Then the phenomenon in the 2000 dpi scan is not grain but artifacting. Grain is a characteristic of the film, not the scanner, and if you experience one scanner producing more visible graininess than another you need to ask yourself why. Of course, the answer was already provided. Whenever you scan you must use an AA filter if there's detail beyond the Nyquist limit. Virtually all digital cameras do this. Why don't film scanners?
35mm digital SLR's approach the resolving power of the lenses available for those cameras (particularly the full frame ones). In light of that and the total absense of grain, it's hard to argue that 35mm film offers resolution that is as good, much less a lot better, than digital does. If film resolution is much better then where's it coming from? It's certainly not getting though the lens. Perhaps all the extra detail you're referring to is that fabulous film grain.
Many are here are claiming (in addition to the author of the article) that larger sensors are more sensitive when they are not. Good to hear you aren't among them. Noise is a greater problem with smaller sensors.
Smaller sensors are not inherently more sensitive to "light refraction" (whatever you are referring to), particles in the air, etc.
I agree that there's a limit to the useful resolution of a sensor. Care to say what that is? Exactly what limits it? People argue this ignorantly everywhere there are digital photography discussions and no one puts forth a convincing argument as to what that limit is. Without knowing the limit how do we know whether we're there yet?
Pinhole cameras lose sharpness due to diffraction. That's true but so what? We already know we shouldn't use apertures that are too small. If your point is that lens performance limits resolution then you are right. Improved lenses raise the limits for smallest useful pixel size.
For those who like grain there are tools that add that to grainless images.
Clearly, those who believe that digital images are superior to film due to lack of grain are judging those images visually. No one would be so stupid as to take that position on specifications alone.
There are those that believe that grain makes an image superior, as irrational as that is.
I agree. Digicams are limited in their aperture due to diffraction. Many believe they have inherently superior DOF compared to SLR's but they are completely mistaken. Larger sensor cameras have greater DOF for macro and shallower DOF when you want it.
The 1DMkII you used as an example is nearly best of breed in DOF, surpassed only modestly by the two full frame DSLR's on the market. Medium format digital backs could theoretically do better but they don't offer lenses really designed for the seriously high magnifications required to compete. I think the 1DMkII should be an outstanding camera.
But larger sensors are not more sensitive than smaller ones. Not in any physics I know.
Anyone who's actually used digicams and DSLR's knows that larger sensors require more light to expose them. You'll find out when you start using strobes.
Sensitivity (specified as ISO) is the amount of light per unit area required to expose the sensor. A typical digicam and an SLR, both rated at ISO 100, have equal sensitivity ratings but the SLR will require many times more light to expose it by virtue of the fact that its sensor is many times larger. If a larger sensor meant better sensitivity then this would not be the case.
A larger photosite requires many more photons to fill it up. How that equates to better sensitivity in your physics I'll never know. Large photosites have superior noise performance. That's where it's at.
Except that the article is wrong on this. The bigger the sensor, the lower the sensitivity. Of course, that may be "better" but certainly not for low light situations. Bigger sensors are better, though. All you need to know is that you should be using a DSLR, not a digicam.
Manufacturers generally do provide this information. Review sites do in any event.
Re:Dynamic Range and the SIZE of the CCD!!
on
Beyond Megapixels
·
· Score: 1
Purple fringing is not just chromatic aberration. CA is different. While the problem may be optics, high resolution sensors expose optical problems more readily.
I don't think smaller sensors inevitably result in worse color artifacting. I think there are practical concerns that may make it so in reality, though.
The Kodak 14n is the king of purple fringing due to its 14MP full frame sensor and lack of AA filter.
"I bet that Canon 1D Mark II has better DOF than the other digi/film cameras."
Curious to know why you said this when it clearly contradicts your previous sentence. It would be important to say what you mean by "better DOF" as well.
You can view DOF under macro (high magnification) and normal conditions. Under macro (where you are looking for the most you can get) DOF is a function of magnification and aperture. Sensor size has a small effect but exactly opposite of what you'd expect. The 1DMkII, by virtue of it's relatively large sensor, is very good there but not quite as good as full framers like the 1Ds, the Kodak SLR/n, and 35mm film cameras. Under normal conditions, this camera can provide shallow DOF very well by virtue of its 35mm lenses. Once again, not quite as good as the full frame cameras but better than the 1.5/1.6 crop cameras and way better than digicams.
Frankly, digicams do nothing as well as DSLR's except fit in your pocket. They take less light to get good exposures (but you trade that for noise). You don't have sensor dust problems, either, I suppose.
If grain is large enough to effect the scan then it's large whether it can be fully resolved by the scanner or not. All aliasing means is that the detail involved is at too high a frequency and "aliases" to a lower frequency due to the absense of a proper anti-aliasing filter.
Going from 2700 dpi to 5000 dpi it's hard to imagine that there's be "a lot less" grain. In any event, the resolving power of the lenses used in the current crop of DSLR's is not a great deal greater than what the DSLR's can resolve so it's impossible for DSLR's to be "far behind" film bodies that use the same lenses. Unlike film, DSLR's have no grain and many believe their images are superior for that very reason.
Yes, it can. I have never heard this criticism before and I can think of nothing that would explain it. Digital cameras have wide color spaces and high dynamic range (from the perspective of this conversation). Color resolution is not and issue in any way.
Films, on the other hand, are frequently unfaithful when it comes to color accuracy and many don't realize it. People often have preferences for film based specifically on the nature of their color "infidelity" and perhaps this explains your issue. Digital cameras are very linear and not hypersaturated. This is not the case for many films.
Regarding ADC resolution, most are 10-12 bits. The new Fuji S3 will have 14 bit ADC's. When an image is converted to a color space (gamma applied), resolution is lost so you can't compare a 16 bit output file to the raw ADC resolution. In any event, output devices don't have the ability to use the extra dynamic range that 12 vs. 16 bit provides so it doesn't really matter. All those bits are for exposure latitude.
Michael Reichman coined the term "pixel peepers" in order to ridicule those who disagreed with him. He's arbitrary and subjective, sloppy and lazy, and whe he's called on his absurd mistakes he dismisses those people as being preoccupied with unimportant technical details (thus "pixel peepers"). Reichman is an incompetent twit. I can't think of anything I've read of his that I consider particularly accurate, much less "very eloquent".
Don't bother with the three editorials. They were written to cover up the flack he was receiving for his glowing review of the considerably flawed Sony F828. Reichman hates being shown to be wrong (but not enough to work on his diligence apparently). He's just an opinionated windbag. I suggest people read Thom Hogan and dpreview instead.
A "a very very long time"? You are quite mistaken. There are many learned film shooters who believe we are already there. There is ample information on the net about comparing digital SLR's to high-res film scans like you mention and the results are quite surprising. Film technology is, in fact, standing still. It's incredibly mature at this point and there is very little inventment in it ongoing. Bye bye, film.
Larger format sensors are most definitely NOT more sensitive than smaller ones. Sensitivity of a sensor is specified using an ISO rating, and ISO ratings are "sensitivity per unit area". That means that a full frame sensor at ISO 100 requires 2.25 times as much light to expose it as a APS (DX) sensor does.
Conveniently, these sensitivity differences are accounted for in the f-number (or f/stop). Rather than specifying aperture directly, aperture is normalized to focal length. This is done so that there is a system for getting exposure right and it works regardless of format. People who think that their digicam and SLR being shot at f/8 are the same are sadly mistaken (and there are plenty of these). The larger format has a larger physical aperture at f/8 because it uses a loner focal length for the same angle of view.
More bothersome to me was the author's claim that larger photosites gather light better. How absurd!
Of course, the F828 is the worst of the 8MP cameras for image quality. What constitutes a better picture you did not say, but I know with absolute certainty that the D30 can't outresolve the F828 or any other 8MP camera. There rest of your comment is just nonsensical hogwash. CCD's don't produce good prints above 7x5"? Please! I'll remind you that the D100/D70 and the Fuji S2 use CCD imagers. The "Larger images don't need to be distorted as much.." comment is completely absurd and larger sensors have nothing to do with faithful color reproduction! You must be Michael Reichman!
Bayer sensor camera manufacturers don't lie about pixel counts. Performing a demosaic on a Bayer sensor image produces a native resolution that matches the raw sensor. That's what counts.
Foveon manufacturers who count separate R,G, abd B photosites are lying however. The image that camera produces is 1/3 the pixel count.
As far as resolving power of the two techniques, that really depends on the nature of the image. Foveon sensors can resolve detail regardless of color whereas Bayer pattern sensors have resolution that varies with color. That's theoretical, of course, since it doesn't take the AA filter into account.
Foveon sensors do have an advantage with color aliasing as you mentioned, but they still require an AA filter to avoid aliasing in general. With a proper AA filter, a Bayer sensor will not suffer from color aliasing either. Problem is that the image will be softer in that case. Color aliasing (moire) is problem that effects different cameras differently since the engineers have to choose what AA filter to use (if any). The current resolution champion, the Kodak SLR/n, has no AA filter. It provides incredible detail but color aliasing can be an issue. Most other digital SLR's don't have serious moire issues.
Patents are public teachings and as such are considered public knowledge. That's the whole point of a patent---you trade making the knowledge of you trade secret public for a limited term monopoly grant on that technology. In light of this, you can see how it would be absurd to say that reverse-engineering a patented technology is prohibited. It's not only not prohibited but its fully disclosed to the public through the patent itself!
Time itself does not make technology enter the public domain when the PTO explicitly recognizes that a patent is still active. I'm not following the reasoning there.
Nevertheless, IP rights exist to provide incentive for creation. Imagine there being no research funding because there's no possibility for payback. Drug companies could not afford the development costs for many new drugs. IP laws may need fixing but they are absolutely essential.
Odd that the grandparent post is modded redundant when there were specific nonredundant comments in it. I guess there was too much reason and not enough hysteria.
The difference is that you already got the ticket. If you did 50 in a 40 10 years ago you don't get a ticket for it next year.
I don't believe your assumptions regarding patent law are valid but IANAL. In any event, it's a civil matter and you do have to notify someone of infringement first.
The good old days of ignorance. This hasn't been true since any of us have been alive.
Giving away knowledge, or anything else, which is not yours to give *is* wrong and who says they didn't "spare" non-commercial entities? Did any OSS developers get put on notice?
IP rights serve a useful purpose in the economy. You may want to learn about that. The IP system may need some serious adjusting but you seem to be looking at this issue strictly as a consumer and not as a producer.
Open sourcing an implementation of a patented technique does not render the patent invalid.
So simply don't ask for a license. Go about our merry way and wait to be put on notice. Since your revenues are zero and their patent is about to expire I don't think there's much to be concerned about (and that assumes their claims can withstand scrutiny to begin with).
There's a big difference between a patent be applicable and there being prior art. With patents you always want to say "we don't infringe" rather than "your patent is invalid". Saying a patent isn't applicable is saying that "we don't infringe". That's not the case here.
While I would agree that CRT projectors can be great and I use one in my theater, I wouldn't agree that the have an "infinitely better picture" (how's that?) or that they "don't quantify the scan lines". They most certainly do. With CRT projectors you will need to match a video processor to the projector and to the screen size and you will need to plan on frequent calibration. With CRT's you get fantastic blacks and tremendous color but it takes a commitment to live with them.
Only mac notebooks power their firewire ports unfortunately.
FirewireDirect used to sell a 2.5" notebook drive chassis with a small builting battery. I loved those since they provided a "UPS" function on the road. In 3rd world areas mains power is unreliable and not external drives are port powered on a PC.
No doubt that $1000 is a pretty sucky price. I'll stick to the 80's until that gets fixed.
At 8 pounds I think you give these machines too much credit. Some are much, much heavier.
Larger drives are inherently faster due to the inherent short-stroking effect. You OS and data don't grow (much) just because your drive is bigger so high capacity drives have shorter seek distances. I would never order a laptop today with a 15GB drive unless it was the cheapest path to a 60 or 80 GB (as it often is with suppliers). I happily replace the boot drive as soon as I get the machine.
Sustained transfer rates on 4200 rpm drives can be good. It's the rotational latency, not transfer rate, that's inherently poor. Large sequential IO performance is not a 4200 rpm problem. Power issues conspire to limit seek times as well.
You won't see any 20+ watt 15K drives in notebooks any time soon.
Slashdot Mirror
Then the phenomenon in the 2000 dpi scan is not grain but artifacting. Grain is a characteristic of the film, not the scanner, and if you experience one scanner producing more visible graininess than another you need to ask yourself why. Of course, the answer was already provided. Whenever you scan you must use an AA filter if there's detail beyond the Nyquist limit. Virtually all digital cameras do this. Why don't film scanners?
35mm digital SLR's approach the resolving power of the lenses available for those cameras (particularly the full frame ones). In light of that and the total absense of grain, it's hard to argue that 35mm film offers resolution that is as good, much less a lot better, than digital does. If film resolution is much better then where's it coming from? It's certainly not getting though the lens. Perhaps all the extra detail you're referring to is that fabulous film grain.
Many are here are claiming (in addition to the author of the article) that larger sensors are more sensitive when they are not. Good to hear you aren't among them. Noise is a greater problem with smaller sensors.
Smaller sensors are not inherently more sensitive to "light refraction" (whatever you are referring to), particles in the air, etc.
I agree that there's a limit to the useful resolution of a sensor. Care to say what that is? Exactly what limits it? People argue this ignorantly everywhere there are digital photography discussions and no one puts forth a convincing argument as to what that limit is. Without knowing the limit how do we know whether we're there yet?
Pinhole cameras lose sharpness due to diffraction. That's true but so what? We already know we shouldn't use apertures that are too small. If your point is that lens performance limits resolution then you are right. Improved lenses raise the limits for smallest useful pixel size.
For those who like grain there are tools that add that to grainless images.
Clearly, those who believe that digital images are superior to film due to lack of grain are judging those images visually. No one would be so stupid as to take that position on specifications alone.
There are those that believe that grain makes an image superior, as irrational as that is.
I agree. Digicams are limited in their aperture due to diffraction. Many believe they have inherently superior DOF compared to SLR's but they are completely mistaken. Larger sensor cameras have greater DOF for macro and shallower DOF when you want it.
The 1DMkII you used as an example is nearly best of breed in DOF, surpassed only modestly by the two full frame DSLR's on the market. Medium format digital backs could theoretically do better but they don't offer lenses really designed for the seriously high magnifications required to compete. I think the 1DMkII should be an outstanding camera.
But larger sensors are not more sensitive than smaller ones. Not in any physics I know.
Anyone who's actually used digicams and DSLR's knows that larger sensors require more light to expose them. You'll find out when you start using strobes.
Sensitivity (specified as ISO) is the amount of light per unit area required to expose the sensor. A typical digicam and an SLR, both rated at ISO 100, have equal sensitivity ratings but the SLR will require many times more light to expose it by virtue of the fact that its sensor is many times larger. If a larger sensor meant better sensitivity then this would not be the case.
A larger photosite requires many more photons to fill it up. How that equates to better sensitivity in your physics I'll never know. Large photosites have superior noise performance. That's where it's at.
Except that the article is wrong on this. The bigger the sensor, the lower the sensitivity. Of course, that may be "better" but certainly not for low light situations. Bigger sensors are better, though. All you need to know is that you should be using a DSLR, not a digicam.
Manufacturers generally do provide this information. Review sites do in any event.
Purple fringing is not just chromatic aberration. CA is different. While the problem may be optics, high resolution sensors expose optical problems more readily.
I don't think smaller sensors inevitably result in worse color artifacting. I think there are practical concerns that may make it so in reality, though.
The Kodak 14n is the king of purple fringing due to its 14MP full frame sensor and lack of AA filter.
"I bet that Canon 1D Mark II has better DOF than the other digi/film cameras."
Curious to know why you said this when it clearly contradicts your previous sentence. It would be important to say what you mean by "better DOF" as well.
You can view DOF under macro (high magnification) and normal conditions. Under macro (where you are looking for the most you can get) DOF is a function of magnification and aperture. Sensor size has a small effect but exactly opposite of what you'd expect. The 1DMkII, by virtue of it's relatively large sensor, is very good there but not quite as good as full framers like the 1Ds, the Kodak SLR/n, and 35mm film cameras. Under normal conditions, this camera can provide shallow DOF very well by virtue of its 35mm lenses. Once again, not quite as good as the full frame cameras but better than the 1.5/1.6 crop cameras and way better than digicams.
Frankly, digicams do nothing as well as DSLR's except fit in your pocket. They take less light to get good exposures (but you trade that for noise). You don't have sensor dust problems, either, I suppose.
If grain is large enough to effect the scan then it's large whether it can be fully resolved by the scanner or not. All aliasing means is that the detail involved is at too high a frequency and "aliases" to a lower frequency due to the absense of a proper anti-aliasing filter.
Going from 2700 dpi to 5000 dpi it's hard to imagine that there's be "a lot less" grain. In any event, the resolving power of the lenses used in the current crop of DSLR's is not a great deal greater than what the DSLR's can resolve so it's impossible for DSLR's to be "far behind" film bodies that use the same lenses. Unlike film, DSLR's have no grain and many believe their images are superior for that very reason.
Yes, it can. I have never heard this criticism before and I can think of nothing that would explain it. Digital cameras have wide color spaces and high dynamic range (from the perspective of this conversation). Color resolution is not and issue in any way.
Films, on the other hand, are frequently unfaithful when it comes to color accuracy and many don't realize it. People often have preferences for film based specifically on the nature of their color "infidelity" and perhaps this explains your issue. Digital cameras are very linear and not hypersaturated. This is not the case for many films.
Regarding ADC resolution, most are 10-12 bits. The new Fuji S3 will have 14 bit ADC's. When an image is converted to a color space (gamma applied), resolution is lost so you can't compare a 16 bit output file to the raw ADC resolution. In any event, output devices don't have the ability to use the extra dynamic range that 12 vs. 16 bit provides so it doesn't really matter. All those bits are for exposure latitude.
Michael Reichman coined the term "pixel peepers" in order to ridicule those who disagreed with him. He's arbitrary and subjective, sloppy and lazy, and whe he's called on his absurd mistakes he dismisses those people as being preoccupied with unimportant technical details (thus "pixel peepers"). Reichman is an incompetent twit. I can't think of anything I've read of his that I consider particularly accurate, much less "very eloquent".
Don't bother with the three editorials. They were written to cover up the flack he was receiving for his glowing review of the considerably flawed Sony F828. Reichman hates being shown to be wrong (but not enough to work on his diligence apparently). He's just an opinionated windbag. I suggest people read Thom Hogan and dpreview instead.
A "a very very long time"? You are quite mistaken. There are many learned film shooters who believe we are already there. There is ample information on the net about comparing digital SLR's to high-res film scans like you mention and the results are quite surprising. Film technology is, in fact, standing still. It's incredibly mature at this point and there is very little inventment in it ongoing. Bye bye, film.
Well, the article was crap.
Larger format sensors are most definitely NOT more sensitive than smaller ones. Sensitivity of a sensor is specified using an ISO rating, and ISO ratings are "sensitivity per unit area". That means that a full frame sensor at ISO 100 requires 2.25 times as much light to expose it as a APS (DX) sensor does.
Conveniently, these sensitivity differences are accounted for in the f-number (or f/stop). Rather than specifying aperture directly, aperture is normalized to focal length. This is done so that there is a system for getting exposure right and it works regardless of format. People who think that their digicam and SLR being shot at f/8 are the same are sadly mistaken (and there are plenty of these). The larger format has a larger physical aperture at f/8 because it uses a loner focal length for the same angle of view.
More bothersome to me was the author's claim that larger photosites gather light better. How absurd!
Of course, the F828 is the worst of the 8MP cameras for image quality. What constitutes a better picture you did not say, but I know with absolute certainty that the D30 can't outresolve the F828 or any other 8MP camera. There rest of your comment is just nonsensical hogwash. CCD's don't produce good prints above 7x5"? Please! I'll remind you that the D100/D70 and the Fuji S2 use CCD imagers. The "Larger images don't need to be distorted as much.." comment is completely absurd and larger sensors have nothing to do with faithful color reproduction! You must be Michael Reichman!
Bayer sensor camera manufacturers don't lie about pixel counts. Performing a demosaic on a Bayer sensor image produces a native resolution that matches the raw sensor. That's what counts.
Foveon manufacturers who count separate R,G, abd B photosites are lying however. The image that camera produces is 1/3 the pixel count.
As far as resolving power of the two techniques, that really depends on the nature of the image. Foveon sensors can resolve detail regardless of color whereas Bayer pattern sensors have resolution that varies with color. That's theoretical, of course, since it doesn't take the AA filter into account.
Foveon sensors do have an advantage with color aliasing as you mentioned, but they still require an AA filter to avoid aliasing in general. With a proper AA filter, a Bayer sensor will not suffer from color aliasing either. Problem is that the image will be softer in that case. Color aliasing (moire) is problem that effects different cameras differently since the engineers have to choose what AA filter to use (if any). The current resolution champion, the Kodak SLR/n, has no AA filter. It provides incredible detail but color aliasing can be an issue. Most other digital SLR's don't have serious moire issues.
Patents are public teachings and as such are considered public knowledge. That's the whole point of a patent---you trade making the knowledge of you trade secret public for a limited term monopoly grant on that technology. In light of this, you can see how it would be absurd to say that reverse-engineering a patented technology is prohibited. It's not only not prohibited but its fully disclosed to the public through the patent itself!
Time itself does not make technology enter the public domain when the PTO explicitly recognizes that a patent is still active. I'm not following the reasoning there.
Nevertheless, IP rights exist to provide incentive for creation. Imagine there being no research funding because there's no possibility for payback. Drug companies could not afford the development costs for many new drugs. IP laws may need fixing but they are absolutely essential.
Odd that the grandparent post is modded redundant when there were specific nonredundant comments in it. I guess there was too much reason and not enough hysteria.
The difference is that you already got the ticket. If you did 50 in a 40 10 years ago you don't get a ticket for it next year.
I don't believe your assumptions regarding patent law are valid but IANAL. In any event, it's a civil matter and you do have to notify someone of infringement first.
The good old days of ignorance. This hasn't been true since any of us have been alive.
Giving away knowledge, or anything else, which is not yours to give *is* wrong and who says they didn't "spare" non-commercial entities? Did any OSS developers get put on notice?
IP rights serve a useful purpose in the economy. You may want to learn about that. The IP system may need some serious adjusting but you seem to be looking at this issue strictly as a consumer and not as a producer.
Open sourcing an implementation of a patented technique does not render the patent invalid.
So simply don't ask for a license. Go about our merry way and wait to be put on notice. Since your revenues are zero and their patent is about to expire I don't think there's much to be concerned about (and that assumes their claims can withstand scrutiny to begin with).
They don't care about revenue-free products. They're looking for money.
There's a big difference between a patent be applicable and there being prior art. With patents you always want to say "we don't infringe" rather than "your patent is invalid". Saying a patent isn't applicable is saying that "we don't infringe". That's not the case here.
While I would agree that CRT projectors can be great and I use one in my theater, I wouldn't agree that the have an "infinitely better picture" (how's that?) or that they "don't quantify the scan lines". They most certainly do. With CRT projectors you will need to match a video processor to the projector and to the screen size and you will need to plan on frequent calibration. With CRT's you get fantastic blacks and tremendous color but it takes a commitment to live with them.
Only mac notebooks power their firewire ports unfortunately.
FirewireDirect used to sell a 2.5" notebook drive chassis with a small builting battery. I loved those since they provided a "UPS" function on the road. In 3rd world areas mains power is unreliable and not external drives are port powered on a PC.
No doubt that $1000 is a pretty sucky price. I'll stick to the 80's until that gets fixed.
At 8 pounds I think you give these machines too much credit. Some are much, much heavier.
Larger drives are inherently faster due to the inherent short-stroking effect. You OS and data don't grow (much) just because your drive is bigger so high capacity drives have shorter seek distances.
I would never order a laptop today with a 15GB drive unless it was the cheapest path to a 60 or 80 GB (as it often is with suppliers). I happily replace the boot drive as soon as I get the machine.
Sustained transfer rates on 4200 rpm drives can be good. It's the rotational latency, not transfer rate, that's inherently poor. Large sequential IO performance is not a 4200 rpm problem. Power issues conspire to limit seek times as well.
You won't see any 20+ watt 15K drives in notebooks any time soon.