was designed for its sensor size, but was designed with a very long register - a mistake Panasonic hasn't repeated with mFT. Nonetheless, the place where PJ's really want fast lenses is the long FL end, and there should be no trouble making them fast and proportionately smaller there, especially with the economic aspherics, exotic glasses and ridiculous amounts of computational power for lens design that is available now and wasn't there when the lenses I referred to were designed.

Edit: Per, I think you and I deserve the 'best steer of an OT sub-thread back on-topic' award for this one.

rovingtim said:

Barring new information I'm content with that answer, though I imagine many will not be.

Click to expand...

I cannot imagine why not, since the post was astoundingly informative and also true. But then, I suppose, the truth can often be more controversial than commonly held myths.