Erik's web

  • Increase font size
  • Default font size
  • Decrease font size

Lot of info in a digital image!

E-mail Print PDF

Note: This article is in early tage of progress

 

 

In a recent article I discussed using HDR tone mapping to enlive a quite boring image. There was some lively discussion on Luminous Landscape regarding making most of the technique, and one of the posters remarked on the DR of the image.

 

This is the initial image

and this is the final one

 

Why does the initial image fail?

The main reason that the initial image fails is that it has low global contrast. The image has actually quite wide DR (Dynamic Range) but once we try to display it on screen it will be flat. Using extensive local adjustments we can make the image more interesting.

Optimizing DR

DR (Dynamic Range) is the range between maximum signal and least discernible signal. The term is coming from signal processing and means maximum signal/signal at SNR=1. In practical terms it is maximum signal / sensor readout noise. The maximum signal we have is limited by what is known as Full Well Capacity (FWC) which is the maximum electron charges a pixel can hold. Readout noise is fixed (by and large). DR is normally defined in stops (or EV). DR = log2(FWC/readout noise). Assuming that we have a sensor with FWC of 25000 electron charges and readout noise of 6 electron charges we would have a DR of log2 (25000/6) = 12.02 stop.
Obviously, to maximize DR we need to maximize exposure. On the other hand we also need to avoid clipping. We have clipping when we have used up the Full Well Capacity (FWC) of the sensor. When the sensor is full, no more photons can be detected and information is lost. To maximize DR we would "expose to the right" using the histogram on the camera. Some cameras also have blinking highlights. The histogram tells how far the sensor is from clipping and the blinking higlights show where clipping will occour.

Processing

Exposing to the right will naturally make the image too bright, so we need to keep down the highlights in processing.
The way I work normally is that I either just start out from "auto" in Lightroom or take a more systematic approach:
  • Start with cropping the image, so we get rid of distracting detail affecting histogram
  • Adjust exposure so highlights don't clip
  • Try to use a graduated filter if it is needed
  • If needed, keep down extreme highlights with "recovery slider"
  • Adjust black level so we just have some clipping
  • Adjust fill light to get some light in the darks
  • Adjust brightness so middle tones are OK
This process is iterative by nature.

Fixing the sky

Normally I try to do as much in Lightroom as possible. The graduated filter for insstance is a great tool but somethimes the horizont will not cooperate. The way I handle this is nowadays to edit the picture in Photoshop. The method is:
  • Duplicate image
  • On top image choose blend mode "multiply"
  • Click alt-cmd-5 (which select blue as lumiosity channel)
  • Click add layer mask
Multiply will darken the image but this will only be applied in the light parts of the layer mask. The mask will be accurate at the pixel level. It may be necessary to paint over some parts of the luminosity mask with black.

Now that we have a nice sky but the image is still boring we can try different tricks

HDR images have very wide dynamic range, far exceeding that of any media. The HDR community has developed tone mapping methods to handle this. Unfortunately these mappings are often not very natural which gave HDR a bad name, but sometimes we can use these effects to our advantage.
Once tonality is decent I go down in the "Basic menu" and add some clarity and vibrance
Last Updated on Monday, 06 February 2012 09:34  

Statistics

OS : Linux u
PHP : 5.2.9
MySQL : 5.0.67
Time : 18:21
Caching : Enabled
GZIP : Disabled
Members : 1510
Content : 72
Web Links : 1
Content View Hits : 365590

Login

Banner1

J!Analytics