Steve Everist, King County Sheriff s Office, WA William Fluit, Sioux Falls Police Department, SD Forensic Photography III, Michael Brooks, January 29, 2007 Footwear & Tire Tread Photography A comparison of digital resolution vs. 35mm film Digital cameras and the use of their images in criminal investigations and prosecutions have become more common as technological advancements have increased resolution and accuracy while bringing down the cost of equipment. Proponents of digital photography believe that this medium has improved to the point of equaling or surpassing the image quality of traditional 35mm photography. Yet there are those that are still critical of its use in law enforcement especially for the purpose of capturing examination quality images. Due to the size of the evidence in forensic footwear and tire tread examination, the focus regarding digital images relates primarily to the resolution of the images, although there still remain those critical of its use in general. For the purposes of this paper, the question that we are looking into is whether or not the current high-end digital SLR cameras are capable of resolving the same level of detail as compared to 35mm Black and White film. Many police agencies throughout the world have recognized that the image quality of digital photographs is sufficient for use in general crime scene documentation. On the other hand, digital photography for examination quality photographic documentation has not been adopted as readily. The field of latent fingerprint examination has begun to adopt digital photography for the photographic capture of latent print evidence, and minimum resolution standards are in place. But the field of footwear and tire tread examination has been more reluctant to begin making the change. The common feeling is that the resolution of digital images is inadequate and that the minute detail needed to come to a conclusion may not be reproduced as it has with traditional film photography. It is the authors contention that although the high end digital SLR imaging systems are now capable of resolutions that can be used for latent print comparison; they have not reached a level where they can resolve the same level of detail as 35mm black and white films. Due to the much larger size of footwear and tire tread evidence in comparison to latent print evidence, 1:1 calibrated resolutions are often less than half the 1000ppi that the latent print field has determined to be their minimum standard. 1 Yet 35mm black and white films have been recognized as having enough resolution for footwear and tire tread comparison. 2 Although the calibrated resolution is less than that used for latent prints, the next question is whether or not current digital SLR systems provide enough detail for footwear and tire tread examiners to come to a conclusion. Additionally, it should be explored as to whether or not using higher resolution combinations of images will be acceptable when resolutions of these images do calibrate to 1000ppi and greater (or to whatever standard the field of footwear and tire tread examination sets as their minimum). In the past ten years digital imaging has made in-roads into modern society. First accepted by the general public and hobbyists, digital photography s progress into widespread use in law enforcement was much slower to evolve. Reasons for this include the relatively low resolution of the images, the quality of available cameras, the 1
acceptance in the courtroom, and the limitations and cost of supporting computer systems and software. Resistance to change has also played a major role. This resistance has come from administrators and technicians alike. Silver-halide photography had become so ingrained in law enforcement that it was much easier to maintain the status quo instead of having to come to grips with a new technology, its validity, and ultimately its ability to stand up in court. The dramatic improvement of the quality of the equipment has made acceptance of digital imaging more commonplace among law enforcement agencies and in the courts. Digital point-and-shoot cameras are now commonly being used by agencies for general documentation of crime scenes and evidence, and the use of digital images in this manner has become established in court. But before many agencies will begin the use of digital photography for the examination of evidence, quality assurance standards are desired. Organizations such as SWGIT, SWGTREAD, and the FBI Laboratory have established and published protocols, but the pace of technology advancement often lead to previous studies becoming quickly irrelevant based on what is currently available only a year or two later. For example, the FBI s Handbook of Forensic Services 2003 states Images should be taken using a 35mm or medium-format camera. Low-cost digital cameras do not provide sufficient image detail for examination-quality photographs. 3 It goes on to suggest using 100 ISO film but 200 or 400 ISO film can be used. The Handbook does not define low-cost digital and it s unknown if any recent efforts have been undertaken to determine if digital cameras are acceptable, given their current level of quality. For example, when the Handbook was released in 2003, the Canon EOS-1D had a 4.2 million pixel CCD sensor. In mid 2004 Canon released the EOS-1D Mark II with a CMOS sensor producing 8.2 million. Today, the Canon EOS-1Ds Mark II, released in late 2005, produces 16.7 million. Since 2003, the resolution has doubled twice from the same product. Although the introduction of the CMOS sensor has played a large part in the resolution increases, technology within the sensors are also playing a big part. Many of the currently available high end digital SLR cameras are producing image resolutions over 10 million, with the Canon EOS-1Ds Mark II producing 16.7 million. Additionally, the sensor sizes are becoming closer to 35mm film s 36X24mm size. 2
The chart below shows a comparison of three high-end digital cameras, currently available, from Nikon, Fuji, and Canon. Make Model Sensor Size Resolution File Type ISO Range CCD 23.7X15.6mm 6.1 million RAW, JPEG 200-1600 3008x2000 px D200 CCD 23.6X15.8mm 3872x2592 px 10.92 million RAW, JPEG 100-1600 Fuji D2Xs FinePix S5 Pro FinePix S3 Pro (UVIR) FinePix S9100 Canon EOS-1Ds Mark II EOS-5D EOS 400D CMOS 23.7X15.7mm 4288x2848 px CCD CCD 23.0X15.5mm 4256x2848 px 23.0X15.5mm 4256x2848 px CCD 1/1.6 inch 3488x2616 px CMOS 36X24mm 4992x3328 px CMOS 35.8X23.9 4368x2912 px CMOS 22.2X14.8mm 3888x2592 px 12.84 million 12.34 million 12.34 million 9.0 million 16.7 million 12.8 million 10.1 million RAW, JPEG TIFF, 100-800 RAW, JPEG 100-3200 RAW, JPEG 100-1600 RAW, JPEG 80-1600 RAW, JPEG 50-3200 RAW, JPEG 50-3200 RAW, JPEG 100-1600 Digital For the purposes of this comparison, the following cameras were used: Nikon D2H (UVIR) Canon EOS-1Ds Mark II Film Nikon N90s Miscellaneous information: Kodak TMax 100 Black and White 35mm film was used in the Nikon N90s. Kodak TMax negatives were scanned using a Nikon CoolScan 5000 at 4000ppi, 1:1 The Fuji SE3 is the UVIR model with a Peca Hot Mirror Filter The Canon EOS-1Ds was only used for the test subject photos as a 100-400mm lens was the only one available. For the remaining cameras, a Nikkor 60mm macro lens was used. The Nikon D2H was available for comparison purposes but was only used on the test subject. ISO used was either 100 or 200 depending on the camera, except for the Canon which was set at 640. 3
The test was a series of scenarios set up to record the same conditions, related to footwear evidence photography, for various digital cameras. The goal was to eliminate as many as possible to allow for meaningful conclusions to be drawn from the image quality. Due to limited time and equipment, a more thorough version of this comparison is encouraged where a larger variety of cameras with matching lenses can assembled and more of the variables reduced. The test subject was new, never-worn boot with a Vibram sole. Various defects were introduced into parts of the sole to make it unique. The first series of photos were taken to document the sole and its defects. The results are shown below, using 2 x2 sections of the sole, taken from an image of the sole that included the entire surface from heel to toe, filling the frame. Test Subject Standards 1 X 1 Sample Canon 1Ds Nikon D2H Nikon N90s 4
Test Subject Standards 1 X1 Sample 2X Enlargement Canon 1Ds Nikon D2H Nikon N90s 5
Test Subject Standards 1 X1 Sample 3X Enlargement Canon 1Ds Nikon D2H Nikon N90s The second series of photos consisted of documenting footwear impressions in two different types of soil. For this project a clay-based soil and a sandy soil were used. The results can be seen below using calibrated 6 x3 samples taken from an image of the impression that included the entire surface that was visible, filling the frame. 6
Hard-packed Sand 4 X 2 Sample Kodak TMax 100 35mm 7
Clay 4 X 2 Sample Kodak TMax 100 35mm The third series of photos were taken of dental stone casts which were cast from the impressions photographed in the second series. After were released from the impressions, they were cleaned of debris to reveal details in the casting. 8
Cast #1 Hard-packed Sand 4 X 2 Sample Kodak TMax 100 35mm 9
Cast #2 Clay 4 X 2 Sample Image Not Available Kodak TMax 100 35mm From examining the images in Photoshop CS, the results appeared to be inconclusive as to the image resolution differences. However looking directly at the negatives produced from the TMax 100 film, they appeared to be of higher quality than what was reproduced once scanned. Since the film was eventually digitized for the purposes of this paper, the actual resolution was limited by the mechanical limits of the scanner. Comparisons were not made with lower scanned resolutions to compare the point at which the scanner was no longer showing a difference in the end result based on the scanning resolution. For the purposes of the primary question presented for this paper as to whether or not the current high-end digital SLR cameras are capable of resolving the same level of detail as compared to 35mm Black and White film, the answer appears to be that it does not. However, once the negatives were digitized the differences appear to be inconclusive. This is most likely a product of the limits of the scanning device. But beyond answering that question, the next question is whether or not the current high-end 10
digital SLR cameras can produce enough resolution for footwear/tire tread examiners to draw their conclusions. It appears as if the resolution currently produced does provide detail that could be used for footwear/tire tread examination. As we did not have a footwear/tire tread examiner available to assist in conducting this study, more research would be encouraged. Additionally, the images presented above were sections of footwear where the image of the entire surface was taken. Higher resolutions would result from photographing smaller sections of the subject and impressions with overlapping photos to go along with the overall shots that would provide size information. Although a standard has not been set as to what a minimum resolution should be for this examination, to incorporate the minimum used for latent prints of 1000ppi, a mathematical formulation can be used to determine how much area can be photographed by a specific camera to resolve 1000ppi. The chart below presents this information. Camera Image Size (in ) Area to Resolve 1000ppi 3008x2000 ~ 3 x2 Nikon D200 3872x2592 ~ 3.87 x2.59 Nikon D2Xs 4288x2848 ~ 4.28 x2.84 Fuji FinePix S5 Pro 4256x2848 ~ 4.25 x2.84 Fuji FinePix S3 Pro (UVIR) 4256x2848 ~ 4.25 x2.84 Fuji FinePix S9100 3488x2616 ~ 3.48 x2.61 Canon EOS-1Ds Mark II 4992x3328 ~ 4.99 x3.32 Canon EOS-5D 4368x2912 ~ 4.36 x2.91 Canon EOS 400D 3888x2592 ~ 3.88 x2.59 This chart gives approximate sizes to resolve to 1000ppi, although in order to include a ruler for calibration and to be able to overlap with corresponding areas, the actual area captured should be somewhat smaller. From these physical dimensions, it appears that the current high-end digital cameras do have the capability of resolving 1000ppi with samples large enough that they could be used for examination. Furthermore, it is possible that resolutions below 1000ppi may be acceptable for this sort of comparison. Additionally, we were concerning ourselves only with digital SLR cameras that are commonly being used by different law enforcement agencies. We did not include any comparison photographs from the digital medium format cameras currently available, as we did not have access to them. But to provide reference as to what is available, we have provided the following specification chart for comparison with those of the digital SLR cameras listed above. Camera Model Sensor Size Resolution Price Hasselblad H3D-22 4080x5440 22 million $26,995 H3D-39 5412x7212 39 million $31,995 Mamiya ZD Not Available 21.5 million * 11
*Pricing information for the Mamiya ZD has not been finalized although some information suggests that for the body-only, the cost is around $12,000. From this information it can be seen that an area of over 5 x7 can be captured using the 39 million pixel Hasselblad H3D-39. The prohibiting factor of these cameras is the cost. At this point there are not many law enforcement agencies able to spend the amount of money needed to put these cameras into service. Additionally, the file size created by cameras with this resolution requires additional consideration for the purpose of storage. The 39 million-pixel H3D creates a lossless compressed RAW file of 50 megabytes and a TIFF file of 117 megabytes. Over the past decade, resolution has been increasing at a rate approximately every 2-3 years. With similar advancements, the question as to whether or not digital images resolve acceptable resolution in comparison to traditional film for the purpose of comparative analysis of footwear and tire tread impressions could become a moot point. However many of the advances have been fueled by the average consumer s demand for increased resolution and competition between the various manufacturers. As the average consumer decides that resolution is no longer an issue for their needs, and may in fact be a detriment to their ability to store the images, manufacturers may begin to scale back their increase in resolution and focus more on the features that they provide in the cameras that seem to be in the highest demand. This could result in a slowing in resolution advances which could render the question of resolution irrelevant. If this is to occur, it would be the demand of the professional photographers and artists that need the higher resolutions in order to assist in bringing the high resolution cameras into a reasonable price range. At this point, the demand of law enforcement for a product that provides for their needs often falls on deaf ears as their share of the market is very small and potentially unprofitable for the camera manufacturers. With this in mind, the future still holds great potential for digital imaging to be used in all aspects of law enforcement, although film will likely not fade away completely. References: 1 SWGFAST. Friction Ridge Digital Imaging Guidelines, Sec 1.3.1 Version 1.0 8/08/01 2 SWIGIT. Section 10 General Guidelines for Photographing Footwear Impressions Version 1.1 2003.06.05 3 FBI. Handbook of Forensic Services 2003 12