Photography Thailand Forum

[Marc Schultz]

This is a shot taken in the early evening along the river in Ayutthaya as a Mahout was riding his elephant out of the river after it was given a bath at the end of a day's work. I purposely exposed for the light in the sky rather than the foreground which was all in shadow. I kind of like the moodiness of the image and the feeling of the sky. It sort of reminds me of the dark style of shooting Steve McCurry did so much of in the eighties in the Middle East and India.

[anna]

Marc, always a joy to see your shots.  Anna

[Marc Schultz]

Thank you Anna.

[Chris Savery]

Excellent. They've got freckles...
I think the dark style works really well to enhance and give it a richness especially in the red clothing.
Chris :)

[gregoire]

wow, dreamlike.

looks like the perfect exposure, that's quick thinking on your part.

[Marc Schultz]

Thanks. Lucky I guess. And RAW files have a lot of lateral movement on exposure as well so it helps during post processing.

Chris, I think those spots are actually traits of a white elephant and are deemed to be of higher standing.

[David Salmanowitz]

Marc--Very nice indeed! The copper/gold on the trunk of the elephant is just great. When you mentioned exposure, did you just spot meter the sky?

[Marc Schultz]

Thank you David. I never actually spot meter anything. I normally just use exposure compensation and leave the meter in the matrix mode. Since RAW files from the 1DS Mark II I use normally have at least 1 stop of lateral movement on exposure either over and under, you really need to either adjust the exposure at least 2/3 of a stop or more over or under a normal exposure to put the dynamic range of a file to a point where the file would go past the threshold of actually already having 1 stop over or under of play based upon a normal exposure. I am not sure how to explain this so well, but I hope that makes some sort of sense.

I am also pretty good with gauging exposure based upon what I see on the LCD screen and the historgram after taking a shot. So I often take one shot, quickly determine what the shutter speed should be, and then throw it into fully manual and set the aperature and shutter speed to a fixed level of exposure, assuming the light is not changing quickly.

I just checked the Metadata though on the RAW file for that shot and I see that I was shooting at ISO 400, F/5.6 and 1 Stop under on exposure compensation. Again, the light meter on my camera never leaves the matrix metering mode.

There is actually a method of exposing on digital cameras known as "Exposing to the Right", which referes to the 1/3 zone on the right side of the histrogram. You may already be familiar with this, but if not you may want to read about it a bit. There is a good general explanation here: http://www.luminous-landscape.com/tutorials/expose-right.shtml

I am afraid if I tried to explain that I would probably confuse things even more :)

[gregoire]

Thanks for the link, Marc. At first, fearing blown out highlights were the worst thing in digital photography, I systematically exposed to the left. I quickly noticed there was something wrong with that. Pictures were flat, noisy, and lacking color even after post-processing. So I went on to expose to the right and, though I must watch out for blown highlights, my pictures are ten times better from it. It's good to read the technical explanation as to why this is.

[Marc Schultz]

I was exactly the same way in the beginning. The misunderstanding I made was thinking that when they were talking about exposing to the right they meant to blow out some highlights in the process. The reason I still exposed to the left was because I know that any blownout highlight detail is not recoverable. But where the misunderstanding is made is that when they talk about exposing to the right they are not talking about shooting to the right to the point of blowing out highlights which would be the quadrant furthest on the right of the historgram. The idea is to have most of your highlights weighted in the quadrant right before the last one on the right hand side of the histogram as opposed to aiming towards that camel hump in the middle histrogram that was belived to be the way to shoot before.

This way there is no loss of highlight detail, but yet you are bringing out more light and better saturated colors in the shadows and avoiding noise in the shadows towards the left side of the historgram in the process. Then later you can always darken more if you want, but you have more dynamic range to play with on the left in the shadows without having blown out the highlights either. Hopefully what I wrote makes some sort of sense. :-\

[David Salmanowitz]

Marc--Funny how I just posted about overexposing being better then I read your post here. Guess I should have gone to elephants before larb, but I am typing with a partly empty stomach. 

[Marc Schultz]

I am glad you saw this one David because I was about to post a reply on your last reply on the Larb to tell you there is a discussion going on here about it.

Yes, laptop screens are the worst for trying to see true colors and exposures and the angles of the screen can be funny. I actually use a 20" Apple LCD Cinema Display which is also on a bit of a tilt like a laptop LCD but the nice thing is the colors and brightness don't change with the tilt.

[David Salmanowitz]

Marc--Saving my $ to buy two of the Apple monitors in the spring. Hopefully in late spring Photoshop CS3 will be available, which will scream on the the new Intel Macs.

[gregoire]

Just showed the luminous-landscape "expose to the right" thing to my bro and this is what he replied (he usually knows what he's talking about):

"he's right except for two things. RAW format is linear, which means there are as many levels for shadows as for highlights, so this is only true for shooting in JPEG, and it's much worse to clip highlights than it is to get noise in the shadows, so unless you have rgb histograms, be very very carefull..."

Yeah 'cause on our cameras (a Nikon d70s and a Canon 350d) we get a histogram that doesn't differentiate red, green, and blue. So you could be blowing out a color (most of the time red) and not even see it in your histogram.

[Chris Savery]

It does seem a bit warped - though in that article he does seem to bring in some notes from people who really ought to know exactly what's going on. I was trying to get my head around it as well.

After thinking about it some more I came to the conclusion that it doesn't matter about file format at all. The reason being that in a linear format like RAW the binary values would indeed be equally distributed across the intensities (as we would expect from a linear sensor) but that doesn't matter. If you have ever looked at a RAW linear image you would immediately know it's not something we can use as is. The fact is, to use that RAW linear image, we have to convert it to a distribution that corresponds to our human perception which is non-linear (logarithmic). In essence we take linear and re-map it back to f/stop (non-linear) scaling so that our eyes perceive the image as we expect.

So what matters is that our eyes see non-linearly and regardless of the intensity values stored in the file we still need to redistibute the values for viewing. On the linear scale we may not be bunching image values at all but on an f/stop scale the result is the bunching up of information into (linearly) unequal bands of intensity. I'm only half convinced by this as the argument that there are more discrete steps available and hence S/N is improved seems a little obscured.

If you try to expose your image in the upper region of the linear sensor then when that image is converted to a viewable form the range of values will be more spread out, which is what I think people are after here. As for S/N ratio well that another thing. I think a higher S/N ratio stems from the fact that in the higher end of the intensity scale the signal is by defintion higher so moving more value to the high range of course results in higher signal values overall and hence also in a ratio to noise.

But one point that is mentioned but glossed over by that whole discussion is that by moving your exposure to the higher range your are in essence forcing the camera into a lower ISO setting. Since you have to expose higher, you would need more light - defeating the purpose of using any higher ISO setting you may have chosen. I guess this is just a natural side effect but when you expose high at 100 ISO in essence it's like pushing your ISO below 100.

I guess in the end it's proof is in the pudding and I'd like to know if real visible gains are made by exposing high like that. Marc seems to say yes - he sees better results by exposing high - as close as possible to the limit but not going over. Is that a real consensus by photographers? Or something just imagined by expectations?

Chris :)