I don't understand them either. I agree, I should have phrased it better.
Something like this is a much better representation of what I was trying to convey -
"we know that it is a sapphire cover that produces these artifacts because it is the only element (apart from a surface of a sensor itself) that can create flat ( = undistorted) reflections. Since this is exactly the type of artifacts we observe, it is safe to say that that sapphire cover is the source of those artifacts".
I don't understand them either. I agree, I should have phrased it better.
Something like this is a much better representation of what I was trying to convey -
"we know that it is a sapphire cover that produces these artifacts because it is the only element (apart from a surface of a sensor itself) that can create flat ( = undistorted) reflections. Since this is exactly the type of artifacts we observe, it is safe to say that that sapphire cover is the source of those artifacts".
Perhaps. On a built-in cover as this one, the cover is computed as part of the lens system. So the internal reflections are able to be minimized. If it's multi coated, as it likely is, then reflections should be well controlled.
Depends on the sensor hardware. If you look at spec sheets of sony sensores, only some of those say RGB, and also RAW. Mostly it is just RGB
All RAW means it that it's the pure data off the sensor, without any manipulation. Unfortunately, some companies, such as Nikon and Sony, do some manipulation, and even compression of their "RAW" files.
Perhaps. On a built-in cover as this one, the cover is computed as part of the lens system. So the internal reflections are able to be minimized. If it's multi coated, as it likely is, then reflections should be well controlled.
Yes. that is why there aren't so many reflections seen on images
But there is no magic to it - coating can elliminate only some amount (%) of the photon energy that would otherwise be reflected back. If the reflection is too bright (bright point-like light source), then most of it would go away, but not all of it. Hence there are still some noticeable, albeit very weak residual reflections of light sources visible on SOME images )
I'm pretty sure it could be. Even better would be a DNG.
All RAW means it that it's the pure data off the sensor, without any manipulation. Unfortunately, some companies, such as Nikon and Sony, do some manipulation, and even compression of their "RAW" files.
1. DNG is just a format that packs RAW data. A container, in other words. I agree that it is an attempt to keep RAW standardized. But it doesn't change quality of packed data.
2. RAW data means - unproccessed (not demosaiced or tonemapped or noise-reduction treated). For example, removing of dead pixels is done routinely and all companies still call it - RAW images. I doubt you would be glad to see dead pixels in your otherwise very clean RAW files. Nikon, Sony, Canon all do that.
3. Compression (lossless or not) is not a bad thing at all if compression is done on RAW files. The most important point of RAW files is that they are efficient at storing higher than 8-bit depth of color information, unlike TIFF or JPEG that not only stores tone-mapped images, but also truncates 16,14,12 or 10 bit image to 8bit.
For example. RED cameras use proprietary wavelet based codec that can vary compression rates from 1:3 to 1:30 while still providing stunning quality (yes, they also allow for uncompressed RAWs' if one wishes to use that option)
1. DNG is just a format that packs RAW data. A container, in other words. I agree that it is an attempt to keep RAW standardized. But it doesn't change quality of packed data.
2. RAW data means - unproccessed (not demosaiced or tonemapped or noise-reduction treated). For example, removing of dead pixels is done routinely and all companies still call it - RAW images. I doubt you would be glad to see dead pixels in your otherwise very clean RAW files. Nikon, Sony, Canon all do that.
3. Compression (lossless or not) is not a bad thing at all if compression is done on RAW files. The most important point of RAW files is that they are efficient at storing higher than 8-bit depth of color information, unlike TIFF or JPEG that not only stores tone-mapped images, but also truncates 16,14,12 or 10 bit image to 8bit.
For example. RED cameras use proprietary wavelet based codec that can vary compression rates from 1:3 to 1:30 while still providing stunning quality (yes, they also allow for uncompressed RAWs' if one wishes to use that option)
A DNG is better than a RAW file because it's become a standard, which RAW files are not. Every DNG is directly readable, as is the data within it by any software that reads DNG, which is about every image editor today. It's an Adobe invention, which is why it's become popular. So e camera manufacturers have that as their standard file format, which is a very good thing.
Unfortunately, all RAW files are not equal. Nikon, for example, refuses to disclose its RAW information, which is why Adobe and other RAW converter developers have a hard time with it. We often do see dead pixels in RAW Files. When they exist. It's far less prevalent than it used to be. You can have 16 bit per color JPEG files. It's unusual, but it can be done.
REDs compression is pretty good, but not that much better than Han other methods, though you'd never know that by reading their BS.
A DNG is better than a RAW file because it's become a standard.
Unfortunately, all RAW files are not equal. Nikon, for example, refuses to disclose its RAW information.
Looks like you were busy looking things up.
1. It is very far from being a standard, say, like PDF. Plus, the majority of editors can only view DNG files but not open RAW in it, because they can read JPG version of the image that is stored next to RAW image itself in the same DNG container. But I agree, DNG is a little better and it also has nice lossy and lossless options for archival storing, for example. I used it a lot when did photography for a living (about 5 years ago).
2. You indeed can create 16 bit JPGs, however I didn't look at the compression algorithm. It might be the case that the algorithm will severly degrade in performance due to the fact that on 8 bits, it works only with 8 bit values with exactly a type of SIMD instructions that moderns CPUs love (4 8 bit values processed in a single instruction). Doing the same with 16 bit values will cut performance in half. Second problem is that you don't actually need 16 bits, since it is a very rare occasion that a camera produces that precision of quantization. Usually it is 10 or 12. 14 is only for good cameras and 16 - almost never is the case. It depends on ADC of a given camera, of course and sensors noise level.
3. Looks like you were busy looking things up.
Not really - none of what I discussed required that. It just happened that I simply knew that from me experience. I worked with RAW professionally about 8 years, until I changed my occupation). And change of my occupation and getting a second (advanced) degree helps me to discuss 2)
1. It is very far from being a standard, say, like PDF. Plus, the majority of editors can only view DNG files but not open RAW in it, because they can read JPG version of the image that is stored next to RAW image itself in the same DNG container. But I agree, DNG is a little better and it also has nice lossy and lossless options for archival storing, for example. I used it a lot when did photography for a living (about 5 years ago).
2. You indeed can create 16 bit JPGs, however I didn't look at the compression algorithm. It might be the case that the algorithm will severly degrade in performance due to the fact that on 8 bits, it works only with 8 bit values with exactly a type of SIMD instructions that moderns CPUs love (4 8 bit values processed in a single instruction). Doing the same with 16 bit values will cut performance in half. Second problem is that you don't actually need 16 bits, since it is a very rare occasion that a camera produces that precision of quantization. Usually it is 10 or 12. 14 is only for good cameras and 16 - almost never is the case. It depends on ADC of a given camera, of course and sensors noise level.
3. Looks like you were busy looking things up.
Not really - none of what I discussed required that. It just happened that I simply knew that from me experience. I worked with RAW professionally about 8 years, until I changed my occupation). And change of my occupation and getting a second (advanced) degree helps me to discuss 2)
)))
Every pro photographer I know would prefer that camera companies make DNG standard. Hassleblad did so years ago. So have some others. It's a standard for Phase One as well. These are the two heavyweights in the medium format industry.
I don't know of anyone who would use a JPEG from a DNG. That's not the purpose. A manufacturer can embed a JPEG for thumbnail usage, but not for editing. All RAW converters can read, and edit a DNG, and we all immediately convert our RAW files into DNG upon import.
I don't know what kind of work you used to do, but you have much of this backwards. We ALWAYS bring images in as 16 bits. That's basic. Adobe, and some others, even support 24, and as much as 32 bits per color, and have for years. All pro level cameras are 14 bits, in "full frame" types, and mostly in APS-C types as well. Sony still uses 12 bits, but they don't make a pro level camera, and there can be problems with some of their files. All medium format cameras are 16 bits out of camera.
Editing in 8 bits gives many problems with gradation, and posterization, so we don't use it. Good printers allow printing in 16 bits, and you can see the difference. I only print in 16 bits.
Modern CPUs and GPUs have no problems with performance. But no one uses JPEGs if they are interested in serious editing anyway. At best, it's a convenient storage format when ultimate quality isn't needed.
I don't know what kind of work you used to do, but you have much of this backwards.
I never said anything that would contradict what you just said. Can you quote where did I get specifically "much of this backward"?
1. I didn't said that one needs to bring RAWs as 8 bit images, hence your pontification on "16 bit basics" seems uncalled for. It is obvious that you need to use more or equal amount of color depth in order not to lose information.
2. I never argued about DNG being worse (or not) than RAW. I simply stated that it is a container for RAW data. You can check that yourself by looking at DNG file format specifications (I mean, structure of data in the file). No need to argue for DNG or defend it, since I am not the one who criticizes or opposes.
3. About modern CPUs - that was not my point. I simply pointed out that if one wanted to implement 16 bit JPEG algorithm, than he would have to deal with performance degradation at least by a factor of 2. And that might be the reason why we still don't see this format.
Sure, use last year's information with not all the phones out for that cycle and you can twist it so that the iPhone looks as if it beat everyone. Once you add in a couple more of the phones from that cycle, the iPhone isn't the top camera anymore. Which is unfortunate since the iPhone has been the best for many years running.
I'll be interested to see how the 6S gets rated by DxO Mark this year. Sony is currently in first place but the 6S might unseat them.
DxO Mark updated it's rankings with the new iPhone 6S and iPhone 6S Plus. Unfortunately they don't rate higher than last year and trail Sony and Samsung. The camera has some new pluses with the 12 MP and 4K video, but disappointing that its not better overall. Think I'll wait another year before upgrading. I guess this isn't "The Best iPhone Camera ever made". I wonder how good the Lumia 950XL's camera is with its 1/2.4" sensor size.
Comments
I don't understand your last two sentences.
I don't understand them either. I agree, I should have phrased it better.
Something like this is a much better representation of what I was trying to convey -
"we know that it is a sapphire cover that produces these artifacts because it is the only element (apart from a surface of a sensor itself) that can create flat ( = undistorted) reflections. Since this is exactly the type of artifacts we observe, it is safe to say that that sapphire cover is the source of those artifacts".
Another vote for RAW support. Could that not even be possible with a mere software update?
Depends on the sensor hardware. If you look at spec sheets of sony sensores, only some of those say RGB, and also RAW. Mostly it is just RGB
Oh, I see, thanks for the heads up!
Perhaps. On a built-in cover as this one, the cover is computed as part of the lens system. So the internal reflections are able to be minimized. If it's multi coated, as it likely is, then reflections should be well controlled.
I'm pretty sure it could be. Even better would be a DNG.
All RAW means it that it's the pure data off the sensor, without any manipulation. Unfortunately, some companies, such as Nikon and Sony, do some manipulation, and even compression of their "RAW" files.
Perhaps. On a built-in cover as this one, the cover is computed as part of the lens system. So the internal reflections are able to be minimized. If it's multi coated, as it likely is, then reflections should be well controlled.
Yes. that is why there aren't so many reflections seen on images
But there is no magic to it - coating can elliminate only some amount (%) of the photon energy that would otherwise be reflected back. If the reflection is too bright (bright point-like light source), then most of it would go away, but not all of it. Hence there are still some noticeable, albeit very weak residual reflections of light sources visible on SOME images )
I'm pretty sure it could be. Even better would be a DNG.
All RAW means it that it's the pure data off the sensor, without any manipulation. Unfortunately, some companies, such as Nikon and Sony, do some manipulation, and even compression of their "RAW" files.
1. DNG is just a format that packs RAW data. A container, in other words. I agree that it is an attempt to keep RAW standardized. But it doesn't change quality of packed data.
2. RAW data means - unproccessed (not demosaiced or tonemapped or noise-reduction treated). For example, removing of dead pixels is done routinely and all companies still call it - RAW images. I doubt you would be glad to see dead pixels in your otherwise very clean RAW files. Nikon, Sony, Canon all do that.
3. Compression (lossless or not) is not a bad thing at all if compression is done on RAW files. The most important point of RAW files is that they are efficient at storing higher than 8-bit depth of color information, unlike TIFF or JPEG that not only stores tone-mapped images, but also truncates 16,14,12 or 10 bit image to 8bit.
For example. RED cameras use proprietary wavelet based codec that can vary compression rates from 1:3 to 1:30 while still providing stunning quality (yes, they also allow for uncompressed RAWs' if one wishes to use that option)
A DNG is better than a RAW file because it's become a standard, which RAW files are not. Every DNG is directly readable, as is the data within it by any software that reads DNG, which is about every image editor today. It's an Adobe invention, which is why it's become popular. So e camera manufacturers have that as their standard file format, which is a very good thing.
Unfortunately, all RAW files are not equal. Nikon, for example, refuses to disclose its RAW information, which is why Adobe and other RAW converter developers have a hard time with it. We often do see dead pixels in RAW Files. When they exist. It's far less prevalent than it used to be. You can have 16 bit per color JPEG files. It's unusual, but it can be done.
REDs compression is pretty good, but not that much better than Han other methods, though you'd never know that by reading their BS.
Looks like you were busy looking things up.
A DNG is better than a RAW file because it's become a standard.
Unfortunately, all RAW files are not equal. Nikon, for example, refuses to disclose its RAW information.
Looks like you were busy looking things up.
1. It is very far from being a standard, say, like PDF. Plus, the majority of editors can only view DNG files but not open RAW in it, because they can read JPG version of the image that is stored next to RAW image itself in the same DNG container. But I agree, DNG is a little better and it also has nice lossy and lossless options for archival storing, for example. I used it a lot when did photography for a living (about 5 years ago).
2. You indeed can create 16 bit JPGs, however I didn't look at the compression algorithm. It might be the case that the algorithm will severly degrade in performance due to the fact that on 8 bits, it works only with 8 bit values with exactly a type of SIMD instructions that moderns CPUs love (4 8 bit values processed in a single instruction). Doing the same with 16 bit values will cut performance in half. Second problem is that you don't actually need 16 bits, since it is a very rare occasion that a camera produces that precision of quantization. Usually it is 10 or 12. 14 is only for good cameras and 16 - almost never is the case. It depends on ADC of a given camera, of course and sensors noise level.
3. Looks like you were busy looking things up.
Not really - none of what I discussed required that. It just happened that I simply knew that from me experience. I worked with RAW professionally about 8 years, until I changed my occupation). And change of my occupation and getting a second (advanced) degree helps me to discuss 2)
)))
Every pro photographer I know would prefer that camera companies make DNG standard. Hassleblad did so years ago. So have some others. It's a standard for Phase One as well. These are the two heavyweights in the medium format industry.
I don't know of anyone who would use a JPEG from a DNG. That's not the purpose. A manufacturer can embed a JPEG for thumbnail usage, but not for editing. All RAW converters can read, and edit a DNG, and we all immediately convert our RAW files into DNG upon import.
I don't know what kind of work you used to do, but you have much of this backwards. We ALWAYS bring images in as 16 bits. That's basic. Adobe, and some others, even support 24, and as much as 32 bits per color, and have for years. All pro level cameras are 14 bits, in "full frame" types, and mostly in APS-C types as well. Sony still uses 12 bits, but they don't make a pro level camera, and there can be problems with some of their files. All medium format cameras are 16 bits out of camera.
Editing in 8 bits gives many problems with gradation, and posterization, so we don't use it. Good printers allow printing in 16 bits, and you can see the difference. I only print in 16 bits.
Modern CPUs and GPUs have no problems with performance. But no one uses JPEGs if they are interested in serious editing anyway. At best, it's a convenient storage format when ultimate quality isn't needed.
I don't know what kind of work you used to do, but you have much of this backwards.
I never said anything that would contradict what you just said. Can you quote where did I get specifically "much of this backward"?
1. I didn't said that one needs to bring RAWs as 8 bit images, hence your pontification on "16 bit basics" seems uncalled for. It is obvious that you need to use more or equal amount of color depth in order not to lose information.
2. I never argued about DNG being worse (or not) than RAW. I simply stated that it is a container for RAW data. You can check that yourself by looking at DNG file format specifications (I mean, structure of data in the file). No need to argue for DNG or defend it, since I am not the one who criticizes or opposes.
3. About modern CPUs - that was not my point. I simply pointed out that if one wanted to implement 16 bit JPEG algorithm, than he would have to deal with performance degradation at least by a factor of 2. And that might be the reason why we still don't see this format.
I hope that clarifies my position.
Unfortunately it did not unseat anyone this year.
Unfortunately it did not unseat anyone this year.
This really show the author's fact twisting skill!
Sure, use last year's information with not all the phones out for that cycle and you can twist it so that the iPhone looks as if it beat everyone. Once you add in a couple more of the phones from that cycle, the iPhone isn't the top camera anymore. Which is unfortunate since the iPhone has been the best for many years running.
I'll be interested to see how the 6S gets rated by DxO Mark this year. Sony is currently in first place but the 6S might unseat them.
DxO Mark updated it's rankings with the new iPhone 6S and iPhone 6S Plus. Unfortunately they don't rate higher than last year and trail Sony and Samsung. The camera has some new pluses with the 12 MP and 4K video, but disappointing that its not better overall. Think I'll wait another year before upgrading. I guess this isn't "The Best iPhone Camera ever made". I wonder how good the Lumia 950XL's camera is with its 1/2.4" sensor size.