Image Sensor Size & Pixel Size Comparison Calculations

K

Kevin Doe

Pulling my weight
Jan 24, 2021
124
106
Ohio, USA
I was recently comparing several cameras for a location I was interested in, and I wanted to assess light sensitivity beyond what is just listed in the spec sheet. I did a rough calculation and thought others may find it interesting.

I calculated the sensor size using the number of vertical and horizontal pixels, and diagonal sensor size. All three sensors I was comparing were all 16:9 aspect ratio.

  • Dahua 1/2.8" 2MP - 4.45mm x 7.91mm, ~35 sq mm area. 0.000017 sqmm / pixel
  • Dahua 1/1.8" 4MP - 6.92mm x 12.30mm, ~85 sq mm area. 0.000021 sqmm / pixel
  • Dahua 1/1.2" 8MP - 10.38mm x 18.49mm, ~191 sq mm area. 0.000023 sqmm / pixel

We talk a lot about pixel size as it relates to the abililty to gather light. The larger the pixel size, the better it can capture light and the better the sensor will be at low light situations. Now I can see why the 1/1.2" sensor is so good. The pixel count went way up to 8MP, but it's physical size also went up considerably, and still resulted in larger pixels. Thoughts?
 
  • Like
Reactions: Treatwise, Lincos, Parley and 2 others
lane777smith said:
Any regrets on the IPC-color4k-x model? i haven't gotten mine yet, I'm excited to say the least.
Click to expand...

No, not specifically. I do have a few regrets, which I will be correcting at some point. My mounting locations and focal lengths aren't great. I went with 2.8mm in all locations, and 3.6 in just one. I should have went 3.6 in all of them, and went with a much longer focal length for the one where I put the 3.6mm. Many of my cams I also mounted too high. They're great for an overall picture, but not great for facial identification. One I just need to move down, and two others I need a longer focal length (those I may end up adding cams to get the zoomed in face ID I'm wanting). I just bought a 2MP 5-60mm (IPC-HFW5241E-Z12E) varifocal to zoom in close on a parked car on the side of my house. I plan on running that thing into like 50mm to get a tight view of the car. Based on a few calculations and the IVPM site, I will have more pixels across someone's face vs my Color4K cameras (based just on the field of view). Will be interesting to see the video quality comparison.
 
  • Like
Reactions: mat200 and sebastiantombs
I went with a 3.6mm and think my placement is pretty simple and straight forward at only about 10 feet high in the corner of a 50ft x 50ft square covering my covered drive and front yard

1654888341158.png
 
  • Like
Reactions: mat200
Kevin Doe said:
  • Dahua 1/2.8" 2MP - 4.45mm x 7.91mm, ~35 sq mm area. 0.000017 sqmm / pixel
  • Dahua 1/1.8" 4MP - 6.92mm x 12.30mm, ~85 sq mm area. 0.000021 sqmm / pixel
  • Dahua 1/1.2" 8MP - 10.38mm x 18.49mm, ~191 sq mm area. 0.000023 sqmm / pixel
Click to expand...
Those stated sensor sizes are somewhat fictional. For example, 1/1.8" is 0.555...5", or about 14.11 mm. The actual type 1/1.8 IMX347 sensor diagonal size is 9.04mm.

According to some Sony data, the pixel size of the 2 Mpix 1/2.8, 4 Mpix 1/1.8, and 8 Mpix 1/1.2 sensors is the same.
Products｜Image Sensor：Security｜Products｜Sony Semiconductor Solutions Group

As an approximation, the actual sensor diagonal size is (camera spec size * 0.667)
The good news is that the sensor size comparisons using the incorrect camera specs are approximately correct proportionally, just not as absolute numbers.

IMO, the sensor and pixel size comparisons don't tell the whole story, as different sensors can have different light sensitivity. For example, Dahua's 5442 series with the IMX347 sensor has much better low light sensitivity than the 4231/5231 series with the IMX290 sensor, all having the same pixel size (2.9 um x 2.9 um).
 
  • Like
Reactions: bigredfish, samplenhold, Griswalduk and 5 others
I don't analyze pixel sizes, just compare the sensor sizes to "everyday" objects.

Disclaimer - These sizes are what the manufacturers advertise and may, or may not, be the true size of the sensor in the camera.
1/3" = .333" Great for 720P
1/2.8" = .357" (think a .38 caliber bullet) Great for 2MP
1/1.8" = .555" (bigger than a .50 caliber bullet or ball) Great for 4MP
1/1.2" = .833" (bigger than a 20mm chain gun round) Great for 8MP
 
  • Like
Reactions: bigredfish, Griswalduk, Parley and 1 other person
tigerwillow1 said:
Those stated sensor sizes are somewhat fictional. For example, 1/1.8" is 0.555...5", or about 14.11 mm. The actual type 1/1.8 IMX347 sensor diagonal size is 9.04mm.

According to some Sony data, the pixel size of the 2 Mpix 1/2.8, 4 Mpix 1/1.8, and 8 Mpix 1/1.2 sensors is the same.
Products｜Image Sensor：Security｜Products｜Sony Semiconductor Solutions Group

As an approximation, the actual sensor diagonal size is (camera spec size * 0.667)
The good news is that the sensor size comparisons using the incorrect camera specs are approximately correct proportionally, just not as absolute numbers.

IMO, the sensor and pixel size comparisons don't tell the whole story, as different sensors can have different light sensitivity. For example, Dahua's 5442 series with the IMX347 sensor has much better low light sensitivity than the 4231/5231 series with the IMX290 sensor, all having the same pixel size (2.9 um x 2.9 um).
Click to expand...

Thanks for the reply. My math checked out ok, but I didn't realize the sensor size was just notional. Like you said though, nearly the same relative result. Thanks for the links!
 
  • Like
Reactions: tigerwillow1
Kevin Doe said:
My math checked out ok
Click to expand...
I fact-checked a little of the math. I was a trig whiz in high school and could have belted out all of the answers almost instantly. Now 50+ years later when I tried to do it my brain started hurting and I cheated with an online calculator.
 
  • Like
  • Haha
Reactions: Lincos, looney2ns, xplorer and 5 others
The "inch sensor sizes" are called optical format - they're the outside diameter of a vacuum image tube with a similar image area. Easy to imagine how that's not exactly a precise way to put things. So the exact size of a 1/3" sensor varies a little bit. Wikipedia has a list of some common sizes: Image sensor format - Wikipedia

If you go to Sony Semicon's site, they have a chart of their STARVIS sensors with factsheets: Products｜Image Sensor：Security｜Products｜Sony Semiconductor Solutions Group

The real exact size of the active area is in those, but they also just tell you the exact pixel size (pitch). For example, IMX585 has a 12.84 mm diagonal, and a 2.9 µm pitch. So pixel area is simply 2.9² = 8.4 µm² - no trig required :)

Not all Dahua/Hik cameras use STARVIS sensors though, but for those that do it's pretty straightforward to find more-or-less the model they have to be using.

Light sensitivity claims are often made using idiosyncratic conditions. Like "1/30s 30 IRE" meaning that full-white in their image will be just 30% brightness and 1/30s is of course way too slow a shutter speed for most applications. Also they're made using a particular lens f-stop, but light transmission varies between lenses. E.g. a simple fixed focal lens has fewer elements than a complicated vari-focal lens with a huge zoom range, so if both are f/1.6, the fixed focal might be T/1.8 and the big vari-focal could be T/2 - a 25 % difference in the amount of light reaching the sensor.
 
  • Like
Reactions: Lincos, alastairstevenson, bigredfish and 2 others
Since this thread has migrated from the original topic to sensor size let me ask the following.

What would be the FOV, both horz/vert for a given lens focal length for lets say a 6mm IP camera lens with different sensor sizes?

If this same lens is used on a camera with sensor sizes of 1/2.8 and 1/1.8 and 1/1.2 then the resultant image size should be different. Correct? But is the difference large enough to be noticeable to the end user?

What would be a good number to use as a multiplication factor to determine the actual FOV (for a given lens) for each of these different sensor sizes?

From a 35mm film camera perspective, the FOV is different when based on sensor size (crop factor). A full frame digital vs a camera with a crop factor of 1.3 x would result in a 50mm lens (full frame) becoming a 65mm lens when used with a smaller sensor (1.3 X).

Would this also not be the same with the sensor size for an IP cam? Thus one could not actually calculate how wide an image is at 100 feet given the lens focal length unless the sensor size is known. Is this correct?

Trying to clarify the overall issue in that a 3.8mm or other lens will not present the same FOV due to sensor size differences.
 
Dahua and friends seem to quote the focal lengths for their fixed-lens cameras in roughly 1/2" equivalent focal lengths. So 2.8 mm is around 100-110° H FoV, 3.6 mm is around 80-90° H FoV etc.

I guess that's because 1/2" CCDs were common in the analog days and so that became the reference point for focal lengths, just like 35mm for photographers.

A "3.6 mm" (1/2" equivalent, ~85°H FoV) lens marketed on a 1/1.2" sensor camera has an actual focal length of around 6 mm (Wikipedia's table).
 
  • Like
Reactions: Parley and sebastiantombs
The only reliable way I've found to get a camera's FOV is to measure it myself, or pull it from the camera's spec sheet, which has been pretty close for a few cameras that I measured.
 
  • Like
Reactions: samplenhold, Parley and sebastiantombs
Alaska Country said:
If this same lens is used on a camera with sensor sizes of 1/2.8 and 1/1.8 and 1/1.2 then the resultant image size should be different. Correct? But is the difference large enough to be noticeable to the end user?
Click to expand...
See this thread.

Comparison of a Fixed Lens and a Varifocal Lens Camera's FOV

I was testing out a new Dahua T5442T-ZE 2.7-12mm Varifocal camera (from @EMPIRETECANDY ) and was going to compare it's performance to the Dahua HDW5231R-ZE and the HDW5431R-ZE cameras, which are 2.7-13.5mm varifocals, along with the T5442TM-AS fixed lens at 6mm camera. The HDW5431R-ZE has been...
ipcamtalk.com ipcamtalk.com
 
  • Like
Reactions: Alaska Country and sebastiantombs
Kevin Doe said:
No, not specifically. I do have a few regrets, which I will be correcting at some point. My mounting locations and focal lengths aren't great. I went with 2.8mm in all locations, and 3.6 in just one. I should have went 3.6 in all of them, and went with a much longer focal length for the one where I put the 3.6mm. Many of my cams I also mounted too high. They're great for an overall picture, but not great for facial identification. One I just need to move down, and two others I need a longer focal length (those I may end up adding cams to get the zoomed in face ID I'm wanting). I just bought a 2MP 5-60mm (IPC-HFW5241E-Z12E) varifocal to zoom in close on a parked car on the side of my house. I plan on running that thing into like 50mm to get a tight view of the car. Based on a few calculations and the IVPM site, I will have more pixels across someone's face vs my Color4K cameras (based just on the field of view). Will be interesting to see the video quality comparison.
Click to expand...

@EMPIRETECANDY said the 4K/X will have a 6mm option available in about a month!!!
 
  • Like
Reactions: Griswalduk and sebastiantombs
One more thing on the 4K cameras is that they will soon over power an NVR. My older NVR's would not handle that many 4K cameras even at 15fps or less. I now have a total of 20 cameras and 4 NVR's with 9 of those cameras being 4k. If I add more 4K cameras I may have to upgrade and NVR or two. Here is the latest specs from Hikvision. So by lowering the FPS to 15 or less it helps the situation.

2-ch@12 MP (20 fps)/4-ch@8 MP (25 fps)/8-ch@4 MP (30 fps)/16-ch@1080p (30 fps)
 
samplenhold said:
See this thread.

Comparison of a Fixed Lens and a Varifocal Lens Camera's FOV

I was testing out a new Dahua T5442T-ZE 2.7-12mm Varifocal camera (from @EMPIRETECANDY ) and was going to compare it's performance to the Dahua HDW5231R-ZE and the HDW5431R-ZE cameras, which are 2.7-13.5mm varifocals, along with the T5442TM-AS fixed lens at 6mm camera. The HDW5431R-ZE has been...
ipcamtalk.com ipcamtalk.com
Click to expand...

A very nice read. Thanks for posting the link.

It would appear that there is no reasonable way to determine a lens focal length due to too many unknown or proprietary variables.

My goal is to add this data to a spreadsheet that would determine the focal length based on measuring camera lens distance to a fixed object. Knowing the width of the image in pixels and the width of the measured object in pixels one can determine the Horizontal Angle of View, Vertical Angle of View, Diagonal Angle of View and Field of View. From this point a fudge factor could be applied to calculate the focal length. However, as pointed out, determining the exact lens focal length does not seem possible or repeatable using different image sensors.

Based the spreadsheet on the work done by Mr Charles Lohr.
Calculation of Horizontal Angle Of View & Field Of View
www.chucklohr.com/808/AOV
 
Last edited:
  • Like
Reactions: sebastiantombs and EMPIRETECANDY
You must log in or register to reply here.
Top Bottom