Foveon

[Marshal]

[font color=\'#000000\']The Sigma mount would not be as big a problem for me as the 1.7X lens magnification. That IS a problem. I had to buy a 14mm ED for my 1.5X mag D1X in order to get as close as possible a 20mm view. That set me back about $1,500, but I am happy to say that the 14mm being an ED is exceptionally sharp and better than the 20mm I had been using. A 1.7X camera would not work for me at all since I need very wide angle views much of the time. It may work for a lot of people, but not me.[/font]

[Quentin]

[font color=\'#000000\']Michael et. al.

What do you think of the images from the Sigma SD9 with the Foveon X3 chip?......

  Are you all as impressed with this camera as Phil and I are?  [/font]

[font color=\'#000000\']Not really.

Its like the AMD Athlon XP v Pentium 4 battle.  The Athlon rates its processors at Pentium equivalent speed.  Does the fact that the Athlon is superior in this regard render the Pentium obsolete?  Not as long as Intel keep pushing up the P4 speed.

Of course, its a bit more complex than that, and I like the look of the Foveon images, but then I also like the look of the new 1Ds images too.  Nothing stands still.  But the traditional CCD / Cmos technology is not just getting more pixels, but better image quality too.  

And there are inherent probelms with the Foveon technology at the moment.  For example, it is complex, and prone to noise at high ISO.

I'd like to see what a 6 or 9mp Foveon could do.  Then it might take off (being roughly equivalent to 12 or 18 mp Bayer respectively).  But by then, will "Bayer" pattern sensors be at 24 or more mp - like the Pentium analogy?  who knows.

Quentin[/font]

[Stefan klocke]

[font color=\'#000000\']>>To me, the huge win for the foveon chip is the lack of
>>artifacts.

The biggest being moire.

>>DPReview has some extensive threads on this on this topic.
>> According to many there, since the Foveon images lack the
>>Bayer interpolation artifacts and have true pixel-level color
>>rendition, they upsize and print much better than Bayer
>>images with twice the resolution.

like most things, the truth is sort of grayish. The foveon has to do some guesswork of its own to sort out what responce belongs to what color. Reds seem to suffer most, even on ISO100 setting from what I can see. It remains of course true that they are doing more measurements, so one would hope that that shows in the result. If X3 can get the horrible shadow color noise (not so much luminance noise) under control, then we are in for a great future.

It is kind of sad, that the discussion on dpreview has ended in fundamentalistic statements, like the X3 does measure better therefore Bayer interpolation is bad and will have no future,  instead of looking at results, I tend to look at prints. If a bayer sensor with all its shortcommings delivers a very good result, then it is nice to know that a much more complex sensor can deliver a potentially better one with less pixels, but does it matter? The same holds of course the other way round, if an X3 can deliver in less storage MBs a better image than a larger Bayer sensor with more storage requirements, why bother with a "larger" imager. So far all printing ends in upsampling.

First X3 results seem very positive to me, but actual comparisons, like the one of the japanse Photographer Yamada, leave quite mixed feelings. Having super perfect pixels  is nice but upsampling (guessing the missing data) is like water in wine. At least the Bayer source has some information from those extra pixels that still have to be invented. For me the whole reasoning that X3s have to invent less data, is only a major factor if one talkes about the same pixel count, then an x3 is a great great step forward.

I am sure that soon we will all know what the X3 delivers and what not, hopefully the emotions, will go out of the discussion.  

>>I'm just stunned by the amazing pixel-level clarity, which I
>>have never seen from any Bayer image, even the D60.

Bright future ahead it seems. Did you ever look at Kodak 760? pictures without AA filter? There is more potential in Bayer then meets the eye. (If I only could believe that from >14MP one does not need an AA filter)

Stefan Klocke[/font]

[mmurph]

[font color=\'#000000\']

Did you ever look at Kodak 760? pictures without AA filter? There is more potential in Bayer then meets the eye. (If I only could believe that from >14MP one does not need an AA filter)

Stefan,

I'm not familiar with the 760 images - I'll go do some research.  

What do you predict for the 14n given that it has no AA filter (perhaps in relation to the D60 or 1Ds)?

Michael T.[/font]

[b2martin]

[font color=\'#000000\']Anyone aware of the Foveon sensor being used for a film scanner?  I think it could reduce the apparent film grain since the sensors are more optimized for the wavelength of light being measured.  It would work best if the Nikon LED light source was used.  Any comments?[/font]

[b2martin]

[font color=\'#000000\']I read that there is an apparent increase in film grain of scanned negatives versus what you get if prints are made directly from the negatives.  I also understand that scans of positives don't show the same apparent increase in film grain. I do see more grain in scans of negatives compared to prints from the same negative.  I don't have any experience with scan and prints from positives.  Scanner noise was given as the reason for an apparent increase in the film grain.  Since all films scanners I am aware of use a CCD sensor where all sensor elements for all three colors have the same junction depth for the sensing element, this means the signal to noise ratio for all colors is not optimized.  This would show up first with the blue light since it is adsorbed at the shallowest depth.  The Foveon sensor is different since the junction depth of the elements varies for each of the three colors.  We should see an improved signal to noise ratio relative to the current sensor.  I don't know if this will improve the apparent grain (due to noise), but it is going in the right direction.[/font]

[Matthew Cromer]

[font color=\'#000000\']Michael et. al.

What do you think of the images from the Sigma SD9 with the Foveon X3 chip?

The images that Phil Askey took with foliage are, to me, the most realistic images of fine-grained leaf foliage I have ever seen at 100% pixel magnification.  I've never liked the look of foliage on a Bayer chip, but seeing the Foveon rendition just floored me.  For my flavor of landscape photography the Foveon technology has now become a "must have" technology with my next camera.

I was completely skeptical of the SD9 and Fovean, and felt that a 3MP camera was hopelessly outdated technology.  After reviewing some of Phil's images, I'm now a believer.  Let's hope they can get the MP count up soon -- I feel that the 3MP Fovean is at least up to the standard of a D60 in image quality after ressing up some fovean images to 6MP.


Curious what the rest of you think.  Are you all as impressed with this camera as Phil and I are?  Yes I know most of you have no interest in getting a Sigma mount SLR, but I am referring to the images themselves.[/font]

[Marshal]

[font color=\'#000000\']Those sample pics do look good though. The SD-9 may make people who don't need wide angle shots very happy. Telephoto fans may be very happy because it'll make a 300mm a 510mm lens.[/font]

[Matthew Cromer]

[font color=\'#000000\']To me, the huge win for the foveon chip is the lack of artifacts.

DPReview has some extensive threads on this on this topic.  According to many there, since the Foveon images lack the Bayer interpolation artifacts and have true pixel-level color rendition, they upsize and print much better than Bayer images with twice the resolution.

I'm just stunned by the amazing pixel-level clarity, which I have never seen from any Bayer image, even the D60.[/font]

[Matthew Cromer]

[font color=\'#000000\']I'm not set up for print comparisons, as I don't have a large-format printer, but to my eye the Foveon images just look massively better.  Especially when looking at foliage detail.

There is more noise at certain luminance levels, but the artifacts from Bayer (now that I have seen the alternative) just bother me to no end now.

For my sense of image quality, I have to reduce a Bayer image by 50% in each direction to have the clean look of the Foveon images.[/font]

[Stefan Klocke]

[font color=\'#000000\']>>I'm not familiar with the 760 images - I'll go do some
>>research.

On Rob Gatbraith side there was last year a review, the effect  positive and negative of the AA filter was substantial. Also outbackphoto had a review or uses it?
 
>>What do you predict for the 14n given that it has no AA
>>filter (perhaps in relation to the D60 or 1Ds)?
I printed the Tokio shopping mall pictures from Yamada the D60  Version and the 1Ds Version. I upsampled the D60 to 1Ds size and applied some USM to both, whatever seemed approriate. Everytime sometimes comes along, i show both prints, 4 out of 4 picked the D60 print as being "better", I think it has more contrast, more color? I do not know. Putting my nose on the paper I can clearly see that the 1Ds has much more detail.
Same with screen compares, a simple glance is not sufficient.

Now the Kodak, my stomac feeling is that > ISO400 performance will be better on the 1Ds, if the AA filter assumption is true, then the Kodak, should make the better pictures in ISO100-200, since the resolution is closer to the lens limits, it will catch less/no moire (I see too many artifacts in the 1ds beta samples, not really bad but more then expected after the D60) and has more resolution. The above D60-1ds print example just wants to say that 25% more pixels are maybe not that relevant. In other words, the 1Ds might be the more practical camera. But the weight, the workflow, the clever JPG formats from Kodak, so much more to consider that I at least will want to wait for samples, it is all too speculative.

Stefan[/font]

[Marshal]

[font color=\'#000000\']it is all too speculative.

Stefan[/font]

[font color=\'#000000\']You got that right!![/font]

[Rainer SLP]

[font color=\'#000000\']Hi b2martin,

From where did you get that idea? A scanner just reads what is presented to him nothing more, nothing less.

If there is grain in the film, the grain will be read. If there is no grain, it will not be read.

It is that simple.[/font]

[bobtrips]

[font color=\'#000000\']Remember that with a Bayer type sensor there is a sort of averaging that goes on between sensor elements.  This would possibly 'smear' the edges of individual 'grains'.

The Foveon sensor might, in fact, be more successful at resolving the edges of the chemical particles.[/font]