Re: filmscanners: RE: filmscanners: RE: filmscanners: Pixels per inch vs DPI
As a result of the continuing and escalating acrimony between Austin and myself, and his incessant nitpicking of my postings, I do not intend to respond directly either publicly or privately to his postings in the future. I bring this to the attention of the other members so that you understand that my silence to Austin's challenges is not necessarily because I am unable to defend my position on either technical or other merits, but because I simply have decided his challenges are not worth my time to pursue. Further, the issues he has brought up to question below were asides and tangential to the main points I was making in my post which were concerning the discussion comparing color dye clouds and capture of images digitally, not black and white developing, and my principle point was that grain was randomly distributed throughout the film emulsion and no process allowed for dye clouds to be moved or lined up within the emulsion during processing, and therefore there was also a built in error factor in grain/dye clouds as there is in digital imagine with its fixed pixels. As I also explained, the order of magnitude of error related to the size and density of grain versus pixels, and as pixels were made smaller and packed more densely, this error factor would lessen. Art Austin Franklin wrote: Austin Franklin wrote: Very simply, grain, or dye clouds are predetermined in their location and shape and are not relocated by picture content. What about development? Also, some developing techniques can somewhat alter the shape or size of the dye clouds... Somewhat? However, most of this type of thing is done in custom film development of black and white film, You can alter the grain of BW film by at least two to four times simply by developer choice, dilution, temperature and technique. It certainly isn't custom, most anyone who uses BW has their favorite developer/dilution/temperature and technique that suits their needs/style/experimentation. It is VERY critical when talking about film grain to discuss development AND even exposure (as you mentioned push/pull too)...since the same film can give such drastically different results...and more so even if you are using Zone system compensation development. because the need to control so many other variables within color film development doesn't allow for much playing around. Most color film processing is fairly uniform in its method... Not quite true...see below... This is why almost all color film is souped in one of two basic color chemistry types (C-41 or E-6). There are different E-6 and C-41 processes. Different chemical AND entirely different developments, as well as techniques. E6 can be 3 bath or 6 bath, and C-41 can be 2 bath or 3 bath. All of this plays a SIGNIFICANT role on the shape and size of the dye clouds. It can be far more significant than you made it out to be. However, I know of no color development technique that is capable of moving film grain or dye clouds within the emulsion so that they can line up the grain as a result of the image content. If you do, I'd like to here about it. I don't believe anyone ever suggested that at all... .
RE: filmscanners: RE: filmscanners: RE: filmscanners: Pixels per inch vs DPI
Further, the issues he has brought up to question below were asides and tangential to the main points I was making in my post which were concerning the discussion comparing color dye clouds and capture of images digitally, not black and white developing, I DID talk about color (see below), not BW exclusively. I find it funny that you ignore that fact. You really believe the size and shape of the film grain is tangential to the capture of images digitally? How do you arrive at that conclusion? and my principle point was that grain was randomly distributed throughout the film emulsion and no process allowed for dye clouds to be moved or lined up within the emulsion during processing, I do believe that is common knowledge, and I don't believe anyone disagreed with that. My correction to your statement was that development has a LOT to do with grain size. Austin Franklin wrote: Very simply, grain, or dye clouds are predetermined in their location and shape and are not relocated by picture content. What about development? Also, some developing techniques can somewhat alter the shape or size of the dye clouds... Somewhat? However, most of this type of thing is done in custom film development of black and white film, You can alter the grain of BW film by at least two to four times simply by developer choice, dilution, temperature and technique. It certainly isn't custom, most anyone who uses BW has their favorite developer/dilution/temperature and technique that suits their needs/style/experimentation. It is VERY critical when talking about film grain to discuss development AND even exposure (as you mentioned push/pull too)...since the same film can give such drastically different results...and more so even if you are using Zone system compensation development. because the need to control so many other variables within color film development doesn't allow for much playing around. Most color film processing is fairly uniform in its method... Not quite true...see below... This is why almost all color film is souped in one of two basic color chemistry types (C-41 or E-6). There are different E-6 and C-41 processes. Different chemical AND entirely different developments, as well as techniques. E6 can be 3 bath or 6 bath, and C-41 can be 2 bath or 3 bath. All of this plays a SIGNIFICANT role on the shape and size of the dye clouds. It can be far more significant than you made it out to be. However, I know of no color development technique that is capable of moving film grain or dye clouds within the emulsion so that they can line up the grain as a result of the image content. If you do, I'd like to here about it. I don't believe anyone ever suggested that at all...
RE: filmscanners: RE: filmscanners: Pixels per inch vs DPI
Austin, you criticise Art, then do it yourself..? How's about we all try to attack the ball, not the man.. Woah, Mark...where did I make a personal attack on Rob? I DID stick to the ball...please point it out...I am interested. At 11:31 AM 28/10/01 -0500, you wrote: .. I don't think there's any point in my responding to an argument like this. That's the point, it isn't an argument! It's like asking why the number 9 is larger than the number 4. It's just the way it is. No, it's not 'just the way it is'. There are five incremented integers between 4 and 9, and the term 'larger' than is NOT ambiguous.. :-) But why IS 9 larger than 4? You didn't explain why. Saying a dye cloud has more information content than A pixel is NOT ambiguous at all, it's just a fact...and for the same reason why 9 is more than, say, 1. Rob's question of how a dye cloud can contain more information than a pixel still stands.. Use your engineering skill and draw a picture! I'm certainly interested. I believe I've explained this about as thoroughly as I can, without holding a class... It's just a fact of simple physics that a pixel does not contain near the same amount of information as a dye cloud. That's not what was being asked. You left out 'pixel of the same or smaller size'. I didn't leave anything out...it doesn't matter WHAT size pixel, a pixel is but a single value of tonality, period. A pixel does NOT contain the same amount of information as A dye cloud. As I said, dye clouds are variable in shape, and a pixel is only a square (or some fixed shape), and the data in A pixel (he said A) does not represent this shape. Rob also raised the perfectly valid point of the rapid development in the number/size of detectors and the amount of color info they can detect. Are you suggesting that dye clouds are so small, ie molecular or atomic :-), that there is no way to create a detector that small? I said that physical limitations prohibit sensors from being as small as dye clouds. AS I said, these sensors have to get light to them, and they have to have wires in and out of them, as well as these wires require some level of separation due to noise. I also said that the consumer digital cameras that have sensors of the same resolution, but the sensor arrays are smaller in size, and give a worse image, for the noise reason, than say a D-30 with a larger sensor and same (if not less) resolution.
RE: filmscanners: RE: filmscanners: RE: filmscanners: Pixels per inch vs DPI
Thanks Harvey...but I really don't know what more I can explain...and I don't know how much more basic I can get... Sigh. -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of SKID Photography Sent: Sunday, October 28, 2001 11:14 PM To: [EMAIL PROTECTED] Subject: Re: filmscanners: RE: filmscanners: RE: filmscanners: Pixels per inch vs DPI Austin, Most of what you are saying in this latest missive was brought up before and rejected by Rob. It was at that point that I gave up. But, kudos to you for your tenacity and deep knowledge on this subject. I feel like I've been vindicated, and by someone with far more skill than I. Harvey Ferdschneider partner, SKID Photography, NYC Austin Franklin wrote: Austin wrote: That's the point, it isn't an argument! It's like asking why the number 9 is larger than the number 4. It's just the way it is. It's just a fact of simple physics that a pixel does not contain near the same amount of information as a dye cloud. I suspected I should have chosen a word other than argument. The number 9 is larger than the number 4 because it is a convention that 9 is 5 integer values larger than 4. Other than that, the digit 9 or the word nine are simply labels to represent an idea. Saying it is because it is does not constitute any sort of meaningful explanation. Some things just are, and the truth is manifested in and of it self. A basket that has 25 eggs in it has MORE eggs than a basket with 4, right? All semantics aside. Here is (one of) your original question(s)/statement(s), which I have been answering: I don't see why stochastic or random dye clouds inherently provides more information than a pixel. The point of contention appears to be more information. I believe we agree on what more and information mean. Pixels ONLY represent the tonal value of the area which the sensor sees, which does NOT represent the physical characteristics of the dye cloud, unless the dye cloud is perfectly square and happens to line up perfectly in the field of view of that one pixel. In fact a pixel MAY represent many dye clouds, or only a portion of a single dye cloud, but there is NO way you can represent the amount of information in a single dye cloud by a single pixel, when A pixel ONLY contains tonal information. Dye clouds are irregular in shape, and dye clouds do NOT line up 1:1 with pixels. Even if you did characterize each and every dye cloud digitally, you would need more than spot tonal information, You would also have to use many pixels, or characterize the shape, because it's irregular. Characterizing the shape will be very consuming (as in a lot of data) to represent. Given all that, I believe it is obvious why a dye cloud provides inherently more information than a pixel. If you don't see that, I can't explain it any further without sitting down at a white board and drawing it out step by step... Claiming that a pixel has anything to do with physics is an odd thing to do. Now that's an odd thing to do...claim a pixel has nothing to do with physics... I don't know about your scanner, but mine is not Gnostic. A pixel is a number or a set of numbers that represent a mixture and intensity of light. It's not limited by physics. A pixel has an analog to digital origin in our case. This analog to digital conversion has limitations, which ARE limitations of physics. That's just a fact. If you created a drawing with Adobe Illustrator, then your pixels would not have an analog origin. A dye cloud has a certain dimension and a certain behviour with light. A pixel is not limited in the same way. Er, a pixel is FAR more limited, since it is only representing a single characteristic of a regular patterned point source (as in a single element in a regular grid pattern of equal sized elements). A pixel could represent an area the size of an atom, or the size of a galaxy; *any* dimension Except for the fact that we are talking about film scanners, and the are a pixel can represent is limited by physics... and it may be an 8 bit number or you could pick any number of bits. Yes, and it ONLY represents tonality, NO other characteristic at all is represented by a pixel. How small would you like to make the area represented by the pixel and how many bits of RGB would you like to use until you exceed the data contained in a chemical representation of an image? Then you said it's just a matter of increasing the resolution of the grid... Which is where the physical characteristics come in play. There are physical limitations as to how many pixels you can practically use in a scanning system. You can't just make a sensor of infinite density (or infinite size and use optics
Re: filmscanners: RE: filmscanners: RE: filmscanners: Pixels per inch vs DPI
Dye clouds are a double edged sword. On the one hand, due to the random positioning and their transparent nature, they can make for a very small apparent resolution because they can overlap in all sorts of random patterns making areas much smaller than a fixed array of pixels which would read the R G B or C M Y components of the pixel area within the same area (the sensor reads all three color separations from the same locations). A single cyan dye cloud, for instance, might partially overlap with a magenta one in one area, a yellow in another and perhaps both or none in yet another. Defining all of this via a pixel array would require very, very, small pixels. Dye clouds, however, being randomly positioned and shaped allow for all sorts of irregular information to form, much of which is smaller than one dye cloud itself, although it may also not be accurate in either color or location. In this case, the grain or dye clouds contain a certain level noise (errors), but when it gets that small, our eye would rather see randomized, inaccurate information than non-random, geometric forms or total lack of this filler. Dye clouds are, in effect, positioned at the point of manufacture, when their progenitor (a silver grain) is laid down in the specific color emulsion layer, long before anyone knows the image content. The grain that eventually allows for the creation of the dye cloud in color films is completely randomized in it's position and to a lesser extent, its shape, and within manufacturing limits, its size. The only reason any specific shaped and positioned dye cloud does not become an impediment or degrade the image characteristic is because each are so small and jumbled around, so in most cases, one rarely can one see an individual dye cloud. They are very small, and are clumped with parts of other dye clouds, both of the same color, and because of their transparency, other emulsion layers which contain other colored dye clouds. Now, if some brilliant engineer (is that an oxymoron? (that's a joke!)) figures out a way to allow for silver grains to literally migrate within the emulsion or change shape as the image is formed or during development and do nice things like, say, all line up perfectly when I'm taking a picture of something with a straight line, well then, yes, the dye clouds will have something going for them that would not theoretically be possible even with very tiny pixels (however, I actually imagine those brilliant engineers will come up with a way to make pixels mobile before making grain do so ;-)... Very simply, grain, or dye clouds are predetermined in their location and shape and are not relocated by picture content. Pixels would be required to be very small to reproduce the perception of current film technologies. Fuji has a hexagonal/star shaped pixel array, so that might reduce the rectangular elements, but we still are faced with the fact that the red, green and blue separations are all taken from an identically positioned array. As to the future of digital capture technologies, who knows what might be stumbled upon. Both film and pixel based captures have inherent errors built into the process, and for now pixel based have many more limitations, but that could change. Art
Re: filmscanners: Re: filmscanners: Pixels per inch vs DPI
Thanks, It would appear the C70 hasn't made it over the great water yet. It does look like a less expensive version of the C80. Hope it comes our way soon. Art Rob Geraghty wrote: Is the C70 being sold anywhere around the world now? http://www.epson.com.au/products/home_and_office/C70.html Yes. Rob Rob Geraghty [EMAIL PROTECTED] http://wordweb.com .
Re: filmscanners: Re: filmscanners: Pixels per inch vs DPI
I see you folks recommending these other Epsons a lot, that aren't advertised with the six color photo printing.Is there any real advantage to going with something like the 890 or 1280 over one of the less expensive office color inkjets? I'm using a HP 722C right now, and I actually get pretty good results from it, although it's only 300dpi. I would like whatever I get next to be a significant improvement. Do I need to go all the way with Epson to get that? Ken
Re: filmscanners: RE: filmscanners: RE: filmscanners: Pixels per inch vs DPI
Austin, I personally have really enjoyed and learned a lot from your last several posts (after my last post) and I suspect that there comes a point where one has to realize that unfortunately, with some people, 'you can lead a horse to water...' Harvey Ferdschneider partner, SKID Photography, NYC Austin Franklin wrote: Thanks Harvey...but I really don't know what more I can explain...and I don't know how much more basic I can get... Sigh. -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of SKID Photography Sent: Sunday, October 28, 2001 11:14 PM To: [EMAIL PROTECTED] Subject: Re: filmscanners: RE: filmscanners: RE: filmscanners: Pixels per inch vs DPI Austin, Most of what you are saying in this latest missive was brought up before and rejected by Rob. It was at that point that I gave up. But, kudos to you for your tenacity and deep knowledge on this subject. I feel like I've been vindicated, and by someone with far more skill than I. Harvey Ferdschneider partner, SKID Photography, NYC
Re: RE: filmscanners: RE: filmscanners: Pixels per inch vs DPI
Austin wrote.. ..I DID stick to the ball...please point it out...I am interested. I'm away from my normal PC right now, so I can't quote the lines that I felt were getting personal (a convenient cop-out, I know!), but comments like this: '..but I really don't know what more I can explain...and I don't know how much more basic I can get... Sigh.' ..are a bit of a put-down in my book. Perhaps I am just the overly sensitive type, but I would be a bit offended if they were comments directed towards me. Anyway, back to the debate, which I hope isn't getting too off-topic.. But why IS 9 larger than 4? You didn't explain why. We have to some basic 'given's', otherwise nothing can be discussed... :-) Saying a dye cloud has more information content than A pixel is NOT ambiguous at all, it's just a fact.. Yes, agreed. But my (and I think Rob's) point is that fact is not of much importance unless you are heading for a useful conclusion, eg saying that 'electronic sensors will never give higher resolution than film.' There are many factors involved in that question, eg over what size area? (Why do we use 6x7 instead of 35mm? - Because of those darned too-big dye clouds, that's why!) And I'm concerned about recording the image, not dye clouds. In the same way that we use larger format films, and smaller dye-clouds as methods to get better images, we can keep reducing the size of the 'pixels', and if we meet a physics limit (or more often an expense limit), then we can increase the area over which that image is recorded. And that *doesn't* necessarily mean huge cameras/lenses, if you think laterally.. I said that physical limitations prohibit sensors from being as small as dye clouds. As above, this is only an issue if you are trying to match up your sensor with some pre-determined film size. AS I said, these sensors have to get light to them Which can be bent, magnified, reflected, spread... and they have to have wires in and out of them hmm. Maybe using current technology they do.. :-) Anyway, as soon as a decent *affordable* 8Mp or better digicam arrives, preferably with interchangable lenses and decent battery life, I'll be jumping ship and only dragging out the film scanner for the 'archives'.. I won't be pining for the days of dye-clouds.. mt This message was sent through MyMail http://www.mymail.com.au
Re: filmscanners: RE: filmscanners: RE: filmscanners: Pixels per inch vs DPI
Austin Franklin wrote: Very simply, grain, or dye clouds are predetermined in their location and shape and are not relocated by picture content. What about development? I could just answer this with an Austinism and say what about it?, but I'll afford you a little more respect than you hand out. Development obviously is where dye clouds are created from the photon activated silver grains. Depending if the film is a positive or negative type, those grains activated by photons either are used to create dye clouds or to be areas where dye clouds are ultimately removed via a bleaching process in creating the final processed film. Also, some developing techniques can somewhat alter the shape or size of the dye clouds, as the grain edges can be eaten away or softened by some developers or the dye cloud could be enlarged during the chemical reaction which creates it, and development could also theoretically alter what level of grain activation will develop into dye clouds by change of chemical concentration, temperature of the baths and timing. However, most of this type of thing is done in custom film development of black and white film, because the need to control so many other variables within color film development doesn't allow for much playing around. Most color film processing is fairly uniform in its method, assuming it is being done correctly (unfortunately, it often isn't controlled as well as it should be, which is evident by the number of people dealing with poor color processing). This is why almost all color film is souped in one of two basic color chemistry types (C-41 or E-6). It is true that some of the above mentioned grain alterations also occur during pushing of color films, but usually the intent of these processes is to extend film sensitivity more than to increase grain, for instance. Some photographers do pull some color films, which creates an apparent slower film and might reduce grain size in some cases. However, I know of no color development technique that is capable of moving film grain or dye clouds within the emulsion so that they can line up the grain as a result of the image content. If you do, I'd like to here about it. Art Austin Franklin wrote: Very simply, grain, or dye clouds are predetermined in their location and shape and are not relocated by picture content. What about development? .
Re: filmscanners: Re: filmscanners: Pixels per inch vs DPI
I think the model many of the major multinational high tech companies use is to get their RD money back first via selling to markets that are less price sensitive. Then they introduce the product into the US, pretty much paid for through other international sales, and can compete more easily in the very cutthroat price sensitive environment there. We get some advantage by this here in Canada, on some goods, but certainly not all. And importation can be quite daunting, due to exchange rates, taxes, shipping costs and possible loss of warranty. Art Rob Geraghty wrote: Arthur Entlich [EMAIL PROTECTED] wrote: Thanks, It would appear the C70 hasn't made it over the great water yet. It does look like a less expensive version of the C80. Hope it comes our way soon. I just wish I could buy printers at prices as cheap as in the US. I understand about discounts for bulk, but isn't it odd that Epson printers in the US are cheaper after a much longer voyage? :( Rob .
Re: filmscanners: RE: filmscanners: Pixels per inch vs DPI
Austin Franklin [EMAIL PROTECTED] wrote: How is a randomly sized and shaped dye cloud a useful characteristic of shape and position? How is it more useful than a precise position in an array? Because it is. It's the way the world works. It IS additional information, plain and simple. Usefulness is a completely different issue, but it is ALSO useful, to a point. I don't think there's any point in my responding to an argument like this. Rob
RE: filmscanners: RE: filmscanners: Pixels per inch vs DPI
Austin Franklin [EMAIL PROTECTED] wrote: How is a randomly sized and shaped dye cloud a useful characteristic of shape and position? How is it more useful than a precise position in an array? Because it is. It's the way the world works. It IS additional information, plain and simple. Usefulness is a completely different issue, but it is ALSO useful, to a point. I don't think there's any point in my responding to an argument like this. That's the point, it isn't an argument! It's like asking why the number 9 is larger than the number 4. It's just the way it is. It's just a fact of simple physics that a pixel does not contain near the same amount of information as a dye cloud.
filmscanners: RE: filmscanners: RE: filmscanners: Pixels per inch vs DPI
Austin wrote: That's the point, it isn't an argument! It's like asking why the number 9 is larger than the number 4. It's just the way it is. It's just a fact of simple physics that a pixel does not contain near the same amount of information as a dye cloud. I suspected I should have chosen a word other than argument. The number 9 is larger than the number 4 because it is a convention that 9 is 5 integer values larger than 4. Other than that, the digit 9 or the word nine are simply labels to represent an idea. Saying it is because it is does not constitute any sort of meaningful explanation. Claiming that a pixel has anything to do with physics is an odd thing to do. A pixel is a number or a set of numbers that represent a mixture and intensity of light. It's not limited by physics. A dye cloud has a certain dimension and a certain behviour with light. A pixel is not limited in the same way. A pixel could represent an area the size of an atom, or the size of a galaxy; *any* dimension and it may be an 8 bit number or you could pick any number of bits. How small would you like to make the area represented by the pixel and how many bits of RGB would you like to use until you exceed the data contained in a chemical representation of an image? I'm astonished that you could believe the fact you have stated above. But please Austin, let's drop this since it isn't helping anyone with anything to do with filmscanning as far as I can make out? Rob Rob Geraghty [EMAIL PROTECTED] http://wordweb.com
RE: filmscanners: RE: filmscanners: RE: filmscanners: Pixels per inch vs DPI
Austin wrote: That's the point, it isn't an argument! It's like asking why the number 9 is larger than the number 4. It's just the way it is. It's just a fact of simple physics that a pixel does not contain near the same amount of information as a dye cloud. I suspected I should have chosen a word other than argument. The number 9 is larger than the number 4 because it is a convention that 9 is 5 integer values larger than 4. Other than that, the digit 9 or the word nine are simply labels to represent an idea. Saying it is because it is does not constitute any sort of meaningful explanation. Some things just are, and the truth is manifested in and of it self. A basket that has 25 eggs in it has MORE eggs than a basket with 4, right? All semantics aside. Here is (one of) your original question(s)/statement(s), which I have been answering: I don't see why stochastic or random dye clouds inherently provides more information than a pixel. The point of contention appears to be more information. I believe we agree on what more and information mean. Pixels ONLY represent the tonal value of the area which the sensor sees, which does NOT represent the physical characteristics of the dye cloud, unless the dye cloud is perfectly square and happens to line up perfectly in the field of view of that one pixel. In fact a pixel MAY represent many dye clouds, or only a portion of a single dye cloud, but there is NO way you can represent the amount of information in a single dye cloud by a single pixel, when A pixel ONLY contains tonal information. Dye clouds are irregular in shape, and dye clouds do NOT line up 1:1 with pixels. Even if you did characterize each and every dye cloud digitally, you would need more than spot tonal information, You would also have to use many pixels, or characterize the shape, because it's irregular. Characterizing the shape will be very consuming (as in a lot of data) to represent. Given all that, I believe it is obvious why a dye cloud provides inherently more information than a pixel. If you don't see that, I can't explain it any further without sitting down at a white board and drawing it out step by step... Claiming that a pixel has anything to do with physics is an odd thing to do. Now that's an odd thing to do...claim a pixel has nothing to do with physics... I don't know about your scanner, but mine is not Gnostic. A pixel is a number or a set of numbers that represent a mixture and intensity of light. It's not limited by physics. A pixel has an analog to digital origin in our case. This analog to digital conversion has limitations, which ARE limitations of physics. That's just a fact. If you created a drawing with Adobe Illustrator, then your pixels would not have an analog origin. A dye cloud has a certain dimension and a certain behviour with light. A pixel is not limited in the same way. Er, a pixel is FAR more limited, since it is only representing a single characteristic of a regular patterned point source (as in a single element in a regular grid pattern of equal sized elements). A pixel could represent an area the size of an atom, or the size of a galaxy; *any* dimension Except for the fact that we are talking about film scanners, and the are a pixel can represent is limited by physics... and it may be an 8 bit number or you could pick any number of bits. Yes, and it ONLY represents tonality, NO other characteristic at all is represented by a pixel. How small would you like to make the area represented by the pixel and how many bits of RGB would you like to use until you exceed the data contained in a chemical representation of an image? Then you said it's just a matter of increasing the resolution of the grid... Which is where the physical characteristics come in play. There are physical limitations as to how many pixels you can practically use in a scanning system. You can't just make a sensor of infinite density (or infinite size and use optics), since these bring up physical limitations. These are just facts of physics, and why physics is involved. I'm astonished that you could believe the fact you have stated above. Because what I have stated ARE facts. It would take MANY MANY pixels to represent the physical characteristics of a single dye cloud, and one could argue for quite some time what is the correct number of pixels to do this...and NO, because of physical limitations on sensor element sizes (that are NOT the same as faster processors, larger memory etc...those aren't analog sensors, so advances in those areas are not entirely applicable to advances in digital imaging sensors in this case) you can not just increase the resolution of the grid.
Re: filmscanners: RE: filmscanners: Pixels per inch vs DPI
I agree about the eventually partbut not yet. I am talking about what is now, not what is theoretically possible, and probable. We essentially, are in agreement. Harvey Ferdschneider partner, SKID photography, NYC Rob Geraghty wrote: SKID Photography [EMAIL PROTECTED] wrote: While I agree that the pixels will be 'smoother' because of the inkjet dither pattern, film grain still contains/imparts more information (on a one to one basis) than a pixel, not matter how it is dithered by the printer. Why? So far I've heard this claimed a number of times, but I've still heard nothing which backs it up. I'm prepared to be convinced, but you haven't explained the facts behind the statement you make above. Are we talking about any theoretical pixel, or the average 24 bit pixel? If we're just talking 24bits per pixel, and 2700 or 4000dpi then absolutely the film contains more information. QED. But if the area represented by the pixel is similar to the area of the smallest dye cloud in the film, and the pixel has enough bits to represent colour, I can't see any reason why the film would contain any more useful information than the digital representation, or the why the digital image would be inherently worse. :-7 Everyone has been telling me that we will all inevitably end up using filmless systems. What changed? I know a digital image is only a representation of an analogue event of light, but an image on film is much the same - a representation with limits. Eventually, the digital image will become as good or better than the film image according to market demands. I suspect this discussion is not really relevent to film scanning any more, other than to say that film scanning is a stopgap between film with chemical production of prints and digital imaging. Eventually it will become redundant except for scanning historical material. But we all know that, don't we? :-7 Rob
Re: filmscanners: RE: filmscanners: RE: filmscanners: Pixels per inch vs DPI
Austin, Most of what you are saying in this latest missive was brought up before and rejected by Rob. It was at that point that I gave up. But, kudos to you for your tenacity and deep knowledge on this subject. I feel like I've been vindicated, and by someone with far more skill than I. Harvey Ferdschneider partner, SKID Photography, NYC Austin Franklin wrote: Austin wrote: That's the point, it isn't an argument! It's like asking why the number 9 is larger than the number 4. It's just the way it is. It's just a fact of simple physics that a pixel does not contain near the same amount of information as a dye cloud. I suspected I should have chosen a word other than argument. The number 9 is larger than the number 4 because it is a convention that 9 is 5 integer values larger than 4. Other than that, the digit 9 or the word nine are simply labels to represent an idea. Saying it is because it is does not constitute any sort of meaningful explanation. Some things just are, and the truth is manifested in and of it self. A basket that has 25 eggs in it has MORE eggs than a basket with 4, right? All semantics aside. Here is (one of) your original question(s)/statement(s), which I have been answering: I don't see why stochastic or random dye clouds inherently provides more information than a pixel. The point of contention appears to be more information. I believe we agree on what more and information mean. Pixels ONLY represent the tonal value of the area which the sensor sees, which does NOT represent the physical characteristics of the dye cloud, unless the dye cloud is perfectly square and happens to line up perfectly in the field of view of that one pixel. In fact a pixel MAY represent many dye clouds, or only a portion of a single dye cloud, but there is NO way you can represent the amount of information in a single dye cloud by a single pixel, when A pixel ONLY contains tonal information. Dye clouds are irregular in shape, and dye clouds do NOT line up 1:1 with pixels. Even if you did characterize each and every dye cloud digitally, you would need more than spot tonal information, You would also have to use many pixels, or characterize the shape, because it's irregular. Characterizing the shape will be very consuming (as in a lot of data) to represent. Given all that, I believe it is obvious why a dye cloud provides inherently more information than a pixel. If you don't see that, I can't explain it any further without sitting down at a white board and drawing it out step by step... Claiming that a pixel has anything to do with physics is an odd thing to do. Now that's an odd thing to do...claim a pixel has nothing to do with physics... I don't know about your scanner, but mine is not Gnostic. A pixel is a number or a set of numbers that represent a mixture and intensity of light. It's not limited by physics. A pixel has an analog to digital origin in our case. This analog to digital conversion has limitations, which ARE limitations of physics. That's just a fact. If you created a drawing with Adobe Illustrator, then your pixels would not have an analog origin. A dye cloud has a certain dimension and a certain behviour with light. A pixel is not limited in the same way. Er, a pixel is FAR more limited, since it is only representing a single characteristic of a regular patterned point source (as in a single element in a regular grid pattern of equal sized elements). A pixel could represent an area the size of an atom, or the size of a galaxy; *any* dimension Except for the fact that we are talking about film scanners, and the are a pixel can represent is limited by physics... and it may be an 8 bit number or you could pick any number of bits. Yes, and it ONLY represents tonality, NO other characteristic at all is represented by a pixel. How small would you like to make the area represented by the pixel and how many bits of RGB would you like to use until you exceed the data contained in a chemical representation of an image? Then you said it's just a matter of increasing the resolution of the grid... Which is where the physical characteristics come in play. There are physical limitations as to how many pixels you can practically use in a scanning system. You can't just make a sensor of infinite density (or infinite size and use optics), since these bring up physical limitations. These are just facts of physics, and why physics is involved. I'm astonished that you could believe the fact you have stated above. Because what I have stated ARE facts. It would take MANY MANY pixels to represent the physical characteristics of a single dye cloud, and one could argue for quite some time what is the correct number of pixels to do this...and NO, because of physical limitations on sensor element sizes (that are NOT the same as faster
Re: filmscanners: RE: filmscanners: Pixels per inch vs DPI
Rob Geraghty wrote: SKID Photography [EMAIL PROTECTED] wrote: Are you saying that because inkjet printers employ a schoastic dithering pattern to represent pixels that film grain and scan pixels (samples, whatever) are equivalent in regards to the amount of information they impart to an inkjet printer? I think Art was saying that the relationship between pixels in the file and dots on the page isn't clear cut because the dither pattern used by the printer driver is random and therefore undoes some of the regularity of the pixels. The print ends up looking smoother than say a monitor image because the printer shadings aren't constructed as rectilinear sharp edged objects but random spots of colour. Rob While I agree that the pixels will be 'smoother' because of the inkjet dither pattern, film grain still contains/imparts more information (on a one to one basis) than a pixel, not matter how it is dithered by the printer. Harvey Ferdschneider partner, SKID Photography, NYC
Re: filmscanners: RE: filmscanners: Pixels per inch vs DPI
SKID Photography wrote: Art, I'm not trying to be difficult, but I don't understand what you are trying to say with the below post relative to film grain. Are you saying that because inkjet printers employ a schoastic dithering pattern to represent pixels that film grain and scan pixels (samples, whatever) are equivalent in regards to the amount of information they impart to an inkjet printer? I wouldn't mind if you want to be difficult ;-) But no, I'm, not saying that they are equivalent. But I am saying that inkjet printers get around the defined pixel array and rectangular aspect of pixels (and their defined borders) by using a printing method which randomizes the edges and placement of the dots (as much as by 1/2880th of an inch). Therefore, some of the rigidity of the pixelized nature of digital images is mitigated by the printing process used. One could argue that film isn't continuous tone either, since the size of the dye clouds are random but relatively predetermined by the silver grain size, and they also do not have a full variation of levels of color density. It is the random nature of the three different layers of the grain/dye clouds that creates the analogue result which we find more pleasing, because, as you stated, an array with rigid sized and shaped steps is more obvious to our eye. Unless one works hard at trying to create steplike diagonals, (for instance), which certainly can be visible in a screen image at certain magnifications, it is relatively hard to reproduce these in the inkjet printed image, unless one works at very low resolution. I both ran a color lab and did custom Cibachrome images using fairly good Nikon EL lenses on a Beseler enlarger at home, so I have a fairly good sense of what good prints look like. Using a several year old Epson 850 printer with there photo paper I've been producing 8x10s from scanned 35mm slides, and I'd be hard pressed to pick them out from a reasonable custom type C lab print. Maybe not as good as a Ciba, but fairly close. Other knowledgeable people I show them to agree. The secret for me is to use 1440 dpi, the photo paper setting and the microweave/super printing feature to prevent banding, and error diffusion, and make sure the heads have no clogged nozzles. Sure, with a loupe you'll see some dots (the printer uses down to 4 picolitre dots), but that's smaller than most photo grain appears on a 8x10 print. Art
Re: filmscanners: RE: filmscanners: Pixels per inch vs DPI
Harvey, Just to clarify, my original comments about the randomization of the pixel edges, etc. was in response to your comment below. I was not implying that current pixel resolution could achieve photographic grain randomness or resolution at current. However, I would agree with Rob that should the resolution of digital scans become high enough, the issue would become moot. Grain is just another dot pattern of random size and placement. Seems a bit of fuzzy logic could replicate it should that be desirable, once the resolution is high enough. We humans seem to like analogue qualities in our visual info, sound, etc. Probably because our own cell structure (and that of most living things) is pretty chaotic, and our retinas are certainly not rigid arrays. Film mimics this quality, so we prefer the results. As I said earlier, the placement and relative size of the grain which creates the dye clouds within the film emulsion is predetermined during the manufacturing process. It isn't like they move around after the picture is taken to produce the image, so they could just as easily be detrimentally located as augmenting to the image quality. It's just that they are so small and there are so many that they allow for more precise positioning than do pixels currently. If pixels were small enough, and closely enough spaced, I don't think the issue of their array position would be very important, if there delineation's became fully invisible to the naked eye. Art QUOTED: I think that part of it, is that pixels are aligned in a grid and have a rectilinear shape, whereas the film grain is (for lack of a better description) schoastic in arrangement and irregular in shape, thereby providing more tonal information than pixels. SKID Photography wrote: While I agree that the pixels will be 'smoother' because of the inkjet dither pattern, film grain still contains/imparts more information (on a one to one basis) than a pixel, not matter how it is dithered by the printer. Harvey Ferdschneider partner, SKID Photography, NYC .
Re: filmscanners: RE: filmscanners: Pixels per inch vs DPI
I couldn't (and probably didn't) say it better myself ;-) Art Rob Geraghty wrote: SKID Photography [EMAIL PROTECTED] wrote: Are you saying that because inkjet printers employ a schoastic dithering pattern to represent pixels that film grain and scan pixels (samples, whatever) are equivalent in regards to the amount of information they impart to an inkjet printer? I think Art was saying that the relationship between pixels in the file and dots on the page isn't clear cut because the dither pattern used by the printer driver is random and therefore undoes some of the regularity of the pixels. The print ends up looking smoother than say a monitor image because the printer shadings aren't constructed as rectilinear sharp edged objects but random spots of colour. Rob .
Re: filmscanners: RE: filmscanners: Pixels per inch vs DPI
SKID Photography [EMAIL PROTECTED] wrote: While I agree that the pixels will be 'smoother' because of the inkjet dither pattern, film grain still contains/imparts more information (on a one to one basis) than a pixel, not matter how it is dithered by the printer. Why? So far I've heard this claimed a number of times, but I've still heard nothing which backs it up. I'm prepared to be convinced, but you haven't explained the facts behind the statement you make above. Are we talking about any theoretical pixel, or the average 24 bit pixel? If we're just talking 24bits per pixel, and 2700 or 4000dpi then absolutely the film contains more information. QED. But if the area represented by the pixel is similar to the area of the smallest dye cloud in the film, and the pixel has enough bits to represent colour, I can't see any reason why the film would contain any more useful information than the digital representation, or the why the digital image would be inherently worse. :-7 Everyone has been telling me that we will all inevitably end up using filmless systems. What changed? I know a digital image is only a representation of an analogue event of light, but an image on film is much the same - a representation with limits. Eventually, the digital image will become as good or better than the film image according to market demands. I suspect this discussion is not really relevent to film scanning any more, other than to say that film scanning is a stopgap between film with chemical production of prints and digital imaging. Eventually it will become redundant except for scanning historical material. But we all know that, don't we? :-7 Rob
RE: filmscanners: RE: filmscanners: Pixels per inch vs DPI
Austin Franklin [EMAIL PROTECTED] wrote: Rob wrote: I don't see why stochastic or random dye clouds inherently provides more information than a pixel. Actually, FAR more. It's their position and size, not their color, that is far more information than pixels are. Pixels (in current implementations) must fall on a grid pattern, and are a fixed size. But the information in a pixel is limited only by the number of bits used to represent the colour. But a dye cloud is more than color. It is ALSO shape and position. Those characteristics (information) are NOT represented by color. Can you explain your claim in some way that shows me how one is better than the other? If you're comparing say a 2700dpi pixel grid with a film emulsion, then I agree. But as far as I can tell, it's just a matter of increasing the resolution of the grid and/or the number of bits in each pixel and you should be able to meet or exceed the amount of information stored in the film. Yes, but increase it to what? You would need to be able to scan the exact edges of every randomly placed dye cloud...it's about three orders of magnitude more information than is currently possible. For practical purposes there has to be a point where the difference becomes irrelevent, or people wouldn't use scanning back cameras or really high res CCD cameras in professional situations. Well, there's yet another problem. You can only make a CCD (or CMOS) pixel just so small, and you are limited also by size of the array. This is a physics limitation. Each and every pixel has to have wires running to and from it. That is not the case with dye clouds ;-) Also, the smaller you make them, the more noise you get. This is the reason the cheapo digital cameras use the small pixel arrays and they are not as good as the larger arrays (physical size, not more pixels) as far as picture quality. Hence, the Canon D30 is FAR better than a same or more sized cheapo digital camera.
Re: filmscanners: RE: filmscanners: Pixels per inch vs DPI
Austin Franklin [EMAIL PROTECTED] wrote: But a dye cloud is more than color. It is ALSO shape and position. Those characteristics (information) are NOT represented by color. How is a randomly sized and shaped dye cloud a useful characteristic of shape and position? How is it more useful than a precise position in an array? Yes, but increase it to what? You would need to be able to scan the exact edges of every randomly placed dye cloud...it's about three orders of magnitude more information than is currently possible. Now you're talking scanning, which I wasn't. I was talking about representing an image with a pixel array, not the process of getting the image into the array. If you scan film, you're making the process a lot harder than other methods of digitally capturing an image. We've already spent a lot of bandwidth talking about aliasing and other limitations of scanning! Well, there's yet another problem. You can only make a CCD (or CMOS) pixel just so small, and you are limited also by size of the array. This is a physics limitation. Dye clouds are subject to limitations of size as well. This is a spurious argument. Hence, the Canon D30 is FAR better than a same or more sized cheapo digital camera. And the technology used in today's base model computers was top of the line or didn't exist a year or two ago. When I was at university, people were convinced that physics severely limited the size of transistors and the speed at which they operated. The first IBM PC ran at what, 4.77 MHz? And now you can buy a 2GHz PIV? If the market is there, the technology will be developed to meet the demand. Absolutely I agree that Provia 100F and Reala in their own way can store more information in an image than a D30. But look at how quickly the technology has advanced to produce the D30. The rate of improvement in digital imaging is much greater than the rate of improvement in film technology. I don't think anyone is doubting that film is doomed for the majority of consumers are we? Surely it's just a question of time? But meanwhile, I'll continue to try to get the most out of my scanner, and improve my skills at achieving a good result. :) Rob
RE: filmscanners: RE: filmscanners: Pixels per inch vs DPI
Austin Franklin [EMAIL PROTECTED] wrote: But a dye cloud is more than color. It is ALSO shape and position. Those characteristics (information) are NOT represented by color. How is a randomly sized and shaped dye cloud a useful characteristic of shape and position? How is it more useful than a precise position in an array? Because it is. It's the way the world works. It IS additional information, plain and simple. Usefulness is a completely different issue, but it is ALSO useful, to a point. Yes, but increase it to what? You would need to be able to scan the exact edges of every randomly placed dye cloud...it's about three orders of magnitude more information than is currently possible. Now you're talking scanning, which I wasn't. I was talking about representing an image with a pixel array, not the process of getting the image into the array. If you scan film, you're making the process a lot harder than other methods of digitally capturing an image. Absolutely not true. Film scanning is STILL better than any digital camera in existence. We've already spent a lot of bandwidth talking about aliasing and other limitations of scanning! Certainly NOT with me you haven't. How come a drum scanner can scan at 10k/inch if film scanning is so limited? Well, there's yet another problem. You can only make a CCD (or CMOS) pixel just so small, and you are limited also by size of the array. This is a physics limitation. Dye clouds are subject to limitations of size as well. This is a spurious argument. It's important to understanding what is going on here. It's really simple, and why you are fighting it, I can't understand. Hence, the Canon D30 is FAR better than a same or more sized cheapo digital camera. And the technology used in today's base model computers was top of the line or didn't exist a year or two ago. BZZT. Wrong answer. Entirely different issue. We are talking about sensors, which are analog data acquisition devices...and that is apples and oranges compared to computer technology. The limitations are entirely different. I described the limitations of digital camera sensors. These limitations are fact. I don't think anyone is doubting that film is doomed for the majority of consumers are we? Surely it's just a question of time? That may be true, but it has nothing to do with the issues above. General consumers were happy with Polaroid pictures for God's sake! But meanwhile, I'll continue to try to get the most out of my scanner, and improve my skills at achieving a good result. :) To get the most out of your scanner, I would suggest getting good a using good films, exposing them accurately and developing them for low grain. These are really key to getting good images out of scanners.
Re: filmscanners: RE: filmscanners: Pixels per inch vs DPI
Rob Geraghty wrote: I think that's an important point - we all have different standards. I have a photographic print on my wall at home which everyone I know loves, yet it was made from ordinary 100ASA Kodak print film back in about 1982. It's quite grainy! The point is you would normally view it from halfway across the room, not at reading distance. For me, this is the sort of situation where a print with less than 240 ppi would work. I think it's important to remember that film grain and pixels are not interchangeable terms. One can have a really grainy image, blown way up and still have a full rich tonal range and luminescence, where as the same cannot be said for a digital output that has too few pixels. I think that part of it, is that pixels are aligned in a grid and have a rectilinear shape, whereas the film grain is (for lack of a better description) schoastic in arrangement and irregular in shape, thereby providing more tonal information than pixels. I know that there are those out there that think grain is a dirty word and that the presence of it, limits the possible size or viewing distance of a print. But go to any museum with a good photo collection and you will see that the masters were easily able to get beyond those artificial limitations. That is not to say that the grainy images will be the same as an 8x10 contact print. Separate but equal. Harvey Ferdschneider partner, SKID Photography, NYC
Re: filmscanners: RE: filmscanners: Pixels per inch vs DPI
SKID Photography [EMAIL PROTECTED] wrote: I think it's important to remember that film grain and pixels are not interchangeable terms. I didn't mean to imply that they were. I was simply trying to make an analogy about expected viewing distance. I think that part of it, is that pixels are aligned in a grid and have a rectilinear shape, whereas the film grain is (for lack of a better description) schoastic in arrangement and irregular in shape, thereby providing more tonal information than pixels. I don't see why stochastic or random dye clouds inherently provides more information than a pixel. A single pixel can represent one of 16 million colours (more or less depending on your bit depth). A single dye cloud can't. The problem is that our pixel sizes aren't down to dye cloud sizes yet (not in the consumer scanners!), and the printers we print with still can't reproduce quite the same subtlety of colour or tone yet. Owners of 1290's may take me to task there. :) In any case until we get affordable digicams with resolution similar to film, most of us are scanning film at resolutions that result in aliasing of some kind, and therefore don't get all the possible information out of the film. I know that there are those out there that think grain is a dirty word and that the presence of it, limits the possible size or viewing distance of a print. Grain is only a dirty word for me because when I scan grainy film I get aliasing. If I could scan at (say) 8000ppi and print a 20x30 image, I might be able to reproduce that photographic enlargement. Actually, for A3 sized prints I've already improved on it because after scanning I was able to remove the dust and scratches from the neg. :) Rob
RE: filmscanners: RE: filmscanners: Pixels per inch vs DPI
I don't see why stochastic or random dye clouds inherently provides more information than a pixel. Actually, FAR more. It's their position and size, not their color, that is far more information than pixels are. Pixels (in current implementations) must fall on a grid pattern, and are a fixed size.
Re: filmscanners: RE: filmscanners: Pixels per inch vs DPI
Art, I'm not trying to be difficult, but I don't understand what you are trying to say with the below post relative to film grain. Are you saying that because inkjet printers employ a schoastic dithering pattern to represent pixels that film grain and scan pixels (samples, whatever) are equivalent in regards to the amount of information they impart to an inkjet printer? Harvey Ferdschneider partner, SKID Photography, NYC Arthur Entlich wrote: Pixels are pretty much only in an array and rectangular on a monitor or a continuous tone printer output. Since inkjet printers use a sub-array of randomized dots to create the illusion of a specific pixel color (usually blended into its surrounding pixel neighbors, as well), there is rarely any true delineation of rectangular pixels in inkjet prints, which use either dithering or an error diffusion pattern to create blends. Art SKID Photography wrote: Rob Geraghty wrote: I think that's an important point - we all have different standards. I have a photographic print on my wall at home which everyone I know loves, yet it was made from ordinary 100ASA Kodak print film back in about 1982. It's quite grainy! The point is you would normally view it from halfway across the room, not at reading distance. For me, this is the sort of situation where a print with less than 240 ppi would work. I think it's important to remember that film grain and pixels are not interchangeable terms. One can have a really grainy image, blown way up and still have a full rich tonal range and luminescence, where as the same cannot be said for a digital output that has too few pixels. I think that part of it, is that pixels are aligned in a grid and have a rectilinear shape, whereas the film grain is (for lack of a better description) schoastic in arrangement and irregular in shape, thereby providing more tonal information than pixels. I know that there are those out there that think grain is a dirty word and that the presence of it, limits the possible size or viewing distance of a print. But go to any museum with a good photo collection and you will see that the masters were easily able to get beyond those artificial limitations. That is not to say that the grainy images will be the same as an 8x10 contact print. Separate but equal. Harvey Ferdschneider partner, SKID Photography, NYC .
Re: filmscanners: RE: filmscanners: Pixels per inch vs DPI
Austin Franklin [EMAIL PROTECTED] wrote: Rob wrote: I don't see why stochastic or random dye clouds inherently provides more information than a pixel. Actually, FAR more. It's their position and size, not their color, that is far more information than pixels are. Pixels (in current implementations) must fall on a grid pattern, and are a fixed size. But the information in a pixel is limited only by the number of bits used to represent the colour. I don't see why a random pattern of dyes would represent an image better than a grid pattern of pixels with a point size equivalent to the size of the smallest dye cloud. Can you explain your claim in some way that shows me how one is better than the other? If you're comparing say a 2700dpi pixel grid with a film emulsion, then I agree. But as far as I can tell, it's just a matter of increasing the resolution of the grid and/or the number of bits in each pixel and you should be able to meet or exceed the amount of information stored in the film. For practical purposes there has to be a point where the difference becomes irrelevent, or people wouldn't use scanning back cameras or really high res CCD cameras in professional situations. Rob
Re: filmscanners: RE: filmscanners: Pixels per inch vs DPI
SKID Photography [EMAIL PROTECTED] wrote: Are you saying that because inkjet printers employ a schoastic dithering pattern to represent pixels that film grain and scan pixels (samples, whatever) are equivalent in regards to the amount of information they impart to an inkjet printer? I think Art was saying that the relationship between pixels in the file and dots on the page isn't clear cut because the dither pattern used by the printer driver is random and therefore undoes some of the regularity of the pixels. The print ends up looking smoother than say a monitor image because the printer shadings aren't constructed as rectilinear sharp edged objects but random spots of colour. Rob
Re: filmscanners: RE: filmscanners: Pixels per inch vs DPI
Chapter 14 of Professional Photoshop - Resolving the Resolution Issue: printed dots per inch consist of grids of spots per dot - of differing picoliter sizes depending on the printer. Apples and oranges? Maris - Original Message - From: Rob Geraghty [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Friday, October 26, 2001 8:17 PM Subject: Re: filmscanners: RE: filmscanners: Pixels per inch vs DPI | SKID Photography [EMAIL PROTECTED] wrote: | Are you saying that because inkjet printers employ a schoastic dithering | pattern to represent pixels that film | grain and scan pixels (samples, whatever) are equivalent in regards to the | amount of information they impart | to an inkjet printer? | | I think Art was saying that the relationship between pixels in the file and | dots on the page isn't clear cut because the dither pattern used by the | printer driver is random and therefore undoes some of the regularity of the | pixels. The print ends up looking smoother than say a monitor image because | the printer shadings aren't constructed as rectilinear sharp edged objects | but random spots of colour. | | Rob | | |
filmscanners: RE: filmscanners: RE: filmscanners: Pixels per inch vs DPI
Austin wrote: Why would you want to output at a fixed 300 PPI? Because that's the requirement of the offset printer which many of my recent photos are going to. Aside from that, 300 dpi is as a general rule of thumb the best resolution *most* printers (pc and otherwise) work with. Some are more, some are less, and the manufacturers muddy the water by talking about the size of individual ink dots not the size at which a pixel is reproduced as a dot on the print. If you are outputting to an inkjet printer, you are best to just choose your image output size and let the PPI output to the printer fall where it may. So what do you set the dpi to in the file? If you create a TIFF file, there will be a figure for the dpi embedded in it. I use 300dpi. When I actually print from PSP the real dpi is hardly ever precisely 300dpi - it depends on the page layout and how the picture is cropped. So yes, I'm effectively doing what you suggest above when prniting on my own printer. But I have to set the file's dpi to something, and it makes more sense to set it to 300dpi than 2700dpi or 100dpi for a full frame 2700spi ( :) scan. If you do any resizing of the PPI to make some fixed number, then you are resampling, which degrades the image. I was talking about the setting in the file. You can set the output dpi of Vuescan (or Nikonscan I think) to anything you want. It makes no difference to the number of pixels. Setting it to 300 dpi means that you'll get a meaningfully sized print (roughly a page) out of a 2700 spi scan on most printers. Leave it at 2700dpi and you'll have a print the size of the neg frame. Set it to 100dpi and the size will be silly for printing. Epsons seem to work quite well at 240dpi because of the integer relationship with the 1440dpi native dot size. That?s pretty much been proven to by a myth. It is true, to some degree, for lineart, but not at all for halftoned images. In my own personal experience it's true that prints having an integer relationship between the output dpi and 1440 on my Epson 1160 will be sharper and have less visible dithering than at other scales. It was also true on my previous Photo 700. I expect that newer printers, particularly those with more than 4 colours, will give better results. I don't know for sure about other printers - for instance the 12x0 series probably have fine enough patterns from 6 colours at 1440 or 2880 dpi that variations in the source dpi make much less difference. I don't know because I don't have one. What I *do* know is that Epson had on their own web site an equation for calculating the ideal source resolution which was based on an integer relationship with the printer's native resolution. Epson themselves said it was the best thing to do. The story may have changed since they wrote that FAQ. Rob Rob Geraghty [EMAIL PROTECTED] http://wordweb.com
RE: filmscanners: RE: filmscanners: Pixels per inch vs DPI
2. If you want to print the picture, the maximum size you can print is limited to the number of pixels expressed at 300 ppi. I always set the output resolution of Vuescan to 300dpi. I have no idea what you mean by that...would you please elaborate? OK. I don't have any files to work with here so I'll work with some theoretical numbers. Suppose you scan a 35mm frame at 2700dpi (spi if you prefer :) and get a file containing 3000x2000 pixels. If you want to print the file at 300dpi (or 300 pixels per inch) Why would you want to output at a fixed 300 PPI? If you are outputting to an inkjet printer, you are best to just choose your image output size and let the PPI output to the printer fall where it may. If you do any resizing of the PPI to make some fixed number, then you are resampling, which degrades the image. If you just let the PPI fall where it may (providing you are above 180+, if even 240), then you only get one processing of the image, namely the halftone algorithm in the printer driver (converting PPI to DPI ;-)...and therefore less image degradation. Am I misunderstanding you perhaps? 2. As I mentioned earlier, some printers give quite good results at lower dpi. Epsons seem to work quite well at 240dpi because of the integer relationship with the 1440dpi native dot size. Thats pretty much been proven to by a myth. It is true, to some degree, for lineart, but not at all for halftoned images.