Confused: Why are we calling this a 1920x1080 sensor?

David Ziegelheim · December 6th, 2006, 12:03 AM

One related question: on the Sony site, the comparison picture shows big black lines around the edges of window panes, white cornerstones, etc for the V1 that are not there in the other images and undoubtly not there in real life. What caused that? Only the Canon picked up detail in the window lock mechanism.

Philip Williams · December 6th, 2006, 06:24 AM

Quote:

Originally Posted by David Ziegelheim

One related question: on the Sony site, the comparison picture shows big black lines around the edges of window panes, white cornerstones, etc for the V1 that are not there in the other images and undoubtly not there in real life. What caused that? Only the Canon picked up detail in the window lock mechanism.

It looks like its just way over-sharpened. I never really understood the usefullness of that comparison chart. Its provided by Sony, so it might have the appearance of being biased. We have no clue whatsoever what the camera settings are (that alone nullifies the comparison). And frankly, when I first saw that graphic I wasn't even sure why Sony bothered; to me it looks like the HVX is resolving a pretty decent amount of detail (after being poo pooed by many for being so "low resolution") and the Canon shot looks the best to a lot of people anyway.

You know, just looking at that "comparison" is a reminder of how similarly all these cameras perform. What's the point of endlessly shooting still life and then zooming into frame grabs to see how the 700 line camcorder supposedly looks better than the 540 line camcorder. Ugh.. I think if there's another HD camcorder "shootout" I'm going to petition that no resolution charts are used. Lets look at colors, lattitude, smear, ergonomics, price.. etc.. anything other than resolution. OK, I'm going on a tirade.. better get back to work.

Graeme Nattress · December 6th, 2006, 07:33 AM

For the numbers I gave I used a 75% factor horizontally and vertically on the bayer CFA, which is about right with a good demosaic and decent optical low pass filter. For the Sony, looking at how it works, they seem to trade extra horizontal for vertical, and it seemed that, given how such things work, a 150% horizontal and a 75% vertical factor would be appropriate. AS I say, it's a guesstimate.

Yes, the lines are just over-zealous sharpening. Yes, the difference between the three is small, with to my eyes, the Canon looking superior and to have the least amount of sharpening too. It still has too much, but, less is better than more.

Graeme

Greg Boston · December 6th, 2006, 07:49 AM

Quote:

Originally Posted by Philip Williams

Ugh.. I think if there's another HD camcorder "shootout" I'm going to petition that no resolution charts are used. Lets look at colors, lattitude, smear, ergonomics, price.. etc.. anything other than resolution. OK, I'm going on a tirade.. better get back to work.

We did all of that with the Texas HD Shootout. Our goal was to not only show res charts which give a camera's potential, but to employ different scenes which would stress the codecs to their limits on each camera and show how each handled highlights and low lights. We also did res charts to see how they matched up against mfgr's claims.

-gb-

David Ziegelheim · December 6th, 2006, 07:57 AM

Isn't the premise of the Foveon that all this interpolation isn't a good thing? It is trading off 40% resolution (effectively a little less than 5MP) for effectively having 3 sensors.

Graeme Nattress · December 6th, 2006, 08:16 AM

Well, good interpolation is good, and bad interpolation is bad :-)

What Foveon doesn't tell you is that the colour you get from them is heavily interpolated as silicon is a poor colour filter. You have to do a fair bit of mangling of the data from the 3 pixels that lie on top of each other to get a real colour out. In contrast, the RAW colour from, as I know it well, Mysterium, looks pretty ok, just a little under-saturated, before we put it through a small amount of correction.

In tests a Bayer CFA looks to have an equivalent Foveon resolution of about 1/2 the mp. So, a 10mp Nikon, will look to have equivalent to a 5mp Foveon. But, for those 5mp, the foveon actually has 15mp of pixels as they're stacked. So, 10mp bayer =15mp foveon, so to me, foveon is inefficient, if rather clever. I'd also remember that Sigma cameras don't have the correct amount of anti-alias filtering on their sensor (they make this out to be a good thing, but it's not) hence half the detail you see is really aliassed, and this shows when you blow up or scale up a Sigma image as it looks like everything is stair-steppy rather than smooth.

Graeme

David Ziegelheim · December 6th, 2006, 08:50 AM

Those ratios make sense since there is excessive color information.

Why hasn't anyone used asymmetric sensors: 1920x1080 green and 960x1080 red and blue?

Graeme Nattress · December 6th, 2006, 08:58 AM

Developing and building a sensor is expensive. It makes sense to have 3 identical sensors, rather than go to the expense of making a special sensor for red and blue. Also, I'd hazard that having different sensors would make colourimetric matching more tricky, but I think the first agurment is the main one.

Having bigger pixels for the red and blue would make them less noisy, but then again, averaging pairs of pixels would do the same thing....

Graeme

Dave F. Nelson · December 6th, 2006, 02:11 PM

Quote:

Originally Posted by David Ziegelheim

One related question: on the Sony site, the comparison picture shows big black lines around the edges of window panes, white cornerstones, etc for the V1 that are not there in the other images and undoubtly not there in real life. What caused that? Only the Canon picked up detail in the window lock mechanism.

You only posted one of the images on the Sony site. There is a 3 picture series of images on the Sony site. I prefer either of the other 2 images (Sony or Panasonic) over the Canon image. The numbers don't tell it all. The proof is in the pudding. What I see is much red fringing on the vertical edges of the window in the Canon image. The Sony image in this same series appears to me to be superior. It's just as clear to my eye and does not have the red fringing problems of the Canon image. How can you overlook the red fringing and talk about resolution?

There is more going on here, and it isn't the numbers. I think this is why Sony used this image. It appears to point out problems with the lens or the processor in the Canon camera.

Graeme Nattress · December 6th, 2006, 03:27 PM

Or problems in the test itself. I'd hardly call it a controlled test, and we only get to see a small section of the image that was shot. You can't evaluate a camera from a single image.

Graeme

David Ziegelheim · December 6th, 2006, 07:35 PM

Actually, I photoshop'd the windows together an enlarged it. There was a definite purple fringe in the Canon. And a much higher level of detail. Enlarged, the Sony had pronouced edge enhancement (I assume thats what those black streaks are) and loss of surface texture (which may be an overactive noise reduction). To me, the Canon had the edge in resolving vertical lines, and the Sony had a edge on horizontal lines. However, what do I know?

Thomas Smet · December 6th, 2006, 08:08 PM

Come on, this is advertising 101.

Any company will make their product look better when compared to the competition. Is Tide really that much better then "product B", or are the ads on TV set up to make the other laundry soap not work as well?

Well over a dozen things could have been adjusted alone on any of those cameras to adjust how the image looks. The images could have also been adjusted later in photoshop.

I'm not calling SONY anything but this is just how marketing and advertising work. I would do the same thing as would anybody else on the planet if they were marketing a product that competed against another product.

My point is to not take the image from the SONY site for any decent information to judge the cameras. Either check them out yourself or wait until somebody such as Adam Wilt can compare the results and put the cameras through a fair test.

David Ziegelheim · December 6th, 2006, 11:56 PM

Generally, there have been few comparitive ads between camera manufacturers. If the add is comparative, it has frequently been against other or previous models from the same manufacturer.

Typically comparitive ads are done for products with a poor image, to properly place them in the mind of the target audience. For example, a Hyundai, Kia, Suzuki, or Mitsubishi may have comparative ads. Toyota and Honda are very unlikely to have comparative ads.

Sony would generally ignore its competition. This ad is unusual, and probably reflects information that engineering provided marketing that their CMOS progressive camera was better than its competitors.

Douglas Spotted Eagle · December 7th, 2006, 08:11 AM

Quote:

Originally Posted by Thomas Smet

Come on, this is advertising 101.

Any company will make their product look better when compared to the competition. Is Tide really that much better then "product B", or are the ads on TV set up to make the other laundry soap not work as well?

Well over a dozen things could have been adjusted alone on any of those cameras to adjust how the image looks. The images could have also been adjusted later in photoshop.

I'm not calling SONY anything but this is just how marketing and advertising work. I would do the same thing as would anybody else on the planet if they were marketing a product that competed against another product.

My point is to not take the image from the SONY site for any decent information to judge the cameras. Either check them out yourself or wait until somebody such as Adam Wilt can compare the results and put the cameras through a fair test.

No Photoshopping was done on any of those images whatsoever. I'd hate to think that Canon or Panasonic would Photoshop an image for their best interests either. Then again...some companies have done camera marketing DVDs that didn't use the camera they're marketing for the shot.

Chris Hurd · December 7th, 2006, 09:59 AM

Beware, conspiracy theorists! None of these manufacturers would ever photoshop a comparison image. They'll simulate an LCD display image on a brochure, but they won't touch a comparison shot.

December 6th, 2006, 12:03 AM	#31
David Ziegelheim Major Player Join Date: Jul 2003 Location: Warren, NJ Posts: 398	One related question: on the Sony site, the comparison picture shows big black lines around the edges of window panes, white cornerstones, etc for the V1 that are not there in the other images and undoubtly not there in real life. What caused that? Only the Canon picked up detail in the window lock mechanism. Attached Thumbnails

December 6th, 2006, 07:33 AM	#33
Graeme Nattress RED Problem Solver Join Date: Sep 2003 Location: Ottawa, Canada Posts: 1,365	For the numbers I gave I used a 75% factor horizontally and vertically on the bayer CFA, which is about right with a good demosaic and decent optical low pass filter. For the Sony, looking at how it works, they seem to trade extra horizontal for vertical, and it seemed that, given how such things work, a 150% horizontal and a 75% vertical factor would be appropriate. AS I say, it's a guesstimate. Yes, the lines are just over-zealous sharpening. Yes, the difference between the three is small, with to my eyes, the Canon looking superior and to have the least amount of sharpening too. It still has too much, but, less is better than more. Graeme __________________ www.nattress.com - filters for FCP

December 6th, 2006, 08:16 AM	#36
Graeme Nattress RED Problem Solver Join Date: Sep 2003 Location: Ottawa, Canada Posts: 1,365	Well, good interpolation is good, and bad interpolation is bad :-) What Foveon doesn't tell you is that the colour you get from them is heavily interpolated as silicon is a poor colour filter. You have to do a fair bit of mangling of the data from the 3 pixels that lie on top of each other to get a real colour out. In contrast, the RAW colour from, as I know it well, Mysterium, looks pretty ok, just a little under-saturated, before we put it through a small amount of correction. In tests a Bayer CFA looks to have an equivalent Foveon resolution of about 1/2 the mp. So, a 10mp Nikon, will look to have equivalent to a 5mp Foveon. But, for those 5mp, the foveon actually has 15mp of pixels as they're stacked. So, 10mp bayer =15mp foveon, so to me, foveon is inefficient, if rather clever. I'd also remember that Sigma cameras don't have the correct amount of anti-alias filtering on their sensor (they make this out to be a good thing, but it's not) hence half the detail you see is really aliassed, and this shows when you blow up or scale up a Sigma image as it looks like everything is stair-steppy rather than smooth. Graeme __________________ www.nattress.com - filters for FCP

December 6th, 2006, 08:58 AM	#38
Graeme Nattress RED Problem Solver Join Date: Sep 2003 Location: Ottawa, Canada Posts: 1,365	Developing and building a sensor is expensive. It makes sense to have 3 identical sensors, rather than go to the expense of making a special sensor for red and blue. Also, I'd hazard that having different sensors would make colourimetric matching more tricky, but I think the first agurment is the main one. Having bigger pixels for the red and blue would make them less noisy, but then again, averaging pairs of pixels would do the same thing.... Graeme __________________ www.nattress.com - filters for FCP

December 6th, 2006, 03:27 PM	#40
Graeme Nattress RED Problem Solver Join Date: Sep 2003 Location: Ottawa, Canada Posts: 1,365	Or problems in the test itself. I'd hardly call it a controlled test, and we only get to see a small section of the image that was shot. You can't evaluate a camera from a single image. Graeme __________________ www.nattress.com - filters for FCP

December 6th, 2006, 07:57 AM	#35
David Ziegelheim Major Player Join Date: Jul 2003 Location: Warren, NJ Posts: 398	Isn't the premise of the Foveon that all this interpolation isn't a good thing? It is trading off 40% resolution (effectively a little less than 5MP) for effectively having 3 sensors.

December 6th, 2006, 08:50 AM	#37
David Ziegelheim Major Player Join Date: Jul 2003 Location: Warren, NJ Posts: 398	Those ratios make sense since there is excessive color information. Why hasn't anyone used asymmetric sensors: 1920x1080 green and 960x1080 red and blue?

December 6th, 2006, 07:35 PM	#41
David Ziegelheim Major Player Join Date: Jul 2003 Location: Warren, NJ Posts: 398	Actually, I photoshop'd the windows together an enlarged it. There was a definite purple fringe in the Canon. And a much higher level of detail. Enlarged, the Sony had pronouced edge enhancement (I assume thats what those black streaks are) and loss of surface texture (which may be an overactive noise reduction). To me, the Canon had the edge in resolving vertical lines, and the Sony had a edge on horizontal lines. However, what do I know?

December 6th, 2006, 08:08 PM	#42
Thomas Smet Trustee Join Date: Mar 2004 Location: Milwaukee, WI Posts: 1,719	Come on, this is advertising 101. Any company will make their product look better when compared to the competition. Is Tide really that much better then "product B", or are the ads on TV set up to make the other laundry soap not work as well? Well over a dozen things could have been adjusted alone on any of those cameras to adjust how the image looks. The images could have also been adjusted later in photoshop. I'm not calling SONY anything but this is just how marketing and advertising work. I would do the same thing as would anybody else on the planet if they were marketing a product that competed against another product. My point is to not take the image from the SONY site for any decent information to judge the cameras. Either check them out yourself or wait until somebody such as Adam Wilt can compare the results and put the cameras through a fair test.

December 6th, 2006, 11:56 PM	#43
David Ziegelheim Major Player Join Date: Jul 2003 Location: Warren, NJ Posts: 398	Generally, there have been few comparitive ads between camera manufacturers. If the add is comparative, it has frequently been against other or previous models from the same manufacturer. Typically comparitive ads are done for products with a poor image, to properly place them in the mind of the target audience. For example, a Hyundai, Kia, Suzuki, or Mitsubishi may have comparative ads. Toyota and Honda are very unlikely to have comparative ads. Sony would generally ignore its competition. This ad is unusual, and probably reflects information that engineering provided marketing that their CMOS progressive camera was better than its competitors.

December 7th, 2006, 09:59 AM	#45
Chris Hurd Obstreperous Rex Join Date: Jan 2001 Location: San Marcos, TX Posts: 27,368 Images: 513	Beware, conspiracy theorists! None of these manufacturers would ever photoshop a comparison image. They'll simulate an LCD display image on a brochure, but they won't touch a comparison shot. __________________ CH Search DV Info Net \| 20 years of DVi \| ...Tuesday is Soylent Green Day!