View Full Version : lens resolution

3-May-2005, 19:36
In a the March 2005 issue of Scientific American, pages 94-95, there is a quote:

"MEGAPIXEL MAX: A so-called four megapixel camera has a CCD with 4 million pixels. The Higher this resolution, the sharper the photograph. But affordable lens systems cannot resolve light into more than six or eight megapixels, Kodaks Gary Hallenbeck says. So consumers touting 10 to 12 megapixel sensors will not create an overall sharper image. The higher count does help make images clearer when using digital zoom."

So, my questions are: If lenses cant resolve more clearly than the equivalent of 6 - 8 megapixels, then why does large format do soooo much better?? Seems to me that medium format, 4x5 and anything larger can out perform anything the digital world can provide. Are the lenses we use for large format really so bad?? What about the newer medium format CCD digital cameras? doesn't this apply to the above quote also? And last, I think this topic directly applies to large format, as there are already LF dig backs out there.

3-May-2005, 19:52
Most of the articles you find equating resolution, megapixels, and sharpness are seriously confused.
For some good tutorials on image quality (put in the context of digital images and real world optics) check out:


Be prepared to spend some time ... as is often the case in these issues, the ansers aren't simple ones.

For what it's worth, some of the statements you're quoting are probably out of context, because you can easily get over 100 megapixels worth of useful image data from a good quality 4x5 neg. Your lenses are actually really good.

3-May-2005, 19:54
What Kodak considers "affordable lens systems" is probably the key to understanding his statement.

3-May-2005, 20:44
i'm not gonna pay a lot for this lens!

Will Strain
3-May-2005, 21:35
I've almost given up on Kodak's official opinions when it comes to digital anything... they may have worked to invent it... but the minute they decided that Wilhelm Researche's methods did not apply to their papers - they lost me.

The statement does not take into account the overall sensor size, and the individual photosite size, which should have more bearing on resolving factor than the number of them. True, 12 million in a 1/8" chip won't help anything... but in a full frame sensor, they are quite well utilized.

Now, I'm heading back to my 4x5 and Tmax...

4-May-2005, 05:14
From my experiance with the kinds of cameras that the quote is refering to (compact digital cameras), I would agree. Most lenses on current 5-7 mega-pixel compact cameras do not resolve enough to give any actual extra image information, despite the higher pixel count - an image from a 4mp camera is often just as good as an image from a 7mp when up-sized in Photoshop. But they are talking about VERY small sensors - 7x10mm or so in size. You could fit 180 of them on a 4x5 film sheet, so they are incredibly small.

LF resolves more in absolute image quality due to the side of the film, not the lenses. In fact many 35mm lenses have much higher resolution per line-pair then LF, but do not produce as clear an image for just that reason - LF just has so much more film area to resolve!

Leonard Evens
4-May-2005, 07:53
When discussing lens resolution, you have to look at pixels per unit length, not total number of pixels. Consider for example, my 2000 x 3000 Nikon D70, which has a frame just about 16 x 24 mm. That comes out to 125 pixels per mm. In terms of standard resolution figures, you have to divide by 2, to get about 63 lp/mm. That is a theoretical maximum and in practice, you will get somewhat less than that. It should not be hard to design a lens which attains that resolution, but since the resolution of a compound system is always lower than that of any component, you would probably have to resolve significantly higher in the lens to take advantage of what the sensor array can capture. If you increase the number of pixels while keeping the same frame size, you put progressively stronger constraints on lens resolution if the aim is to take full advantage of what the sensor array can deliver.

For 4 x 5, the arithmetic is more forgiving. Suppose you scan a 96 x 120 mm frame to produce an 8000 x 10,000 scan. That would be 80 Megapixels. But it would be only 10,000/120 ~ 83 pixels per mm. Half of that is about 42 lp mm. But good LF lenses can do rather better than that.

Perhaps a better way to think about this is to take the lens resolution as the basic limiting factor. Then the higher the number of pixels, the closer you will come to the lens resolution in the compound system. There are different ways to estimate the compound resolution. One way is to add the squares of the reciprocals of the components and take the square root of the reciprical of that. So, for example, if the resolution in lp/mm of the sensor array is twice that of the lens, the combined system would deliver approximately 90 percent of that of the lens. But since it is still pixels per unit length that counts, LF still has an advantage since you can get away with lower resolution lenses.

Janko Belaj
4-May-2005, 09:28
(as 1st, I hope I will be able to write in english what I'm thinking about this problem - resolution, pixels, calculation...)
In most cases people do forget one simple advantage of film over digital medium: positioning of silver halide crystals and positions of bayer-pattern (most used) ccd/cmos chip. (in my everyday job I use both photosensitive medias, so I'm not writing this to start some one-vs-other verbal fight)
In film emulsion (b&w and color) we are dealing with thousands of small crystals randomly layered in 3 dimensional space. one over the other, thousands over the other few thousands and we can not tell where one "ends" and where will the other "start". One resolved line from our test chart will be placed on several silver halide crystals. On any chip (except B&W 3 shot cameras with separate R, G, B filters and Foweon chip) one resolved line have to be placed on 2 raws of RG/GB pixels. To get reed of "light bumpers" each and every pixel have to be separated from his "neighbors" with something I will call electron-vacoom-cleaner (ouch... I hope you will get what I mean).
So, to understood why film can create better non-streat lines imagine that your film is sand on the beach and that chip is road of small stone bricks.

Neal Shields
4-May-2005, 12:12


I thought that the heights of mis-information on digital photography had been reached long ago. Obviously I was mistaken.

Alan Davenport
4-May-2005, 13:45
... affordable lens systems cannot resolve light into more than six or eight megapixels, Kodaks Gary Hallenbeck says.
So, my question: If lenses cant resolve more clearly than the equivalent of 6 - 8 megapixels, then why does large format do soooo much better??

I think the key word is "affordable." As a manufacturer of konsumer-grade kameras, Kodak is probably talking about a pea-sized blob of plastic, not a real lens. Remember that every dollar they don't spend in manufacturing is another dollar in the stockholders collective pocket.

John Berry ( Roadkill )
4-May-2005, 14:36
I just got my new ,to me, 8X10 deardorff. When I put my 2 sided image senser in I found it was only one pixel, but I'm learning to live with it.

Kevin Crisp
4-May-2005, 14:53
Oh, I think this is true for lenses that are a lot more than a blob of plastic. I occasionally use digital consumer grade cameras for snap shots. I recently replaced my broken Nikon Coolpix 950 with the model 5400. The lenses on both are multicoated zooms, on the newer one I think it is even ED glass. The megapixel count doubled, more or less. When taking a photo at a high quality setting, the difference is a difference, but not a significant one. When greatly enlarged, the small megapixel camera's image will suffer from showing its pixels. The higher megapixel count camera can make pictures which can be displayed larger before the pixels become detectable BUT before you get to the point where you start seeing the pixels the photo is so fuzzy it really doesn't make a difference. The lens doesn't have the resolution to deliver and take advantage of the higher megapixel count. The new camera was originally in the $500-$600 range, though with the $200 rebate it came down to under $200, net. It has been replaced with a new model. I am sure for more money there are lenses for pros which can take advantage of large file sizes, but for 95% or more of what is being sold, I think Kodak has a point, and not just about Kodak product. The megapixel count has become a marketing technique that has little if any bearing on the quality of the end product as used by most of the people buying them.

Gene Crumpler
8-May-2005, 13:04
I have to share this. In the last couple of months, I've talked with two digital "photographers?" who saw me using my hasselblad and they told me that they liked digital better than film as long as you don't look TOO closely at the prints!!!!!