Subscriber Discussion

3MP And 5MP Low Light Performance Equal?

Mainstream megapixel cameras predominantly use 1/4" - 1/2" sensors in the 1-5MP range.

Most people will agree that 5MP cameras perform worse than 3MP cameras (which are now very much mainstream) in low light because the sensor size is typically around 1/3" for both resolutions which means the pixels on the 5MP sensor are much smaller and hence less sensitive to light.

However, after studying the datasheets of Aptina's main 5MP (MT9P006) and 3MP (AR0331) sensors for the surveillance market, one may draw a different conclusion:

MT9P006 (5MP, 1/2.5")

• OpticalFormat:1/2.5-inch

• Pixel Size: 2.2μm x 2.2μm

• Responsivity: 1.76V/lux-sec(550nm)


• OpticalFormat:1/3-inch

• Pixel Size: 2.2μm A-pix

• Responsivity: 1.9V/lux-sec(550nm)

Because Aptina has chosen to make the 5MP significantly larger at 1/2.5", the per pixel size between the two the 3MP and 5MP sensors are THE SAME. Now the 3MP sensor does on paper have a slight 6% sensitivity advantage at 1.9v vs 1/76v - but this is only because the AR0331 is 1 year newer and uses Aptina's next gen A-Pix pixel structure that allows more light to hit the pixel.

Now Omnivision's 5MP surveillance sensor is a horrendously small 1/3.2", and I believe they are used in the Avigilon (1/3.2") and Axis (1/3.2") products since that is the only major chip maker to make such a oddly small 5MP sensor. In this case there is little wonder why the low light performance is not stellar

However, Stardot, Arecont, Dahua, Hikvision, Dynacolor and many others are using the Aptina 1/2.5" sensor - so their low light performance may not be so much worse then the 3MP products.

The reason I am talking about this is because, in the market we cater for (full integrated all in one outdoor cameras with lens and IR illumination included) there is a shift away from Omnivision's 1/2.7" 2MP and Sony's 1/2.8" 2MP to the similarly priced Aptina AR0331 3MP - because the Aptina does 60FPS/double exposure 100db WDR and has similar light sensitivity - so the shift to 3MP will occur for economical (lower cost per megapixel) and functionality (WDR!) reasons.

So with the move of the mainstream product to 2.2μm pixels while maintaining the same low light performance one begins to wonder if it would be wise to start investigating the use of 5MP 1/2.5" products which potentially have the same low light performance.

Has anyone deployed these 1/2.5" sensor 5MP cameras in the field as found the low light performance to be surprising acceptable?

[IPVM Editor's Note: A claim about IPVM test results has been deleted because it is inaccurate.]

Yes I know 1.3MP 1/3" sensors are still king for night vision but 5MP does have its merits.

Bohan, I removed your reference to our 2009 Stardot test as validation of your thesis. Most certainly, that 2009 model was not using the imager you are referring to here. Also, that test was not a shootout and should not be used for comparisons.

Ultimately, a test needs to be done as speculating on the impact of pixel pitch, given all the other factors involved, is risky.

Among other things, differences in gain control / enhancement (like Axis's lightfinder) and lens f-stop (some 5MP use f/1.2, others use f/1.4 or even f/1.6) will make a difference.

We are open to doing a shootout but we should look at what specific models to test head to head.