1.) In-camera dynamic range (DR) settings.

The S3 gives you the ability to influence dynamic range by either selecting a certain film behaviour or by manually choosing one of the pre-defined tone mapping curves.

These are the possible DR settings. The red color indicates overexposed areas. Click the respective image to view a larger version.

	Extended Dynamic Range switched OFF
	Extended Dynamic Range switched ON:
	AUTO dynamic range increase This setting seems to select the best fitting DR method for the captured image.
	Simulation Film F1 looks like "Negative Film"... Film simulation includes pre-defined color, tone and sharpening settings
	Simulation Film F2 obviously "Slide Film"... also includes pre-defined color, tone and sharpening settings
	DR setting WIDE1
	DR setting WIDE2 This setting tries to squeeze the most out of the RAW sensor data - even at the cost of decreased contrast.

2.) Using the extended DR capabilities of the camera.

As already stated in the beginning of this article: If you want to learn about the DR capabilities of the S3, imagine an extra -2EV bracket done by the camera. It is up to you if you give the camera the chance to use this extra image data or not.
Personally, i am used to slightly underexpose my images if i don't want to loose highlight detail. With the S3 i  have to modify my rule of thumb - and to make things even harder - add a good amount of random Voodoo to my assumptions, especially when shooting with matrix-metering:

- Basically, the camera tries to behave like any other digital camera
- No special preference to expose towards shadows
- Highlights will be mapped from the additional sensor data depending on the camera settings, but expect small bright areas to be overexposed while shadows will still be dark.
- Sometimes the camera shifts exposure data downwards, resulting in more detailed shadows instead of highlights (the Voodoo factor)

Some test i found on the net came to the conclusion, that there is virtually no difference to other digital cameras (e.g. Luminous Landscape - section S3 vs. 1Ds MkII). I think the problem in such tests is the assumption, that the camera will always try to shoot the widest possible DR and do a highly sophisticating tonemapping afterwards. This is not true. First, this would overstress the in-camera CPU, second, it would result in a strange looking image that wouldn't be accepted by most photographers.
Take a look at the lively tonemapping discussion to find out more about mapping High Dynamic Range into a viewable color space.

WRONG testing - replayed:
Here is an sample of a "typical" test shot that will NOT show more DR with the S3. All cameras use AUTO-settings. For your convinience, the thumbnail colors are normalized.

	Nikon D70
	Fuji S2
	Fuji S3 (DR settings WIDE2)

CORRECT testing:
This is a shot done on an overcast day.

	Nikon D70 Parts of the sky are burnt out.
	Fuji S3 (DR settings WIDE2) No sky detail is lost. If you think the image looks dull: pleas stop by at the tonemapping discussion.

2.) Shooting with the extended DR settings.

< Felix the Cat, sunbathing (JPEG mode, DR settings WIDE2, 1 F-stop underexposed, shadows restored and local brightness increased) (click image to view large)

Yesterday, we had a sunny Saturday afternoon, the best opportunity to capture a real challenging town scenery - the Mariahilfer Street in Vienna. Crowds of people moving around and the bright sun, combined with deep shadows in the small lanes. As a panographer, you'll normally try to shoot (RAW) brackets to manually blend them. This is very time consuming - and - if you have something moving around in your images - sometimes simply not possible (See a typical HDR shot at the Neubaugasse Panorama).

	Exposure leaned towards hightlights, restored shadows. Click thumbnail to open the panorama.
	Exposure like negative film (towards shadows) Click thumbnail to open the panorama.

---

3.) Comparison of the Dynamic Range capabilities of different cameras

It turned out to be a little tricky to estimate the DR of the S3 compared to other cameras. As a first idea, i thought about direct RAW to HDR conversion but this was not possible due to software (compatibility) reasons.
So i ended up shooting a bracketed series of images to determine, when the camera will begin to under- respectively overexpose.

The brackets are done with a 1EV stepping. Note that the images use in-camera JPEG mode, something i am not really comfort about, as it introduces an unknown post-processing step which may falsify results. So, please take the results with a grain of salt.

Median values of the R-G-B channels
Red color: overexposed, blue color: underexposed

S2/S3 exposure time (D70 exposure time)	Nikon D70 ISO 200 (1sec - 1/4000sec)	Fuji S2 ISO 100 (2sec - 1/2000sec)	Fuji S3 ISO 100 (2sec - 1/2000sec)
2" (1")	255 255 255	252 252 252	255 255 255
1" (1/2")	255 255 255	252 252 252	250 255 255
1/2" (1/4")	255 255 255	252 252 252	238 246 249
1/4" (1/8")	231 254 251	233 252 251	213 233 238
1/8" (1/15")	169 215 240	175 211 224	167 197 210
1/15" (1/30")	115 171 193	120 153 165	119 151 168
1/30" (1/60")	61 111 140	77 107 117	75 105 118
1/60" (1/125")	28 58 80	39 66 75	43 65 76
1/125" (1/250")	15 32 42	19 33 40	24 37 45
1/250" (1/500")	9 14 17	12 16 21	14 22 26
1/500" (1/1000")	3 7 6	3 8 9	8 13 16
1/1000" (1/2000")	2 3 2	1 2 4	4 8 9
1/2000" (1/4000")	1 1 1	0 0 0	4 5 7

Still there is some room left for interpretation. After some time of concurrent use of the S2 and D70, i know that the useable DR with the S2 is significantly higher than with the D70 (in therms of image quality). After some shootings with the S3, i can see that this is also true for the S3 compared to the S2. It is up to you how seriously this may tamper your DR estimation of the different cameras.

4.) Conclusion

Don't expect the camera to reward its purchase by delivering spectacular eye-popping HDR shots that are already tonemapped with your favourite mapping algorithm. The S3 is able to capture an extra dynamic range, worth a 2EV bracketed shot.

Different from my original intention, i now decided to leave out the comparison to film. I see no sense in it, as even the digital vs. digital comparison causes me serious headaches about measuring the useable dynamic range in a standardized way.