Recording calculator
How much storage and bandwidth do your cameras need?
Enter cameras, resolution, codec and days of recording. You get bandwidth, terabytes, disks and how far you can see each thing.
- Storage and RAID
- Network and WAN
- Video walls and workstations
- Replication
- DORI in 2D and 3D
Everything is calculated in your browser. Nothing is sent to any server, and you do not need to sign up.
Sizing report, . Scenario: A, Scenario A.
Cameras and retention
Disks and protection
Annual disk workload
The annual workload is within the disk rating.
Surveillance disks list their TB-per-year rating on the datasheet: about 180 for standard models and 550 for AI and enterprise models. Reads cover playback, exports and searches. If the workload goes over the limit, data disks are added. Capacities in decimal TB (1 TB = 10¹² bytes): the operating system shows about 9% less.
Video walls and operator workstations
Each row is one type of display: a video wall, a group of operator workstations or a full-screen monitor. The automatic stream requests the smallest resolution that fills each tile of the layout.
54 cameras shown, 640×360 px tiles, stream 640×360: 16.4 Mbps, 6.0 equivalent streams and 2 stations.
With the main stream 4K it would be 414 Mbps and 216 equivalent streams: 36 times more decoding.
32 cameras shown, 960×540 px tiles, stream 1K: 33.9 Mbps, 14 equivalent streams and 4 stations.
With the main stream 4K it would be 245 Mbps and 128 equivalent streams: 9.0 times more decoding.
Decoding is measured in equivalent 1080p 25 fps streams: match it to the real hardware at each station. This traffic goes from the IRIS server to the workstations. It is not added to the bandwidth from the cameras.
Replication and continuity
Remote access
If IRIS recodes, only the phase 1 stream travels to the second site, and the recoding is done there. An overnight window frees up the WAN during the day, but multiplies the bandwidth during the window. The second site uses the same disk, RAID and margin settings.
Network
What the network carries, segment by segment, and a table to size any installation by resolution and codec.
Network use in this scenario
| Segment | Throughput | Volume per day | Volume per month |
|---|---|---|---|
| Cameras → server Camera LAN, 24 h | 383 Mbps | 4.1 TB | 124 TB |
| Server → workstations Live view on video walls and workstations | 50.4 Mbps | 544 GB | 16.3 TB |
| Replication to second site WAN, daily average over the window | 383 Mbps | 4.1 TB | 124 TB |
| Remote access WAN, users connected at once | 5.5 Mbps | 59 GB | 1.8 TB |
| Total | 822 Mbps | 8.9 TB | 266 TB |
Network table by resolution and codec
Estimated bitrate per camera and what they all use together. Change the cameras, codec or fps and the table updates.
| Resolution | MP | Mbps per camera with each codec | 50 cameras in H.265 | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
| MJPEG | H.264 | H.264+ | H.265 | H.265+ | AV1 | Throughput | Volume per day | Volume per month | ||
| Substream 640×360 | 0.2 | 4.34 | 0.55 | 0.36 | Chosen codec: 0.30 | 0.17 | 0.22 | 15.2 Mbps | 164 GB / day | 4.9 TB |
| 1K 1280×720 | 0.9 | 16.2 | 1.93 | 1.25 | Chosen codec: 1.06 | 0.58 | 0.77 | 53.0 Mbps | 573 GB / day | 17.2 TB |
| 2K 1920×1080 | 2.1 | 35.0 | 4.00 | 2.60 | Chosen codec: 2.20 | 1.20 | 1.60 | 110 Mbps | 1.2 TB / day | 35.6 TB |
| 4K 3840×2160 | 8.3 | 130.6 | 13.9 | 9.05 | Chosen codec: 7.66 | 4.18 | 5.57 | 383 Mbps | 4.1 TB / day | 124 TB |
| 8K 7680×4320 | 33 | 487.5 | 48.5 | 31.5 | Chosen codec: 26.7 | 14.6 | 19.4 | 1.33 Gbps | 14.4 TB / day | 432 TB |
| 16K 15360×8640 | 133 | 1,819.4 | 168.9 | 109.8 | Chosen codec: 92.9 | 50.7 | 67.6 | 4.64 Gbps | 50.2 TB / day | 1.50 PB |
Volume = throughput × hours streaming × 3,600 s ÷ 8,000. A month counts as 30 days. A camera that records only on activity keeps streaming: network volume does not drop, disk use does. This is the calculator's own estimate: a bitrate measured on site takes priority over this table.
Scenario comparison
There is only one scenario.
Duplicate it, then change what you want to compare: resolution, codec, days, RAID, disk or replication.
The selection with each codec
Estimated bitrate per camera, in Mbps, for each phase set up. The marked cell is the codec chosen in the group.
| Group | Phase | MJPEG | H.264 | H.264+ | H.265 | H.265+ | AV1 |
|---|---|---|---|---|---|---|---|
Group 1 50 cameras, 25 fps, medium activity, continuous recording | Phase 1: 4K · 25 fps days 1–30, Main stream | 130.6 | 13.9 | 9.05 | Chosen codec: 7.66 | 4.18 | 5.57 |
| Phase 2: 2K · 25 fps days 31–90, Recoded by IRIS | 35.0 | 4.00 | 2.60 | Chosen codec: 2.20 | 1.20 | 1.60 |
Totals if the whole scenario used each codec
| Codec | Bandwidth | Video disk | Disks | Disk vs selection |
|---|---|---|---|---|
| MJPEG | 6.53 Gbps | 3.25 PB | 215 × 20 TB | +1,563 % |
| H.264 | 696 Mbps | 355 TB | 24 × 20 TB | +82 % |
| H.264+ | 453 Mbps | 231 TB | 17 × 20 TB | +18 % |
| H.265 current selection | 383 Mbps | 195 TB | 16 × 20 TB | Reference |
| H.265+ | 209 Mbps | 107 TB | 8 × 20 TB | −45 % |
| AV1 | 279 Mbps | 142 TB | 10 × 20 TB | −27 % |
If a group has a measured bitrate, the other codecs are scaled from that value. Not every camera or workstation supports H.265+ and AV1: confirm support before including them in a quote. MJPEG compresses each frame separately: it takes up far more space and grows in proportion to fps.
Breakdown by phase
| Group | Phase | Period | Resolution | FPS | Source | Mbps per camera | Days stored | Storage |
|---|---|---|---|---|---|---|---|---|
Group 1 50 cameras, H.265 | Phase 1 | days 1–30 | 4K | 25 | Main stream | 7.66 | 30 | 124 TB |
| Phase 2 | days 31–90 | 2K | 25 | Recoded by IRIS | 2.20 | 60 | 71.3 TB | |
| Total: 50 cameras | 383 Mbps | 90 | 195 TB | |||||
Pixel density (DORI)
What you want to see and why: how many pixels the object needs, how far each resolution reaches and which one is the minimum.
Cautious: light changes, occasional backlight and moving objects. Requires 2.0 times the ideal pixels.
Compression and bitrate
| Resolution | On the object at 25 m | Detect at 25 m | Maximum distance with a 60° angle | |||
|---|---|---|---|---|---|---|
| Detect 25 px | Observe 63 px | Recognize 125 px | Identify 250 px | |||
| 1K 1280 px | 22×75 px | Falls short 22 of 25 px | 22 m | 8.9 m | 4.4 m | 2.2 m |
| 2K 1920 px | 33×113 px | Meets | 33 m | 13 m | 6.7 m | 3.3 m |
| 4K 3840 pxIn use | 67×226 px | Meets | 67 m | 27 m | 13 m | 6.7 m |
| 8K 7680 px | 133×452 px | Meets | 133 m | 53 m | 27 m | 13 m |
| 16K 15360 px | 266×905 px | Meets | 266 m | 106 m | 53 m | 27 m |
Pixels are counted on the shortest side of the object seen from the front: the worst case. For a person they follow EN 62676-4: 25, 62.5, 125 and 250 px/m over 0.5 m width. The safety margin multiplies them. You can change the ideal pixels if you have your own measurements. If the analytics rescale the image before analyzing it, what counts is the pixels of the analyzed image.
Rough simulation: below the recommended bitrate, usable detail drops with the square root of the ratio. At half the bitrate, 71% of the detail remains; at a quarter, half. Never less than 15%.
Legal notice
The results of this calculator, and in particular the pixel density results, are theoretical, indicative estimates obtained with a simplified geometric model. They do not constitute an offer, commitment, performance guarantee or technical specification, and they create no contractual or pre-contractual obligation for North Point Solution, S.L. (Infinity Neural).
The actual ability to detect, classify, recognize, identify or read depends on many factors that this calculation does not take into account: lighting (day, night, infrared, backlight, glare), weather (rain, fog, snow), sensor and lens quality, focus, compression and bitrate, installation height and angle, perspective, object speed, vibration, dirt on the lens, occlusions and the configuration of the AI models.
Any design must be validated with on-site tests, at the actual location and with the final equipment, before making any commitment.
Calculation settings
For cameras with other resolutions or with bitrates measured on site. Changes apply to all scenarios.
Resolutions
| Name | Width (px) | Height (px) | Megapixels |
|---|---|---|---|
| 1K | 0.9 MP | ||
| 2K | 2.1 MP | ||
| 4K | 8.3 MP | ||
| 8K | 33 MP | ||
| 16K | 133 MP |
Reference bitrate per codec
Mbps at 1920×1080, 25 fps and medium activity. H.264+ and H.265+ are the smart codec modes (dynamic GOP and regions of interest) offered by most manufacturers.
How it is calculated
Mbps per camera = reference × (pixels ÷ 2,073,600)^0.9 × (fps ÷ 25)^0.85 × activity
Exponents below 1 reflect that the codec compresses better the more pixels and frames there are. Activity factors: low 0.6; medium 1; high 1.5; very high 2.2. With a measured bitrate, phase 1 uses that value and the later phases are scaled in the same proportion.
MJPEG is handled separately: its bitrate grows in proportion to fps (exponent 1), pixels weigh 0.95 and activity only 30%.
Storage = cameras × Mbps × 10.8 GB per day × days stored × share of time recorded. If IRIS recodes, each phase stores only its own days. If the camera sends another stream, that stream is recorded from day one to the end of its phase.
Workload per disk = video written per year × (1 + reads) ÷ data disks. RAID parity is spread across all the disks in the array: each one carries the same load as a data disk.
How the math works
Each camera's bitrate
It starts from a reference per codec at 1080p and 25 fps: 4 Mbps in H.264, 2.2 in H.265, 1.6 in AV1.
It scales with pixels and fps, using exponents below 1, and with scene activity.
Storage
One Mbps sustained for a whole day is 10.8 GB. Multiply by cameras, days stored and time recording.
Each retention phase keeps its own resolution and days. The margin is added at the end.
Disks and RAID
Data disks are calculated by capacity and by annual workload, and the larger number wins.
RAID 5 and 6 add one or two parity disks per array; RAID 10 doubles every disk.
The network
It adds up what leaves the cameras, live view to video walls and workstations, replication and remote access.
For each segment it gives the throughput in Mbps and the volume per day and per month.
DORI
Pixels per meter = image width in pixels ÷ scene width at that distance.
Scene width = 2 × distance × tan(angle ÷ 2). A low bitrate takes away usable detail.
Frequently asked questions
How much space does one day of recording from one camera take?
A 1080p camera in H.265 at 25 fps streams about 2.2 Mbps: about 24 GB per day. In 4K, about 7.7 Mbps and 83 GB per day.
What is DORI?
Detect, Observe, Recognize and Identify. The EN 62676-4 standard sets 25, 62.5, 125 and 250 pixels per meter for these tasks.
Which RAID is best for video surveillance?
RAID 6 is the usual choice in large recorders: it survives two failed disks at once. RAID 5 only survives one.
Does recording only on motion save bandwidth?
No. The camera keeps streaming all the time. What drops is disk use, in proportion to the time with activity.
How much does H.265 save over H.264?
With this calculator's references, H.265 needs 45% less bitrate than H.264 for the same image and scene.
Does lowering the bitrate affect what you see?
Yes. With fewer bits per pixel, the codec drops fine detail first. The calculator simulates this in the DORI section.
Next step
Public transport
Who watches a whole station at rush hour?
17 cases: 11 raise an alert when something happens and 6 only measure. You see each image as it is, and then with what IRIS understands on top.
Retail
Who sees the sales that never reach the register?
10 cases: 4 raise an alert when something happens and 6 only measure. You see each image as it is, and then with what IRIS understands on top.
And the best part
The rule does not decide anything.It just never forgets, and it tells the person who decides.
You write once what needs watching, and it holds at three in the morning just as at noon. What to do with the alert is still a person’s decision.
Tell us your problem and we’ll say whether IRIS understands it, or not yet.
A person answers, the same working day.