Cctv Fps Comparison Video

[Rulland]

Obv you make the 'settings' suit the application!-but generally imho 8 fps is normally suitable for most surveillance.

Richard.

[Adi]

IF you have the hd sapce then id agree, but i think you would be better of dropping the fps rather than the quality

Agreed, 6 FPS here @ D1

You gain no more usable info with more FPS unless your targeting a fast moving image and even more so at close range.

[digitalwitness]

For me what resolution and frame rate completely depends on application (as others have said) but also what compression is available.

More modern compression technologies like MPEG4 and H.264 are more suited to higher frame rates, due to the nature of the compression to save a reference image and then only save the difference in subsequent frames, long times between each shot may result in additional blocking or the reconstructed image (i.e. playback) or a completely new reference image resulting in little or no space savings over the likes of JPEG.

Regardless of how many fps, each frame needs to use usable for the intended purpose (Identification, recognition or observation), so resolution and quality are first, then comes fps and recording time with the bias towards fps.

Agreed, 6 FPS here @ D1

You gain no more usable info with more FPS unless your targeting a fast moving image and even more so at close range.

Have you noticed any interlacing issues with using D1? Recently downloaded from a DVR for the police (don't want to mention make but its a well recognised manufacturer), the static picture was very nice indeed until the target ran through it and interlacing problems rendered the footage unusable, target was split in two, best described as interlaced fields were slightly out of phase with each other on the fast moving target.

Edit: Not that there is any interlacing issues with the above video but not sure what is being used for capture.

[arfur mo]

i can't reference it (like i did on the 30 days lol!)

seem to recall i read or heard somewhere CCTv footage was thrown out as the defence claimed "the compression software altered so tampered with the original evidence'?

might have been an image from a speed camera when digitally blown up to prove the driver

don't burn me guys, just something i trawled up from the deeper recesses.

Arfur

[Adi]

For me what resolution and frame rate completely depends on application (as others have said) but also what compression is available.

More modern compression technologies like MPEG4 and H.264 are more suited to higher frame rates, due to the nature of the compression to save a reference image and then only save the difference in subsequent frames, long times between each shot may result in additional blocking or the reconstructed image (i.e. playback) or a completely new reference image resulting in little or no space savings over the likes of JPEG.

Regardless of how many fps, each frame needs to use usable for the intended purpose (Identification, recognition or observation), so resolution and quality are first, then comes fps and recording time with the bias towards fps.

Have you noticed any interlacing issues with using D1? Recently downloaded from a DVR for the police (don't want to mention make but its a well recognised manufacturer), the static picture was very nice indeed until the target ran through it and interlacing problems rendered the footage unusable, target was split in two, best described as interlaced fields were slightly out of phase with each other on the fast moving target.

Edit: Not that there is any interlacing issues with the above video but not sure what is being used for capture.

No mate, but will look more closely at it, ive burned quite a bit of evidence lately and thought it was good.

The IT chap who set up some dixons/currys remote view said my dvrs were better than theirs and a tesco manager said mine was better than theirs. Admitted with the tesco chap, he only saw live images on the hd monitor and not playback.

These were adata units, are you refering to these??

[digitalwitness]

seem to recall i read or heard somewhere CCTv footage was thrown out as the defence claimed "the compression software altered so tampered with the original evidence'?

There has to be more to that story Arfur, taken at face value it would mean that every DVR is inadmissible as evidence. Possible circumstance maybe were enhancement software is used on a picture that predicts the pixels where there was insufficient resolution in the original and used as evidence.

These were adata units, are you refering to these??

No mate they weren't Adata, I did not want to mention the make because I do not want to been seen as knocking a particular make when it is quite a common issue and not limited to that make.

[james.wilson]

Very Valid point regarding interlacing etc. AFAIK any digital capture of a std pal source will result in interlacing.

Brief info on interlacing Here

image to show horizontal motion

This occurs because the of interlacing. ie the chip captures all the odd lines, ie 1 to 625 then comes back to the evens after it has finished the odds. Obviously there is a time diffrence between line 1 and 2 but none between 1 and 3. If the image isnt moving then the image will look fine, if the imgae only has veritcal movement it wont be so noticable. BUt if you have horizontal movement it will look awful and IMO unusable.

That is why most machines are limited to 2 cif (only capture the odd or even lines and interlacing isnt an issue). Once you go above 288 vertical lines then this issue will appear.

The better dvrs deal with this in various ways, But as noted any machine that alters the image BEFORE it is dropped to disk is modifying the image and MAY make it inadmissible. (If unsure dont go above 2CIF)

You will usually find on higher end dvr's a PLAYBACK deinterlace filter. This means that interlaced frames are being stored with the above combing problem. This means an unmodified image is being stored AND is available for evidence. The playback filters then deinterlace, whether that be on the machine i/f or on the pc software.

Deinterlacing is a whole subject in its own right and their are various methods, none IMO perfect. But then PAL is an analouge transmission system, the only way to remove all these problems is to use IP where PAL isnt usually relevant (some manafactures bodge a converter onto an existing PAL pickup).

You will notice deinterlacing at work on your flat panel when watching football etc and the pitch lines loose their straight line look on camera pans.

[hpotter]

More modern compression technologies like MPEG4 and H.264 are more suited to higher frame rates, due to the nature of the compression to save a reference image and then only save the difference in subsequent frames, long times between each shot may result in additional blocking or the reconstructed image (i.e. playback) or a completely new reference image resulting in little or no space savings over the likes of JPEG.

Spot on.

notion of dvr taking n (1-25) photo's per second "just like how they make cartoons" is misleading. and have found you do need to up frame rate to give relative quality, and D1 ime aint worth the sacrifice. eg 25fps @ 2cf far better than 12 @D1 on some of the machines been working with recently.

[Adi]

thou shall play

[breff]

I will always sacrifice frame rate to get better quality, better to have 5 good images of someone than 25 poor images