The merits of photographing a latent at 1000 ppi

[antonroland]

Hello all

I need some advice from you all in the above. The relevant SWGIT and corresponding SWGFAST guides all want latents to be photographed at 1000 ppi. Now if I do this with the equipment at my disposal I end up with a block of 2" x 3". This is clearly too big for a single latent finger print and too small for even a portion of an average adult latent palm such as thenar, hypothenar or writer's palm.

This is using a Pentax *IST DL at 6Mp and a Pentax 50mm macro lens capable of 1:1 reproductions. When the frame is filled with a block measuring 2" x 3" the lens is somewhere around 1:3.5 in magnification. These are only quick shots out of hand and not ultra-precision on a repro table.

For the purpose of this post I will not even touch on the merits of Jpeg vs Raw - that will be part two of this set of tests and the basic idea has been discussed enough elsewhere.

My question now then, is this: What is the reason for wanting a 1000 ppi capture or do we simply want a minimum of 1000 ppi? Is it for the sake of having a workable 300dpi 3x enlargement of the initial capture?

I do not really see the merit in this either as a full frame 35mm format D-SLR is only capable of capturing a single finger at 1:1 and latent palms could never be captured at 1:1

I can see the idea of speeding up capturing latents at a standard ppi suitable for AFIS inputs but practical considerations will dictate that there will always be a fixed pre-determined distance between the camera and the latent. Nice and doable in the lab but less than a pleasure on scenes.

I guess this question will still grow in complexity but let's start at some point at least

[antonroland]

Now if I were to use my EOS 5D to do the same the situation would worsen in that I would have a 1000 ppi image block of 4.368" x 2.912" that would easily swallow all four finger tips made in the fashion of simultaneous plain impressions.

If I were to do this with my EOS 1Ds Mk II I would have a 1000 ppi image block measuring approximately 4.992" x 3.328". This is nearly big enough to contain a complete adult latent palm.

If we consider a Nikon D90, the 1000 ppi image block would measure 4.288" x 2.848"

I will do some tests over the weekend and see what type of ppi size I obtain when I photograph a single latent finger tip at around 1:1.5 magnification.

I guess where I am going with this is the fact that (in the case of latent fingers at least) it is very possible - and in my world preferable - to capture latents at a much higher resolution than 1000 ppi and then work them back to 1000 ppi or resize the image in some other way for AFIS input or any other specific end result.

[Gerald Clough]

As I understand it, and as SWGFAST says, resolution is to be 1,000ppi or finer. This means that there is, as you found, some target area that will be the largest area that can be captured by a sensor/lens combination and still meet the 1,000ppi minimum. But that's just the largest area. The FBI document provided a step-by-step procedure for determining the maximum allowable distance by testing.

http://www.fbi.gov/hq/lab/fsc/backissu/ ... latent.htm

Repeat steps 5, 6, 7, and 8. Once verified, this becomes the camera-to-subject maximum distance to provide 1000 ppi.

I think at the time all this was written, there was still some possibility that some cameras might be considered that would have difficulty meeting the minimum in common situations. (I mean by people who should actually know better. We're doomed to forever see impression evidence photographed with simple cameras.) We're rapidly moving to sensor densities where the 1,000ppi minimum would be met in any rational situation.

[Steve Everist]

Hello all
My question now then, is this: What is the reason for wanting a 1000 ppi capture or do we simply want a minimum of 1000 ppi? Is it for the sake of having a workable 300dpi 3x enlargement of the initial capture?

The 1000ppi referenced is usually listed as a minimum resolution after calibration. As to why, I'm guessing it's a vendor-driven minimum.

Take all of the responses as to why there isn't a "point minimum" to ID a print and apply them to the 1000ppi resolution 'standard'. There's no reason that I'm aware of, based on the comparison of images, that there needs to be any standard. I consider it an administrative standard. That being said, I still recommend filling the frame and letting the image calibrate to whatever it ends up calibrating to.

Unfortunately what happens when these numerical standards are put in to place is that they're no longer applied as a minimum, but as a goal. So instead of using best practices to get the most out of your image, people are setting themselves up to meet the minimum. A few years ago I went to some training where they were trying 'calibrate the lens' for a specific camera so that you could put it in a certain spot so that you'd always shoot at 1000ppi. They went so far as to mark the lens with a pencil once that magical resolution had been achieved.

[antonroland]

As I understand it, and as SWGFAST says, resolution is to be 1,000ppi or finer. This means that there is, as you found, some target area that will be the largest area that can be captured by a sensor/lens combination and still meet the 1,000ppi minimum. But that's just the largest area. The FBI document provided a step-by-step procedure for determining the maximum allowable distance by testing.

http://www.fbi.gov/hq/lab/fsc/backissu/ ... latent.htm

Repeat steps 5, 6, 7, and 8. Once verified, this becomes the camera-to-subject maximum distance to provide 1000 ppi.

I think at the time all this was written, there was still some possibility that some cameras might be considered that would have difficulty meeting the minimum in common situations. (I mean by people who should actually know better. We're doomed to forever see impression evidence photographed with simple cameras.) We're rapidly moving to sensor densities where the 1,000ppi minimum would be met in any rational situation.

Hello Gerald and thanks for the response.

I had a look at the link you provided and this info is practically a carbon copy of the similarly dated Section 8 SWGIT guideline pdf. I downloaded this off their site not more than a week or two back.

Could it be because of this outdated info that jpeg is so victimized?

Also, the calibration of a lens is awfully awkward in comparison to how easy it is to do in Photoshop...

Surely these guidelines can be updated sometime? In terms of current camera technology the 6th replacement model has already been launched since the Canon D30 was launched in May 2000.

Who of us would still happily work with a pc that was assembled in May 2000?

ps, am I correct in assuming that the standard of AFIS inputs is 1000dpi for the purpose of data movement upstream and downstream or was it the chosen resolution for the sake of image quality?

[antonroland]

I still recommend filling the frame and letting the image calibrate to whatever it ends up calibrating to.

Thank you and AMEN!

Unfortunately what happens when these numerical standards are put in to place is that they're no longer applied as a minimum, but as a goal. So instead of using best practices to get the most out of your image, people are setting themselves up to meet the minimum. A few years ago I went to some training where they were trying 'calibrate the lens' for a specific camera so that you could put it in a certain spot so that you'd always shoot at 1000ppi. They went so far as to mark the lens with a pencil once that magical resolution had been achieved.

Sad that that should always be the case, aspiring to meet minimum standards...only thing worse than that...ooh no, I will not go down that road on here

[Gerald Clough]

Could it be because of this outdated info that jpeg is so victimized?

While the vilification of JPEG has a component of hysteria to it, the objections are valid for forensic photography in which fine judgments must be made from the image. I think it's not so much that there's a high risk of misinterpretation on account of some effect of conversion but rather a loss of control with an admitted potential for mischief. But the questions did start to become seriously considered when digital imaging became common, although some of the problems discussed have analogues in film processes. In one respect, digital might be said to be able to provide a more certain record of the primitive image than film, but I'm not sure that's entirely true. Film has a chemical development step (solvent versus accutance developers, for instance), but to utilize a digital capture, something still has to happen to generate a visible image, since few of us were born with memory card slots. One can always pick infinitely smaller nits.

There are just very, very few properly composed general crime scene photos that cannot fill any requirement solely on account of file format. By that, I mean that, if we're trying to interpret the nature of some small object that happened to fall within a scene ("Is that an earring?"), and we're arguing over whether or not a conversion effect makes the interpretation questionable, the fault is in not addressing the object appropriately on the scene. There will likely be far more questions raised on account of simply not properly working the scene photographically than on account of resolution and algorithm issues.

Of course, it's certainly not that a RAW file image is the viewed image, anyway. It merely preserves the sensor record and provides a base for replicating or testing the validity of whatever happened to the data downstream. And that's really the nature of the whole SWGFAST paper. Whether any such things are going to be argued with any sort of vigor will depend on the nature of the image and difficulty of interpretation, and even then, rarely. (Very few attorneys have any notion of any of these issues. Often a good thing, since I could imagine bogging down in the fact that the RAW is a pretty unsatisfying visual image and is, as are all photos, "enhanced" for practical viewing.) But it's just a matter of best practice. Even if the instant case isn't much attackable on these technical issues, one shouldn't have to admit that their standard practice doesn't provide the best possible record in cases where they may be issues. The preserved RAW file can settle the question of how an apparent sharp edge was or was not an effect of any process that increases accutance, a potential issue in the use of an image of any grayscale object. Such things hardly arose when film was universal. Far fewer non-specialists had any orientation to resolution, contrast, and accutance. Today, many more people understand (more of less) these concepts, if for no other reason than they wish to argue with some degree of knowledge in their photography forums.

[rbostrum]

Hello all
My question now then, is this: What is the reason for wanting a 1000 ppi capture or do we simply want a minimum of 1000 ppi? Is it for the sake of having a workable 300dpi 3x enlargement of the initial capture?

The 1000ppi referenced is usually listed as a minimum resolution after calibration. As to why, I'm guessing it's a vendor-driven minimum.

Take all of the responses as to why there isn't a "point minimum" to ID a print and apply them to the 1000ppi resolution 'standard'. There's no reason that I'm aware of, based on the comparison of images, that there needs to be any standard. I consider it an administrative standard. That being said, I still recommend filling the frame and letting the image calibrate to whatever it ends up calibrating to.

Unfortunately what happens when these numerical standards are put in to place is that they're no longer applied as a minimum, but as a goal. So instead of using best practices to get the most out of your image, people are setting themselves up to meet the minimum. A few years ago I went to some training where they were trying 'calibrate the lens' for a specific camera so that you could put it in a certain spot so that you'd always shoot at 1000ppi. They went so far as to mark the lens with a pencil once that magical resolution had been achieved.

To answer the original question, the reason the standard says 1000 ppi (as a minimum that you can happily exceed if you want to) is that this resolution was deemed to be the minimum necessary to capture the desired detail from the original. Meeting this minimum gives an image with the desired detail. If that's your 'goal', then there is no need to do more. Higher resolution may provide additional information but that extra data is unnecessary digital overhead - at least for the purpose addressed by the standard.

Not being a LPE, let me ask a few questions. First, what size are the 'features' of interest that need to be captured? According to the standard, they will be adequately rendered if you use 1000 ppi. I'm sure that using a higher resolution would give images that more fully represent the original being imaged but to what end? When Steve advocates "using best practices to get the most out of your image" what does this mean? If the imaging being done is that which the standard addresses, then "the most" isn't really needed; it's over-kill. The standard says 'minimum' but, once that has been achieved, the point is whether or not 'more' actually means 'better'.

Just my 2 cents worth...

[Steve Everist]

To answer the original question, the reason the standard says 1000 ppi (as a minimum that you can happily exceed if you want to) is that this resolution was deemed to be the minimum necessary to capture the desired detail from the original.

By whom and how did they come to this resolution number? Not that I couldn't have missed it, but I have yet to have anyone show me any research done to show that 1000ppi is the resolution at which the minimum necessary for capturing desired detail had been determined. I have yet to see a standard with a footnote for reference to where this number was established. I have seen it referenced relative to latent prints relative to digital transmissions standards. And known prints, for some reason, only need half the resolution to capture their necessary detail which comes from the same source - I don't get that.

Meeting this minimum gives an image with the desired detail.

Usually, however I've often found myself wanting more. In many instances, much less than 1000ppi can also give sufficient detail to come to a conclusion. This is why I am hesitant to declare a standard minimum unless I can see any valid proof that in all cases this is what is required - again, I parallel it with the idea of declaring a minimum number of 'points' in order to conclude the source of a latent print. I see these minimum numbers as administrative if nothing else.

If that's your 'goal', then there is no need to do more. Higher resolution may provide additional information but that extra data is unnecessary digital overhead - at least for the purpose addressed by the standard.

True, but whatever camera is being used is going to produce the same size file with the same pixel by pixel resolution regardless of if the latent fills the frame or not. So why not fill the frame with the latent? It's better to have more information that you need than to need more than you have, considering the file size will be equal regardless.

Not being a LPE, let me ask a few questions. First, what size are the 'features' of interest that need to be captured?

They will vary from latent print to latent print, as will the amount based on a quality/quantity that is sufficient.

According to the standard, they will be adequately rendered if you use 1000 ppi.

But what if they are not? And on the other side, there are many instances where the necessary resolution is below this standard. Once again, I haven't seen any research into how 1000ppi became the established standard.

I'm sure that using a higher resolution would give images that more fully represent the original being imaged but to what end? When Steve advocates "using best practices to get the most out of your image" what does this mean?

Best Practices are just that - procedures put into practice to best support the evidence that may be admitted to trial. In many cases they may be beyond what is necessary, however they should stand up to outside scrutiny or legal challenges.

It means that we should be trying to get the highest quality image available from the latent print we have to work with. In the case of a latent print that may be destroyed with a subsequent development technique, why assume that the magical 1000ppi number will be enough when we can fill the frame with the latent (and ruler, of course), knowing that regardless of the number - we'll have as much as we can get. Since the file size is going to be the same from every exposure of the camera - why not put it to good use? Ultimately this may be well beyond what we need, but it's better than not having as much as we need. It's much easier to sample down than up.

It also means that, in the case of a flatbed scanner, that we may want to scan at a multiple of 1200ppi so as not to automatically introduce interpolation into our image since most scanners have a native resolution that is a multiple of 1200. To automatically choose 1000ppi may be less representative of the image than choosing 600ppi - which wouldn't meet the standard.

Best Practices extend from capture to processing to printing to documentation of what was done with the image after capture. I'd rather try to get as much as I can in practice than find out in court that I didn't do enough.

If the imaging being done is that which the standard addresses, then "the most" isn't really needed; it's over-kill. The standard says 'minimum' but, once that has been achieved, the point is whether or not 'more' actually means 'better'.

It may not mean better, and in instances half of that minimum may be sufficient, however I'd much rather have more than may be necessary when it won't make a difference in the file size.

Even though I'm employed by a police agency, I work for the victims. If I were to ask my neighbors, had they been the victim of crime, if they'd rather have me do my job at or around the minimum that's been established, or for little to no difference in cost go above and beyond; I'm going to make a guess that they'd rather have me do more.

Just my 2 cents worth...

I think there are a lot of issues to address with digital capture, however there are many people who were capturing images that were used in court long before a single fingerprint could even be calibrated to anything near 1000ppi. With today's cameras, resolution isn't much of an issue. And in those cases where you have an entire palm latent (if we could be so lucky regularly), then do an overall of the entire palm and then specific close-up images of specific areas that contain enough quantity, based on the quality, that a comparison can be done should the overall not be sufficient on its own.

I'm just not a fan of this arbitrary 1000ppi number. I believe that in many instances lower is sufficient, yet in practice I'm going to continue to default to exceeding this number unless I'm unable to for whatever reason.

[rbostrum]

By whom and how did they come to this resolution number? Not that I couldn't have missed it, but I have yet to have anyone show me any research done to show that 1000ppi is the resolution at which the minimum necessary for capturing desired detail had been determined. I have yet to see a standard with a footnote for reference to where this number was established. I have seen it referenced relative to latent prints relative to digital transmissions standards. And known prints, for some reason, only need half the resolution to capture their necessary detail which comes from the same source - I don't get that.

Good questions. Since SWGIT references NIST I suppose you would have to track back through the history of that development to find out. As I had nothing to do with the process I can't help with the answer. It would not surprise me to find out that setting the figure was an essentially arbitrary decision.

Meeting this minimum gives an image with the desired detail.

Usually, however I've often found myself wanting more. In many instances, much less than 1000ppi can also give sufficient detail to come to a conclusion. This is why I am hesitant to declare a standard minimum unless I can see any valid proof that in all cases this is what is required - again, I parallel it with the idea of declaring a minimum number of 'points' in order to conclude the source of a latent print. I see these minimum numbers as administrative if nothing else.

If that's true, then the standard is not adequate. Or you are using the image for something other than that intended for the standard.

The parallel you make is interesting. Let me present it a little differently. What if there was a minimum number of points that, if met, would reliably and accurately allow a conclusion of common source? Note that I'm not saying there is, just proposing a hypothetical. If there were, then you could set a minimum standard for that purpose and it would work.

The argument in SWGIT seems to be that it is possibe for these images and the minimum value is 1000 ppi. Is this correct? I would suggest that it is not from your comments.

True, but whatever camera is being used is going to produce the same size file with the same pixel by pixel resolution regardless of if the latent fills the frame or not. So why not fill the frame with the latent? It's better to have more information that you need than to need more than you have, considering the file size will be equal regardless.

I agree, why not? The standard doesn't stop someone from doing this so what possible reason would there be to not fill the frame? Beats me.

Not being a LPE, let me ask a few questions. First, what size are the 'features' of interest that need to be captured?

They will vary from latent print to latent print, as will the amount based on a quality/quantity that is sufficient.

Fair enough. But any standard would have to aim at the smallest feature to be captured. In this instance it seems they are saying that any feature smaller than 1/500 inch doesn't not need to be detected (since 2 pixels are needed for simple detection).

According to the standard, they will be adequately rendered if you use 1000 ppi.

But what if they are not? And on the other side, there are many instances where the necessary resolution is below this standard. Once again, I haven't seen any research into how 1000ppi became the established standard.

Again, according to that standard, anything that isn't resolved at 1000 ppi isn't important. My question would be is that true? Your comments suggest to me that it is not. In which case, the figure is simply wrong and should not be used.

I'm sure that using a higher resolution would give images that more fully represent the original being imaged but to what end? When Steve advocates "using best practices to get the most out of your image" what does this mean?

Best Practices are just that - procedures put into practice to best support the evidence that may be admitted to trial. In many cases they may be beyond what is necessary, however they should stand up to outside scrutiny or legal challenges.

It means that we should be trying to get the highest quality image available from the latent print we have to work with. In the case of a latent print that may be destroyed with a subsequent development technique, why assume that the magical 1000ppi number will be enough when we can fill the frame with the latent (and ruler, of course), knowing that regardless of the number - we'll have as much as we can get. Since the file size is going to be the same from every exposure of the camera - why not put it to good use? Ultimately this may be well beyond what we need, but it's better than not having as much as we need. It's much easier to sample down than up.

It also means that, in the case of a flatbed scanner, that we may want to scan at a multiple of 1200ppi so as not to automatically introduce interpolation into our image since most scanners have a native resolution that is a multiple of 1200. To automatically choose 1000ppi may be less representative of the image than choosing 600ppi - which wouldn't meet the standard.

Best Practices extend from capture to processing to printing to documentation of what was done with the image after capture. I'd rather try to get as much as I can in practice than find out in court that I didn't do enough.

Understood and agreed. But then the question can be asked what resolution should you use? As much as humanly possible? As much as one can afford? Modern, cheaper technology makes higher resolution readily available but higher resolution doesn't always translate into better resolving power (which I would suggest is the real issue).

If the imaging being done is that which the standard addresses, then "the most" isn't really needed; it's over-kill. The standard says 'minimum' but, once that has been achieved, the point is whether or not 'more' actually means 'better'.

It may not mean better, and in instances half of that minimum may be sufficient, however I'd much rather have more than may be necessary when it won't make a difference in the file size.

Even though I'm employed by a police agency, I work for the victims. If I were to ask my neighbors, had they been the victim of crime, if they'd rather have me do my job at or around the minimum that's been established, or for little to no difference in cost go above and beyond; I'm going to make a guess that they'd rather have me do more.

We all work for the victims. But the point is whether or not you need to go beyond, isn't it? IF the standard is adequate, then doing your job at or above the minimum would be fine. IF it is not adequate, then of course you need to use more.

Please understand that I'm not defending the standard. I'm pointing out how it 'should' work. If it doesn't then the onus is on the LPE community to get it changed. Or, I suppose, ignore it.

Just my 2 cents worth...

I think there are a lot of issues to address with digital capture, however there are many people who were capturing images that were used in court long before a single fingerprint could even be calibrated to anything near 1000ppi. With today's cameras, resolution isn't much of an issue. And in those cases where you have an entire palm latent (if we could be so lucky regularly), then do an overall of the entire palm and then specific close-up images of specific areas that contain enough quantity, based on the quality, that a comparison can be done should the overall not be sufficient on its own.

I'm just not a fan of this arbitrary 1000ppi number. I believe that in many instances lower is sufficient, yet in practice I'm going to continue to default to exceeding this number unless I'm unable to for whatever reason.

I agree. Any truly arbitrary number is silly at best; outright wrong at worst.

[Steve Everist]

Good questions. Since SWGIT references NIST I suppose you would have to track back through the history of that development to find out. As I had nothing to do with the process I can't help with the answer. It would not surprise me to find out that setting the figure was an essentially arbitrary decision.

I don't know that it's completely arbitrary, however I think there's a certain push by a combination of those that oversee the standards (ANSI/NIST) and those that are vendors producing the equipment by which the data will be transmitted and processed (Livescan vendors, AFIS vendors, etc...) to adopt one. In doing so, they're looking at speed of transmission, cost of storage of the data being stored, and things I'm sure I know nothing about. Somewhere in the mix are those that conduct comparisons.

If that's true, then the standard is not adequate. Or you are using the image for something other than that intended for the standard.

It could just be that 1000ppi as the standard is adequate in most instances. But with the power of digital imaging, there may be those instances where a latent print that previously would have been set aside may be usable - but increased resolution may be needed.

The parallel you make is interesting. Let me present it a little differently. What if there was a minimum number of points that, if met, would reliably and accurately allow a conclusion of common source? Note that I'm not saying there is, just proposing a hypothetical. If there were, then you could set a minimum standard for that purpose and it would work.

If there were a scientific foundation that were used for the determination, then I would say that it can. The problem is the lack of scientific basis for this support. Maybe that will change with some of the new probabilistic models that are being researched.

The argument in SWGIT seems to be that it is possibe for these images and the minimum value is 1000 ppi. Is this correct? I would suggest that it is not from your comments.

To be honest, I don't know how SWGIT (or SWGFAST) came to this number. My guess is it came to SWGIT first and SWGFAST adopted it without reason for why its best. I sent comments to both when relevant drafts were available for comment. They acknowledged receipt of my comments, but neither gave (or gives) feedback to commenters.

Fair enough. But any standard would have to aim at the smallest feature to be captured. In this instance it seems they are saying that any feature smaller than 1/500 inch doesn't not need to be detected (since 2 pixels are needed for simple detection).

I think one problem is that the features aren't always cut and dry after development. And the poorer the quality of the latent the greater the quantity. And quantity isn't always Level 2 details.

But what if they are not? And on the other side, there are many instances where the necessary resolution is below this standard. Once again, I haven't seen any research into how 1000ppi became the established standard.

Again, according to that standard, anything that isn't resolved at 1000 ppi isn't important. My question would be is that true? Your comments suggest to me that it is not. In which case, the figure is simply wrong and should not be used.

This is where I have issues with the standard. Unless it is stated as an administrative standard, and in some instances that lower resolution is sufficient, then it's ultimately an arbitrary number. Over 10 years ago, most digital cameras weren't going to resolve even a finger latent at 1000ppi. And these are the cases we reference in our SOP's Prior Case Law sections.

Understood and agreed. But then the question can be asked what resolution should you use? As much as humanly possible? As much as one can afford? Modern, cheaper technology makes higher resolution readily available but higher resolution doesn't always translate into better resolving power (which I would suggest is the real issue).

The courts don't require anything of digital that wasn't asked of traditional (is it too soon to call it legacy?) based film. All that means is that the image must represent what it purports to represent. So we're the ones who are setting minimum resolution standards. In some instances, I think we really are doing this administratively so as to cover things for those who are going to be capturing images but don't necessarily have the knowledge or skill (whether ignorantly or purposely) to understand the issue. So to parallel my previous comparisons with latent prints - how much is enough? As much as it takes.

As an aside, one of the most common requests I see relating digital images to latent print work is "which printer is best for printing digital images of latent prints at 1:1?" Since most printers don't even resolve at 400ppi on output, it really doesn't matter if the original was calibrated to 1000, 1200, or 2400ppi if the comparison is being done from a printout of that image at 1:1. What's equally interesting is that the same person printing at 1:1 of their reduced resolution image is going to put their magnifier on top of it. Of course this doesn't matter if the comparison work is done on-screen with the higher resolution image.

Please understand that I'm not defending the standard. I'm pointing out how it 'should' work. If it doesn't then the onus is on the LPE community to get it changed. Or, I suppose, ignore it.

Very true. As all of what used to be called Guidelines evolve into being called "Standards", my concern is with those agencies that do a blanket acceptance of all of them by putting them into their SOP's. Yet in many cases they don't read the documents themselves. Down the road they may find themselves with issues in court or when being inspected for accreditation. If the Standard is 1000ppi and your office has in its SOP's that SWIGIT or SWGFAST Standards are complied with, and then captured images are found to be under this number - you just violated your own standards. Could this lead to failing an inspection from ASCLD/LAB or worse yet - the images being thrown out of court?

I agree. Any truly arbitrary number is silly at best; outright wrong at worst.

...if it's going to be stated as a standard.

[Gerald Clough]

Wouldn't you imagine that 1,000ppi was pretty much a hip shot at a bare minimum for recording, influenced to some degree by the resolution of other devices, but really just a reasonably attainable adequate number, without demanding achievement of film resolution. Making the conversion from typical color film resolution of 1,000 to 1,600 line pairs per inch, theoretically requiring four pixels to distinguish two pairs on a digital sensor, to at least 4,000 ppi to match the line resolution of the color film. Or somewhere in the direction of the high resolution film end, the Polaroid T-55 negative resolution of 4,000 line pairs per inch, or a 16,000 ppi equivalent. But it's pretty clear that when we try to record fine detail on film, having a high-resolution-capable sensing surface doesn't mean we are always entirely delighted with the result.

Decent lenses can mostly hit the equivalent of better than 1,700 line pairs per inch, which is right in there with the pixel pitch of the better digital sensors (EOS 40D, EOS 1Ds MkIII). So, even without considering more flexible adjustment for subjective accutance, most digital camera/lens combinations we would use for latent photography will resolve about as well as typical color film and, given limitations on lens resolution, not practically worse than the same optics with Technical Pan film (film resolution 2,500+ line pairs per inch). When we consider that 1,000ppi is 250 line pairs per inch, it's quite a modest minimum easily attainable for generations.

Of course, lab test numbers don't tell all that much of the story. And film and digital sensors are more different than just the chemical and physical nature of recording light, and each is subject to different artifacts and differences in the most fundamental steps of making the record usable.

But I have to think that the SWGFAST standard of "captured at 1,000ppi" addresses nothing more than an archival standard. That is, the standard tacitly recognizes that (1) images are recorded and transmitted for a variety of purposes, (2) the common resolutions of the formats and devices used to display, analyze, and transmit images will increase, to different degrees and at different points in time for each application. There are two issues to consider. One is that, if 1,000ppi is accepted as adequate (I know - the validity of that assumption is another matter), it's a reasonable number to declare as a minimum. The other is that, if that assumption is valid, it declares that one need not challenge an image captured to at least that standard, even if a few or many operations are working to a far higher standard. In other words, if 1,000ppi is adequate, an archived image cannot be reasonably challenged in the future, merely because some future imaging technologies can discriminate individual powder particles or <ludicrous mode on> individual molecules of media <ludicrous mode off>. (We could, in fact, achieve that today with other technologies than photography.)

I have to ask, too, how one would objectively establish an actual threshold minimum, a minimum such that anything less would be inadequate. I think that would be substantially subjective, so long as the record could resolve sufficiently for machine reading. So, the 1,000ppi hip-shot may be quite reasonable, without being extravagantly demanding.

[Steve Everist]

I guess this question will still grow in complexity but let's start at some point at least

Mission accomplished

[antonroland]

I guess this question will still grow in complexity but let's start at some point at least

Mission accomplished

Not that I meant to open a can of worms of this magnitude but some things just need to be questioned, I have just shot a few test shots that should settle a few of the arguments on here, maybe not all...

I have always been an advocate of over-engineering, not for any reason but for the sake of capturing the best possible latent for future use because tomorrow it will be gone forever and I will be committed to the work I did today...If my efforts today were mediocre, my comparisons tomorrow or next year will be limited in terms of quality and so will be my court testimony also.

Now to figure out how to post attachments on here...

[antonroland]

Hope this works, more to follow if it does...

I quickly made up a quick sample chart. The block measures 2" x 3" which gives me 1000 ppi on my work Pentax (* ist DL)

Woohoo, it did...more of the test coming up