The Importance of Megapixel Lenses

Published Feb 22, 2009 00:00 AM
PUBLIC - This article does not require an IPVM subscription. Feel free to share.

Selecting the right megapixel lens is as confusing as it is important. The wrong choice of lens can make an expensive megapixel camera no better than a cheap analog camera. Unfortunately, manufacturers generally do not provide sufficient information on their lenses and general educational information on lenses is scarce.

[UPDATE: October 2011: We have completed the first ever Megapixel lens shootout comparing performance of Computar, Evetar, Fujinon, Tamron and Tuss lenses.]

Here are the key issues with megapixel lenses:
  • Many systems designers are not used to evaluating lens details because lens quality was generally not a significant issue with standard definition cameras.
  • Manufacturers now label their lenses "megapixel lenses" but generally do not disclose a quantitative measure of how 'megapixel' their lenses are
  • Line Pair per millimeter (LP/MM) is a widely accepted quantitative measurement of the resolving power of a lens; however it is rarely disclosed
  • Even with a given LP/MM, the performance of a megapixel camera still varies depending on the size of the camera's pixels
Background Information
In this report, I draw on a number of excellent papers. I recommend you read these papers:
  • General Overview of Lenses [link no longer available]: A nice tutorial by Gizmodo on why lenses are important and how they work (hat tip to Ari)
  • Quality Criteria of Lenses: This paper provides a technical introduction to the key factors that impact lens generally
  • How to Choose a (Megapixel) CCTV Lens: A practical tutorial that walks through the specific challenges in selecting megapixel lenses
  • A Visual Introduction to Lens Quality and MTF: You should read this article if you understand MTF or if you already understand MTF. This article is important to provide additional depth on evaluating lens quality.
  • Axis Camera Comparison with Different Lenses: This post shares pictures from the Axis 211m using different lenses. While it does not cite the technical differences between the lenses, it does demonstrate the visual impact lens changes can make.

Not A Traditional Concern

When you are recording video at CIF or lower (which most of the world's surveillance cameras probably do today), any stock lens is likely to be good enough. For many years, limitations in analog transmission technology and storage placed a natural constraint on the quality of lens needed.

While a designer might be concerned about the range of focus for a varifocal lens or whether or not a lens was day/night capable, the resolution of a lens was not normally a critical issue to consider.

With megapixel, you suddenly switch from a world where 320 horizontal pixels was the norm to where 2000 horizontal pixels (a 3MP camera) is not uncommon. A 200-700% increase makes lens quality suddenly much more important.

Manufacturer's Technical Information is Unhelpful

Publicly disclosed technical information on lenses generally lack a quantitative measure that designers can verify. For instance, when you buy a megapixel camera, you always know how many megapixels the camera is (1, 1.3, 2, 3, 5 MP, etc.). With lenses, there is rarely any equivalent resolution stated. Worse, many manufacturers have started to label their lenses with big 'Mp' logos. This may help in differentiating it from a standard definition lens but it does little to identify how 'megapixel' it is.

However, manufacturers certainly have this information internally. It is not simply an issue that it does not exist. 

Part of the problem is that a lack of technical disclosure can help manufacturers cut corners and make their products seem better than what they are. (I am not saying that all manufacturers are doing this but it must be at least a partial factor for some manufacturers).

LP/MM is a Widely Accepted Technical Measure

Line pairs per millimeter is the widely accepted quantitative measure for the resolution of a lens. Basically, this metric describes how tiny the distance betweens lines can be for a lens to transmit. See this article for a good visual demonstration of this.

From what I can tell, standard definition cameras need about 30 lp/mm where megapixel lenses need at least 60 lp/mm (or more depending on the amount of megapixels).

Another interesting and important issue is that the resolving power of a lens (the lp/mm) decreases towards the edge of the image. In other words, a lens may have 60 lp/mm in the center but only 30 lp/mm toward the edges.  Making this more confusing, the standard way of expressing lp/mm in the surveillance industry seems to be quoting the lp/mm in the center of the image. As such, someone may tell you a lens is has a lp/mm for 1.3MP but it may only be capable of that resolution in the center of the image. On the exterior, the lp/mm may only be sufficient for standard definition imaging.

Lens Performance Impacted by the Size of Pixels

To make matters more confusing, the size of pixels for various megapixel cameras differ and that size impacts the selection of lenses. LP/MM expresses resolution for a given physical width (millimeters). However, the physical width of pixels in megapixel cameras varies because of different image size formats (1/3",1/2",2/3" etc.) and the number of pixels that need to fit on the imager (1.3, 2, 3, 5 MP etc.).

The practical importance is that the lp/mm metric is only relative to the size of the pixels on the camera. For example, a lens with 120 lp/mm might be great for a 1.3MP camera using a  1/2" imager but very poor for a 2MP camera using a 1/3" imager. 

This will be especially difficult to measure because you need to know both the size of a pixel for a given camera and the lp/mm for the lens - neither of which are generally provided.

Conclusion

Megapixel lens selection matters greatly yet megapixel technical disclosures are lacking for this element. I am looking forward to a discussion to help improve the level of eduction on this.