Technical question on CA/PF & F828

[Mike Par]

Hi folks,

Having read zillions of messages out there on the chromatic aberration/purple fringing problem, I'm still not quite sure what causes it (especially in the case of the F828).

According to the literature, purple fringing traditionally occurs because the short wavelength components of white light (e.g. blue, purple) are refracted to a greater extent than the longer wavelength components (e.g. red, orange), and thus have different focal points.

But why then (in the case of the F828) does the problem only occur at the edges of over-exposed whites. Why not also at the boundary of dark areas and normally exposed whites (albeit then fainter)? On the contrary, the photos I've seen on this forum without any overexposed areas are often of stunning quality, with not a trace of CA/PF. But if CA/PF is an issue related to the lens, why should the level of exposure make a difference? The physical properties of the lens should be the same regardless of the amount of incident light passing through, shouldn't they?

This ties in with the alleged response from Carl Zeiss in another message in this forum that the lens couldn't be the culprit, because the F828 CA/PF problems only occured with overexposed regions. Also (as an aside), you would have thought that the lens could be accurately unit tested without having an F828 attached, and that Carl Zeiss should be well-established masters of their own game by now.

Maybe someone can shed a bit of light on this issue?

Cheers,
Mike.

P.S. Don't really want to become embroiled in the debate as to whether the F828 is merely a good camera or an excellent one. But it's just a real shame that the noise and CA/PF issues are just the things you DON'T want on a camera that was (before its launch) hyped up to be a great low light/night shot digicam. Now if Sony could just solve these issues (and do a few cosmetic changes to give people the impression that they were buying a completely new camera) they could quickly release an F838 (or whatever) and have a real winner in their hands...

 

[Evil Eggplant]

Here's part of your answer. The 828 sensor is made up of 8 million photosites. Each photosite is like a bucket that fills up with photons. Beacuse there are so many photosites on such a small chip the photosites are small, and fill up rather quickly. When the bucket is filled to the top with photons (saturated) the excess photons spill over into neighboring photosites, and such is the nature of blooming.

My explanation is intentionally simplistic, for the sake of clarity.

For this reason most digicams behave like slide film, to expose properly it is necessary to expose for the highlights. A saturated photosite contains no useful information. Plenty of 828's are being accused of being faulty because their owners don't care to understand why things work the way they do. The photosites saturate quickly and the pictures get ugly. The camera does have its faults, but lack of understanding by the photographer makes the existing problems much worse.

The small size of the photosites also explains why the camera is noisy. There is always noise to be dealt with, but when the photosite fills with photons the signal (picture information) is higher than the existing noise. When you shoot a dimly lit scene, or recording shadow detail, the noise becomes a larger component of the photosites' output, and you see noise in the image. This whole thing is called Signal to Noise ratio. Because the 828 sensor photosites are so small there is only room for a few photons, the noise is always there (called "dark current"), the S/N ratio is not that good, you get a noisey picture. The 717 has les photosites, but they are bigger. That is why the 828 is noisier than the 828.

Back to your question (sorry..)

Because the 828's photosites are so small this puts a large burden on the lens. The optical system has to behave almost perfectly in the 828, or all kind of bad things may happen. All kinds of bad things do happen..

Look at Phils 828 resolution test comparisons and see how close the 300D (6mp) with its "touristy" kit lens stacks up against the 828. You really cant tell the difference. That is because the 300d photosites are 7 times the size of te 828. The lens does not need too be as sharp. Consider too, that the difference between 6mP and 8mP is not as meaningful as you may think it is.

There are many reasons why a large sensor is better than a small one. There is only one (practical) reason why a smaller sensor may be desirable, and that would be the reason; because it is smaller. Becaue the sensor is smaller, the camera can be made smaller, and so can the lens. If image quality is the criteria for what sensor you will end up with, a bigger sensor will always outperform a smaller one. There may be a rare exception, but it would be a rare one.

The 828 is chosen because it is small and sexy, it is well built and has a good lens (good enough? the jury is still out), the zoom ratio is good. The 828 will never be known for it's stellar image quality, but many are satisfied and are willing to do the PS thing for the fringing and the NR thing for the noise. Personal experience over the years has shown me that there may be a big difference in performance between two similar cameras. I went thru this with my OLY. It was really noisey, I exchanged it. I still have that camera. Big difference.

I held the 828 in my hands, it is lovely, but did not meet my needs. Shame..

Hi folks,

Having read zillions of messages out there on the chromatic
aberration/purple fringing problem, I'm still not quite sure what
causes it (especially in the case of the F828).

According to the literature, purple fringing traditionally occurs
because the short wavelength components of white light (e.g. blue,
purple) are refracted to a greater extent than the longer
wavelength components (e.g. red, orange), and thus have different
focal points.

But why then (in the case of the F828) does the problem only occur
at the edges of over-exposed whites. Why not also at the boundary
of dark areas and normally exposed whites (albeit then fainter)?
On the contrary, the photos I've seen on this forum without any
overexposed areas are often of stunning quality, with not a trace
of CA/PF. But if CA/PF is an issue related to the lens, why should
the level of exposure make a difference? The physical properties of
the lens should be the same regardless of the amount of incident
light passing through, shouldn't they?

This ties in with the alleged response from Carl Zeiss in another
message in this forum that the lens couldn't be the culprit,
because the F828 CA/PF problems only occured with overexposed
regions. Also (as an aside), you would have thought that the lens
could be accurately unit tested without having an F828 attached,
and that Carl Zeiss should be well-established masters of their own
game by now.

Maybe someone can shed a bit of light on this issue?

Cheers,
Mike.

P.S. Don't really want to become embroiled in the debate as to
whether the F828 is merely a good camera or an excellent one. But
it's just a real shame that the noise and CA/PF issues are just the
things you DON'T want on a camera that was (before its launch)
hyped up to be a great low light/night shot digicam. Now if Sony
could just solve these issues (and do a few cosmetic changes to
give people the impression that they were buying a completely new
camera) they could quickly release an F838 (or whatever) and have a
real winner in their hands...

--
rich
'beware the eggplant'
c-7oo, d-51O, DSC-F7O7, 3OOD
'it's not having what you want, it's wanting what you got'
http://www.pbase.com/iceninevt

 

[xpostal]

 

[Evil Eggplant]

--
rich
'beware the eggplant'
c-7oo, d-51O, DSC-F7O7, 3OOD
'it's not having what you want, it's wanting what you got'
http://www.pbase.com/iceninevt

 

[nand]

Mr. Eggplant’s explanation is one correct possibility for the cause of blooming but at the same time this bucket doesn’t appear to hold all the water as far as blooming is concerned for the F828.

The link below shows the CA test done in the review for the F828 that shows no sign of blooming or CA at all in the centre hole of the picture. From looking at CA tests for other cameras it shows that the F828 and the Fujifilm FinePix S7000 are perhaps the only cameras that absolutely do not show any blooming or CA in the centre of the picture.

http://www.dpreview.com/reviews/SonyDSCF828/Samples/Fringing/DSC03559.jpg

Nand.

Here's part of your answer. The 828 sensor is made up of 8 million
photosites. Each photosite is like a bucket that fills up with
photons. Beacuse there are so many photosites on such a small chip
the photosites are small, and fill up rather quickly. When the
bucket is filled to the top with photons (saturated) the excess
photons spill over into neighboring photosites, and such is the
nature of blooming.

My explanation is intentionally simplistic, for the sake of clarity.