Live View at Scale
Multi-camera live viewing with PTZ control from the platform, grouped by map, location, or venue rather than by which recorder holds the stream.
Video Management Built for a City, Not a Building
IREX replaces the rack of recorders with a private cloud you control. Live view, instant playback, and retention work the same whether the estate is one site or a hundred thousand cameras, and every access is scoped by role and logged.
Why Retire the VMS
One System Instead of a Recorder per Site
Traditional Windows-server video management systems were designed for a building. They are not cloud-native, they cannot operate at big-data scale, and each site becomes its own island with its own storage, its own failure modes, and its own upgrade schedule.
IREX stores raw video in a distributed data layer on infrastructure you own, with triple replication and automatic failover so an individual server or drive failing is an operational event rather than a lost case. Every component is hot-swappable and can be replaced under load without operator-visible downtime, and a version change is a rolling restart rather than a maintenance window.
Because the same platform runs the analytics, video management stops being a separate product to integrate. Live view, playback, search, alerts, and the audit trail are one system with one permission model.
Capability
What the Operator Gets
Instant Playback
Jump straight from a map pin, an event, or a search result into the archive. Snapshot previews open without loading the whole stream.
Forensic Review
A browser-based player for frame-level review, with reverse playback, frame stepping, digital zoom, and the detection annotations drawn back over the video.
Retention You Size
Retention is an independent scaling dimension. 30-day instant search is proven at 100,000+ cameras; the architectural ceiling is a one-year period.
Permission-Driven Access
Six roles from Viewer to System Manager, over a hierarchical catalog of groups, each scoped to specific cameras, watchlists, locations, and monitors.
A Non-Erasable Trail
Views, playbacks, searches, and exports are logged to an append-only record that supervisors and oversight bodies can read.
Camera Integrity
An estate is only as good as the cameras still working. TamperTrack alarms on signal loss, a low frame rate, a shifted view, an obstructed, dirty or defocused lens, and a scene too bright or too dark, per camera, and a faulty camera carries a Video issues status on the Cameras page.
See Camera Tamper DetectionForensic Review
The Player an Investigator Works In
Video management is judged on one question: how long it takes a person to get from a report to the frame that answers it. These are the two screens where that happens.
Find the Moment without Scrubbing the Archive
Click a camera and the player opens on the live stream. Under it, the archive timeline shows in blue exactly which intervals hold recorded video, and clicking a point starts playback there, so nobody hunts for the end of a gap. The events the analytics found are on the timeline too, which turns the density of marks into the answer to where to look; the events panel on the right lists the same events as cards, and clicking one plays it. The scale control fits ten minutes, an hour, or a longer window into the width of the screen, the calendar jumps to a date and time, and LIVE returns to the present. Since version 4.40 the player also reaches archive that never reached the server: where cameras sit behind a registered ONVIF device with archive access enabled, the toolbar’s Video source option switches between Server and Camera archive, so recordings held on the camera or on an existing NVR play back in the same window. Close in on a frame with reverse playback through the segment already buffered, which is the fast way to find where something started, then step a frame back or forward to pick the cleanest view of a plate or a face. Digital zoom works on live and recorded video, PTZ controls appear for a PTZ camera, playback speed is adjustable in both directions, annotations draw the detected bounding boxes and the lines and boundaries the modules were given back over the video, and Go to Events opens the Events screen already filtered to this camera. Where CarTrack Pro, MotionTrack Pro, or ObjectTrack Pro is saving tracks on the camera, a Search in archive tab lets an operator draw a line or an area on the picture itself and ask which crossings, movements, or abandoned items happened in it.
Docs: Explore Your Workspace
Save the Clip, Share the Link, Log the Download
Save to Gallery in the player options captures a segment: set the two time bounds, which cannot sit closer than ten minutes to the present, and the save runs in the background with a progress indicator while the operator carries on elsewhere. The clip then lives in the platform’s own storage rather than on somebody’s desktop, listed with its channel, the period it covers, its duration, and a download status, so a case file cannot quietly go missing or corrupt. The row menu offers Download to a local file, Link sharing to hand the clip to another user of the same instance, rename, and delete. Two lighter exports sit in the same menu: Download current frame saves the full frame as a JPEG, and Copy video link produces a short link pinned to the exact playback time, so a colleague opens the same second rather than a description of it. Event cards carry their own Download snapshot and Copy link. Downloading a clip from the Gallery is classed as an ethically sensitive action, so it needs a Case ID and is recorded in full: the user and IP address, the subject, the reference document, and the timestamp.
See Ethical AI & Case ID
Live Monitoring
A Video Wall for Every Desk
Give Each Desk the Video Wall It Actually Needs
A traffic desk, a transit desk, and a night shift do not want the same twelve cameras on the screen, and in a recorder-per-site estate they end up arguing over one wall. In IREX a layout is created from a template that allocates the screen space, named, filled with cameras, and saved. Layouts are user-specific rather than room-specific, and access to one is granted to the user groups that should have it, both when the layout is created and afterwards on the Layouts tab of the User Groups and Resource Groups pages. One layout carries up to 25 video streams. Two things are worth knowing before sizing a deployment around it. Layout mode is for watching many cameras at once, so video analytics is disabled while a layout is active: zoom, switching cameras, and viewing streams stay available, the analytics features do not, and an operator returns to a single camera to work an event. And a layout plays the camera’s low-resolution secondary stream, which is why the camera configuration asks for an optional low-resolution secondary stream alongside the 1920 × 1080 or 1920 × 1440 primary. That is the difference between a wall that stays smooth on a citywide estate and one that does not.
See the Global Map & Floor Plans
Accountability
Every Playback Is an Action with a Reason
A recorder logs that a file was opened. On this platform the reason is captured at the moment of the action, by the person who has it, and it stays attached to the action permanently.
-
A Case ID Before the Video Opens
Viewing live video in the media player, and loading archived video into it, are on the platform’s list of Case ID-gated actions, beside event search, person and vehicle database management, alarm-monitor management, and Gallery management. The operator enters the reference to the lawful grounds in a standard format, [object] - [reason] - [document number], and the action does not run without it. Case ID is a platform-level mandate rather than a setting an administrator can switch off, and it follows the action rather than the interface: the same request made through the API or through Ask IREX meets the same gate. Ask IREX is shipping in beta on selected instances.
-
The Logbook Records the Interval, Not Just the Click
For Live video loaded and Archived video loaded entries the Logbook stores the start and the end of the interval that was played, writing the entry when playback ends or every three hours through a long session. A supervisor can then filter by played-back video date and time and get every session that overlaps a window: ask for 10:00 to 10:30 and a playback that ran 09:50 to 10:40 comes back with it. Results are exported as a CSV report on demand, and searches of the Logbook are themselves logged, so reviewing the record leaves a record.
Docs: View the Logbook -
Video from the Field Becomes Searchable Video
Footage that never came from a fixed camera still belongs in the case. Upload the file through Gallery > Uploads with its recording date and time, the user groups allowed to see it, and its place on the map, then run Analyze video and the same analytics modules that run on a live camera run on the file. Its events land on the Events screen beside camera events and answer the standard filters, so a phone clip or a body-camera export becomes searchable evidence instead of an attachment somebody has to remember. A second pass with a different module adds its events to the same set, which is how one file yields both the abandoned bag and the person who left it. Uploading and analyzing external video is Gallery management, so it is Case ID-gated too.
See Video Search & Investigations
Compatibility
Your Cameras, Your Codecs
- Any camera that supports ONVIF Profile S, or streams RTSP with H.264 or H.265 at a valid RTSP URL on a static IP, connects.
- Reusing the existing fleet is the standard deployment model, not an exception. Resolution requirements are set per module, so a survey checks each camera against the module it will run.
- One camera runs several analytics modules at the same time, for example face recognition, object tracking and traffic monitoring on the same stream, with no extra hardware and no duplicated video stream. A second module is enabled from that camera’s module list, and the platform lists what is already running before the change takes effect.
- Cameras connect directly over fiber or high-availability broadband, or through an edge server where bandwidth is constrained.
- Offline recording at the edge with automatic event synchronization once the network recovers.
- The recommended primary stream is 1920 × 1080 or 1920 × 1440, with an optional low-resolution secondary stream that exists only to make multi-camera layouts play efficiently.
- Non-camera devices (ATMs, legacy CCTV, intercoms) sit on their own map layers.
- There is no plug-and-play connection, by design: a qualified network engineer configures each camera and router, and per-module camera requirements are published in the platform documentation.
- An ONVIF device such as an existing NVR can be registered whole, with per-stream selection of which of its feeds IREX analyzes. Access to the archive on the cameras behind it is a per-device option, and a camera can be added for archive access only, which leaves it disabled for analytics while the recordings held on it stay playable.
From the Record
What IREX Has Published on Video Management
The case for retiring the recorder rack is not new, and it was costed rather than asserted: a system-integration partner ran a three-year total cost of ownership on a 1,000-camera, 100-user program against IREX and two traditional non-cloud platforms. The player and the edge architecture have their own public trail.
Demo: Forensic Video Analysis with IREX Media Player Unlike most VSaaS, IREX is built from scratch as an AI-driven platform for Smart Cities. IREX video analytics is the core of the platform rather than an add-on to a cloud video… Read More 
Why IREX is retiring Traditional VMS - Total Cost Ownership Analyzed Recently, one of IREX system integration partners has estimated the total cost of ownership for a smart city project with an initial size of 1,000 cameras and 100 users. The integrator has compared… Read More 
IREX Expands its Platform with Edge Devices to Reduce Bandwidth Usage Alexandria, VA - April, 19 2021 - IREX announces today its edge video analytics solution for reduced bandwidth usage and offline recording. The IREX platform is expanding to… Read More FAQ
Do we keep our existing cameras and recorders?
The cameras, yes: that is the standard model. The recorders generally become unnecessary, because retention moves into the private cloud where the analytics and search already live. An existing ONVIF NVR can also be registered as a device, which makes its connected streams available to IREX analytics with per-stream selection. Where recordings already sit on the cameras or on that NVR, the platform can play that archive back rather than re-record it centrally, and a camera can be registered for archive access only, which leaves it disabled for analytics while its own recordings stay reachable.
What happens if a server or drive fails?
Storage is triple-replicated with automatic failover, and every component is hot-swappable, so a failure is handled without operator-visible downtime. There is no single point of failure by design.
What happens if a camera fails?
That is the other half of the question, and it is what TamperTrack is for. It runs as a detector alongside whatever analytics module a camera already carries, so watching camera health costs nothing in module capacity, and it alarms on signal loss, low frame rate, high bitrate, a shifted view, an obstructed, dirty or defocused lens, and a scene too bright or too dark, each switched on and tuned per camera. The faulty camera also carries a Video issues status here, filterable by issue type. Every alarm and how it is tuned is on the camera tamper detection page.
How do we get video out for a prosecutor or a court?
Through the Gallery, and the export is part of the record rather than a side channel. In the player, Save to Gallery captures the segment whose bounds you set, and the clip is then held in the platform’s own storage with its channel, period, duration and download status listed, where it cannot be lost or corrupted on a desktop. From the row menu it downloads to a local file or is shared as a link with another user of the same instance. Single frames come out as JPEG with Download current frame, and Copy video link produces a short link pinned to the exact playback moment. Downloading a clip is classed as ethically sensitive, so it requires a Case ID and is logged in full with the user, the IP address, the subject, the reference document, and the timestamp.
Can we analyze video that did not come from one of our cameras?
Yes. Upload the file through Gallery > Uploads with its recording date and time, the user groups that may see it, and its location on the map, then run Analyze video. The events appear on the Events screen under the standard filters, and you can run a second module over the same file and review both sets together. Suitability is per module rather than universal: FaceTrack Pro and StreamVLM™ support uploaded and non-fixed sources, while FireTrack, ObjectTrack Pro, CarCount and MotionTrack Pro are documented as fixed-camera only, and person re-identification inherits that limit. Uploading and analysis are Case ID-gated. StreamVLM™ is shipping in beta on selected instances.
How long is video kept?
Retention is sized per deployment and scales independently of camera count. The default is 30 days of raw footage, configurable upward, with critical alert images kept twelve months or longer. A single instance is proven at 100,000+ cameras with 30-day instant search; the architectural ceiling is a one-year retention period.
Can a site keep recording if the network drops?
Yes, where an edge server is deployed. The edge records offline and synchronizes events automatically once the link is restored, and the high-availability model runs redundant edge servers so the site keeps recording and analyzing while its uplink is down.
Compare It against Your Current Stack
The three-year total cost of ownership against what you run today is a standard deliverable, not a rebuttal we produce only when challenged.