Is thermal right for me?

[Dis7]

Long-time lurker and appreciate the incredible knowledge shared here. Due to recent crime, I am working on upgrading my family's home security, including cameras. After a lot of planning I am thinking about adding thermal cameras as well, but would appreciate advice on whether they'd be good for me.

To give an overview, my house is separated from others by a few hundred feet on both sides. There is some flood lighting in between. But in the back there is nothing behind but a fence up to 200' away. There is open farmland beyond that with no lighting at all anywhere behind the house.

I have invested into some cameras including a couple color4K, but mostly T54IR ZEs, ASs, etc. The approach taken was mainly due to focal lengths, pixel sizes, min focus range, etc, for the various points around the house from which someone might pose a threat or need to be identified.

However, I found that none of them will do a very good job for Overview at night back there. For example, with half a moon in the sky, the Color 4Ks can't seem to track someone 75 feet away even with 16.67ms shutter. I know the other cameras have IR but their IR doesn't seem to get out that far with sufficient coverage.

All this has me thinking that maybe thermal might be good for my use case. Specifically: 2x IPC-TPC124X-AI in 2mm for ~170º Overview. The idea would be to know if someone is crossing the fence to get within an area 20-200' behind the house. My thought was that the other cameras (mainly T54IR ZE) would come into play if they got closer and they (ZEs) would Identify if the person got close enough.

Anyway Im not experienced with thermal so I don't know if this strategy makes sense. If it did though, and the thermal cameras could do what I wanted and the range/resolution/thermal sensitivity is as advertised, what about the ability to observe/recognize? Or would I likely be happy with the setup as described? Also, on a tangent, would the visual side of the IPC-TPC124X-AI be sufficient for Overview observe/recognize in the day?

Any advice would be appreciated.

 

[luk8899]

I was in similar circumstances and somewhat similar environment.

I bought a thermal camera with a 3.5mm lens and it is on its way back. The answer for me is firmly no.

Why not? Because I stumbled on a visible light (and ir) camera that comfortably sees about 60m (~180ft) in complete darkness with its own Illumination. How come? Because it has a huge lens (1in+ or 30mm). See the this thumbnail. This is a moonless, fully overcast night where I can see nothing despite accomodating my eyes for few minutes. But this camera sees everything it needs. This is near IR, not thermal (the camera is mounted fairly high up at about 8m elevation, this thumbnail view is with 1x zoom).

The furthest down the road is about 60m (180ft) and I can pan it around.

The thermal Dahua I bought was ok-ish. It too would see a human at comparable distance, but not much further. The quality of the picture was orders of magnitude worse.

A month or so ago I bought a different USB C thermal imager for my phone. It is xinfrared x02. It has a big 11mm lens and when I saw pictures it produces it I immediately thought. I want this in one of my cctv cameras. Few examples of what it can see:
A car 170m away (510ft)

A bunch of deer 300m (1200ft) away:


I can easily spot my cat at night from 100m with it and it has little felt latency. So when I got my Thermal dahua (that seemed to have the same 256x192 12um pixel microbolometer sensor) I expected similar performance (even better with regards of detection as it has a lens of f1.0 and my xinfrared is f9.0), but no. My x2 can comfortably see my thermal footprints on a ceramic tile floor for good 10s after I walked over it(in socks!). The dahua camera I had (TPC-DF1241-D3F4) couldn't see any footprint immediately after I lifted my foot regardless of settings. This means, for some reason this camera is not using its sensor to its potential. Would it detect a human better than the near IR I show on the top in fog? Yes, but it would be just a blob of light. If you need to reliably detect humans in fog, when it rains/snows heavily, perhaps thermal is for you.

I'm staying with near IR. Possibly adding illuminators where necessary. The visual light camera I talked about is a generic $200 30x optical zoom imx415 4K "hk compatible" aliexpress camera. It is the model with no wiper and no visible illumination diodes, just one huge lens on the front.

 

[Phil.g00]

@luk8899 I would like to know more details about this alternative model.

 

[wittaj]

@luk8899 - what does motion look like on that first camera? I can make any camera look like noon at midnight with slower shutters and higher gain, but motion is a complete ghost blur.

And comparing a 30mm optical zoom lens to a 2,3.5, or 7mm lens isn't a fair comparison. Nobody here would say that is a fair comparison.

And yes that USB C thermal imager for your phone looks cool, but how long will it last outside in 24/7 weather LOL?

Most here don't buy a thermal camera for the outstanding picture quality - it just doesn't happen.

It is used to pick up a heat signature and see things the visual camera cannot. And I assure you that first image you have someone could sneak thru that without the camera triggering under the right conditions.

 

[luk8899]

@luk8899 I would like to know more details about this alternative model.

Do you mean the visible light camera, or the thermal usb c thermal imager? The optical/near IR camera I bought is this one:

https://www.aliexpress.com/item/1005005537996405.html

Many sellers sell them under various names. The key thing for the large lens is that they say it has 30x optical zoom and that it is Hikvision compatible.

The thermal imager is not a standalone device, it plugs into a mobile phone. I have this one:

https://www.amazon.pl/dp/B0CN2ZHKQ4

Sorry for the polish language amazon link, but for some reason I have trouble finding it on the English language site, perhaps because of my location it only shows me ones that ship here. Hopefully if you copy the relevant part of the title "Android USB-C Xinfrared One XH09" and search for it on amazon, it will find one for you.

 

[luk8899]

@luk8899 - what does motion look like on that first camera? I can make any camera look like noon at midnight with slower shutters and higher gain, but motion is a complete ghost blur.

Surprisingly OK-ish. My cat walking around looks perfectly normal(I haven't seen it running fast yet). When I use the PTZ, there is signifficant motion blur during motion, when I stop it clears up very quickly. Zoom on the other hand can take few seconds to acquire sharp image, especially at high zoom levels.

And comparing a 30mm optical zoom lens to a 2,3.5, or 7mm lens isn't a fair comparison. Nobody here would say that is a fair comparison.

Of course. I'm just showing it from a practical sense. I was really surprised how well the near IR picture looked. I also tried finding a bigger replacement germanium lens for the dahua thermal camera(it can take a 9mm M12 germanium lens).... lets say its easier to buy a complete thermal imager and scavenge a lens from it, than to buy an optical quality thermal imager lens online for reasonable price.

And yes that USB C thermal imager for your phone looks cool, but how long will it last outside in 24/7 weather LOL?

It is IP65 rated... I'm not sure how, considering the USB C port, but still, they claim it is.

Most here don't buy a thermal camera for the outstanding picture quality - it just doesn't happen.

It is used to pick up a heat signature and see things the visual camera cannot.

Indeed, everyone has different circumstances and I'm not saying the thermal camera is not good at all. I'm just saying for me that dahua made no sense at all and what the op wrote resonated with me.

And I assure you that first image you have someone could sneak thru that without the camera triggering under the right conditions.

That's why it is not the only camera looking at that scene But, in principle, yes, heavy snow fall or fog makes it useless. It all depends on what does one need.

My primary purpose for the thermal camera wasn't just detection (I would have to buy 6 of them to cover my perimeter), but situational awarness. I planned to put it in a PTZ enclosure to look around, hoping to see what's going on if there is activity on nearby fields in a middle of a night. But if I can't tell a human from a deer at 150m that's not good enough for me.

 

[bigredfish]

Thats a 5 year old 1/2,8" sensor on a 4K camera.

Sony to Release Two Types of 4K-Resolution CMOS Image Sensors for Security Cameras: World's Smallest*1 type 1/2.8 and Enhanced Low-Light Performance Type

Tokyo, Japan — Sony Corporation today announced the upcoming release of two new models of CMOS image sensors: the IMX415 type 1/2.8 4K-resolution stacked CMOS image sensor, which is the smallest of its kind in the world*1; and the IMX485 type 1/1.2 4K-resolution back-illuminated CMOS image...

www.sony.com

Many of us have 1/1.8" sensor 4MP PTZs (much better sensor to MP ratio) that have no problem seeing that far and further at night with IR. The proof is in the actual video of moving objects.

While I dont have a thermal cam, I can see how they would be beneficial in a rural scene as an overview means of detecting critters or people, coupled with a high quality (not that one) PTZ with powerful zoom that can be activated/directed by the thermal cam

 

[wittaj]

I was interested in the thermal that connects to a cellphone until I saw the price LOL ($369USD)

https://www.amazon.com/Xinfrared-Monocular-Resolution-Inspection-Wilderness/dp/B0CHRVJMJ7?th=1

 

[luk8899]

Thats a 5 year old 1/2,8" sensor on a 4K camera.

Sony to Release Two Types of 4K-Resolution CMOS Image Sensors for Security Cameras: World's Smallest*1 type 1/2.8 and Enhanced Low-Light Performance Type

Tokyo, Japan — Sony Corporation today announced the upcoming release of two new models of CMOS image sensors: the IMX415 type 1/2.8 4K-resolution stacked CMOS image sensor, which is the smallest of its kind in the world*1; and the IMX485 type 1/1.2 4K-resolution back-illuminated CMOS image...

www.sony.com

Many of us have 1/1.8" sensor 4MP PTZs (much better sensor to MP ratio) that have no problem seeing that far and further at night with IR. The proof is in the actual video of moving objects.

While I dont have a thermal cam, I can see how they would be beneficial in a rural scene as an overview means of detecting critters or people, coupled with a high quality (not that one) PTZ with powerful zoom that can be activated/directed by the thermal cam

Pictures would be nice... (prefferably on a dark night)

But in principle, yes, I didn't even buy that 30x zoom camera for night vision. It's main purpose is the 30x optical zoom, nice night vision is a bonus. I'd love to see some comparison pictures. What is the best lens/sensor combination for wide angle visible/nearIR night vision today?

I was interested in the thermal that connects to a cellphone until I saw the price LOL ($369USD)

https://www.amazon.com/Xinfrared-Monocular-Resolution-Inspection-Wilderness/dp/B0CHRVJMJ7?th=1

Considering the 11mm lens and 50hz refresh rate... I say they would still sell many of them if they cost 10x the money (not to me however).

 

[bigredfish]

For 4MP you want the 1/1.8" sensor and for 4K you'd want the 1/1.2" sensor

I'll have to get some IR video later tonight. As I have plenty of visible light, I run color at night mostly. There are PLENTY of folks here who have similar Dahua 1/1.8" PTZs running in IR at night at 300-500+ ft

45X optical zoom picks them up at about 500ft
View attachment HOA Rear_ch1_20211005111115_20211005111251.mp4

 

[wittaj]

This chart was developed based on the collective input of many members here with real world experience with many/all of the different cameras out there.

You also will see a price correlation between those in red (cheaper) versus those in green (more expensive than those in red).

Generally the larger the sensor, the more light that can be captured per pixel.

So we look for MP/sensor ratios that are in the green:



So your 8MP camera on the 1/2.8" sensor means it needs over 4 times the light a 2MP on the same size sensor needs.

That tells me that your first image is a slowed down shutter like maybe 1/3 to 1/12 with a high gain that will result in motion blur of a person or car.

 

[bigredfish]

Here's an example of 4MP cameras side by side with 1/2.8" sensor and 1/1.8" sensor. It (1/2.8") will by much worse on a 4K camera
(Hint: the smaller number on the right after the / is the bigger sensor)

Both are running 1/120 exposure to eliminate motion blur. I can make them MUCH brighter, cool trick, but motion will be a blurry mess.
What exposure are you running? It matters

1/2.8 on left >>> 1/1.8" sensor on right


And video
1/2.8"
View attachment HOA Entr_ch1_20201220200103_20201220200114.mp4













1/1.8" sensor
View attachment HOA Entr_ch4_20201220200102_20201220200115.mp4

 

[bigredfish]

^^^^^
look at them full screen on a real monitor. Postage stamp size images on a phone don’t show the real difference

 

[wittaj]

To get back to the OP question.

You are correct - for a visual camera to detect at 75 feet for a 2.8 or 3.6mm fixed lens will be spotty if there isn't enough light if you are running shutter speeds intended to minimize blur.

If you want that camera to be an overview camera to get to 75 feet, you need to slow the shutter down to let more light in, recognizing that comes at the expense of not getting clean captures.

You can't use a camera to IDENTIFY at close range and also be a reasonable overview at any distance with a faster shutter. You gotta pick one or the other. Especially if light is lacking.

Many of us have found the DORI numbers to be way too large. Maybe under ideal conditions with a stationary object in the middle of the day.

I usually cut the DORI numbers in half during the day and cut that half number in half or 75% for night numbers.

Same with infrared distances - once you increase the shutter speed, those numbers get cut in at least half if not 75%.

So short of getting a varifocal optically zoomed to your pinch point, the thermal camera would be a reasonable choice to trigger for people movement back there.

 

[luk8899]

For 4MP you want the 1/1.8" sensor and for 4K you'd want the 1/1.2" sensor

I'll have to get some IR video later tonight. As I have plenty of visible light, I run color at night mostly. There are PLENTY of folks here who have similar Dahua 1/1.8" PTZs running in IR at night at 300-500+ ft

45X optical zoom picks them up at about 500ft
View attachment 189988

What camera model is it? What are the lens parameters?

This chart was developed based on the collective input of many members here with real world experience with many/all of the different cameras out there.

You also will see a price correlation between those in red (cheaper) versus those in green (more expensive than those in red).

Generally the larger the sensor, the more light that can be captured per pixel.

So we look for MP/sensor ratios that are in the green:

View attachment 189995

Thanks, the chart could use sensor types too. Not every sensor of the same pixel size will have same performance. I fly FPV drones(mostly analog/hdzero) and there are cameras with the same lens, resolution and sensor size, but one is unusable in the dark, while another is amazing. Knowing what sensors to look for would be nice.

So your 8MP camera on the 1/2.8" sensor means it needs over 4 times the light a 2MP on the same size sensor needs.

If we assume both sensors are using the same tech . That's why seeing sensor types on this chart would be useful. Pixel size is not all there is (at the lower end especially)

That tells me that your first image is a slowed down shutter like maybe 1/3 to 1/12 with a high gain that will result in motion blur of a person or car.

1/3? Definitely not. I've had it for a couple of weeks. In this time there was dust, snow and rain in front. at 1/3s this would turn into a blurry mess, but it didn't. We are talking about a varifocal lens with an objective about an inch in diameter. I have a number of other imx415 cameras and this one is signifficantly better thanks to its lens.

I'll play with the shutter speeds tonight (they have been on auto ) and we'll se how it looks at various settings.

Here's an example of 4MP cameras side by side with 1/2.8" sensor and 1/1.8" sensor. It (1/2.8") will by much worse on a 4K camera
(Hint: the smaller number on the right after the / is the bigger sensor)

Both are running 1/120 exposure to eliminate motion blur. I can make them MUCH brighter, cool trick, but motion will be a blurry mess.
What exposure are you running? It matters

Auto.... I'll do some experiments tonight.

 

[wittaj]

Thanks, the chart could use sensor types too. Not every sensor of the same pixel size will have same performance. I fly FPV drones(mostly analog/hdzero) and there are cameras with the same lens, resolution and sensor size, but one is unusable in the dark, while another is amazing. Knowing what sensors to look for would be nice.


If we assume both sensors are using the same tech . That's why seeing sensor types on this chart would be useful. Pixel size is not all there is (at the lower end especially)



1/3? Definitely not. I've had it for a couple of weeks. In this time there was dust, snow and rain in front. at 1/3s this would turn into a blurry mess, but it didn't. We are talking about a varifocal lens with an objective about an inch in diameter. I have a number of other cameras and this one is signifficantly better thanks to its lens.

I'll play with the shutter speeds tonight (they have been on auto ) and we'll se how it looks at various settings.


Auto.... I'll do some experiments tonight.

True, but comparing the sensors of these types of cameras to drones or iphones or any other device isn't comparing apples to apples. The optics in an iphone are much better than any of these cameras.

You are mistaking the benefit of varifocal and its ability to OPTICALLY zoom in on an object compared to a digital zoom of a fixed lens camera .

The imx415 sensor is a fairly common sensor used. Comparing a 2.8mm focal length to a 30mm focal length on that same sensor isn't a fair comparison. The 2.8mm strength will be within 15 feet or so whereas the 30mm strength is about 60 feet or so. Then the firmware plays an even important role.


Sure sensor size/tech plays a role, but in our experiences, there really are not a lot of competition with these types of sensors as it relates to what manufacturers are putting into these types of cameras, so generally the sensor size is a sufficient factor one can use to determine the quality of the image the camera will produce.

The MP/sensor ratio is part of the equation and the firmware is the rest of the equation.

Generally, the cameras that are on what we call ideal MP/sensor ratios tend to be better cameras with better firmware. But they all don't perform the same. Between Hikvision, Dahua and Axis, they would be comparable. Some cheap no name probably not.

But back in the day Reolink got in the mix and had a 2MP - 1/2.8" starlight sensor that showed how important the firmware is. A great sensor on poor firmware results in poor performance.

So they had the same sensor as the top of the line Dahua in the day, but this is an example from Reolink's marketing videos of their Starlight camera - do you see a person in this picture...yes, there is a person in this picture. Will give you a hint - the person is in between the two columns:

Their firmware is designed to give a nice bright static image at night. That comes at the cost of poor performance with an object in motion. And the user can do nothing to correct this even with settings because it has been shown that Reolink (and most consumer grade cameras) favor nice bright static images at night over performance. So at some point even if you can set shutter settings, the camera will override your input in favor of a nice bright image. This is done by slowing down the shutter and increasing the gain. So then you see what Reolinks are notorious for - ghost blur invisible person images at night and inability to capture plates.

So the difference between a better camera like say a Dahua and a Reolink or some no-name camera on Amazon with a Starlight sensor is that you can set parameters on the Dahua and it will hold. If you set parameters on these other cameras that would result in a darker image the algorithm internally says "idiot alert" and it won't let you set parameters that the firmware thinks will result in not displaying a nice bright image. Don't believe me, set the shutter to 1/10,000 at night and the image should be completely black. It won't with the reolink...or any cheap camera. It will override your 1/10,000 shutter and favor a bright image. It is a good test to determine how good the camera is.

But most consumer grade camera manufacturers know that consumers chase MP, so to keep costs down, they will put 1/3" or 1/2.8" sensors in the cameras. And 8MP on a 1/3" sensor looks great on default settings for a static image at night....and that is what the consumer looks at. No consumer then tests it with motion.

It is always comical to me when a neighbor buys a camera and asks me to help them set it up. I set the shutter to 1/60 or 1/120 and the image gets darker (because the faster the shutter, the more light that is needed but the faster shutter is what gets clean captures with motion). And they always complain the image is too dark. So at some point they go back to auto settings and then capture a blur motion of the perp that door checked their car.


And now Reolink has done it with the 1/1.8" sensor.

The CX410 is one such camera. Let's take a look.

While it is on the proper MP/sensor ratio (4MP on a 1/1.8" sensor) with a large f1.0 aperture, at $95 it would be a steal if it worked like other full color cameras.

Sadly, it appears to suffer from the same issues of reolink, which is poor night quality and ghosting, etc. And look how washed out the camera on the right is when the lights kick on.

Is missing a leg normal?

Are these captures that could IDENTIFY a person? Looks like the person is close enough that a Dahua could provide IDENTIFY quality.

But it benefits from the light turning on with motion. Wonder if the camera gets momentarily blinded.

So are there exceptions, yes, but it takes having experience with these types of cameras and the various manufacturers to then know if you are getting a camera that can perform.


I assure you on auto/default settings of your camera, that shutter speed is slow - maybe not 1/3s slow, but slower than most of us would run a camera.

At night do the following:

• Make it 1/60 (16.67ms) and post the same field of view - I bet it will be a lot darker.
• Then make it 1/120 (8.33ms) and post the same field of view. It should be even darker
• Then make it 1/10,000 and that image should be completely black - if it isn't then you know the firmware is playing with the image and at that point it doesn't matter how good the sensor is in the camera.

As a comparison, here is a shot from my PTZ at full zoom of ole glory 900 feet away riding the 20MPH wind gusts with a 1/60 shutter where the only light the camera is getting is from the single flag lamp lighting up the flag:

 

[bigredfish]

Willing to bet "auto" is 1/30 like everyone else

 

[bigredfish]

Hey @wittaj do you have those samples of your thermal testing? I thought they were quite interesting for the OP to see how far out you can detect movement?

Seems using a thermal to direct a good PTZ might be a good combo for detecting perimeter intrusions?

 

[wittaj]

Hey @wittaj do you have those samples of your thermal testing? I thought they were quite interesting for the OP to see how far out you can detect movement?

Seems using a thermal to direct a good PTZ might be a good combo for detecting perimeter intrusions?

Yeah the thermal is a way better spotter cam than a visible camera, especially at night if you live in an area without much light.

Regardless of day or night, rain or snow, the thermal image looks the same.

Here was my thread on it:

Review - EmpireTech IPC-TPC124X-AI 1/2.7'' CMOS 4MP Thermal Network Mini Hybrid Eyeball Camera

As @EMPIRETECANDY pointed out in this thread, I have been playing with and testing the thermal camera IPC-TPC124X-AI. In particular, we wanted to see if they performed better than the previous thermal cameras that had proven unreliable if being used in mission critical situations and could...

ipcamtalk.com

Here was another test I did.

The white blob near top right is a person at 200 feet and it triggered the IVS rule that was for human only.

It stands a much better chance at DORI distances than a visible camera.

In spot checking about 20 people, the camera did trigger out at that distance.

Just keep in mind that at 200 feet it will only be a little blob, but it is great at seeing people/movement in complete darkness compared to the visible camera with a 1/120 shutter.

 

[Dis7]

I was in similar circumstances and somewhat similar environment.
...
I'm staying with near IR. Possibly adding illuminators where necessary. The visual light camera I talked about is a generic $200 30x optical zoom imx415 4K "hk compatible" aliexpress camera. It is the model with no wiper and no visible illumination diodes, just one huge lens on the front.

Thanks for sharing your experience with thermal. Your comment led me down a rabbit hole on thermal sensors, and one thing I noticed was that the TPC124X-AI-S2 (maybe the same as TPC-DF1241-S2?), seems to have different advertised sensitivity vs the DH-TPC-DF1241-D3F4. Specifically the former is rated with a Noise Equivalent Temperature DifferenceI of 40mK while the latter is rated at 50mK. I wonder if that's meaningful, or reflective of a more broad difference. Coming back to your recommendation, I'm not sure if PTZ would be good in my specific case. I'm thinking the alternative would be adding IR emitters?

To get back to the OP question...

Thank you for the guidance! Because there's a porch my plan was actually to use the variofocals near the rear entries and windows for identify. Then, use additional cameras for overview to see further out. That's where I thought that maybe the Color4Ks could fit that overview role. But once I saw the image at 16ms, I knew it wasn't going to work out.

I've read (and reread!) your prior review of the TPC124X-AI-S2 and had a couple questions. Wisdom here seems to be that you can't really trust DORI distances on spec sheets. That makes sense. But you've referenced being able to trigger on a person at over 400 feet with the 3.5mm version while the spec sheet says it can do something like 479 feet. Does that mean that the thermal performance is close to what's on paper? Related- the thermal perimeter distance for a human is specified as a completely different number - 80 feet. That appears to be very different vs all of your testing, but I'm not sure what the difference is. What am I missing?