This thread has developed beyond a discussion about exposure to include the topic of visible detail and resolution. Otherwise I would not post the following here. My intention is to make the concept easier to understand but if I have failed then perhaps you know why I'm not a teacher
When we look with our eyes we see an angle of view. The level of details that we can resolve depends on the level of contrast of those details and also on the angle of view they cover. This is a physiological characteristic of our eyes and brain. If the details cover a wide angle of view then we can hardly miss them. If they cover a small angle of view then they become harder to resolve.
The angle of view can be measured in various units much like temperature or distance can. Most of us know about degrees, minutes and seconds, in which a second is 1/60 of a minute and a minute is 1/60 of a degree. 360 degrees covers a full circle of view (fisheye). Another measure is gradians (400 grads to a circle) and another is radians (approximately 6.3 (2 x Pi) radians to a circle) but to keep it simple I'll stick with degrees.
There are a number of ways that we can use to increase the angle of view covered by each detail. e.g.:
1. we can get closer (but too close and we cannot focus, or the thing bites us on the nose or just flies away). As we get closer the angle of view that it covers gets bigger.
2. we can use a lens that magnifies the size of the details more than our eyes can (binoculars, telescope, etc.)
3. we can take a photo and enlarge it and look at that instead of the real thing (so long as the photo captured the necessary detail in the first place)
I've been using the term angle of view, but I could easily call it an angle of arc. If you imagine stright lines from the extreme sides of an object to your eye, then the angle betweeen those lines where they meet at your eye is the angle or view covered by that object. An arc-second defines an arc that covers an angle equal to 1 second, or (1/60) x (1/60) x 1 degree. An arc-minute covers (1/60) x 1 degree. We can see things that are bigger than about 0.3 arc-minutes or 18 arc-seconds. Roger was explaining that each pixel on the various cameras with long lenses sees a part of the image that can resolve the equivalent of an angle of view of about 3 arc seconds. That means that with a tele lens on the camera we can capture details that are one sixth of the smallest size that we can see with the naked eye. Yay. But if we had a shorter lens then each pixel may cover less detail than we can see with the naked eye. That's partly why wide angle shots never seem to capture the same level of detail that we saw - lots of items but each one is a bit small and so it becomes vague and undetailed. Even a standard lens on these cameras is struggling to do better than good eyes.
What Roger did not mention is that his camera / lens angle of view was only theoretical to the extent that it assumed a perfect lens. It was about lens angle of view rather than actual real-life lens resolution. A real lens has its own optical limitations that further reduce the amount of detail captured by the camera. The camera pixels still see the same angle of view through the lens but the view is somewhat clouded.
Of course it doesn't matter what the camera captured until we display it or print it. Until then it is invisible. So how much detail can we see in a print ?
It has been published somewhere (probably by Norman Koren in one of his highly detailed articles on scanning and resolution - I'll try to find a reference) that people with excellent eyesight can resolve printed details in good lighting as long as the details are big enough to cover at least 0.3 arc minutes or 5 thousandths of a degree. How big is that ? Well, it depends how far away those details are. At 10 inches it is about one thousandth of an inch. It would need a printer at about 1000 dots per inch to show it. [At 100 yards it is about 0.3 inches. Have you noticed how hard it is to see a fence wire at 100 yards, other than when the sun is glinting off it ? At 10 yards we are struggling to see 0.03 inches or 0.8mm. That's why we can't see all the fine detail of a feather at that distance.] Back to our print at 10 inches... We can see something that is bigger than 0.001 inches. If we put two of those somethings beside each other then they become a big something of 0.002 inches. We need a gap between them if we are to see them separately in a group or pattern, but the gap also has to be at least 0.001 inches or we won't see it either. So now our print resolution is down to about 0.002 inches. That means we can see a repeating pattern of maybe 500 lines per inch. If the source material happens to not line up with the printer dots then we can resolve less. Say 300 lines per inch. This is why 300 pixels per inch is about as good as most of us need when sending images to the printer. We let the printer use 1200 or whatever dots per inch so that it can use a pattern of dots to make up the required colour and tone for each image pixel, but the pixels that are smaller than 1/300 inches start getting too hard to separate and resolve consistently. Sometimes it works and sometimes it doesn't. 300 ppi is pretty reliable.
Now we can get to the other factor touched on previously. As well as physical size we need optical contrast in order to separate and resolve the details. Two adjacent dots with almost identical colour and tone are effectively one big dot. We cannot tell them apart. Our bright white egret feathers are represented in our photos by a bunch of almost identical pixels. We are limited in how much we can enlarge the detail by the resolution of the photo. Too much enlargement adds no new info and doesn't help. To see the details in a smaller print or on screen we need to exaggerate the contrast between adjacent pixels that are very similar in colour and tone. We can use curves in photoshop to do this. One of those "S" curves that enhance contrat at the bright end of the scale but reduces it at the middle tones. Alternatively we apply lots of sharpening until we start to see halos but that looks rather too unnatural.
Mention has been made of using a maximum value of 254 instead of 255 for whites. By using something less than 255 we are sure to get at least some ink on the page. That might prevent those horrible bare patches where the glossiness of the page and the glossiness of the ink are so different that we can immediately tell where there no ink at all. As long as some ink is there the glossiness is consistent over the whole page.
I hope this has clarified some of the concepts for some of the readers but the fact is that many switch off as soon as they see numbers. That's partly why most photographers are point and shooters. No disrespect intended or needed, it's simply the way of the world. If you can understand the concepts then you'll be better for it than someone who does what they're told without knowing why. That's what learning is all about.
- Alan
