Foveon/Bayer comparison

The problem has essentially nothing to do with the number of colors
that can be represented in the output medium.
Then shrink it to 2 and show me what you mean.
???
jpeg compression was designed so that at low compression levels,
the information discarded is not visible to humans.
Utterly absurd.
It's funny that you find so much of reality to be absurd. The world must be very confusing to you. Go to google and search on "compression faq" and "jpeg faq". You have a lot of reading to do.

--
Ron Parr
FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/
 
Moire is caused by an interaction between the frequency of the
pattern photographed and the arrangement of the pixels. There's a
clear and consistent explanation for what's going on. Why try to
explain it away by attributing it to forces you obviously don't
understand?
JPEG shrinks the colorspace by 8000x, then introduces artifacts and
dithering that eliminate any hope of seeing and/or quantifying any
subtle detail in any image. It works fine for identifying Bayer
moire, because that is anything but subtle.
Moire in the luminance channel isn't necessarily subtle and jpeg doesn't dither. Do your reading.

--
Ron Parr
FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/
 
JPEG shrinks the colorspace by 8000x, then introduces artifacts and
dithering that eliminate any hope of seeing and/or quantifying any
subtle detail in any image. It works fine for identifying Bayer
moire, because that is anything but subtle.
jpeg does not dither and as long as the moire is in the luminace data (obviously caused by the lack of an AA filter) you can't blame jpeg for it. The point some people (including me) made long ago, that jpeg is better suited for Bayersensors, was more or less in reference to the lack of chrominance resolution of Bayer Sensors and therefore the not so bad impact of jpegs compression of the chrominace data, luminace is a different story...

--
Regards from Old Europe,

Dominic

http://www.pbase.com/sigmasd9/dominic_gross
 
JPEG shrinks the colorspace by 8000x, then introduces artifacts and
dithering that eliminate any hope of seeing and/or quantifying any
subtle detail in any image. It works fine for identifying Bayer
moire, because that is anything but subtle.
jpeg does not dither
It does effectively dither, it doesn't formally call it dithering, but cutting the unique colors in half (sort of like 23-bit) and rearranging whats left to compensate, is really just glorified dithering.
and as long as the moire is in the luminace
data (obviously caused by the lack of an AA filter) you can't blame
jpeg for it.
Did you see all the odd-ball color artifacts in the JPEG of the same image I posted?

Also realize that JPEG's biggest whack is its limited colorspace. We never see that because even a 16-bit TIF is displayed as an 8-bit TIF (since we are viewing in 24-bit colorspace) on our monitors--and you can't produce a JPEG without truncating 4095/4096ths of the colorspace.

Here's a 24-bit representation of the relative difference...





When you round each color to the neareast one up to 2048 shades in either direction for each component of R, G, and B, you are going to introduce artifiacts in areas of very sensitive patterns, artifacts that aren't present in the SD-9 RAW file.
The point some people (including me) made long ago,
that jpeg is better suited for Bayersensors, was more or less in
reference to the lack of chrominance resolution of Bayer Sensors
and therefore the not so bad impact of jpegs compression of the
chrominace data, luminace is a different story...
JPEG is definitely better suited to Bayers simply because the 3x up-sampled, interpolated image can absorb the damage better. Bayers also produce several slightly different color pixels for each full set of RGB data, so there are already plenty of artifacts to camo the new artifacts. JPEG is also harshest by far on sharp lines; a Foveon image isn't about to enjoy that.
 
jpeg compression was designed so that at low compression levels,
the information discarded is not visible to humans.
Utterly absurd.
It's funny that you find so much of reality to be absurd. The
world must be very confusing to you. Go to google and search on
"compression faq" and "jpeg faq". You have a lot of reading to do.
So you see no relative difference at all in the split images below? Are you serious?



 
First of all the most important point remains, the moire is a luminace artifact.

You claimed it is jpeg and to support that you throw tons of color / chrominace things on us. This is just BS, and if your color claims below are right or wrong does not even matter. This is about the luminace data not the chrominance data. And jpeg is simply not causing this in the luminace data!
It does effectively dither, it doesn't formally call it dithering,
but cutting the unique colors in half (sort of like 23-bit) and
rearranging whats left to compensate, is really just glorified
dithering.
It is not, dithering introduces patterns of colored dots to simulate colors that can not be represented with the current colors.

And if you think you can fool me by putting screenshoots of ATI demos that are about advantages of newer mapping methods (normal mapping) and the other one about HDRI ???

--
Regards from Old Europe,

Dominic

http://www.pbase.com/sigmasd9/dominic_gross
 
What exactly do you think these show?
Comically, they are non-sequiturs which has no bearing to the discussion, Ron.

The original was to show high dynamic range scene mappings; the stuff that is the rage at the last SIGGRAPH conference and the one before.

JPEG has obviously nothing to do with left/right differences, since the composite came across the web as... a single JPEG!

Ron, if you don't get the SIGGRAPH proceedings, I am pretty sure you can find the conference proceeding in one of your colleague's office. Researchers in that area has recently been spending more time (and our taxes :-) with topics that is related to high dynamic range photography. Should be interesting bedtime reading for you. Heck, graphics types are now worried about things like scotopic and photopic sight :-).
  • kc
 
What exactly do you think these show?
Comically, they are non-sequiturs which has no bearing to the
discussion, Ron.

The original was to show high dynamic range scene mappings; the
stuff that is the rage at the last SIGGRAPH conference and the one
before.

JPEG has obviously nothing to do with left/right differences, since
the composite came across the web as... a single JPEG!
Maybe that's why I was very careful to say "relative" difference, unfortunatley you were not as careful reading it.

Obviously you didn't know that JPEG'ing truncates 12-bits from the colorspace (and then reduces the remaining unique colors by about half, which we haven't even got to). To say that, alone, is not humanly perceivable is absurd. Why do you think 36-bit colorspace exists?

And we haven't even gotten to PEG'ing artifacts yet. But let's see if you can understand this first.
 
First of all the most important point remains, the moire is a
luminace artifact.
We'll get to that if we can ever get past the first point, which is that JPEG'ing, without a doubt, destroys the camera's colorspace...
You claimed it is jpeg and to support that you throw tons of color
/ chrominace things on us. This is just BS, and if your color
claims below are right or wrong does not even matter. This is about
the luminace data not the chrominance data. And jpeg is simply not
causing this in the luminace data!
It does effectively dither, it doesn't formally call it dithering,
but cutting the unique colors in half (sort of like 23-bit) and
rearranging whats left to compensate, is really just glorified
dithering.
It is not, dithering introduces patterns of colored dots to
simulate colors that can not be represented with the current colors.

And if you think you can fool me by putting screenshoots of ATI
demos that are about advantages of newer mapping methods (normal
mapping) and the other one about HDRI ???

--
Regards from Old Europe,

Dominic

http://www.pbase.com/sigmasd9/dominic_gross
--
http://www.pbase.com/imageprocessing/sd9
 
It does effectively dither, it doesn't formally call it dithering,
but cutting the unique colors in half (sort of like 23-bit) and
rearranging whats left to compensate, is really just glorified
dithering.
You don't understand what jpeg is doing, or you don't undersatnd
dithering - probably both.
"Yes," you, a human, can see the relative difference there.

or...

"No," you, a human, can see no relative difference.

Simple question.
 
I can't believe you dragged out these pictues again. Anybody can see the difference between the left and the right. The real question is, why do you think these pictures have anything to do with this discussion, JPEG compression, or digital photography??

These pictures were released by ATI over a year ago to demonstrate the difference between a scene rendered with a 128-bit floating point pipeline vs. a 32-bit integer pipeline. You can follow the link below for a more complete explanation (including the exact same images).

http://www.tomshardware.com/graphic/20020718/radeon9700-06.html#floating_point_precision_color
So you see no relative difference at all in the split images below?
Are you serious?



 
Obviously you didn't know that JPEG'ing truncates 12-bits from the
colorspace (and then reduces the remaining unique colors by about
half, which we haven't even got to). To say that, alone, is not
humanly perceivable is absurd. Why do you think 36-bit colorspace
exists?

And we haven't even gotten to PEG'ing artifacts yet. But let's see
if you can understand this first.
LMAO. Please explain the PEG'ing process for us. Hee hee...

(Just so everybody can share fun: JPEG stands for Joint Photographic Experts Group and reductions in the number of colors is not an explicit part of the jpeg standard. The "Group" is a reference to the individuals who created the standard and not come "grouping" process. At high compression levels, some software will elect to reduce the colorspace explicitly before compression, but most reductions in the range of colors come as a side effect of the quantization pase, which occurs indpependently in the individual components of YCbCr space [usually - other spaces are possible] after the discrete cosine transform. Short version: sg10 has absolutely no idea what he's talking about.)

--
Ron Parr
FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/
 
What exactly do you think these show?
Is that a yes you can see a difference, or no you cannot?
They are different images. Unfortunately they have nothing to do with the topic. Please start by apologizing to the people from whom you've stolen those images, then tell us what they have to do with moire in the luminance channel. Alternatively, you could apologize for wasting everybody's time...

--
Ron Parr
FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/
 
First of all the most important point remains, the moire is a
luminace artifact.
We'll get to that if we can ever get past the first point, which is
that JPEG'ing, without a doubt, destroys the camera's colorspace...
Just for everybody's amusement and to make the substance of your claim clear, please give your definition of colorspace and then explain in a methmatically precise manner how it gets "destroyed" by the jpeg process.

If you are unable to do this, then explain why anybody should continue to listen to you...

--
Ron Parr
FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/
 
What exactly do you think these show?
Is that a yes you can see a difference, or no you cannot?
They are different images. Unfortunately they have nothing to do
with the topic. Please start by apologizing to the people from
whom you've stolen those images, then tell us what they have to do
with moire in the luminance channel. Alternatively, you could
apologize for wasting everybody's time...
I fully agree, always introducing new points to distract from the false assumptions in the old ones is really odd.

This and the wrong claims Mr. sgIO made waste a lot of time and people that are new to the subject matter might end up very confused...

--
Regards from Old Europe,

Dominic

http://www.pbase.com/sigmasd9/dominic_gross
 

Keyboard shortcuts

Back
Top