[Eugene]

Originally Posted by Powerdoc Eugene. If my memory is correct, you were much interested by the foveon technology. Are you going to invest in a new sigma camera, like the one who are going to be shipped ?

A Sigma with a Foveon X3 is still a Sigma.

[stevegong]

Originally Posted by Eugene A Sigma with a Foveon X3 is still a Sigma.

Is this meant to be derogatory? I don't know too much about this thing so please explain.

[NosferaDrew]

My little coupe.

[Moogs]

Re: the new Foveon, it's way underpowered. It's still a 4.6 effective MP chip (4.6 per layer, 4.6 once composited inside the camera), despite the marketing hype. So great color but very low resolution. They screwed up / blew their chance to make some revolutionary IMO.

Right now the D200 is probably the best DSLR value out there though not the best outright in every category of performance. Still, in skilled hands you can do almost anything with it.

[Powerdoc]

In a resolution point of vue, the new foveon is equal to a 8 mpixels Bayer sensor.
I am ready to bet, that this sensor will outperform any 10 megapixels compact camera in term of resolution. PIxels are not everything, and in the compact camera market, the race to more pixel is stupid : more noise, no more sharpness.

And yes the D200 is a great camera.

[Dorian Gray]

NosferaDrew, great angle of your timeless beauty! The lens flare reduces the overall contrast nicely and compliments the old-timer aesthetic, but did you add that in PS? You could also do without the JPEG artefacts, particularly severe in the car's shadow.

Moogs, the colour resolution of a Bayer filter type sensor is much lower than the pixel count suggests, because of two main factors: the image must be heavily interpolated (i.e. made up) by a demosaicing process, and the resulting artefacts must be minimised by a strong anti-aliasing filter, which reduces sharpness. The result is that a 4.6 MP Foveon type sensor may well have colour resolution equal to the 10 MP Nikon D200, and with fewer artefacts, although strict black and white spatial resolution will still be higher on the Nikon. If anything, it is the cameras fitted with Bayer type sensors that are most reliant on "marketing hype". Besides, 4.6 MP true resolution is high enough for nearly everyone not making poster-size prints. People with 5 MP digicams who complain about poor sharpness don't realise that it's the lens, not the megapixel count, which is limiting the sharpness. One could put 100 megapixels behind a bad lens (i.e. any small zoom) and the result will still be blurred (and as Powerdoc notes, the image will also suffer from very bad noise).

That said, the Nikon D200 - a truly great DSLR - is likely a far better camera than the Sigmas for other reasons.

[julesstoop]

My rabbit when he was still alive (obviously)

[eulersmethod]

was there an oversized magicians hat involved in your rabbits death?

[turtle]

Originally Posted by chucker Yeah, forget about using your camera's B&W filter, just shoot in color. Aperture's monochrome mixer is a much more powerful tool than merely setting saturation to zero; you should look at it.

I can't find this tool. I know it is right in front of my face, but can I get some direction here please? Thanks.

[chucker]

Originally Posted by turtle2472 I can't find this tool. I know it is right in front of my face, but can I get some direction here please? Thanks.

Sure.

Do you have the Adjustments panel open? Type control-A to get it if not. Should be in the top right.

In there, you should have sub-sections like Exposure, White Balance, and such. You can customize which ones. To add one, see the top right of the panel header; there's a plus button and an actions one.

So, with the plus button, you can add Monochrome Mixer. And then you have various presets, or can customize manually.

[NosferaDrew]

Originally Posted by Dorian Gray NosferaDrew, great angle of your timeless beauty! The lens flare reduces the overall contrast nicely and compliments the old-timer aesthetic, but did you add that in PS? You could also do without the JPEG artefacts, particularly severe in the car's shadow.

Thanks!
Didn't add the lens flare - I'm a noob photographer and suck at Photoshop.

Originally shot in color with my 7 megapixel camera, I made it grayscale, sharpened it, cropped it a bit, then reduced the image size.

Full-sized shot here, if you (or someone) wants to give a brief how-to for achieving better results. It would be greatly appreciated!

Drew

[Moogs]

Originally Posted by Powerdoc In a resolution point of vue, the new foveon is equal to a 8 mpixels Bayer sensor.

Show me the math. This sounds like something you read at the dpreview forums, which are notoriously bad for spreading half-truths and misinformation, because 80% of the people in there will say anything to make thsemselves feel better about a purchase they wish to make.

The specs on this sensor are very clear. Each R,G,B layer on the sensor captures 4.6 MP worth of data, and then green doesn't count double in the final interpolation AFAIK. This is a 4.6MP camera. The physical dimension of the image is not even 3000x2000, which is 6MP. You can't just believe the hype and say "oh it's at least as good as 8MP", because there's no physical property of the camera to back that up.

Now, maybe its 4.6MP files stand up to enlargement better than most cameras in its resolution range, and maybe you can safely enlarge up to the equivalent of 8MP without blurring or artifacting, but so too can you take a clean 8MP and enlarge up to 10 or 12 if you're careful in how you do it.

I'm sure the colors will be great but truthfully, there are 4 or 5 DSLRs out there right now that give you outstanding color when used properly. Good enough that most people will never be able to tell the difference. And ultimately, it's going to be the printer and paper, not the sensor that determines how much people like the final output.

[Eugene]

Originally Posted by NosferaDrew Originally shot in color with my 7 megapixel camera, I made it grayscale, sharpened it, cropped it a bit, then reduced the image size. Full-sized shot here, if you (or someone) wants to give a brief how-to for achieving better results. It would be greatly appreciated!

Using one of channel-mixing the techniques listed above, I'd probably reduce the red influence in the photo to blacken the brake lights, ignore green altogether and push up the blues. So, 30% red and 70% blue, or even more lopsided.

Also resist the urge to sharpen photos. Your camera already does it for you, and your original looked incredible already. The standard Photoshop sharpen filter is usually way too drastic and can magnify noise, contrast, etc. If you must, use Unsharp Mask instead.

[Powerdoc]

Originally Posted by Moogs Show me the math. This sounds like something you read at the dpreview forums, which are notoriously bad for spreading half-truths and misinformation, because 80% of the people in there will say anything to make thsemselves feel better about a purchase they wish to make. The specs on this sensor are very clear. Each R,G,B layer on the sensor captures 4.6 MP worth of data, and then green doesn't count double in the final interpolation AFAIK. This is a 4.6MP camera. The physical dimension of the image is not even 3000x2000, which is 6MP. You can't just believe the hype and say "oh it's at least as good as 8MP", because there's no physical property of the camera to back that up. Now, maybe its 4.6MP files stand up to enlargement better than most cameras in its resolution range, and maybe you can safely enlarge up to the equivalent of 8MP without blurring or artifacting, but so too can you take a clean 8MP and enlarge up to 10 or 12 if you're careful in how you do it. I'm sure the colors will be great but truthfully, there are 4 or 5 DSLRs out there right now that give you outstanding color when used properly. Good enough that most people will never be able to tell the difference. And ultimately, it's going to be the printer and paper, not the sensor that determines how much people like the final output.

Resolution wise the latest foveon equal to a 8 mpixels camera for several reason.
No resolution loss due to moire filter
No resolution loss due to interpolation of the Bayer matrix, who create artifacts and other cons

This have been backed up by many reviews, and not only DP review.

Never forget that the number of pixels is not the maximum resolution of a camera. My canon 20 D has far more resolution than my coolpix P3, dispite the fact, that both have the same number of pixels.
Pixels are not everything. Resolution is more important.

PS : I don't have any foveon camera, so I don't care.
Personnaly I think that the major cons of the sigma cameras are :
- sigma mounting lenses
- high iso noise (very high)

[Moogs]

Maybe my definition of resolution and yours are different.

Do you consider the numbers

3872 x 2592 pixels

and

2896 x 1944 pixels

to be "image resolutions"?

The first is from a 10MP camera, the second a "half size" setting from the same camera. The SD14 images are 2,652 × 1,768. That is, when you open a document in Photoshop and go to "image size"... the above number is what you will see and what you will have to work with in terms of data. The above number is less than 5MP. You can spin it any way you want but it's a relatively low resolution IMO.

It may stand up to 50% Bicubic Smoother enlargements better than any other 4, 5 or 6 MP camera for example, but that doesn't make it "equivalent" to a good 8MP, because that 8MP is what it is. You don't have to stretch the image data to get there the way you would with an SD14. And don't get me wrong: I don't think it's "bad" at all; quite the contrary its colors are probably very compelling. And if all you need to do is make prints up to about 11x14. If you're regularly making prints larger than this, I wouldn't buy the SD-14, especially given the fact they're not really giving a price break despite the subpar specs.

And now back to your regularly scheduled BW photo expose.

[World Leader Pretend]

Kudos to anyone who can figure out what this is!

[billybobsky]

Originally Posted by Moogs Maybe my definition of resolution and yours are different. Do you consider the numbers 3872 x 2592 pixels and 2896 x 1944 pixels to be "image resolutions"? The first is from a 10MP camera, the second a "half size" setting from the same camera. The SD14 images are 2,652 × 1,768. That is, when you open a document in Photoshop and go to "image size"... the above number is what you will see and what you will have to work with in terms of data. The above number is less than 5MP. You can spin it any way you want but it's a relatively low resolution IMO. It may stand up to 50% Bicubic Smoother enlargements better than any other 4, 5 or 6 MP camera for example, but that doesn't make it "equivalent" to a good 8MP, because that 8MP is what it is. You don't have to stretch the image data to get there the way you would with an SD14. And don't get me wrong: I don't think it's "bad" at all; quite the contrary its colors are probably very compelling. And if all you need to do is make prints up to about 11x14. If you're regularly making prints larger than this, I wouldn't buy the SD-14, especially given the fact they're not really giving a price break despite the subpar specs. And now back to your regularly scheduled BW photo expose.

I think the comparison shouldn't be resolution but rather quality. A 4.6 MP cmos photo will have a lower quality than a 4.6 MP foveon photo... I don't believe it will survive blowing up anymore than a cmos photo; and while it might be a little less fuzzy at full zoom, that fuzziness is caused by our brain and not by the sensor. However, they are offering 10.2 and 14.1 mp sensors... so the revolution will probably continue.

I also like the fact that the pixels are individually addressed, which might allow for higher contrast control with smart programming...

[Eugene]

Originally Posted by World Leader Pretend Kudos to anyone who can figure out what this is!

A lane marker?

Originally Posted by Moogs The first is from a 10MP camera, the second a "half size" setting from the same camera. The SD14 images are 2,652 × 1,768. That is, when you open a document in Photoshop and go to "image size"... the above number is what you will see and what you will have to work with in terms of data. The above number is less than 5MP. You can spin it any way you want but it's a relatively low resolution IMO.

If you were in RAW you will have 14mp worth of "data" to work with, same as Fuji's 12mp SuperCCD. If you check the size of the RAW files, you'll be able to confirm this.

It may stand up to 50% Bicubic Smoother enlargements better than any other 4, 5 or 6 MP camera for example, but that doesn't make it "equivalent" to a good 8MP, because that 8MP is what it is. You don't have to stretch the image data to get there the way you would with an SD14.

The bottom line is a RAW file from a conventional camera must first be run through bayer interpolation to create a usable image. This process *is* stretching to some extent. Luminance affects sharpness...and with incorrect levels of red light being measured by a blue photodiode, do you think the luminance will be accurate?

[Majost]

Originally Posted by Moogs Maybe my definition of resolution and yours are different. Do you consider the numbers 3872 x 2592 pixels and 2896 x 1944 pixels to be "image resolutions"? The first is from a 10MP camera, the second a "half size" setting from the same camera. The SD14 images are 2,652 × 1,768. That is, when you open a document in Photoshop and go to "image size"... the above number is what you will see and what you will have to work with in terms of data. The above number is less than 5MP. You can spin it any way you want but it's a relatively low resolution IMO. It may stand up to 50% Bicubic Smoother enlargements better than any other 4, 5 or 6 MP camera for example, but that doesn't make it "equivalent" to a good 8MP, because that 8MP is what it is. You don't have to stretch the image data to get there the way you would with an SD14. And don't get me wrong: I don't think it's "bad" at all; quite the contrary its colors are probably very compelling. And if all you need to do is make prints up to about 11x14. If you're regularly making prints larger than this, I wouldn't buy the SD-14, especially given the fact they're not really giving a price break despite the subpar specs. And now back to your regularly scheduled BW photo expose.

Now, I may be a bit of a novice here, but I have to disagree with you. The only reason the traditional 8 megapixel camera is 8 megapixel is because extra pixels were interpolated. By the same rationale, you could apply a similar computation to the raw data of the Foveon to increase the pixel count, making it a similar quality to a higher "resolution" traditional sensor.

Edit: Bah, I got interrupted with the phone, and Eugene beat me to essentially the same point. I don't mean to beat it into the ground.

[World Leader Pretend]

Originally Posted by Eugene A lane marker?

Nope, no roads involved.

[Jamie240]

A mirror panel?

[World Leader Pretend]

Ding Ding Ding!!

[Powerdoc]

Moogs : When I speak of resolution, I speak of the ability of a camera to capture the maximum of information in a pic.
Shoot a resolution pattern , and see how much resolving power your camera has. If the camera is able to record very tiny details, then it has a good resolving power, otherwise not.
This ability to have a good resolution is based upon several factors :
- the sensor
- the lens in front of it
- the camera post processing (in the camera, or in your computer in case of raw)

Imagine for example that you have a 100 millions pixels sensor, with a terrible lens in front of it : the resolving power would be very low.

Now for the sensor, the resolution is not on par with the maximum number of photosites, because there is various loss even with a perfect lens in front of it. Basically you have various factors of resolution loss :
- the sensor itself : moire filter, depth of the photosites, quality of the micro lenses ...
- the post processing : dematrixing (Eugene explained why it hurt resolution), and image enhancement (suppression of noise, sharpening ...)

That's why a classical DSLR camera has more resolving power than the compact counterpart, and why foveon camera has a 30 % pixel resolution advantage compared to Bayer based matrix camera.
I am ready to bet, that you will be able with the new foveon sensor to do nice A2 print.

[Jamie240]

Originally Posted by Jamie240 A mirror panel?

Originally Posted by World Leader Pretend Ding Ding Ding!!

I thought that the minute I saw it.

I'm not sure why exactly.

Maybe I'm a genius or something.

[World Leader Pretend]

Maybe...

Here's a different angle of the same picture:

[Moogs]

Powerdoc: I see what you're saying now. I might agree with your argument as you just described it, but it's wait and see at this point, since no one out in the real world other than reviewers (many of whom are afraid to give bad reviews) has the camera.

"Resolving Power" might be the phrase I would use to describe what you're talking about, but there is no official term, so I can understand why you used "resolution" earlier. So IOW, I concur that the camera might have the resolving power of some 8MP cameras when standing 6 feet away from a chart, particularly non-DSLR models with fixed lenses like the Canon S and G series. But I still question whether the images will have the "enlargeability" (new word!) of a true 8MP DSLR shot (from a good lens) when shooting real world subjects. Especially with the noise trade-off, as you mentioned. Enlarging noise... not so good. I guess we'll find out how that camera's images stand up to interpolation.

Either way it's disappointing to me that the Foveon sensor itself is not very "scalable" it seems. I had hoped that one day many cameras might use next generation Foveon sensors that were in the 8 to 12 MP (actual) ballpark, along with all the other benefits it brings. This chip was under development at least 4 years and it's still in the same spot basically. Not encouraging for the future.

[Eugene]

Originally Posted by Moogs But I still question whether the images will have the "enlargeability" (new word!) of a true 8MP DSLR shot (from a good lens) when shooting real world subjects. Especially with the noise trade-off, as you mentioned. Enlarging noise... not so good. I guess we'll find out how that camera's images stand up to interpolation.

The fact that you still question all of this is somewhat annoying. And what noise trade-off are we talking about here?

Either way it's disappointing to me that the Foveon sensor itself is not very "scalable" it seems. I had hoped that one day many cameras might use next generation Foveon sensors that were in the 8 to 12 MP (actual) ballpark, along with all the other benefits it brings. This chip was under development at least 4 years and it's still in the same spot basically. Not encouraging for the future.

It's not scalable because Foveon doesn't have a purse like Sony, Canon or even Fujifilm. Also you're talking about mass-market sensors with 24 and 36 million pixels. Economies of scale at work here in addition to the technology just being far more complex.

[Moogs]

Originally Posted by Eugene The fact that you still question all of this is somewhat annoying. And what noise trade-off are we talking about here?

Sorry to annoy you, but maybe you're taking it kind of personally. You don't work for Foveon do you? Maybe
"trade off" was the wrong choice of phrase. Prior generations of Sigma cameras produce fairly noisy results on shots with relatively long exposures (1/8, 1/4, 1/2, 1", etc), or low light exposures. Hence, when enlarging them out of camera, the results aren't likely to = 8MP (or 10MP or whatever) when you have the same shot from another high-end, higherMP camera. Another point is that when you push the ISO on the older Sigmas, the noise doesn't increase as dramatically as other cameras, but the color accuracy that is the hallmark of the camera, drops off noticeably.

We'll see with the SD-14. Maybe they've made great strides in the ISO/noise department, but fundamentally it's the same technology that's creating the image so I'm a skeptical bastard until I see evidence from out in the real world. And, FWIW, I'm equally skeptical with Nikon when they introduce cameras like the D2x and D200, touting great noise performance. So far only the D200 and a couple of Canon DSLRs have really beaten the noise game IMO. The S3 Pro is pretty good also. Just about everything else suffers badly once you get to about ISO 400, unless it's a very well lit scene.

It's not scalable because Foveon doesn't have a purse like Sony, Canon or even Fujifilm. Also you're talking about mass-market sensors with 24 and 36 million pixels. Economies of scale at work here in addition to the technology just being far more complex.

I recognize they don't have the money those other companies do, but they have enough that if it were technically practical to scale it up to say 6.5 MP per layer, they would've done that. It's obviously a technical barrier where to go to the next level (and by that I mean just a few extra MP per layer, not some huge amount), the cost would be gi-normous. Now that we know that fine, but two months ago nobody knew what the next generation was going to bring and many expected more.

Again I'm not saying Foveon is a bad technology I'm just saying as a photographer, it would've been exciting if they could've figured out a way, given their 4 year hiatus, to bring a new Foveon chip to the *pro* (not mass) market that other manufacturer's could license for use in their cameras. I think at low MP counts, Foveon is probably the best technology, but the simple fact is none of its competition are using low MP counts, and in some cases they're playing from a point of near-parity in terms of color fidelity and noiseless images at low ISO.

That's all the more I'll say of it. To the others, sorry for my part in the hijacking.


And now for something completely different.... a BW picture.

http://forums.applenova.com/attachme...1&d=1163172006

[Powerdoc]

Originally Posted by Moogs Powerdoc: I see what you're saying now. I might agree with your argument as you just described it, but it's wait and see at this point, since no one out in the real world other than reviewers (many of whom are afraid to give bad reviews) has the camera. "Resolving Power" might be the phrase I would use to describe what you're talking about, but there is no official term, so I can understand why you used "resolution" earlier. So IOW, I concur that the camera might have the resolving power of some 8MP cameras when standing 6 feet away from a chart, particularly non-DSLR models with fixed lenses like the Canon S and G series. But I still question whether the images will have the "enlargeability" (new word!) of a true 8MP DSLR shot (from a good lens) when shooting real world subjects. Especially with the noise trade-off, as you mentioned. Enlarging noise... not so good. I guess we'll find out how that camera's images stand up to interpolation. Either way it's disappointing to me that the Foveon sensor itself is not very "scalable" it seems. I had hoped that one day many cameras might use next generation Foveon sensors that were in the 8 to 12 MP (actual) ballpark, along with all the other benefits it brings. This chip was under development at least 4 years and it's still in the same spot basically. Not encouraging for the future.

It seems that we almost agree now.

The foveon could be a very nice portrait or landscape camera (I have seen some great pics, color wize from the previous sigma SD 10 on the web), but unfortunately, I am not ready (like many other people) to invest in a niche market.

There is no problems for interpolating foveon files, or any great pics DSLR files. I have seen examples of 6 mpixels cameras, printed in A2 who look very nice (D70). If the initial file is very sharp, you can enlarge it without any problems. People have reported an amazing butterfly giant poster coming from a nikon D2H (at a photo expo). So personally, and from the advice from many advanced amateurs or pro, it's not really a big deal

If Sigma was making, Canon or NIkon foveon camera, I would have been interested, for some kind of pics. Unluckily, you can only use Sigma for sigma cameras.

Currently my 5D does the job

[stevegong]

yay or nay?

(taken with an infrared filtre, more on my site www.stevegongphoto.com)