The fake look comes from the fact that most humans are not used to seeing a HDR image. We are used to looking at limited dynamic range images because that is what can be made with a single exposure on most cameras. The HDR image has a much greater tonal range...much closer to real life. But to see it in 2d looks funny.
Tone mapping will produce very funny looking images if not done correctly. But producing the actual HDR image is only one of the steps. Simply merging a set of images to HDR is not enough. At this point you have a "tonal rich" image. What this really means is you now have a lot and I mean "a lot" of pixal data to work with. You now have enough data to edit it in photoshop without loosing so much data that the image suffers from data loss and again becomes wierd looking. Every time you perform an edit in photoshop like levels, curves, sharpen, color correct you are removing some data from the image. PS interpolates the missing data and redraws the histigram to replace the missing data. The more editing that is done the worse things get. If the image is not an HDR image depending on the exposre of the image you will have some or very little editing room before you hack the image to death.
The other thing is you MUST take at least 3 exposures at 2EV ( EV=Exposre Value...1ev is equal to one stop. But we use EV so we dont have to say 1/60 to 1/30) if the scene has a very high dynamic range.
ALL panos have the highest range possible because you are shooting an entire 360 area that includes all tonal ranges from white to solid black. IF you are shooting a still image then the dynamic range is limited to the specific lighting situation that is framed and often only 2 exposures at 1 EV are required. It is the panos that are the hardest to do using HDR.
About doing the RAW trick to get an HDR image: Smooth is very correct in how this is done as a RAW image is only data that if not shot raw is processed by the cameras "algorithm" into a JPG or TIFF. How this is done is a closely guarded secret specific to the camera manufacturer. So by using your own RAW converter you are doing the same thing except that you have control over how this happens. This is related to how color is created from basically a black and white image. In RAW conversion you determine several things related to exposure and two of those are color intensity and tonal range. You are really "developing" the image. If we were using film we could in the lab change some of the these parameters by using different solutions and developing time. So we could make the image over expose or under expose. BUT...and this is the key....we are limited to the original exposure as to "how much we can push this either way."
Altering a single image in RAW either way "is not the same thing" as bracketing seperate exposures. Even though RAW is only capturing the data there is a dynamic range in which this data lives. The dynamic range is a function of how the sensor performs. For example a full frame sensor will have many more diodes and will therefor be capable of much more dynamic range than a 1.5x sensor.
A simple example is a bracket set shot at 1/8, 1/500 and 1/2000 vs 1/30, 1/90, 1/250
The fist set will have a dymaic range that far exceeds any sensor made today ( except large format cameras). The second set will have a dynamic range that is close to what could be coxed out of a single RAW exposure.
We often bracket in still photogrpahy because we are not sure what the "exact" perfect exposure is. Do we expose for shadows or highlights???. Cant tell until we get back to the house. Assuming you make a "perfect" exposure then it is very possible to use a single RAW exposure to cox out the missing data and merge to a HDR. But if you are off then you wont quite have enough data in either shadows or highlights to make it work out like you want.
So, wheather you can make it work from a singel RAW image or 3 bracketed RAW images depends on the dymanic range of the scene and the dynamic range of your cameras sensor.
Smooth has demonstrated how to do this with one RAW image....and I might add very well. However this will not work for extremly high dynamic range scenes.
General Lee
|