# The Naked Scientists Forum

### Author Topic: how many Megapixels would the human eye have?  (Read 47141 times)

#### paul.fr

• Guest
##### how many Megapixels would the human eye have?
« on: 31/10/2007 07:57:32 »
Not sure if it should go here, but: If the human eyesight was measured in megapixels (like camera's) how many would it have?

#### another_someone

• Guest
##### how many Megapixels would the human eye have?
« Reply #1 on: 31/10/2007 09:28:22 »
Don't know the answer to that, but one qualifier I can add is that the megapixels would not be evenly spread out - there is much better definition in the central region of the eye than in the periphery.

#### techmind

• Hero Member
• Posts: 934
• Un-obfuscated
##### how many Megapixels would the human eye have?
« Reply #2 on: 31/10/2007 13:36:31 »
Don't know the answer to that, but one qualifier I can add is that the megapixels would not be evenly spread out - there is much better definition in the central region of the eye than in the periphery.
Furthermore you can only see colour properly in the central visual field.

Even so, you can fairly easily measure the resolving power of the eye. Print off a pattern of fine parallel black and white stripes and pin it to the wall. Cover one eye, and step backwards until you can no-longer resolve the stripes and they just become grey. (yes it will make your eyes shimmer a bit). From the pitch of your stripes and the distance they disappeared, you can calculate the angular resolving power of your eye. Note that the resolving power will be decreased in low levels of illumination.

You should find that it is roughly equivalent to 300dpi at reading distance (around 14 inches).
(As an aside, it is beneficial to have 600 dpi laser-printers for text because sometimes you'll look at it more closely! Also you need higher resolution to do half-toning for good greyscale images.)

If you held a piece of A3 paper (about 17x12 inches for the benefit of Americans) at reading distance -which is a good field of view- that's about 18 million pixels @ 300dpi.
Owing to the misleading way that camera manufacturers count their (sub)pixels, I'd say that'd be roughly equivalent to 36MPix (in camera megapixels).

But of course you don't actually resolve all the detail on the paper simultaneously, only by moving your gaze around it.

« Last Edit: 31/10/2007 13:46:06 by techmind »

#### lyner

• Guest
##### how many Megapixels would the human eye have?
« Reply #3 on: 31/10/2007 15:01:04 »
The actual resolving power is only half the story. On the surface of the retina, the nerves connect in many different ways to perform very clever image processing, before the signals get to the actual optic nerve, which enhances 'important' features (like edges and familiar patterns) and, probably, ignore others.
I have read that the lens, for instance, is not particularly good quality, but people with the  best vision can resolve to better than what is reckoned to be the limit imposed by diffraction. For instance, astronauts are (or at least were) selected as fine physical specimens and have claimed to see incredible details on the ground, more than 100km below.
As the first posts say, the resolving power in the central bit (the fovea) is far better than elsewhere. Nature is far too efficient to give you high acuity on the edge when you can just move your eyes to see detail.
How much of what you actually 'see' of a scene, depends on how long you look (not just the photographic-type exposure) and how used you are to looking at that type of subject. The actual data rate is quite low but the brain, being pretty smart, makes the best of it. The last thing your brain needs is 30+MPx fed to it  on a regular basis - it just couldn't cope.
Cameras are designed to very different criteria - they need to provide as much info as possible about the  whole of the image, just in case someone wants to look at any particular part. JPEG etc help to reduce the actual amount of data needed to be transmitted or stored but you can't easily beat the interactive way the brain and eye work.

#### another_someone

• Guest
##### how many Megapixels would the human eye have?
« Reply #4 on: 31/10/2007 16:55:27 »
You should find that it is roughly equivalent to 300dpi at reading distance (around 14 inches).
(As an aside, it is beneficial to have 600 dpi laser-printers for text because sometimes you'll look at it more closely! Also you need higher resolution to do half-toning for good greyscale images.)

As you say, you need higher resolution for half-toning/dithering, yet dithering was just about usable at 300dpi, and is certainly very usable (in 4 colours) when at 4 times that definition (1200dpi, which has 16 times the number of dots that 300dpi has, but has to be able to accommodate a wide variety of dithering patters, including anti-aliasing); and I believe most VDU's are only about 96dpi, although they do not need to accommodate dithering patterns.

I would therefore be surprised if the human eye can actually resolve 300dpi at 14 inches.

#### Bored chemist

• Neilep Level Member
• Posts: 8670
• Thanked: 42 times
##### how many Megapixels would the human eye have?
« Reply #5 on: 31/10/2007 20:22:03 »
If the eye were diffraction limited (ie perfect) then for a pupil size of 1cm working with light of 500 nm (to make the maths easier) it would have a resolution of 1.2*10 / 0.0005 ie 1 in 24000.
To see 300 lines per inch at 14 inches it would need a resolution of 1 in 4200 so it's theoretically possible but I still doubt it. Can you see the individual milimetre markings on a ruler that's 4.2 metres (about 14 feet) away?

#### techmind

• Hero Member
• Posts: 934
• Un-obfuscated
##### how many Megapixels would the human eye have?
« Reply #6 on: 31/10/2007 22:59:11 »
I would therefore be surprised if the human eye can actually resolve 300dpi at 14 inches.
Well, do the experiment yourself!

I suggest you make the lines somewhat fatter than 300dpi - maybe aim for 1mm black then 1mm white. Obviously the distances will then be somewhat larger than 14 inches. Do the experiment outdoors in good light (not necessarily direct sun).
BTW to clarify, my definition of "resolve 300dpi" would mean 150 lines per inch, high-contrast black/white, 50% duty-cycle.
Laser-print, and make the lines 6-inches long too.

Apparently "20:20" vision equates to resolving detail at the scale of 1 minute of arc (1/60th  degree). Doing the maths that works out at about 0.1mm at 35cm i.e. 250dpi @ 14 inches. I stand by my original claim!

http://www-staff.lboro.ac.uk/~huph/visionvisualacuity.ppt

By the way, horizontal resolution is actually very important for stereoscopic depth-estimation. It was when I was doing work on autostereograms over a decade ago that I first had cause to ponder the resolution of the eye.
« Last Edit: 31/10/2007 23:13:38 by techmind »

#### lyner

• Guest
##### how many Megapixels would the human eye have?
« Reply #7 on: 01/11/2007 10:12:15 »
Quote
horizontal resolution is actually very important for stereoscopic depth-estimation
That makes  lot of sense. Is there any difference in the structure of the retina to enhance horiz resolution?

#### techmind

• Hero Member
• Posts: 934
• Un-obfuscated
##### how many Megapixels would the human eye have?
« Reply #8 on: 01/11/2007 10:48:12 »
That makes  lot of sense. Is there any difference in the structure of the retina to enhance horiz resolution?
I recall reading that the human eye/brain combination is slightly better at picking out horizontal and vertical features than diagonal ones. I think this is in the image-processing rather than the structure of the retina though.

Another interesting thing to try is to print out a Zoneplate image (concentric circles which get closer together as you get further from the centre) and look at that (with one eye at a time). If you have good vision you may find that you can resolve the rings slightly better horiz and vert than diagonally. But the results can be confounded by astigmatism, if you suffer from that...

There's a zoneplate image (not as big as ideal, but hey it's a start) on my never-finished zoneplate webpage: http://www.techmind.org/zoneplate/ (use the "greyscale" zoneplate)
« Last Edit: 08/11/2007 23:59:24 by techmind »

#### another_someone

• Guest
##### how many Megapixels would the human eye have?
« Reply #9 on: 02/11/2007 12:45:05 »
There's a zomeplate image (not as big as ideal, but hey it's a start) on my never-finished zoneplate webpage: http://techmind.org/zoneplate/    (use the "greyscale" zoneplate)

Sorry, I get a 404 for that site, but am OK with http://www.techmind.org/zoneplate/

#### nikomaster

• Jr. Member
• Posts: 16
• A new world, new technology science for all
##### how many Megapixels would the human eye have?
« Reply #10 on: 04/11/2007 09:39:39 »
Human sight is an analog system, we cannot how many megapixels does the human eye can get, we are not digital systems.

#### Bored chemist

• Neilep Level Member
• Posts: 8670
• Thanked: 42 times
##### how many Megapixels would the human eye have?
« Reply #11 on: 04/11/2007 17:12:06 »
Strictly that's true but you can work out an equivalent resolution. Also there is a fairly close analogy between the photoreceptive cells in the retina and the pixels in a camera.
Anyway, I got round to printing out some black lines on a clear plastic sheet. They are about 0.6mm "wavelength" ie 0.3mm white, 0.3mm black. I will have to wait now until I get to see them in good daylight.

#### lyner

• Guest
##### how many Megapixels would the human eye have?
« Reply #12 on: 04/11/2007 17:40:52 »
Quote
Human sight is an analog system, we cannot how many megapixels
A pixcel can just as easily be an alanogue quantity as a digital quantity. A pixcel is a sample.

#### another_someone

• Guest
##### how many Megapixels would the human eye have?
« Reply #13 on: 04/11/2007 18:59:36 »
Human sight is an analog system, we cannot how many megapixels does the human eye can get, we are not digital systems.

The CCD (or CMOS sensors in more expensive cameras) sensor are also analogue devices.  We get digitize signal by passing the analogue signal through a digital to analogue converter.

#### techmind

• Hero Member
• Posts: 934
• Un-obfuscated
##### how many Megapixels would the human eye have?
« Reply #14 on: 04/11/2007 23:20:02 »
Anyway, I got round to printing out some black lines on a clear plastic sheet. They are about 0.6mm "wavelength" ie 0.3mm white, 0.3mm black. I will have to wait now until I get to see them in good daylight.
Ahh good. Be sure to wear any corrective spectacles you require for the distance you're viewing at

#### another_someone

• Guest
##### how many Megapixels would the human eye have?
« Reply #15 on: 04/11/2007 23:40:33 »
Anyway, I got round to printing out some black lines on a clear plastic sheet. They are about 0.6mm "wavelength" ie 0.3mm white, 0.3mm black. I will have to wait now until I get to see them in good daylight.

Not sure that I would regard this as an adequate test, since there is no objective test that you are not imagining a resolution that is not there.

I think it would be better to have someone print off a number different cards, with a mix or smooth grey, and a small number of lines (each card having a different number of lines); and then be asked to count the number of lines, and judge the distance at which you can consistently correctly count the number of black and white lines within the grey surround.

#### nikomaster

• Jr. Member
• Posts: 16
• A new world, new technology science for all
##### how many Megapixels would the human eye have?
« Reply #16 on: 05/11/2007 01:44:59 »
Well yes, we can make an approach but just an aproximation not more than that.

#### Bored chemist

• Neilep Level Member
• Posts: 8670
• Thanked: 42 times
##### how many Megapixels would the human eye have?
« Reply #17 on: 05/11/2007 20:01:11 »
That would undoutedly be a better test. But I can't be bothered- not very scientific I know but soemtimes the practicalities of life get in the way.

#### techmind

• Hero Member
• Posts: 934
• Un-obfuscated
##### how many Megapixels would the human eye have?
« Reply #18 on: 06/11/2007 00:15:08 »
Anyway, I got round to printing out some black lines on a clear plastic sheet. They are about 0.6mm "wavelength" ie 0.3mm white, 0.3mm black. I will have to wait now until I get to see them in good daylight.

Not sure that I would regard this as an adequate test, since there is no objective test that you are not imagining a resolution that is not there.

Try the experiment yourself: you'll find that when you're close you can resolve the stripes. When you're too far, you see plain grey. When you're close to the critical distance it "shimmers and goes weird" (as long as there's enough light and your eyes are focussing properly). Sure, no-one else can verify what you personally see, but if you try it you'll find that it's not actually very easy to deceive yourself with this test.

There's gonna be too many lines too finely spaced to count them, another_someone.

Just try the damned experiment folks. It's a good one!
Forget "plastic sheet". Plain white office paper will do.
What do I have to do? Supply you with a bitmap?
« Last Edit: 06/11/2007 00:23:10 by techmind »

#### Bored chemist

• Neilep Level Member
• Posts: 8670
• Thanked: 42 times
##### how many Megapixels would the human eye have?
« Reply #19 on: 06/11/2007 19:30:10 »
"What do I have to do? Supply you with a bitmap?"
You might want to get your website to give anything other thah a 404 error, but what you would really need to do would be change the season so I get more daylight.
Incidentally, I'm quite happy to do the experiment, more or less as you sugested. It will have to wait till the weekend. It's the idea of counting zillions of lines and getting someone else to blind the experiment that I'm not going to do.
(and it's on clear plastic so if I get bored I can measure the spacing by using it as a diffraction grating.)

#### techmind

• Hero Member
• Posts: 934
• Un-obfuscated
##### how many Megapixels would the human eye have?
« Reply #20 on: 09/11/2007 00:02:07 »
"What do I have to do? Supply you with a bitmap?"
You might want to get your website to give anything other thah a 404 error
Awww sorry - it seems the forum-software here managed to insert a control-character at the end of the URL (a ^A thing) which caused problems. I've edited my previous post and removed the invisible gotcha - hopefully it works now?

Good luck with the experiment.

#### Batroost

• Sr. Member
• Posts: 276
• There's no such thing as a dirty atom!
##### how many Megapixels would the human eye have?
« Reply #21 on: 22/11/2007 20:14:39 »
I accept that diffraction and dispersion are going to be important to the maximum resolution in vision and that the brain plays a big part in what you recognise of what you see.

But...

They eye is composed of discrete photoreceptor cells, which are finite in size. In this respect we have 'digital' vision.

According to one reference I found, the peak light-recieving cell density in the retina is about 4000 cells/mm2; with a total of around 130 million cells - or 130 megapixels! In reality, the optic never is more limiting at only around 1.2 million fibres, so your retina 'pre-processes' your image down to around a megapixel.... it's like we're seeing in JPEG rather than BMP.

#### Batroost

• Sr. Member
• Posts: 276
• There's no such thing as a dirty atom!
##### how many Megapixels would the human eye have?
« Reply #22 on: 22/11/2007 20:37:37 »
Just had another thought - if the above post is true why can I see the difference between a 1 milllion pixel image and a 4 million pixel image?

Answer: I don't see the whole image all at once unless I'm standing back from it, in which case I probably won't be able to tell the difference...

I remember I saw the way we 'see' demonstrated in a science documentary where the motion of the eyeball was tracked as people looked at images (paintings etc...). The track of the eye over a few seconds traced out the significant parts of the image. An unmoving eyeball rarely picked-out any detail from the edges of the picture..

#### lyner

• Guest
##### how many Megapixels would the human eye have?
« Reply #23 on: 27/11/2007 13:56:40 »
Quote
They eye is composed of discrete photoreceptor cells, which are finite in size. In this respect we have 'digital' vision.
No, it's not the slightest bit digital. It just takes samples of the scene. Each sample is analog. There is a fundamental difference. Just because film has grains, you wouldn't say it was digital, would you?
A straight bitmap of the signals from your retina would be far too wasteful to have evolved successfully. The 'best value' function the eye has is the ability to pan and tilt - intelligently. It puts the hi res part  where it needs it.
And it's more smart, even, because its periphery is particularly to movement - so you can see the  lion creeping up on you from the side but you don't need to count its teeth!

#### Bored chemist

• Neilep Level Member
• Posts: 8670
• Thanked: 42 times
##### how many Megapixels would the human eye have?
« Reply #24 on: 27/11/2007 20:59:31 »
Actually, it is a bit digital. Nerves fire or they don't (also photons hit receptors or they don't- but that's another story).

#### The Naked Scientists Forum

##### how many Megapixels would the human eye have?
« Reply #24 on: 27/11/2007 20:59:31 »