High Resolution Cameras and Modect

[jsylvia007]

Howdy all!

Back again! After tuning a bunch of my cameras and getting a mostly stable system, I've added a high-resolution camera (2560x1440) and I've set both Reference Image Blend %ge, and Alarm Reference Image Blend %ge to 3.125%. Yet, I can't manage to trigger an alert with the whole frame being a single zone.

I've created the attached zones, but now, if motion crosses the zones, (I think), the original event closes, and I miss the data from the beginning of the next event.

Thoughts on possibly getting back to one zone, or perhaps my understanding is incorrect of how the zones work, or the %ges?

PS - I've attached a screenshot of the zones I have created. I'm contemplating creating another couple zones horizontally over the others, but I don't know if that will help.

[zd59]

Hi!

For a speed of motion detection and offload of the server I would suggest you, to create another monitor. The same camera but low resolution stream.
As per documentation: create zones on low stream monitor function = "modect", high resolution monitor function = "nodect" and link to low resolution with modect function. Be careful at link creation. Link only high resolution monitors, low resolution ones must be without links. Understand that as low res do a motion detection, hires DO NOT do motion detection, but instead is triggered (linked to) by low res. So you get high and low res events saved.
Try to solve your issue on low resolution stream. Do not overlap zones and try to create small ones, as modect on big zone eats a lot of CPU time.

I have two 3MB cameras (same res as yours) and do a modect on low res (VGA). Each high stream monitor is nodect linked to low stream from the same camera and works perfectly. I added a filter to delete a low res events after an hour, so I have only hi res events saved - to minimise disk space and get rid of duplicate events created by low and high resolution.

[jsylvia007]

I have two 3MB cameras (same res as yours) and do a modect on low res (VGA). Each high stream monitor is nodect linked to low stream from the same camera and works perfectly. I added a filter to delete a low res events after an hour, so I have only hi res events saved - to minimise disk space and get rid of duplicate events created by low and high resolution.

Thanks so much for the post! I will give this a shot when I get home. Quick question though, I assume this means that I won't get my red outlined motion areas on the hires versions? What other data is lost in doing capture in this manner? (I'm not sure there is any, just asking the question).

[zd59]

Yes you are right. When hires monitor is triggered, it records, but no info regarding detection is transferred from triggering device (logicaly). If you need info regarding triggers (points, group of points blobs or else) do not delete lowres events. I think, those are needed only for fine tuning of zones at setup.

[jsylvia007]

Yes you are right. When hires monitor is triggered, it records, but no info regarding detection is transferred from triggering device (logicaly). If you need info regarding triggers (points, group of points blobs or else) do not delete lowres events. I think, those are needed only for fine tuning of zones at setup.

Thanks so much. My plan is to do exactly as you have suggested. I will report back after I get everything configured. I should just be able to change my existing monitors to "nodect", right? I shouldn't have to restart all over again?

[jsylvia007]

I have two 3MB cameras (same res as yours) and do a modect on low res (VGA). Each high stream monitor is nodect linked to low stream from the same camera and works perfectly. I added a filter to delete a low res events after an hour, so I have only hi res events saved - to minimise disk space and get rid of duplicate events created by low and high resolution.

Thanks so much for the post! I will give this a shot when I get home. Quick question though, I assume this means that I won't get my red outlined motion areas on the hires versions? What other data is lost in doing capture in this manner? (I'm not sure there is any, just asking the question).

Can you please share what your monitor settings are for the lowres MODECTs? I've added the secondary channel to my cameras (704x480@15fps), and on the one camera to test with, BUT, this testing camera can't seem to make an event last longer than 10 seconds...

[jsylvia007]

I have two 3MB cameras (same res as yours) and do a modect on low res (VGA). Each high stream monitor is nodect linked to low stream from the same camera and works perfectly. I added a filter to delete a low res events after an hour, so I have only hi res events saved - to minimise disk space and get rid of duplicate events created by low and high resolution.

Thanks so much for the post! I will give this a shot when I get home. Quick question though, I assume this means that I won't get my red outlined motion areas on the hires versions? What other data is lost in doing capture in this manner? (I'm not sure there is any, just asking the question).

Can you please share what your monitor settings are for the lowres MODECTs? I've added the secondary channel to my cameras (704x480@15fps), and on the one camera to test with, BUT, this testing camera can't seem to make an event last longer than 10 seconds...

Ok! Making progress. I have the monitor producing good events now... I've now linked it to the actual monitor I want recording.

SO FAR SO GOOD!!

Now, I realized that I needed to ensure that all the buffer information matches between the two, which it does.

Now for the most interesting part. My lowres camera will produce a 35 second alert, which is perfect and matches what I expect. The hi-res which is what I care about, produces two 17 second alerts...

Any idea why that might be?

[zd59]

Hi jsylvia007!

The above you mentioned is the easy one.
ZM records event in size as set in monitor settings - Post Event Image Count.
For the same seconds of an event recorded, you have to match "Post Event Image Count" X "frame rate" of low res and highres monitor of the same camera

[jsylvia007]

Hi jsylvia007!

The above you mentioned is the easy one.
ZM records event in size as set in monitor settings - Post Event Image Count.
For the same seconds of an event recorded, you have to match "Post Event Image Count" X "frame rate" of low res and highres monitor of the same camera

So I think I have it all figured out! Thanks so much for the help dealing wth me to get it tuned.

I don't think that I have the hi-res tuned enough, even the low-res modect isn't picking up any motion, but the rest are all working perfectly well.

Thanks again for the help!

[Maximo1970]

Hi!

For a speed of motion detection and offload of the server I would suggest you, to create another monitor. The same camera but low resolution stream.
As per documentation: create zones on low stream monitor function = "modect", high resolution monitor function = "nodect" and link to low resolution with modect function. Be careful at link creation. Link only high resolution monitors, low resolution ones must be without links. Understand that as low res do a motion detection, hires DO NOT do motion detection, but instead is triggered (linked to) by low res. So you get high and low res events saved.
Try to solve your issue on low resolution stream. Do not overlap zones and try to create small ones, as modect on big zone eats a lot of CPU time.

I have two 3MB cameras (same res as yours) and do a modect on low res (VGA). Each high stream monitor is nodect linked to low stream from the same camera and works perfectly. I added a filter to delete a low res events after an hour, so I have only hi res events saved - to minimise disk space and get rid of duplicate events created by low and high resolution.

This sounds like the approach I need to look at. But my knowledge of zoneminder is very limited so forgive me. My setup has four servers, all virtual and take streams from my Floureon DVR. I have a total of six cameras, two per server for Hi-Res and the fourth server has all six lo-res feeds. Can you explain what this linking means and how I go about setting it up? As I read the above does it mean that once motion is detected on the lo-res feed the linking mechanism will start recording the Hi-res feed. Or have I missed the mark and there is more to it?

Sorry for the hi-jack.

Regards,

Garry

[zd59]

Hi Maximo1970!

You understand triggering right.
I'm new to ZM too (1/2 a year), and spend a lot of time to study/test, read posts, to understand it's philosopy. ZM is free, difficult to tune, but excellent when you get it.
How to link a monitor from a different server? - have no idea. Test for your self, you will learn a lot. It is difficult to tune motion detection, but it works well.
Currently I have problems with weather change - sudden shadows from objects, light/sun reflection, fast change in lighting (sun to cloud). I added some preclusive fast detect zones to manage this, but still have a lot of triggers caused by weather. Posted a question about that on this forum but no useable answer. Expected, as this is a free software, and developers are volunteers working on ZM in free time.