Digital photography of crime scenes

[gerritvolckeryck]

If I may ask, why the interest in and move towards full frame, any specific reason?

Anton,

When comparing two camera's with the same number of pixels, the bigger sensor will have bigger pixels. It will therefor be more sensitive to light (higher ISO) and have a better signal/noise ratio. If you want to record more information (more pixels) you will automaticly be driven towards full frame camera's (or higher) : I don't know of any DX size 20 megapixel camera. Most of the time those hi-end camera's have a higher bitdepth (Nikon D3 : 14 bit), which gives you more "colour"information and more exposure latitude if you take advantage of the RAW-format (Jpeg : 8 bit).

When buying some high-end fullframe camera, our goal is not to take some snapshots at crime scenes (our old D70S suffice), but to have the best possible quality for marks photography (although we still may be better off with film based photography or scanning).

Oh and do you run any further software on your photos to validate, verify and or watermark the archive copies?

We don't. We use the RAW-format and embed it into a DNG, which - in our opinion - has better chances to survive the clash of the everchanging (Raw)file formats of every individual camera. We have to keep our records for 30 years and I'm sure, I won't be able to read my D70S Raw files then. I'm not sure about DNG, but I'll let you know in twenty years or so.

The information available in RAW (date, cameratype, serial number, lens, ...) is more than enough for our judicial system. Actually, everything we state as a fact in our police reports, is considered to be true. So, if we state that we took pictures of this and that and attach a copy of those pictures to our report, these are true and accurate photographs. A judge would have to rule that the police report is a fraud, before anyone could toss the pictures out.

All the best,

Gerrit

[gerritvolckeryck]

Personally I cannot see how jpeg cannot work in this day of Megapixel races. Also let's be fair and admit that, unless we buy serious printers capable of printing 16 bit image files, we ultimately will convert the original Raw (or tiff) to an 8 bit jpeg for printing and will therefore be forced into a lossy compression format in anyway. Further to this, printers capable of 16 bit print outputs (as far as I am aware) are built for large format A3+ prints and are ill-suited to printing jumbo and A4 prints as we would typically do in a lab environment.

Anton,

JPEG will do for snapshots, if you don't need any corrections being done.

You're correct in stating that printers can't reproduce all of the information in high quality images (the same goes for film and -paper, btw), but I see the prints we make as an illustration of what we have done, not as the result. We need high-quality images for on-screen comparisons and if we can show the judge what we think is relevant, we probably don't need the highest printquality. If minute detail is relevant, we can enlarge that portion. Judges don't like to use magnifying glasses

All the best,

Gerrit

[antonroland]

JPG is fine ..... provided you keep a log,

Neville, I would like to have a better understanding of your line of thoughts in this, not attacking at all...

What would a log do for jpeg that it cannot or won't do for RAW captures? I agree with justabout all you say but I am just curious to know the basis for your opinion.

Excellent written documentation in our line is admittedly ALWAYS avery good thing

My opinion rests on inter alia the following...

1) Yes, we work for Government / Semi Govt Services, they are budget dependant and their wheels turn slowly.

2) This is 2008 and not 2002, 10+ Mp D-SLR's are now considered entry-level and not 1 or 2 Mp.

3) In line with my point 1 above I understand why this is the case but the SWGIT and similar guidelines practically date from the stone age relative to todays technology.

4) I AGREE - RAW is by far the better option for all reasons but why victimize Jpeg as it is so much easier to use and all the valid arguments against it seem to disappear with time and better hardware...

[antonroland]

If I may ask, why the interest in and move towards full frame, any specific reason?

Anton,

When comparing two camera's with the same number of pixels, the bigger sensor will have bigger pixels. It will therefor be more sensitive to light (higher ISO) and have a better signal/noise ratio. If you want to record more information (more pixels) you will automaticly be driven towards full frame camera's (or higher) : I don't know of any DX size 20 megapixel camera. Most of the time those hi-end camera's have a higher bitdepth (Nikon D3 : 14 bit), which gives you more "colour"information and more exposure latitude if you take advantage of the RAW-format (Jpeg : 8 bit).

When buying some high-end fullframe camera, our goal is not to take some snapshots at crime scenes (our old D70S suffice), but to have the best possible quality for marks photography (although we still may be better off with film based photography or scanning).

Oh and do you run any further software on your photos to validate, verify and or watermark the archive copies?

We don't. We use the RAW-format and embed it into a DNG, which - in our opinion - has better chances to survive the clash of the everchanging (Raw)file formats of every individual camera. We have to keep our records for 30 years and I'm sure, I won't be able to read my D70S Raw files then. I'm not sure about DNG, but I'll let you know in twenty years or so.

The information available in RAW (date, cameratype, serial number, lens, ...) is more than enough for our judicial system. Actually, everything we state as a fact in our police reports, is considered to be true. So, if we state that we took pictures of this and that and attach a copy of those pictures to our report, these are true and accurate photographs. A judge would have to rule that the police report is a fraud, before anyone could toss the pictures out.

All the best,

Gerrit

Gerrit

Thanks for your response, now we are getting to the part I really wanted to explore. I absolutely and fully agree with the 2nd part of this post. Please do tell me, your court system...is it adversarial or inquisitory in the criminal sode of things?

Would like to compare notes re your law of evidence vs ours...

The idea I would like to bounce off you, however, is the idea of resolution in pixel density vs sensor size...

I agree, your D70 cameras are fine for general scene work, heck, I saw an A1 print of an untouched image off a D70s of a mate of mine and it is as sharp as you would want it at a 10-20cm viewing distance and so that should settle the issue right there and then. This should not be possible according to the book.

Besides, how many A1 prints have you taken to the courtroom? I know I have never...A3 on rare occasion but that was it.

As to the lab applications, consider the following: (I trust you will agree on the Nikon D90, D300, D700 and D3 being fair contenders in this comparison)

Noise is pretty much a factor of ambient light or the lack thereof in sufficient quantity in order for the individual photosites to render a high signal to noise ratio value, right?

Now if we agree that, in lab conditions, light is absolutely controllable, then noise should not be such a great factor, right?

Couple this with the fact that a CCD by virtue of it's design always gives a sharper off-the-sensor image than a CMOS sensor would, the only remaining factor of high resolution images should be pixel density.

This could potentially damage the chances of any full frame camera - both Canon and Nikon as they all have CMOS sensors...ahem, sorry, correction, all 4 Nikons quoted above and ALL Canons now have CMOS sensors so there goes that one...

Bact to pixel density,
D90 D300 D700 D3

Sensor photo detectors 12.9million 13.1 million 12.9 million 12.9 million
Click for help Sensor size 23.6 x 15.8 mm (3.72 cm²) 23.6 x 15.8 mm (3.72 cm²) 36 x 24 mm (8.64 cm²) 36 x 24 mm (8.64 cm²)

This gives us a pixel density of 3.3 MP/CM square for the D90 and D300...more than double that of the 1.4 MP/ CM square of the D700 and D3.

This also makes the D200 an excellent choice because it has a CCD sensor with 2.6 MP/cm²...pity it has been discontinued.

You must agree this is difficult coming from me as a great Canon aficionado but these appear to be the facts.

Would like to hear your opinion in this

BTW, This was taken from DP Review here:http://www.dpreview.com/reviews/compare ... 3&show=all

[antonroland]

Personally I cannot see how jpeg cannot work in this day of Megapixel races. Also let's be fair and admit that, unless we buy serious printers capable of printing 16 bit image files, we ultimately will convert the original Raw (or tiff) to an 8 bit jpeg for printing and will therefore be forced into a lossy compression format in anyway. Further to this, printers capable of 16 bit print outputs (as far as I am aware) are built for large format A3+ prints and are ill-suited to printing jumbo and A4 prints as we would typically do in a lab environment.

Anton,

JPEG will do for snapshots, if you don't need any corrections being done.

You're correct in stating that printers can't reproduce all of the information in high quality images (the same goes for film and -paper, btw), but I see the prints we make as an illustration of what we have done, not as the result. We need high-quality images for on-screen comparisons and if we can show the judge what we think is relevant, we probably don't need the highest printquality. If minute detail is relevant, we can enlarge that portion. Judges don't like to use magnifying glasses

All the best,

Gerrit

I will do some tests in the next week or so...oops, starting a stand-by week tomorrow, but anyway, the week after that then if we are spared and live to see the day...

If I understand you correctly, we should print an enlarged relevant portion of the image than the whole image? So far so good.

What does interest me is the possible merits of the limits in resolving power of (a) the camera and lens and (b) the output material (print and paper) and the relationship between the two.

Will do some tests and show the results in a while.

Also looking forward to your comments.

Anton.

[gerritvolckeryck]

Please do tell me, your court system...is it adversarial or inquisitory in the criminal sode of things?

It's inquisitory (we copied it from the French) and imho the system works well (we have our fair share of morons within the police and the justice system, but the chance of them having the majority working on a case is low)

The idea I would like to bounce off you, however, is the idea of resolution in pixel density vs sensor size...

I agree, your D70 cameras are fine for general scene work, heck, I saw an A1 print of an untouched image off a D70s of a mate of mine and it is as sharp as you would want it at a 10-20cm viewing distance and so that should settle the issue right there and then. This should not be possible according to the book.

Besides, how many A1 prints have you taken to the courtroom? I know I have never...A3 on rare occasion but that was it.

None, I hardly ever go to court. I write my report (A4), add the pictures (A4 maximum) and if I'm lucky, I read in the newspapers that X or Y has been convicted. In jury-cases (mostly murder) we use powerpoint presentations.

As to the lab applications, consider the following: (I trust you will agree on the Nikon D90, D300, D700 and D3 being fair contenders in this comparison)

Noise is pretty much a factor of ambient light or the lack thereof in sufficient quantity in order for the individual photosites to render a high signal to noise ratio value, right?

Now if we agree that, in lab conditions, light is absolutely controllable, then noise should not be such a great factor, right?

You've got a point, although low angle lighting and fluorescence, use of filters, ... really are unfavorable conditions.

Couple this with the fact that a CCD by virtue of it's design always gives a sharper off-the-sensor image than a CMOS sensor would, the only remaining factor of high resolution images should be pixel density.

This could potentially damage the chances of any full frame camera - both Canon and Nikon as they all have CMOS sensors...ahem, sorry, correction, all 4 Nikons quoted above and ALL Canons now have CMOS sensors so there goes that one...

True

Bact to pixel density,
D90 D300 D700 D3

Sensor photo detectors 12.9million 13.1 million 12.9 million 12.9 million
Click for help Sensor size 23.6 x 15.8 mm (3.72 cm²) 23.6 x 15.8 mm (3.72 cm²) 36 x 24 mm (8.64 cm²) 36 x 24 mm (8.64 cm²)

This gives us a pixel density of 3.3 MP/CM square for the D90 and D300...more than double that of the 1.4 MP/ CM square of the D700 and D3.

This also makes the D200 an excellent choice because it has a CCD sensor with 2.6 MP/cm²...pity it has been discontinued.

You must agree this is difficult coming from me as a great Canon aficionado but these appear to be the facts.

Would like to hear your opinion in this

BTW, This was taken from DP Review here:http://www.dpreview.com/reviews/compare ... 3&show=all

Comparing the 4 Nikons (all of them with the same number of pixels, give or take a few), the full frames would be the better choice (quality), due to the lower pixel density.

All the best,

Gerrit

[gerritvolckeryck]

If I understand you correctly, we should print an enlarged relevant portion of the image than the whole image? So far so good.

What does interest me is the possible merits of the limits in resolving power of (a) the camera and lens and (b) the output material (print and paper) and the relationship between the two.

I'm not really sure if I'm with you on this. I guess you're talking about interpolation when printing a digital image of, lets say a 300 ppi scan on a 360 dpi printer ? If so, I'm not the right partner for you. There are more knowledgeable people around.

As I said before, a printed image is a mere illustration of what I want to show to the judge.

Our unit will compare all kind of marks and test prints. The comparison is done on-screen. The quality of the images we want to work with, should be high enough to show the smallest relevant detail. In footwear marks this would be the shallamach pattern (around 1000 ppi, compared to live size, is OK for that).
We work with filmbased photographs (24x36 mm up to 8x10" negatives), an 8bit greyscale linescan camera (BVDA GLScan) with a native resolution of 1045,5 (!) ppi, photoscanners up to 4800 ppi (optical) and digital camera's (Nikon D70s - 6 Mpx) with macro lenses (a 1000 ppi image with a 2000x3000 pixel camera will be limited to 2 by 3 inch). We try to avoid interpolation as much as we can (1200 ppi scanning on a 4800 ppi scanner instead of 1000 ppi scanning, for instance) to get the best possible image for comparison.

Once the comparison has been done, I'm not ashamed to use even a laser printer to print the pictures, as long as they show what is relevant. A typical footwear mark comparison, for instance, will show seperate pictures of the mark, the test print and the outsole, an overlay and enlarged portions of one or more details of the mark, test print and outsole with an overview picture to indicate where the detail comes from. The overview pictures will not show enough detail for interpretation, but the enlarged portions will.

Hope this helps, although it's a bit off-topic (crime scene photography)

All the best,

Gerrit

[antonroland]

Gerrit yes thanks that did help a great deal as it tells me that you folks do your comparisons on screen and not on the print output as some of our guys are used to doing.

I have long since appreciated the merits of on-screen comparisons but I am afraid it is a bit of a challenge for some of my older (mentally) colleagues.

The part about staying away from interpolation also helped to clinch it.

Unfortunately in our adversarial criminal court system we occasionally have to show the detail in print output form to the presiding official / magistrate and many of them want to see for themselves. The law of precedent (Stare decisis) is against them in doing so but they do it all the same.

[Neville]

antonroland, sorry mate must fly, on days off will get back to you late next week

[antonroland]

Thank you very much to every single person who contributed on here. I really do appreciate it.

And to those who might still contribute, thank you for your inputs too.

Unfortunately, it is very often the situation that the more technical questions I pose here require a fairly short turnaround time and in this regard you have all been wonderfully helpful.

Regards,

Anton

[George Reis]

What a great thread - full of good information.

Unfortunately, with long threads that cover many sub-topics, it gets a bit overwhelming to read through the whole thing and understand exactly what opinions, advice and suggestions each person is making. I'd like to comment a bit on the general topic of Raw file capture vs. JPEG or TIFF file capture.

First, I would choose not to consider TIFF from digital cameras that offer it as a file format. Camera based TIFF images will be converted to 8 bits per channel, they will have a color profile and white balance assigned to them, the files will be larger than Raw files and contain less information, they will take longer to write to the memory card and to download, etc.

This leaves us with the option to choose JPEG only, Raw only, or Raw plus JPEG. I personally don't see any advantage to choosing both - it will take longer to write, it creates two files, and it is likely that the two will usually NOT match. Plus, one can easily extract the embedded JPEG thumbnail from a Raw file - a recently released free application to do so is here: http://rawworkflow.squarespace.com/inst ... w-utility/.

We all understand that JPEG images from a digital camera are converted to 8 bit per channel, that a color space is applied, a white balance setting is applied, other image processing may be applied (including sharpening), and that the image is compressed using a lossy compression algorithm. Thinking about all of this, it sounds like JPEG is to be avoided at all costs. But is it really? To me the question is one that involves the entire workflow, from original evidence to final output. We need to consider issues like the quality of the optics, the quality of the CCD, the pixel density, the signal to noise ratio, any lossy image compression, the image processing within the camera and in our workflow, the output of the file, etc. What a lot of variables in any workflow! And, I've only listed a few of the variables. Let me suggest a comparison here. Let's compare a file created from a digital camera that uses a poor quality 12 megapixel image but uses a Raw file format, and one that is from a high quality 8 megapixel point-and-shoot digital camera with an A quality chip and the Raw file format, and an 8 bit scan from a flatbed scanner that will scan a latent at a native resolution of 1200 PPI. Which will produce the best file? Which will produce a file adequate for comparison? I bet they all will provide great images.

My point is that it is easy to dismiss JPEG because it is a lossy file format - but there is a lot more to look at. Any agency that is producing images of adequate quality to see what they need to see is meeting the standard they need to meet. We all can go back to using 8 X 10 view cameras to photograph everything, or we can take some test images and make decisions as to what meets our needs, and what doesn't.

Personally, I use Raw for everything - but I work alone, I have plenty of storage, my volume is relatively low, and I understand the Raw workflow. If any of the above were different, I might choose to use a different workflow and format.

George

[Gerald Clough]

Let's face it. I have no doubt that the vast majority of crime scene digital photos in the U.S. are captured as in-camera JPEG. Most are done by officers equipped with P&S cameras, and many or most of the agencies don't own a DSLR or anything else capable of capturing RAW. It makes not the slightest difference in the utility or admissibility of the images. And in most sorts of scenes, if the proper photos are made, there won't be any instances where very fine detail is reasonably questioned on the issue of processing.

And most of those agencies neither provide facilities to process downloaded RAW image files nor have anyone trained in digital processing. If they left the processing of RAW images to the officers, the instances of over-sharpened and otherwise mishandled images would likely be high. It takes only a brief look around forums where folks who believe themselves to be accomplished amateurs post their work to see how poorly they manage post-processing. They would have produced a better image with the in-camera JPEG than failing around in an editor. Most decent P&S cameras do a reasonable and reasonably conservative job of in-camera JPEG production - modest sharpening, color balancing, slight "S" curve, and noise removal, and the images are ready to use just as they come off the memory card. You might lose a stop of headroom at the top of the range, but almost everything of interest is around the middle, if you expose anywhere near right and glance at the preview before going on to the next shot. And they should, as a matter of course, be choosing the highest quality JPEG option and leaving the image alone, rather than running it through multiple rounds of changes and JPEG saves, maybe unknowingly saving in a lower quality compression. But that's just basic proper equipment usage.

Really critical work where detail will have to be interpreted to determine just what it represents deserves critical capture with good equipment and good technique in RAW and lossless formats for processed images, just as that kind of work deserved expert attention with the right film camera equipment and materials. But with digital or film, the same old general crime scene problems are more important than worrying about file format. The fingerprints shot from two feet away with a hand-held P&S or straining at the bitter extreme of digital zoom. The shoe marks shot at odd angles, without a scale, and slightly out of focus. The burned out shots of bruises. The night shot of the wide scene from 100 feet away made with the teeny flash. And the prosecutor's bane - the disembodied close-up of the wound with no other shots to show where it was.

[David Grady]

Our office uses Nikon equipment (a D2X in lab, D2Hs for scene, and some D300s for all uses. All latent prints in lab are shot in RAW. Officers photographing scenes may shoot in JPEG (at fine setting); however, they must switch to RAW when recording evidence that is subject to further analysis (latent prints, tire treads, footwear impressions, bloodstain patterns, etc.).