Dear friends, welcome! The time has come to say farewell to this blog post series! This last post will look back at what we have learned in previous weeks and provide links to all of the 17 posts. We hope that you can use this as a basic knowledge base for future reference and perhaps even share it with your colleagues!
We started this blog series at the beginning of the year (here is the introductory article (Introducing “Video Evidence Pitfalls: Because You Don’t Know What You Don’t Know”). Five months have passed and, week after week, we’ve presented some of the major pitfalls investigators may stumble upon during their daily work with videos and images. Instead of listing them chronologically, we’ll group them by topic in the following sections.
We have seen that caution must be taken when relying on colors in surveillance videos. Beginning with images obtained by infrared (aka “night vision”) cameras, we detailed that you cannot draw any conclusion about the color of objects (Infrared: Don’t Trust the Colors of Most CCTV Footage at Night). Then, we’ve shown that even when images are captured with a standard “visible light”, the reliability of colors is often hindered by compression and other artifacts (Color Fidelity: It’s Hard to Objectively Evaluate Colors on Video).
Size and Shape of Objects
Together with colors, the size and shape of objects are properties that are immediately considered in an investigation. You should not blindly trust what you see, though. We’ve shown that a video presented with a wrong aspect ratio may severely change the shape of things (Aspect Ratio: Be Sure Your Image Is Not Stretched in Either Direction); then we’ve considered the distortion that lenses of surveillance cameras tend to introduce (Lens Distortion: Cameras May Change the Shape of Things) and, finally, we’ve discussed the difficulty of estimating the size of objects or even just comparing them (Perspective: Size Comparison May Be Tricky).
Playback Speed and Time Resolution
Spatial resolution is surely important, but time resolution is no less! We’ve thus shown the pitfalls of dealing with low-frame rate videos (Low Frame Rate: If You Don’t See It, It Doesn’t Mean It Didn’t Happen), and the possible consequences of playing a video at the wrong speed (Playback Speed: If the Video Plays Too Fast or Too Slow, It Can Affect the Interpretation of the Events).
Proper Acquisition of Evidence
Starting your work by dropping data is surely not a good idea. We’ve seen that, when extracting a video from the source DVR, taking screenshots is probably the worst thing you can do (Cell Phone Snaps: Bad Quality and Reliability), and that sometimes even using the proprietary player may be risky (Proprietary CCTV/DVR Players: Often Not Showing the Original Pixels). We then considered the problem of extracting the video from the player and showed that screen capture hides pitfalls as well (Screen Capture: It’s Not the Evidence, It’s a Video of the Evidence). Besides that, we’ve explained that the cause of most ‘seemingly corrupted’ videos is just multiplexing, something that you must be able to handle (Multiplexing: The Cause of Many Seemingly Corrupted Videos). Finally, we explained that, especially when you did not complete the acquisition personally, it’s good to check if you’ve been provided the original data, and if not, ask for it (Receiving Video Evidence: Usually It’s Not the Original).
Timestamps and Metadata
Pixels are exciting, but there’s more. The availability of timestamps can be a game-changer in many investigations, and we’ve dedicated a post to the possible pitfalls behind them (Timestamps: Not Always Showing the Right Time). Other kinds of metadata may also be available, e.g. telling about the source device model, acquisition place, etc. So, we described the difference between filesystem and embedded metadata, and presented the weaknesses of both, in a dedicated article (Metadata: So Useful But Not So Reliable).
Trusting What You (Believe) to See
The leitmotiv of the series was: “always be prudent”. That holds also when dealing with the interpretation of results. We’ve thus presented some typical compression artifacts and their potentially dramatic consequences on the appearance of objects (Compression Artifacts: Hiding or Adding Details to the Scene). Then, we’ve discussed the problem of cognitive bias, which ‘pollutes’ your interpretation with extraneous information, making you act illogically without even realizing it (Cognitive Bias: Steering Conclusions Irrationally).
And that’s all folks!
Needless to say, there are countless more pitfalls out there, but hopefully what we covered was enough to raise your awareness. Although everyone deals with videos and images every day, their use for investigative and forensic purposes is a completely different matter: the objective is different, the technological background is different, and most importantly, the possible implications of a wrong decision are extremely different. When in doubt, look for an expert and do some research – and of course, don’t forget to take a look at our blog 😉