Helpful Posts: 0
14th January 2012, 11:06 PM
I am trying to take a photo of a city skyline (Partially cloudy bright sky and tops of buildings). I am having trouble with proper exposure; i am trying to capture nice details in the clouds but the tops of buildings always are dark/underexposed. Whenever I make adjustments the tops of buildings show up nice but the sky becomes overexposed and I lose the details in the clouds. Any suggestions?
14th January 2012, 11:23 PM
You should look into a graduated neutral density filter. That would cure your problem.
14th January 2012, 11:41 PM
You can combine different (Possibly simulated from a single actual one) exposures in software so as to simulate a GND filter. There is no need to buy additional hardware. Also you can take an exposure with no blown highlights and put shoulders in the curves (In a proper color space like YCbCr).
15th January 2012, 04:44 PM
Mario, a newbie like me does not understand that process you mention and the color space thing. Can you explain it in more detail or give a link for additional info?
15th January 2012, 05:14 PM
The problem is that there is too much difference in luminosity between the sky and the buildings for your camera to capture; a bit like being outside a door in bright sunlight: you can either see what's inside, and then the reflection on the wall hurts (overexposes) your eyes, or you can see the wall, but then the inside is black (underexposed). So you must make a choice: either get the sky correct, or get the buildings correct, or 'cheat'.
Originally Posted by vitaminvee
There are two basic cheats, or rather ways to solve your problem (both have been mentioned ) .
- Use a so-called GND or Graded Neutral Density filter: that's a filter half of which is clear (colourless), the other half is gray, with a fuzzy boundary between the two zones; you should place the filter in front of your lens so that the dark part covers the sky, and the clear part the buildings, and then expose normally. The filter will reduce the luminosity of the sky, so your camera can see both correctly.
- (if you use a digital camera) Use two image, and combine them in post-production (on the computer). The images must be identical, except for the exposition, so you should use a tripod for this. Then expose one image for the sky, the other for the buildings, and combine the properly exposed parts of each image (rather easy with the right program). As you don't tell us what camera you are using, it's hard to give more details.
Feel free to ask, though
15th January 2012, 06:54 PM
As Remco says the problem here is the board dynamic range; presumably too much for the imaging device (Camera) and display device (E.g: computer monitor, print) and maybe also for our eyes. With digital image processing we can easily compress the dynamic range of the image (Make the shadows relatively lighter and the highlights relatively darker) to one suitable for display.
I mentioned two different methods of addressing the problem:
One is exposure combination and requires no specific knowledge of color spaces. You can find an example at http://www.gimp.org/tutorials/Blending_Exposures/. Basically you load both images as layers and replace the blown highlights in the dark image with the proper area of the light image.
If your camera has a board enough dynamic range (So the limiting factor is the display) then you can use a single suitable exposure (Preferably use the exposed to the right method) and compress the dynamic range.
In The GIMP convert to a color space which has a component representing lightness (Luminance in YCbCr, Lightness in LAB, Value in HSV) (Colors->Components->Decompose, check "decompose to layers"). Select the lightness or similar layer (Windows->Dockable dialogs->Layers or C-l keystroke, then click the layer thumbnail). You can do a simple adjustment with Color->Levels or a detailed one with Colors->Curves. Make the curve more steeper at the left (Shadows) and more flat at the right (Highlights). Then convert back to RGB (Colors->Components->Compose or Colors->Components->Recompose).
If you do a non-linear transform on individual RGB components you will usually perceive a hue shift. Color perception is a difficult topic but the principle is the same: These color space attempt to analyze a color as lightness and other components enabling us to modify the former while keep the later closer to the original.
I use The GIMP because it is free as in freedom. I can use it freely and study how it works. Proprietary software is detracting in general but especially a hindrance for learning. Please take a time to read further about in http://www.gnu.org/philosophy/philosophy.html.
16th January 2012, 08:50 AM
Have you taken any shots yet that you can share with us? How close to the skyline are you and what lens are you using?
20th January 2012, 08:35 AM
Wow people, Thanks for all the input. Im really new to photography, and Im glad to hear that I wasnt just setting my iso/aperture/shutter speed completely wrong. Im using a Nikon 5100, with the basic lens it usually is packaged with (18-55mm f3.5-5.6). It has an HDR mode, would that be similar to the post-processing that was previously mentioned? Unfortuneately, I deleted all the photos mainly bc I was just playing around with the settings.
20th January 2012, 09:52 AM
If your not CS5 or LR savvy and you don't want to bother working with multiple images,layers, and processing you would be better off putting out the few bucks for a ND filter. It's a simple tool with great results. I use a Hoya HMC ND8 for bright sky & ND4 for medium and lower light sky. There are cheaper brands out there too, Hoya is just my filter of choice along with a few Tiffen.