Go Back   Steve's Digicams Forums > Digital SLR and Interchangeable Lens Cameras > Sigma dSLR

Closed Thread
 
Thread Tools Search this Thread
Old Sep 25, 2004, 5:43 PM   #51
Member
 
Join Date: Sep 2004
Posts: 66
Default

Well now Lin, if you weren't letting your emotions get in the way, then you wouldn't need to insist that you weren't.

As for "meaningless" full color testswhich show that 3.43 millionRGB triples (SD9) are in fact more plentiful than 2.75million (1Ds), I suppose that all depends on your somewhat peculiar anti-Foveon agenda. Suffice to say, these test prove beyond rational (as opposed to emotional) doubt that the SD9 does in fact have10.3M sensors, and thusit has 10.3MP's worth of full color resolutionwhen imaging full color. Crazy claim, I know.

The only time it is very nearly equalled by a 6MP-monochrome DSLR (only 1.5M RGB triples)is when imaginga monochrometwo colorblack on white subject,which isn't so easy to find in the real world.


SigmaSD9 is offline  
Old Sep 25, 2004, 6:43 PM   #52
Moderator
 
Join Date: Jun 2002
Posts: 1,139
Default

Quote:
Well now Lin, if you weren't letting your emotions get in the way, then you wouldn't need to insist that you weren't.
I see - does this mean you are an M.D. practicing Psychiatry? :-)

You are in more dire need of a colonic than any single individual I've had the pleasure of communicating with.
Quote:
As for "meaningless" full color tests which show that 3.43 million RGB triples (SD9) are in fact more plentiful than 2.75 million (1Ds), I suppose that all depends on your somewhat peculiar anti-Foveon agenda. Suffice to say, these test prove beyond rational (as opposed to emotional) doubt that the SD9 does in fact have 10.3M sensors, and thus it has 10.3MP's worth of full color resolution when imaging full color. Crazy claim, I know.
I have no anti-Foveon agenda, but it's quite obvious you have a "pro" Foveon "agenda". Would you care to furnish a single example of my alleged "anti Foveon" agenda?

You seem not to understand English, would you prefer that I communicate with you in whatever your native tongue is? Your convoluted sense of logic is superceeded only by your ignorance of photography.

Does anyone say that the SD9 has less than 10.3 million sensors? I haven't heard anyone say that and if you can quote where I've said that then please do or stop lyingand living in your fantasy world or whatever is driving your zealotry.

What's "full color mean?" You've already shown your ignorance of film by using that term and the red on blue resolution charts you keep referring to don't seem to qualify either so you're correct in one sense - it's a "crazy claim".

Let's talk about why you've been banned multiple times over at DPReview. Care to comment on that? Could it be that you've been spreading this type of nonsense to the point that no one over there takes you seriously on any subject?

It's been amusing, but a waste of time to continue trying to get you to look at reality. You have you Foveon filter on and can't see anything else (can't imagine why your "handle" is what it is) and trying to get you to see beyond it is fruitless.

As soon as the Canon 1DS Mark II is released I'm quite certain you'll be back with your red on black "resolution charts" and vivid imagination trying to justify that the SD9/SD10 have greater resolution in the "full color" world. Good luck Bozo.....

Just remember that you're doing neither Foveon nor Sigma any favors by your zealotry, simply wasting bandwidth and enjoying your pseudo-science trip through fantasy land. No need for continued discussions because your responses are like a broken record.

Lin

Bye...
Lin Evans is offline  
Old Sep 26, 2004, 5:45 AM   #53
Member
 
Join Date: Aug 2004
Posts: 34
Default


Now this is from a simple discussion on X3 vs Bayer senors on the net: http://www.ddisoftware.com/reviews/sd9-v-bayer/

Quote:
*Bayer sensors have a very significantly reduced resolution when resolving detail comprised of mainly red/blue primaries, such as a red sports car with black pin stripes, a blue sweater with red lettering, red soda can with black lettering, etc. In these cases, resolution of the Bayer sensor is reduced to less than 1/4 of its "image" resolution! Black and white details will show the highest resolving power on a Bayer sensor, while saturated color detail will vary greatly. A Foveon sensor is much more consistent, resolving near the full resolution of the images for every color combination. *Bayer sensors will produce images that are softer and less detailed due to the "smoothing" needed to eliminate artifacts and color distortions. *Bayer sensors tend to omit chrominance (color) information when sampling high frequency detail. If you look at a picture of a tree that has many small branches with a brick wall behind it for example, you will see that many of the smaller branches "morph" into the color of the bricks in the background. This is because the branches are not wide enough to cover the multiple pixels needed to derive accurate color information on a Bayer sensor. Full color sensors completely eliminate this problem. *Bayer sensors tend to produce color moire on high frequency detail like the cuff of the jeans below (more visible in center image).
Did you spot this:

Quote:
In these cases, resolution of the Bayer sensor is reduced to less than 1/4 of its "image" resolution! Black and white details will show the highest resolving power on a Bayer sensor, while saturated color detail will vary greatly.
This is obvious to anyone who knows a little about sensors and their built in problems, and yet you try to trash me writing something like this:

Quote:
Arguing that "color resolution" is somehow "real" and that "black and white" resolution is somehow "different" simply shows a lack of understanding of the term "resolution."
Enough said. Bye! :lol:
code is offline  
Old Sep 26, 2004, 6:55 AM   #54
Member
 
Join Date: Sep 2004
Posts: 66
Default

Lin Evans wrote:
Quote:
Well now Lin, if you weren't letting your emotions get in the way, then you wouldn't need to insist that you weren't.
Quote:
I see - does this mean you are an M.D. practicing Psychiatry? :-)

You are in more dire need of a colonic than any single individual I've had the pleasure of communicating with.
Hey, you lightened up for a second, good job.
Quote:
As for "meaningless" full color tests which show that 3.43 million RGB triples (SD9) are in fact more plentiful than 2.75 million (1Ds), I suppose that all depends on your somewhat peculiar anti-Foveon agenda. Suffice to say, these test prove beyond rational (as opposed to emotional) doubt that the SD9 does in fact have 10.3M sensors, and thus it has 10.3MP's worth of full color resolution when imaging full color. Crazy claim, I know.
Oh darn, the rest of your response was immature nonsense and childish name calling. Too bad, because this is an issue (the difference between effectively sensing color/light,and sensingblack/no light) all digital enthusiasts really need to learn and understand. It'll become even more important of Canon successfully navigates Foveon's patents and release their verison of Foveon's full color CMOS (yes, they are trying to catch up).

As far as this thread,I think you do understand, now,that full color performance is not shown in a test that only images a B&W target. And that all sensors ina Bayer mosaic, R,G, and B, can return an "accurate" black value, no reading/light, which inflates their measured resolution performance vice when light is involved and those sensors are then parsed. Foveon doesn't have all-monochrome photosites, so there is no difference between monochrome vice full color, so no advantage to imaging black.

We could go on ad infinitium, with me posting test after test that proves that 10.3Mis greater than6M, and you throwing Tourettian rage after childish fit, but lets not, ok? It only makes you look bad and I'm not into that. So for the last time, do not reply to me anymore, ok?
SigmaSD9 is offline  
Old Sep 26, 2004, 2:09 PM   #55
Member
 
Join Date: Sep 2004
Posts: 66
Default

code wrote:
Quote:
Now this is from a simple discussion on X3 vs Bayer senors on the net: http://www.ddisoftware.com/reviews/sd9-v-bayer/

Quote:
*Bayer sensors have a very significantly reduced resolution when resolving detail comprised of mainly red/blue primaries, such as a red sports car with black pin stripes, a blue sweater with red lettering, red soda can with black lettering, etc. In these cases, resolution of the Bayer sensor is reduced to less than 1/4 of its "image" resolution! Black and white details will show the highest resolving power on a Bayer sensor, while saturated color detail will vary greatly. A Foveon sensor is much more consistent, resolving near the full resolution of the images for every color combination. *Bayer sensors will produce images that are softer and less detailed due to the "smoothing" needed to eliminate artifacts and color distortions. *Bayer sensors tend to omit chrominance (color) information when sampling high frequency detail. If you look at a picture of a tree that has many small branches with a brick wall behind it for example, you will see that many of the smaller branches "morph" into the color of the bricks in the background. This is because the branches are not wide enough to cover the multiple pixels needed to derive accurate color information on a Bayer sensor. Full color sensors completely eliminate this problem. *Bayer sensors tend to produce color moire on high frequency detail like the cuff of the jeans below (more visible in center image).
Did you spot this:

Quote:
In these cases, resolution of the Bayer sensor is reduced to less than 1/4 of its "image" resolution! Black and white details will show the highest resolving power on a Bayer sensor, while saturated color detail will vary greatly.
This is obvious to anyone who knows a little about sensors and their built in problems, and yet you try to trash me writing something like this:

Quote:
Arguing that "color resolution" is somehow "real" and that "black and white" resolution is somehow "different" simply shows a lack of understanding of the term "resolution."
Enough said. Bye! :lol:
You are exactly right, Lin is holding on to the belief that a monochrome sensoractually resoves more from the same number of individual sensors than a full color sensor can do with the same. All that with the inherent design inefficiencies of a 1 layer mosiac. That's obviously incorrect.

In fact Foveon is way ahead of all other camera-CMOS makers in terms of underlying fabrication technology, so if anything Foveon will extract more bang per sensor than the other manufactures who are still using very outdated fabrications (maybe some of the new Canon's this Photokina will catch up a little, but they are in a pretty deep hole so I doubt it). The 1Ds sensor, for example, uses 0.5u fabrication, similar to a 386 processor. The newer 10D uses 486-era 0.35u fabrication. The Foveon 10M in the SD9 is built with 0.18u fab technology, the same asthe original Pentium4. It's nice to be owned by National Semiconductor, up-to-date fabriction processes/plants are extrememly expensive.

All else being equal (which its not)a monochrome Bayer sensor like the one in his 1Ds does fine in some color areas and not so fine in others, because its one layer monochrome design liimits the way the sensors can be arranged. As long as it is 2 dimensional, 3 primaries won't scale properly into the mosiac, which will have blocks awlays divisible by 2. Bayers are still very simple, using an RGGB 4-sensor iterative pattern, where one color has to be doubled--no choice with 4 slots. This produces a major resolution hit which in the past Bayer manufactures have tried to cover with the "eye is most sensitive to green" bunk. Sony (the world's largest Bayer manufacturer) recently crushed that long time scam with the revaltion that splitting the 4th slot in half blue/half green (cyan, they call it emerald--more catchy) improves color response dramatically, halving theinterpolative errors induced by simple double green. In other words, color bias is bad, you want balance since the final display media forces an equal amount of RGB data is required per presentationpixel (the deficit of Bayer is simply interpolated, made up). Foveon has a sound optical design in this respect, so interpoation is not necessary, all the RGB data presented is real.

Additionally, Foveon needs no blur filter to avoid offset-monchrome-mosiac-induced rainbowing in areas of intricate detail. So optically, Foveon is a direct image sensor while Bayer images are always slightly intentionally blurred.

The only reason I say all of the above is to note that if anything, Foveon extracts slightly more bang fromtheir 10.3M sensors than any of the old Bayer CMOSs on the store shelves now can do per capita. The only way a 6MP (monochrome MPs) can appear competitivewith a 10.3MP Foveon camera is to use a black on white resolution chart, since color interpolation works best when no color is involved. A B&W chart lets any color sensor (filter, actually)in the Bayer mosaic work equally efficiently, since there is nominally nolight there, they all simply read nothing to sense the black areas. This increases their appart optical resolution by about 400% over their worst cases, red and blue, and 200% over their bast case, green. Its not a straight line gainthough, since there is white on the chart too which requires a little work on the part of the Bayer sensor. An all-black image is actually the only time a Bayer sensor accurately outputs is advertised number of MPs--not a lot of value there.

If there is light, i.e. a full color target, Foveon is a 10.3MP Bayer equivalent or slightly better, depending on how old the Bayer CMOS fabrication technology is. The old 1Ds sensor is worst case given the Bayer DLSRs currently for sale, although the 14c/n has overall image quality problems as well, compared to a the more modern sensors, anyway. See:

http://www.pbase.com/dslr_noise/image/32014486/original
SigmaSD9 is offline  
Old Sep 26, 2004, 2:25 PM   #56
Moderator
 
Join Date: Jun 2002
Posts: 1,139
Default

Quote:
Foveon doesn't have all-monochrome photosites, so there is no difference between monochrome vice full color, so no advantage to imaging black.


Then the maximum resolution of the SD9/SD10 is precisely as measured at 1550 lines horizontal and 1550 lines vertical by your own admission.

Lin
Lin Evans is offline  
Old Sep 26, 2004, 3:19 PM   #57
Member
 
Join Date: Sep 2004
Posts: 66
Default

B&W test. Agrees with all the other data absolutely perfectly. The SD9 outresolves an only 6MP-monochrome Bayer only by about 6%when using a B&W chart. Here are the exact numbers in all color channel, as compared to the 10D:

When both DSLR images are enlarged or reduced to the same pixel dimensions, the SD9 very slightly outresolves the Canon 10D in black on white, and has significantly higher resolution in black on color tests. The SD9's maximum line pair resolutions, normalized to the 10D's results were:

Black on White - 106% of the 10D
Black on Red - 127% of the 10D
Black on Blue - 115% of the 10D
Black on Green - 106% of the 10D

Note: The SD9 used a wider focal length to equal the field of view due to slight crop factor differences. The SD9's resolution is slghtly higher than shown when using the same lens, with field of view then being slightly smaller.

Note: It takes twice the MPs to raise resolution 25%, all else being equal.

The SD9's number of indivdual color sensors, presented in the same format is:

Overall - 164% of the 10D
Red - 217% of the 10D
Blue - 217% of the 10D
Green - 109% of the 10D


Thanks for backing up my point. It's great that the dpreview data is exactly in line with all the others. Proves all the numbers a re all as close to perfect as possible.

BTW, the Fuji B&W numbers are skewed a little due to its brick pattern sensor orientation, which is optimized for detecting vert/horizontal lines (i.e. its cooked for a known test) at the direct and corresponding expense of 45 degree diagonal resolution. Vert/Horiz is a few percent higher, 45 degre diagonal is the same percent lower, net gain is 0, net loss is 0.

------

Late add:

Oh and the 1Ds's 11M monochrome sensors should do a lot better than a 6MP-monochrome or a 10.3MP SD9 when imaging a 2 color B&W chart, obviously.

Overall, the 1Ds is a good DLSR with 11M sensors, although it is a horrendousbuy. If it had an up to date sensor it should slightly outresolve the 10.3MP SD9, overall.But unfortunately its image quality per pixel is quite poor due to its obsolete sensorfabrication, small pixels, and resulting high noise. It was pretty good when it came out. Here is how badly the 1Ds loses to the SD9 in Red & Blue:

http://www.outbackphoto.com/artofraw/raw_05/essay.html

Of course you can buy all-stronger lenses to eliminate the 1Ds's full frame FOV advantage, concentrate it's 11M sensors on the same imaged area, and instantly negate all the money spent for that nearly-unique 35mmfull frame advantage. Then you have an $8000 DLSR with a tiny bit moreoverall color resolution (about 7% more MPs, which equates to maybe 4% across the board, but then again the blur filter and full frame glass optically hurt it even more) as compared to the 10.3MP SD9, but with full frame distortion andtoo high ISO 100 noise and up. Not a good deal unless you live and die by wider angle than about 14mm (35mm equivalent). Canon really needs to update their underlying sensor fabrication technology. Hopefully their newer CMOSs will be manufactured by a sufficiently high-tech semiconductor firm.
SigmaSD9 is offline  
Old Sep 27, 2004, 1:43 PM   #58
Moderator
 
Join Date: Jun 2002
Posts: 1,139
Default

Quote:
B&W test. Agrees with all the other data absolutely perfectly. The SD9 outresolves an only 6MP-monochrome Bayer only by about 6% when using a B&W chart. Here are the exact numbers in all color channel, as compared to the 10D:

No, the SD9 doesn't "outresolve an only 6 mp bayer" by any amount when using a B&W chart. Read the numbers......

The Fuji S2 is a six megapixel bayer which produces 1800 lineshorizontal and 1700 lines vertical. You need to be specific. The Canon D60 outresolves the SD9 on the horizontal axis which the SD9 outresolves the Canon D60 on the vertical axis.

The six megapixel Fuji S2 outresolves the SD9 by a good degree on both axis, and is not "cooked" for the test. Real world observations bear out the increased resolution over the 10D and the SD9/SD10 by considerable margins.
Quote:
Overall, the 1Ds is a good DLSR with 11M sensors, although it is a horrendous buy. If it had an up to date sensor it should slightly outresolve the 10.3MP SD9, overall.
It doesn't "slightly" outresolve the SD9/SD10, it blows the SD9/SD10 completely out of the water in real life photography. Only a fool would argue differently after using both cameras side by side. I do this every single day - there is absolutely no comparison in resolution.


Lin


Lin Evans is offline  
Old Sep 27, 2004, 8:46 PM   #59
Junior Member
 
Join Date: Jul 2004
Posts: 2
Default

I have been listening to these arguments and counter arguments for some time. Clearly everyone here has impassioned views and everyone has the right to share their opinions. That said...this is a Sigma forum. It is probably best served by folks who love the Sigma cameras and their capabilities.
ebgreen is offline  
Old Sep 28, 2004, 12:01 AM   #60
Member
 
Join Date: Sep 2004
Posts: 66
Default

You seem to have trouble distinguishing between B&W and color tests. B&W tests are those where only black and white are involved, you can tell right away because they kinda look like a page from a text book, or a speed limit sign, or the Moon. Color tests are those where an unbiased bystander might remark, "hey, isn't that pretty," or, "ouuu, isn't that colorful?"

When imaging2 color B&W targets, a Bayer's performance is inflated by between 300% and 400%, depending on whether you give full or partial weight to the orphaned greensensors. In a real pretty color test,a 6MP Bayer certainlycannot compete well in the 10MP+ class. Sensors=sight, lack of them = blind. Cameras can't see witout sensors,it's justnot possible. Using digital interpolation to invent 66% to 75% of the data presented by inserting computer-generated placeholders after the shutter closes, never influences optical resolution in any way, good or bad.

You can rearrange sensors to help in certain ways at the direct expense ofother areas, like Fuji does when they help vert/horiz resolution by sacrificing 45 degree diagonal, but its a zero sum game so it never changes overall picture quality a bit. Think about it this way, would concentrating 75% of the sensors in the lower left hand corner of an image help optical resolution? Of course. Would it hurt resolution elsewhere? Of course.

The 1Ds is a good camera, I've said that all along. With 11M sensors it will usuallyslightly outresolvethe 10MP SD9 overall. But the old 1Ds is also relatively noisy by modern DSLR standards:

http://www.pbase.com/dslr_noise/image/32014486/original

It also uses full frame glass, which always yeildslower optical quality overall, especially WA. It also needs a blur filter since it has a mosiac sensor. But yes for a2 color B&W target it should do well with all 11M of its sensors black-capable--a novelty that's mostly unrelated to photography, unfortunately. Am I sayingthat to say all digital photography web sites are clueless? Well, uhm, no, if you stick to comparing Bayer to Bayer,there is relative value in any test.

And don't feel bad for buying a 1Ds,of course it has (more like "had" after this Photokina) a solid niche: those who live and die by ultra-wide angle. And Canon's old 1 body is very worthy structurally, though woefully outdated in terms of digital operation. So its not like it is a waste of money. I personally wouldn't be interested in 2.75M relatively noisy RGB triples upscaled into 11MP file sizes. Especially with thelow performance of Canon's RAW software.But that's just me.

And hey, it's better than 1.5M.

SigmaSD9 is offline  
 
Closed Thread


Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off



All times are GMT -5. The time now is 3:59 PM.