Faolan said:I see you subscribe to the 'new' theory that rather than having a hill
in the middle you prefer it to the right,
Not exactly. I've reported my views on this in more detail in the past,
but I didn't really want to get into all that again. I don't agree
completely with the "expose to the right" philosophy. Tests I did with
the G3 (which uses the same type of sensor as the 300D, so it's probably
similar) indicated that the optimum peak for quality is 2/3rds to the
right, so if I take a "single hump" type of picture then I'll try to
place it in that position.
However, the type of photograph that we are discussing here is the
"double hump with wide valley" shot - a very high contrast image with
bright sky and dark land. If you place either hump in the centre you'll
lose all the detail in the other hump. In this instance I try to put
the right hump on the far right in order to keep the left hump as far
away from the left edge as possible. The biggest quality drop is on the
far left edge so I try not to get anything in the left 1/6th if
possible.
Quoted message said:I have tried it with 2
cameras one Fuji (602 Pro) and one Kodak. The Fuji preferred the hill
in the middle type approach which I think is due to it's sensor
array. The Kodak came out with the image you are thinking of. The
method you suggest is more suited to dSLRs I feel.
Or at least the Canon sensors. Bear in mind that these are 12-bit
sensors that produce 48-bit tiffs when used with RAW processing. Thus I
can extract more and better quality detail than any camera that produces
a 24-bit jpeg. The RAW processing can extract perhaps 2 extra stops of
detail from the image, including some highlights that seem to have burnt
out.
Quoted message said:As to the second part, again depends on the camera some cameras can
recover more detail when it's balanced like a hill type histogram
whilst others thrive on a off balance shot.
Yeah, but histograms vary from shot to shot depending on the lighting.
There's no single rule you can apply in all instances. Some shots are
compromises, it's up to you to decide how to best deal with them.
Quoted message said:However all digitals
because of their nature once the line has gone off the edge no detail
can be recovered.
Generally true, but as I said above, if shooting in RAW mode it is
possible to regain some highlight detail that seems to have been lost.
I say "seems" because the histogram will show a vertical line on the far
right and the preview image will flash the burnt out areas. However,
this applies to a 24-bit jpeg - if you save in that format you have lost
the detail. However, if you save in RAW mode you'll find that it has a
wider dynamic range and may still have some highlight detail which is
recoverable (in the RAW processing stage).
I must stress that I'm not in the habit of letting the highlights go off
the far right edge, but if they just go slightly over, you can get them
back in RAW mode. I haven't tried to measure how far you can push this
yet.
Quoted message said:Unlike film it's both easier to blow a shot,
There's a sharp drop off when you hit the limit. It's like digital
audio recording, when you hit 0db the sound wave clips, so you must
never ever go over. With analogue recording you get distortion when it
goes over but you have a little leeway before it starts sounding really
bad, so there's a gradual deterioration rather than a sharp cut off.
This is most noticeable in sunset shots where the sun typically burns
out. With film there's quite a smooth transition to the burnt out area,
whereas with digital it just suddenly goes white. Quite annoying.
Makes it difficult to get really good shots with the sun in the pic.
Possible solution is to take multiple exposures and blend them later.
It's almost impossible to expose for the sun properly without turning
the rest of the picture black, except when it's really low in the sky.
Quoted message said:and
more difficult to recover a shot.
I'd have to disagree with this. As I've often argued, digital (at least
the ones with 12-bit sensors) has a much wider contrast range than film,
therefore a huge amount of detail can be recovered from shadow areas.
Film tends to be more contrasty, especially slide film (which is
generally preferred by professionals for the low grain) so when you lose
shadow detail, it's gone. Even with a 42-bit Nikon slide scanner, I
can't extract as much shadow detail from a slide as from a digital
image.
Granted the darkest shadow detail in a digital image is of much lower
quality, but at least the detail is there, which is better than none at
all.
Quoted message said:http://www.luminous-landscape.com/tutorials/understanding-
series/understanding-histograms.shtml
Oh not again! His comment about digital cameras only being able to
capture 5 stops is just plain wrong. I'm not going to get into another
big argument about this. One point I do want to make about that article
is that he divides the histogram into 5 equal areas each of which he
says represents the five stops. This is quite wrong. Apparently the
histogram on the 1Ds has these divisions, but they do not represent 5
stops.
The histogram is not linear! You can't divide it up equally into stops.
This may be one of the main reasons why people think digital cameras
don't have a wide dynamic range.
My tests have shown that the "widest" stop is the third stop which peaks
at roughly 2/3rds to the right. As the stops get darker they become
more and more bunched up. I can't remember the exact details offhand,
but there are about 5-6 stops located in the left 1/3rd of the
histogram. Since they are so close, people often don't realise how
large a range of brightness is contained on the left edge. The very
darkest stops in the left 1/6th of the histogram are of very low quality
though, so keep detail away from the very left edge if possible.
There's also a small drop off on the right hand side according to my
tests (contrary to the "expose to the right" theory) but it's not much
of a drop off and the quality degredation is nowhere near as bad as on
the left.
Quoted message said:The key thing to remember as long as it's not off the edge then you
may be able to use it.
You can use it. Everything within the histogram is usable, if not all
of equal quality.
Quoted message said:Even if you blow out the sky (fairly common)
It's fairly common because most inbuilt exposure meters give too much
emphasis to exposing the land correctly. Interestingly my 300D doesn't
do this, the auto exposures are much darker, not that I ever use them.
Quoted message said:then you probably mask in a sky from another layer fairly easily
depend on the shot.
If you do need to do this, put the camera on a tripod and take multiple
exposures. Using the auto-bracketing feature is best if you have it.
Paul
--
http://www.wilderness-wales.co.uk
http://www.wildwales.fsnet.co.uk
http://www.photosig.com/go/users/userphotos?id=118749