The Tyre Nichols Videos Show How Police Edit “Raw” Evidence

In January 2023, Memphis officials released four videos showing police officers stopping, pursuing and beating Tyre Nichols, who died after the encounter. Three files came from cameras worn by officers. The other, labeled Video 2, came from a fixed SkyCop surveillance camera mounted above the second scene.

Video 2 is the file to keep in view. Its elevated angle showed officers surrounding Nichols without inheriting any one officer’s movement, obstruction or decision to look elsewhere. It also made the release’s central contradiction visible: the camera that was not attached to a police officer provided the clearest view of what police officers did.

The four files were horrifying evidence. They were also a publication assembled by the government implicated in that evidence, released after officials and prosecutors had reviewed it, after Nichols’s family and their attorneys had seen it, and after criminal charges had been announced. None of that made the images false. It meant the public received a selected set, in a selected order, on an institutional timetable.

“Raw footage” is the phrase agencies and newsrooms often use for material without narration or a broadcast edit. Raw is doing too much work. A camera records from a position, starts and stops according to its settings and the wearer’s conduct, then produces files that enter police custody. By the time one reaches a city website, lawyers, records officers, investigators, commanders and communications staff may all have influenced what you can see.

The release begins before the camera turns on

A body camera is an evidence-collection device governed by agency policy. Officers usually activate it during specified encounters, though the exact rules vary, and many cameras keep a short buffer, meaning the device continuously retains a temporary slice of video before activation. That buffered segment may lack audio. It may show the approach to an encounter, or only the instant before an officer presses the button.

Failures at this stage become holes no editor can repair. A camera may be activated late, pointed into a uniform, knocked loose or muted. An officer can move beyond its field of view. Battery, storage and upload problems exist too, but policy compliance matters more than the gadget advertisements suggest.

The camera sees where the officer’s torso faces, not what the officer notices, knows or does with hands outside the frame.

Then the recordings enter an evidence-management system, usually supplied by a private vendor and controlled through agency permissions. Metadata, the information attached to a file about matters such as time, device and user, helps staff find recordings associated with an incident. Search terms establish the first editorial boundary. Staff may look for a case number, named officers or a defined time window.

If that window ends when the handcuffs go on, the public may never see the conversation afterward, the medical response or officers comparing accounts.

The Memphis release showed why source variety matters. Video 2 did not depend on an officer activating a chest camera at the right moment, and its distance exposed conduct that became fragmented across the body-worn views. Yet even that surveillance file was not omniscient. It was aimed, framed and eventually selected from a larger municipal camera system.

The useful angle can still be an institutional angle.

Selection is where the story acquires walls

Once an agency decides to release footage, someone must determine which files count as responsive and which portions can legally or practically be published. State public-records laws differ. Prosecutors may argue that disclosure could affect an active investigation or trial. Departments also invoke privacy protections involving minors, medical treatment, sexual violence, private homes, uninvolved people and sensitive personal information.

Those concerns are real. They also give institutions broad control over scope and timing, especially when the records concern the institution’s own possible misconduct. Privacy can require redaction, the obscuring or removal of material before disclosure, without explaining why the public receives ten minutes rather than two hours. An investigative exemption can protect a prosecution while shielding the agency from immediate scrutiny.

The same delay can do both.

Staff may blur faces, license plates, screens, documents and exposed bodies. They may silence names, addresses, radio traffic or conversations judged unrelated. Each intervention changes how a viewer understands the scene. Muting a name can be necessary.

Muting the seconds around it can erase tone or an instruction. A blur that follows a civilian through the frame can conceal identity while also hiding the placement of an officer’s hands.

Selection exerts even more power than blurring because missing footage leaves no visible scar. A black box or tone announces that something was removed. An unpublished file announces nothing.

In the Nichols case, the four released videos quickly displaced the Memphis Police Department’s initial account, which had described a traffic stop involving alleged reckless driving, a confrontation and Nichols fleeing. The department’s chief later said an internal review had not found evidence substantiating the reckless-driving allegation. Video 2 mattered here because it widened the public’s view of the second scene, but no released camera could independently reconstruct the claimed driving that preceded the stop.

That absence did not vindicate the initial account. It revealed an asymmetry. Police could put an allegation into the first official narrative without publicly showing evidence for it, while Nichols could not supply his own institutional archive. The release corrected part of the story after the state had already enjoyed the first draft.

Sequence turns files into causation

Chronology looks neutral. It rarely is.

An agency can begin with dispatch audio, an officer’s account, surveillance footage or the first body camera to activate. It can place a suspect’s movement before an officer’s use of force, encouraging viewers to read the second as a response to the first, even when another camera shows simultaneous conduct that complicates that chain. Slow motion, freeze frames and on-screen labels narrow attention further.

Many departments now publish a “critical incident video,” an agency-produced package that combines selected recordings with narration, graphics, transcripts and identifying information. These presentations can help viewers follow a chaotic event. They also borrow the authority of documentary form while letting the subject of scrutiny choose the opening context, the vocabulary and the point where the credits roll.

Even releases without narration have an interface. File names establish order. Thumbnail images determine the first visual impression. Separate players make simultaneous recordings difficult to compare.

A transcript may turn uncertain audio into confident language, and viewers often remember the words on screen more clearly than the noise underneath them. News outlets then compress the package again for television, social feeds and article embeds.

Memphis’s Video 2 resisted some of this control because its wide, sustained view was harder to reconcile with euphemistic description. Still, audiences had to synchronize it mentally with the wearable-camera files. A genuinely transparent release would make that comparison easier, not leave each viewer to operate four video players and guess at overlapping moments.

The incentive sits behind the export button

Police departments have several audiences at once: prosecutors protecting a case, officers and unions protecting employment interests, city attorneys anticipating civil litigation, families demanding answers, reporters working under deadline and residents judging whether an agency can police itself. The release has to move through that conflict. Communications strategy arrives wearing the badge of evidentiary caution.

Timing matters. Officials can publish after charging decisions, before a weekend, during intense public pressure or only after a records lawsuit. They can preview graphic content in language that prepares viewers to see danger, resistance or disorder. They can release enough to satisfy a demand for visibility while withholding surrounding hours under an investigative or privacy rationale.

The money is less cinematic but still present. Cities pay for cameras, storage, vendor software, records staff, legal review and frame-by-frame redaction. Broad disclosure takes labor, and agencies can cite that burden when requests seek many files. Yet departments also spend public money producing polished briefings and defending denials in court.

Scarcity is partly a budget choice. Narrative management tends to find resources.

The cleaner alternative is not indiscriminate dumping. Agencies should publish a manifest identifying every known recording, its source, duration and release status; preserve original files and metadata; explain each omission; provide synchronized copies alongside minimally edited files; and separate factual navigation from agency argument. Independent records custodians or oversight bodies can reduce the obvious conflict, though they need authority, staffing and access to the evidence system.

Video 2 remains the useful object because it demonstrates both the value and the limit of disclosure. It showed conduct the body cameras divided into unstable fragments. It reached the public only because officials selected it, labeled it and placed it among three other files. Transparency did not remove institutional power.

It gave that power an export menu.

Questions people ask

Who chooses which police body-camera footage gets released?

The police agency usually controls the recordings, while records staff, investigators, prosecutors, city attorneys and command personnel may shape the decision. State law, court orders and department policy set boundaries, but those rules often leave substantial discretion over search windows, exemptions, redactions and timing.

Is released body-camera footage really raw?

Usually not in the ordinary meaning of untouched. Even a file without narration has been retrieved from an evidence system, selected from other recordings, reviewed and sometimes redacted or transcoded into a publishable format. Camera position, activation and missing files have already limited the scene before an editor opens it.

How can the public tell whether footage is missing?

A complete release should include an inventory of known cameras, file durations, activation times and reasons for every withheld or shortened recording. Without that manifest, viewers can identify visible cuts and muted passages, but they cannot reliably detect a camera file that the agency never disclosed.

Why did the fixed camera matter in the Tyre Nichols case?

The SkyCop file labeled Video 2 supplied a wider, more stable view of the second scene than the officers’ moving chest cameras. It did not show everything, including the alleged driving before the stop, but it let viewers assess the beating without being confined to one officer’s body and line of sight.