The merits of photographing a latent at 1000 ppi

[Neville]

Hi Gerald
Thank you, at last it has been explained in technical terms about the problem of 1:1 talk. I have had this, at times, rather heated discussion at my last office about 1:1. It took me a while to work out what the staff were meaning because I kept asking at what point is it going to be 1:1 is it on the sensor, on the memory card, on the hard drive, on the screen, or when it is printed. Unfortunately there are some fingerprint offices overseas where comparisons are still only done 1:1, when they come to NZ they still expect to work 1:1, ie the printed scene image is the same size as the impression recorded on the fingerprint form. To achieve this using normal computer equipment is a bit of a challenge when photographs are taken by various different cameras in the field at different distances.

Do you people resize every photo taken at scenes of crime back in the lab before the comparison stage? Is every photo printed at 1:1? How do you achieve this with heavy work loads?

Sorry if this is getting off the subject a bit.

[Gerald Clough]

That gets into issues in printer resolution that I haven't messed with enough to understand much, except that unqualified printer dpi is even less a reliable a measure of resolution than camera ppi. But my feeling is that printing an impression image to 1:1 with the original target would be a pretty poor use of the technology. For any digital image file, there is surely an optimum print size for viewing at a specific distance. (Viewing distance is always an issue in image printing.) If we assume a printer capable of doing it for the desired viewing distance, that point would be where the print output makes maximum use of the file data without resampling or interpolation. In other words, nothing lost, nothing created that wasn't explicit in the file. There really is no "resize" in digital. Digital image files don't have "size." As a computer scientist (whose credentials are sufficiently ancient as to make that a shaky claim at best), I would view a digital image file as simply a single number. It's just a string of binary bits. One might as well ask, "What are the physical dimensions of the number 8.34212?" There is only the physical size of the font character string "8.34212." Printing it ten feet high doesn't make it a bit more precise, or a bit less.

If we talk about printing an impression of, say, a single digit, 1:1 with the original impression, we're not assuming a viewing distance of about two inches. We're assuming viewing at two inches through a loupe. Well, maybe that IS pretty much like viewing at two inches, just with a lens to allow us to focus. That's with direct inspection with a loupe. Dealing with that case first, we would calculate a variety of alternate print sizes that would present the same view from other distances. If the 1:1 print and the larger print for viewing at 12 inches both use the same dot pitch, the larger print would seem to have the greater spatial resolution. In other words, the printer may be capable of producing the larger image without interpolation at its greatest available dot density. If that's true, then the 1:1 print sampled down to represent the image with the limited number of dots available according to the original target dimensions. I don't know yet how to make the calculation. I seem to recall that dpi does not equate to ppi, and printer dpi specifications tend to represent a third or a quarter of what they seem to imply in terms of image ppi. So it seems to me that printing 1:1 throws away image data. And using the magnifier or optical comparitor just takes a closer look at the diminished output. Go back to the number analogy. If I limit my expression to one decimal, I display 8.34212 as "8.3." I lost four places of precision.

I begin with an impression. I capture it with sufficient spatial resolution to preserve anything that matters. I want to view it in such a way that I don't lose anything that matters, and I don't want to alter anything in any way that matters. If I can reproduce it at 1:1 and not lose anything that matters, fine, even if I do lose something. If I can reproduce it very large and don't create something that alters the information that matters, fine. Is one better than the other? I could be missing a concept, but I can't see any advantage at all in working with 1:1 printer output, unless one cannot work well without the stimulus of neck and eye strain. I suppose we tend to be kind of comfortable with 1:1 because most of what we subject to comparison by inspection falls within size limits that are manageable at "life-size." Shoe marks, tire marks. But firearms examiners surely don't work at 1:1.

Perhaps someone with a better grasp of printer resolution can explain, for a typical printer technology, how to determine, for a given file representing n by m pixels, what reproduction ratio (what printed image size) most exactly represents the file contents without lose or interpolation.

[Neville]

Yes

I go for less stimulus.

I have seen enough missed id's or the writing off as not suitable to what the enlarged photos.

[antonroland]

Personally I think that printing recorded latents life-size (1:1) is an exercise in futility and wasteful expenditure.

Officially I have to but working on changing that.

[GEilers]

I would suggest that it might benefit you in your quest to not print latents for comparison to read our article published in the August 2008 issue of Evidence Technology Magazine, "Getting out of the Loupe". The article can be viewed online at the magazine website if you do not have a paper copy. The paperless process for comparison has been working quite well.

Greg Eilers

[antonroland]

Thanks for that Greg

Will have a look and see if I can access it.