Capture, networking, recording, management and analytics planned as parts of the same system.
Defense IA 3.3, Defense IA Lite and iNVD considered according to version, license and processing location.
An enterprise video project must turn a specific need into retrievable footage and clear procedures. Monitoring an entrance, observing a loading area or reviewing an incident calls for different camera, recording and analytics choices. The Intelbras ecosystem brings together devices and software for those functions, but the result depends on the combination of models, versions, infrastructure and operating rules. Since 2011, in São José do Rio Preto, ADD Solution has focused on technology engineering, deployment and operations. In Video Intelligence, the approach is to plan the system as a whole: coverage, networking, power, storage, integration and the routines of authorized personnel. The specification must preserve what already works and identify limitations before expanding the system. Technical basis: Intelbras Defense solutions.
IP cameras in the VIP family send video over the network; resolution, lens, illumination and onboard analytics vary by model. A camera with local analytics can generate events, while another provides only the stream required by the application. The family name does not replace the data sheet. For recorders, Intelbras distinguishes MHDX and iMHDX, associated with the Multi HD architecture, from NVD and iNVD, intended for network video. Intelligent features and the number of processed channels must be checked per product. During modernization, this allows existing camera points to be evaluated for reuse without assuming that every camera or technology will perform equally in the new system. The VMS organizes users, devices, viewing, searches and events. It does not automatically create an analytic function missing from the camera or recorder. The project must record where each function is processed and which information reaches the software. This separation avoids sizing the server solely by the advertised channel count. Sources: official recorder classification, VIP camera example and Defense IA manual.
Defense IA centralizes device and event management in projects that require operational organization and integrations. When consulted on 2 October 2026, the current documentation presented Defense IA 3.3. Features, architecture, server requirements and client limitations must be associated with that version, without automatically reusing old tables. Defense IA Lite has its own positioning and limits. The comparison page describes free operation for up to 16 video channels and licensed expansion to 64, alongside differences in modules and architecture. These figures do not remove the need to verify licensing, approved devices and the computer’s effective capacity. Selection must start with the required functions: how many operators will work together, which sites will be managed, which events will be received and how searches will be performed. Multiview layout, resolution and simultaneous export requirements can affect the client even when the server still has capacity. Available licensing, the product’s nominal limit and validated performance for the scenario must be distinguished. Sources: Defense IA 3.3, Defense and Lite comparison, and 3.3 data sheet.
Virtual lines, virtual fences, person and vehicle classification, and counts can support asset protection and operational monitoring routines when available in the selected system. Camera-side analysis must be distinguished from recorder-side analysis, including simultaneous channel limits. The iNVD 3008-P illustrates this difference: its data sheet separates capabilities by processing location and states operational restrictions. It is therefore incorrect to infer that eight video inputs mean eight channels with any intelligent function active simultaneously. Certain outputs and configurations can also change feature availability. For the exact iNVD 3008-P model, the June 2025 data sheet states input and output network throughput of 256 Mbps with artificial intelligence functions disabled and 165 Mbps when at least one AI function is active. The same data sheet warns that combining more than one AI type reduces the number of available channels and can affect performance. These figures must not be transferred to another recorder, firmware or operating mode. The rule must be tested under actual illumination, distance, motion and occlusion conditions. Alerts need an owner, priority and a human verification procedure. Aggregate counts must be checked by sampling and used within the defined purpose. An automatic classification must not become a conclusion about someone’s intent, emotion or conduct. Sources: iNVD 3008-P data sheet and analytics in Defense IA Lite.
Before installation, survey channels, bitrates, network paths and power points. Continuous recording, live viewing, playback and export compete for resources. A Gigabit port alone does not ensure performance: uplinks, intermediate switches, the server and disks must also support the load. For recorders with PoE, check the electrical standard, per-port power and total limit. For the iNVD 3008-P, for example, the data sheet specifies eight PoE ports and a separate total power budget. The calculation must use the maximum consumption of the selected cameras, including illumination and accessories, rather than multiply a nominal wattage by an arbitrary number of ports. The survey must also address cabling quality, distance, connectors, surge protection and backup power runtime. Extended-range features, where available, have specific conditions and must not be generalized to every Intelbras device. Network segmentation and access between video and corporate systems must be agreed with IT and validated without exposing administrative interfaces to the Internet. Sources: iNVD 3008-P product, recorder data sheet and Intelbras cybersecurity.
Channel count is only one of a recorder’s limits. Input, recording and output bandwidth, decoding, resolution and simultaneous functions also matter. A prominently advertised throughput figure must not automatically be treated as recording capacity: the data sheet must be read by function. Storage sizing starts with measured or estimated average bitrate, recording hours and retention period. As an arithmetic example, eight cameras at 3 Mbps, for 24 hours a day over 30 days, produce approximately 7.78 decimal TB of video. This excludes audio, metadata, reserves, redundancy and differences between raw and usable capacity. The Intelbras storage table allows slots and capacities to be checked by model. The final configuration must also verify approved disks, firmware and operating conditions. Event-based recording can reduce storage use, but depends on observed activity and pre-event and post-event recording; it must not be the only strategy where continuity is required. During operation, monitor the oldest retrievable recording, alerts and disk integrity. Changes in scene or quality can reduce effective retention and must trigger a new measurement. Sources: storage capacity by recorder and recording plans in Defense IA. The example is an engineering calculation, not a retention promise.
Defense IA documentation directs users to consult the compatibility table before adding devices. For Intelbras equipment, record the model, firmware, VMS version and required function. Native integration does not mean that every feature is available without distinction, including in installations with different generations. The Defense IA 3.3 data sheet mentions third-party devices through ONVIF S, G, T and M. This must be understood within the capabilities actually implemented at both ends. S relates to streaming; T to advanced streaming; G to recording and retrieval; M to metadata and analytics events. ONVIF conformance is tied to the registered version, and conditional features exist. At acceptance, test video, time, events, audio where authorized, searches and export. Live video does not prove that analytics information is transported. Where documentation does not clarify a behavior, conduct a proof of concept or obtain technical confirmation before including the feature in the commercial commitment. Sources: Defense IA manual, Defense IA 3.3 data sheet, ONVIF profiles and version-specific conformance.
Deployment must deliver an inventory, point identification, version records, accounts and permissions. Individual credentials and least privilege help control viewing and export. Updates must come from official channels, with compatibility assessment and a recovery procedure. The Defense IA 3.3 guide recommends contacting support in advance when a 3.2 or earlier environment has active integrations. An update must account for dependencies, backups and testing before operations resume. Installing the latest version and assuming every integration remains unchanged is not enough. Database and configuration backup is different from recording backup. Both require designated owners, storage locations and restoration testing. A redundant setup, where supported, does not replace an independent copy. Acceptance must also verify recovery after network loss, power failure, a full disk and service unavailability. Recording recovery features depend on the devices and the configuration actually approved. Sources: 3.3 installation guide, security guidance, and Defense backup and support.
Processing images of identifiable people requires assessing purpose, legal basis, necessity and data protection. Retention periods and access must be defined for the context; choosing a brand does not automatically establish LGPD compliance. The design can reduce unnecessary collection through framing, masks and audio and export restrictions. The controller must assess transparency and responses to data subjects, with legal support where needed. Any adoption of biometrics requires its own assessment and is not part of a generic Video Intelligence promise. Legitimate interest does not apply to sensitive personal data. Sources: LGPD and ANPD guidance on legal bases.
Complete answers for Intelbras architecture, deployment and operational decisions.
The inventory can identify camera points suitable for reuse. Compatibility, physical condition, image quality and support must be checked individually. The decision must preserve required functions and account for the cost of retaining existing limitations. Intelbras recorder classification.
No. The official comparison presents different limits, modules and expansion methods. Selection must consider channels, sites, operators and licensed functions, as well as hardware capacity. Official comparison.
This must not be assumed. For Defense IA Lite, the manual states that analytics are processed by compatible devices. The analytic function, processing location and channel limit must be included in the specification. Lite manual.
No. The common profile, version and conditional features must be checked. Proprietary functions and metadata must be tested in the actual workflow. Checking the official registry does not replace application acceptance. ONVIF conformance.
It depends on usable capacity, bitrate, recording hours and configuration. Calculate using explicit assumptions and verify retention in the environment. A model’s maximum capacity does not guarantee a retention period. Intelbras table.
It must follow the company’s security policy. Assess the supported mechanism, authentication, permissions and updates. Publishing administrative interfaces without an approved architecture increases exposure and must be avoided. Intelbras guidance.
This must be validated. Migration to Defense IA 3.3 carries a specific warning for earlier environments with active integrations. Plan support, backups and a test window before making the change. Guide 3.3.
Equipment, disks, infrastructure, server, clients, applicable licenses, deployment and operations. Modules and expansions must be identified with their entitlements and limits, without assuming they are all included in the base package. License comparison.
Structure capture, retention, management and analytics with clear compatibility and responsibilities.