frame rates

[dharper 0]

I am thinking about putting together one of these systems for a client. I installed his alarm but I have never installed a survellance system before. On cards I see frame rates from 30 - 480. The prices go up as the frame rates go up. So whats a decent frame rate to have.

 

dave

[larry 2]

Hello welcome to cctvforum. I'll give this one a shot.

 

I have read that the frame rate is divided among the number of cameras you have on your system. For example, 60fps divided by three cameras means an adjusted rate of 20fps. Smooth motion is about 30fps. For security purposes, I think an average of 10fps would be fine.

 

Keep in mind though that most software used to run dvr cards allows automatic adjustments to the fps. An example would be - say you have 4 cameras on an 60fps card, but you have the cameras on motion detect; meaning they only record when they sense motion. If only two are picking up motion then your recording fps could be 30fps each.

 

So the higher is better but fps really depends on your layout and how many cameras will be recording at one time.

[AVCONSULTING 0]

The frame rate depends a lot on what the cameras are covering. Is it a retail store, a home, or what? Remember that higher frame rates take up much more disk storage and you will spend a lot of extra money on hard drive space if you go with a higher frame rate even though it may not even be needed. In many instances 2 fps is more than enough for surveillance.

[cctv_down_under 0]

Both the previous posts are correct however there are a few more things to consider, most of the stated frame rates by manufactorers are based on small image recordings, for example 400FPS Pal is real time recording when 16 cameras are connected @ 320x240 Resolution (PAL). howver 640x480 (PAL) would be about 75% of this figure, also the capacity of the FSB (Front Side Bus) of the motherboard and the capacity of the CPU make a big difference too, PC Based systems usually require a P4 with Hyperthreading to be able to achieve real time recording rates and the better the CPU the faster the rates that are recorded, some cards do MPEG4 compression on board then send ity down the bus and uncompress it again making the CPU not work so hard and therefore giving faster frame rates.

There are a compnaies that state the frame rates as Totals instead of individual and there are some that measure in Fields instead of frames so BEWARE!!!

[DavidR 0]

In my experience, 10fps is PLENTY for most applications.

 

1-2fps is too slow, unless you are going to catch people standing still (like at an ATM)

 

If you're in a casino trying to catch someone dealing from the bottom of the deck, you may need 30+.

 

But to identify an intruder, and provide decent evidence, somewhere between 4fps and 10fps is adequate.

[cctv_down_under 0]

I agree with David, I try not to have my systems slower than 12 fps per camera where possible, however price can dictate the speed.

[qman 0]

Damm, Interestig discussion

 

I'm throwing in my own 2 cents..

 

a lot of manufactures say that their cards do XXX FPS,

 

Example:

 

"8 Port DVR Card, 120FPS"

 

but in reality, if you make the math, 120/8=15!!

 

That means that you only get 15fps recording.

 

SO, be very careful on what you buy

[cctv_down_under 0]

Worse than that is that some state Fields instead of Frames and there are 2 Fields to every Frame, therefore a 100Frames Per Second Global Card would be 200 Fields Per Second Card, however they both abbreviate the same (FPS).

 

You really do need to be carefull, also the rates that they boast are calculated at 320x240 resolution when there is an option for 640x480, but if that is used you will not see the promised frame rate.

 

The easy way to check is to count the number of chips on the card (each one is worth 25fps).

[AVCONSULTING 0]

Actually in NTSC each processor is good for 20 fps when more than 1 camera is used. So a 120 fps rated system really is 80 fps but you will rarely find a dvr company stating this in their spec's (although there are some that tell the truth).

[paprotec 0]

15 to 20 FPS used to be the max FPS per chip. However in recent months the major DVR manufactures have written new drivers to support true 30 FPS per chip. It will be some time before we see 480 display and record at 640X480. This PC bus is the bottle neck.

 

Also, as more and more DVR board makers go to hardware compression, frame rates and quality will increase. There are only a handful of companies using hardware compression at this time.

 

PA

[qman 0]

Well, with the new PCI Express coming out soon, that's supposed to be 2.5ghz, (compared to the old PCI that has 33mhz,), then we are going to see some real nice pictures from our DVR;s., imagine a .99C card then that works like a $1500.00 today?

 

We shall see

[cctv_down_under 0]

Yes but imagine the size of the HDDs needed, I think the day of HDD's is running out and we will soon see storage on non volitile ram.

[rory 0]

kalatel gives 30pps, i use 5pps that gives goodplayback (the lowest I use) for retail, and if it is an all in one i use 2.5pps.

 

rory

[cctv_down_under 0]

I dont sell anything under 10FPS per camera but I am fussy, and yes i know you do not need that much, however I like to set mine with the larger resolutions.

 

Geovision buffers unused frames so unlike 100fps divided by 10 cams @ 10fps per camera..... If only five of the ten have motion at one time then it buffers the spare unused 50fps on the fly, they call it S.R.T. or smart recording technology.

 

Chews space but works very well.

[qman 0]

But aside from all this, the only problem that get from clients is when they ask me "why does the picture looks like slow motion?"

 

Hard to make them understand teh technology, but when they ask for the price to make it better, then they go like "oh, never mind"

[cctv_down_under 0]

Thats always the way. the realtime preview cards cost a lot.

Geo makes an 8ch one as well as a 16ch, this way you can at least save when only 8ch's are used.

 

I always quote MUX recording speeds and say " Look your system is 10 times faster".

 

Teh other trick is to use a looping switcher for display then feedback to the DVR then they get full size images and clear picture and no delay.

 

I use looping switchers a lot!

[Th 0]

I am sorry for bad english.

 

1. It's enough to have 1-4 fps in normal processing and 10fps on alarm with 720x576 or 720x288 resolution for recording.

2.This high resolution declared by the manufacturer in pixels not guarantees real high resolution.

3. With small resolutions (such as 320x200) you cant see details at all.

4. Recording of each channel in realtime - 25 (PAL), 30 (NTSC) demands a huge amount of hard disk space. One day record of 16 channels in high resolution with 25 fps per channel - it is more than 1000 Gb for most devices. One month - more than 30000Gb.

5. Recording with motion detection saves not in all cases.

6. You can't get a good quality with MPEG4, so you must use at least MPEG2.

7. Most of video capture boards works without hardware compression.

The compression is carried out by a computer. If you record a lot of video channels in real time, you need a smart computer and there is a problem with PCI bus throughput. So some hardware works fine while recording a lot of channels with small fps and there are a lot of problems in real time.

[rory 0]

I am sorry for bad english.

 

1. It's enough to have 1-4 fps in normal processing and 10fps on alarm with 720x576 or 720x288 resolution for recording.

2.This high resolution declared by the manufacturer in pixels not guarantees real high resolution.

3. With small resolutions (such as 320x200) you cant see details at all.

4. Recording of each channel in realtime - 25 (PAL), 30 (NTSC) demands a huge amount of hard disk space. One day record of 16 channels in high resolution with 25 fps per channel - it is more than 1000 Gb for most devices. One month - more than 30000Gb.

5. Recording with motion detection saves not in all cases.

6. You can't get a good quality with MPEG4, so you must use at least MPEG2.

7. Most of video capture boards works without hardware compression.

The compression is carried out by a computer. If you record a lot of video channels in real time, you need a smart computer and there is a problem with PCI bus throughput. So some hardware works fine while recording a lot of channels with small fps and there are a lot of problems in real time.

 

good answer.

 

Also, if they really wanted real time, like a casino, they would need single channel DVRs for each channel, so if they had 10 cameras, thats 10 single channels DVRs:

 

If you dont need audio recording, you could do this:

 

10 Camera Triplex Multiplexer

10 Kalatel DSR-2000 1 channel 60pps DVRs 40GB

 

Then to access the video, though, you would need 1 IP address for each DSR2000, and you can use a custom multi site software to view on a PC, all in one screen, live, playback, etc. I guess you could split the video signal, from the 10 DSRs, to go into the 1 VCR input on the mux, then just hit play on the one you want to play from.

 

There is probably another way to do this, with another device suchas PS based to play back or view the video stream from all DSRs live. You could use a DVMRe DVR multiplexer, but then you wouldnt be able to access the recorded video from the DSRs, but you would be able to view all live with 1 IP address...

 

Any ideas?? They also have a 1 channel Video Digital Recorder, that has removeable hard drives, but no LAN, but they cost same as a regular time lapse VCR.

[cctv_down_under 0]

Nice points TH

 

One thing to note though is that many manufactorers already provide proprietry onboard video compression, taking the load away from the CPU, but they are too expensive at this stage, and try explaining THAT to the end user!

[ChuckM 0]

Thats always the way. the realtime preview cards cost a lot.

Geo makes an 8ch one as well as a 16ch, this way you can at least save when only 8ch's are used.

 

I always quote MUX recording speeds and say " Look your system is 10 times faster".

 

Teh other trick is to use a looping switcher for display then feedback to the DVR then they get full size images and clear picture and no delay.

 

I use looping switchers a lot!

 

Forgive my ignorance, but I don't know what a "looping switcher" does. This is the definition I found:

 

"Looping switcher: is a type of switcher with looping outputs. These outputs allow the camera to be connected to another device without affecting the usage of the switcher."

 

Is this like using a "splitter" to send the camera signal to both the DVR and a CRT monitor, thereby allowing you to view the image real-time on a CRT while the DVR records at whatever its fps capacity is?

[rory 0]

yes, you can loop the video out to other muxes, switchers, quads, monitors, etc.

 

just better than using bulky T-connectors, most muxes now come this way.

its unprocessed video.

 

however, i havent had a need for a switcher yet. The muxes have switching built in, and id buy a quad / mux before a switcher. I dont like switching, dont see all the cameras at once.

[brevardcam 0]

If not already stated, consider a card with "Smart" Frame Rate option (used with motion detect). This is a nice feature and allows a lot of applications to use a lower fps card.

 

This feature will divert system resourses so as to increase the frame rate of the camera(s) detecting motion, so effectively a 30 fps card divided up by 4 cameras could have one camera processing at 25 fps and the other 3 at 1 fps until motion detect is diverted or max record time exceeded.

 

Sorry if repeated anyones post

[cctv_down_under 0]

good answer.

 

Also, if they really wanted real time, like a casino, they would need single channel DVRs for each channel, so if they had 10 cameras, thats 10 single channels DVRs:

 

Wrong, you can have 400Fps machines (Pal) that is real time on 16 cameras.

 

Also you can get good quality on Mpeg4 otherwise everyone would not be striving to develop the chips, no debate that Mpeg 2 is much nicer but WAY too big on file size unless very heavilly compressed and if it is then it becomes below par.

 

I havent seen many other than Geo that do the Smart frame buffering.

 

You use a looping switcher but do not record from it as the cameras go straight through it to your DVR, Its like interrupting each camera into a box allowing the choosing of any full screen view of a camera at another location (where the switcher is) and then each camera still goes to the DVR as if the switcher was not there..... an example.

 

Imagine a Guard room and a server room, the DVR would sit in the server room and most people on the network in the building can view all cameras on their pc's but the guard room at the front gate may not have a computer in it, therefore you run all cameras to the switcher in the Guard room then back from the loops to the DVR, the DVR works like nothing has changed and the Guard can press any button to view any area he wants on a full screen large view.

 

A switcher is about 1/8th the price of a Mux, therefore it is a very cheap way to have more than one controll of the DVR and you can also wire alarm sensors to it so that movements around the area will trigger a full screened view for the Guard.

[rory 0]

Wrong, you can have 400Fps machines (Pal) that is real time on 16 cameras.

 

edit..

 

Im not saying I love the Kalatel, if someone else had a better product, or the same for less, and even some less features as i dont use them all, I would use it, but there is nothing yet. Im not going to use a PC based, as that involves maintenance, and i may as well then just go into the PC business and make more money than we get per hour in the CCTV business.

 

Using WaveReader you can view 16 DVRs, 64 cameras at once on one monitor, or even better a 40-60' Plasma Display:

 

WaveReader 3.0, the newest version of

our exclusive remote viewing software,

offers more features and versatility for

your remote surveillance monitoring.

Now you can simultaneously connect

to as many as 16 different recorders,

and WaveReader can tile images from

all those recorders to create a

64-camera display. You can drag and

drop images from any recorder onto

WaveReader any spot on the display screen. What’s

more, you can create groups of

recorders, label them and monitor by

group, like “Texas” or “California.”

And you can save files larger than

2 GB to store high-resolution or longrunning

clips more efficiently.

 

Also you can get good quality on Mpeg4 otherwise everyone would not be striving to develop the chips, no debate that Mpeg 2 is much nicer but WAY too big on file size unless very heavilly compressed and if it is then it becomes below par.

 

I havent seen many other than Geo that do the Smart frame buffering.

 

 

Kalatel's Clear Cast Product::

----------------------------

ClearCast™ recording technology

delivers what’s most important and

what’s been missing from many digital

recording systems: Image quality. But

higher quality images are bigger—and

ClearCast can deliver real-time

video—30 frames per second for every

camera.

 

So ClearCast uses a new form

of video compression: eTreppid™

Exclusive to GE Interlogix, eTreppid compresses

files three to five times smaller than

MPEG-4 and up to 20 times smaller

than Wavelet files. As a result,

ClearCast lets you store 30 days or

more of high-resolution images.

 

Maximum storage for largescale

applications

 

ClearCast is designed for large-scale

applications that need to store highresolution

images from hundreds of

cameras. Casinos, financial institutions,

correctional facilities, airports

and other high-security environments

can get the clearest, most detailed

images available with ClearCast.

Most security applications record

multiplexed video.

 

That’s where

MPEG-4 compression falls short.

MPEG-4 only compresses the changes

between scenes. So it performs well

for a single camera, but chokes on

multiplexed video. eTreppid compresses

the entire image, not just the

changes. This allows ClearCast to store

30 days or more of multiplexed video

at 30 frames per second—with

unbeatable image quality.

 

 

Simple, scalable system

architecture

 

Large-scale applications may use

hundreds of cameras, so ClearCast

features a simple modular architecture.

The ClearCast capture station

incorporates an encoder, PC and

storage device. Each capture station

records video and audio from up to

eight cameras to onboard RAID

storage. And the drives are hot

swappable.

 

So if one fails, you can

simply just replace it with a fresh

drive. You won’t lose any video.

Another benefit of ClearCast’s

architecture is expandability. You can

start with only the number of capture

stations you need. As you needs grow,

you can easily add capture stations to

accommodate more cameras.

-----------------------------------

[rory 0]

Ok, checked with Kalatel,

 

Their DVRs use PowerPC Processors, and the Nucleus OS.

 

IBM PowerPC MicroElectronics:

http://www-306.ibm.com/chips/techlib/techlib.nsf/techdocs/76DC9CC2F622BCAE87256E39005925C3

 

Nucleus Embedded OS:

http://www.acceleratedtechnology.com/

 

Ok now can we make our own! I want DVDRW and USB Flash, anyone else ? )