Photographing an evening lighted scene is a challenge on many fronts. The high dynamic range, about 21 EV or stops, of these scenes is the highest we will photograph. And because significant portions of the scene is at low light we need to use the lowest ISO practical which means some exposures will be in the seconds range.
The high EV range of these scenes is a challenge. Many of the lights, particularly in holiday scenes, are point sources. There is no practical way to meter the scene. Trial and error is the only way I've found so far.
Use the lowest F stop available
To achieve the lowest ISO possible we need to use the lowest F stop possible. The front yard scenes we are looking at are between 40 to 80 feet away. The images were taken with a focal length around 25mm. The hyperfocal distance at f/2.8 / 25mm is 24.5 ft. if the focus is at 50 ft, the depth of focus is between 16 feet and infinity. At f/5.6 with the same parameters the near limit drops to 9-3/4 feet. At f/1.4 the near limit is 24.5 feet.
There is no reason not to use the lowest F stop available.
ISO 800 is a reasonable compromise
As a practical matter the ISO needs to be raised up to the 800 or 1600 range. That reduces the longest exposure at f/2.8 to 2 seconds and at f/5.6 to 4 seconds. At ISO 100 the times would be 30 seconds and 60 seconds respectively. Patience would be a factor.
5 images at 1.6 EV spacing or 13 images at 1 EV spacing
My best images were taken with 1 EV spacing starting at 2 seconds marching up to 1/4000 seconds. While 2 EV spacing works they are noisier.
4513 Halmark Drive
f/2.8; ISO 800; 1.6 EV Spacing; 5 images; 0.3 to 1/1000 sec; 8 added + 11 camera @ ISO 800 = 19EV
f/2.8; ISO 800; 1 EV spacing; 13 images; 2 to 1/4000 sec; 13 added + 11 camera @ ISO 800 = 24EV
The second smage is smoother and sharper. Whether it is because of the reduced spacing or the increased 13 EV range needs to be looked into. I suspect the increased range is most important. DXOMark was used to determine the sensor sensitivity at the reduced ISO rating at ISO 800.
When is enough enough?
The challenge is to capture all the details desired in the scene. In holiday scenes as above, maybe we don't want all the detail under the bushes. Maybe under the bushes should go black. Or maybe not, that is one choice. But the colored lights should have their color fully represented.
In the scene below, the sequence did not go dark enough to capture the color in the highlights.
The picture is well composed and very pretty. But looking at the sleigh, the high light is blown out. The red detail has gone to total white.
And many of the lights have blown out. We know the lights are supposed to be red and green but many have gone to white.
Lightroom offers a clue. If the "White Balance Selector" tool says "Cannot set the white ballance . . . ." it means that one or more channels is at 255.
Here is the same scene with a little wider latitude:
Rollovers show the sequence of shots. This was ISO 800, F/5.6, 1.3 EV spacing from 4 to 1/60 sec. In retrospect, this should have been 1.6 or even 2.0 EV spacing starting at 4 seconds and stopping at 1/200 or 1/1000 seconds respectively. With 2 EV spacing it would be helpful to add a second sequence 1 EV higher.
Here is a closeup of the sleigh:
Notice we can see red in all the sleigh.
None of the lights have blown out to white. The picture in the three above images would be improved with a larger EV spacing to maybe 1/1000 second.
With some of the long exposures in these sequences, maintainimg camera stability is imperative. The two factors are mirror slap and flimsy tripod subject to wind and passing truck vibration.
Any picture taken between 1/30 to 1/2 second is sensitive to mirror slap vibration and anything beyond 1/30 seconds is sensitive to any vibration.
With mirror slap the sequence is: the mirror pops up, the focal plane shutter actuates, and the mirror plops back down. The plop down isn't the issue. The pop up creates camera vibration.
Other vibrations to be aware of are wind, heavy machinery moving nearby, and the photographer.
There are three ways to eliminate mirror slap.
- Activate mirror lock up in the camera advanced settings. This activates a two step process. The first trigger raises the morror. The second trigger activates the shutter and then drops the mirror. Rapid fire is not supported with this feature. Press the trigger 14 times for a 7 image sequence.
- Activate permanent mirror lockup in the camera advance settings. This feature is available only in advanced (expensive) cameras like the Canon 1D series. Rapid fire is supported with this feature. Not for the rest of us.
- Open live view, if available. Live view automatically raises the shutter. Take your picture sequence. Close live view. Rapid fire is supported with live view.
All tripods are prone to vibrations to some extent. Less expensive tripods tend to be more susceptible than expensive tripods simply because they are more rigid. Vibration problems can be reduced by adding mass to the center column. Many tripod center posts have a hook for just that purpose. The hook makes it easy to attach a weight such as a camping stuff bag filled with rocks and gravel or a camera bag lying on the ground. Extra weight can also be added using ropes and / or clamps.
Inexpensive tripods will have problems with exposures less than 1/30 seconds. Add weight!
The image below is a good example of mirror slap.
Observing that the image is sharp except for the diagonal lights at the longest exposure. We know those lights are supposed to be points. This is an indication of mirror slap vibration.
HDR software is developing rapidly. In theory a 32 bit HDR file has all the color temperature ranges in it similar to RAW files. If that is true than tone mapping software needs improvement. It appears this may have improved in recent releases. It is best to send the your desired color temperature to the 32 bit HDR engine. If tiff or jpeg files are used they should have the desired color temperature. If RAW files are used, most if not all, HDR engines have the option to set the color temperature.
DXO Optics Pro
All the images I took at Deerfield were preprocessed with DXO Optics Pro software. I do not use DXO all the time because of the time and extra steps. But using DXO offers the fllowing advantages:
- Noise suppression. With high ISO, long duration exposures it is important to reduce noise as much as possible. DXO does an even better job reducing noise on difficult images than Canon's Digital Photo Professional. I use DXO preprocessing almost always with low available light scenes.
- It also removes lens aberations all popular body / lens combinations.
- It has powerful optical distortion removal tools (keystoning, horizons, verticals, etc.: . . . . )
- It exports to a variety of formats is available including RAW DNG which is my favorite.
DXO has a holiday offering of $99 from normally $169. Highly Recommended. With discounts it should be even less.
Photomatix 5 just released.
Photomatix is the grandfather of all HDR / tone mapping software. I have not yet fully checked it all out but it looks like the best release yet. It costs a reasonable $99 initially. An upgrade from 4.0 is FREE, from 3 or earlier is $39 or maybe $29.
Unified Color HDR Expose 3
This is interesting software. Expose 1 was pathetic. Expose 2 was better. Unified Color nailed it with Expose 3. If you are looking for "Natural" images with a minimum of fuss, this is the software to buy. It lists for $119.
Also highly recommended.
Nik HDR Efex Pro
This is getting a little long in tooth. The most powerful feature of this software is NIK's "U-Point" technology that allows emphasizing spacific parts of the image. This is occasionally very useful.
Processing HDR scenes - Some things to try
There are two steps to processing HDR images. The first step is processing the fixed 8 or 16 bit files into a 32 bit floating point file. The second step is tone mapping the 32 bit file down to a useful 8 or 16 bit file. The objective is to compress the 21+ EV scene down to 7 or 8 EV available on lur displays and prints.
The floating point engine creates a 32 bit file has all the radiance and color information available in the source files. The 32 bit files can be saved in a variety of 32 bit formats. I finally chose OpenEXR mostly because it was created by Industrial Light and Magic. The thought being whatever is good enough for Star Wars it is good enough for me.
Why save the 32 bit file? The 32 bit file can be tone mapped by any tone mapping software. Why do that?
That gets to the second part, tone mapping.
Some 32 bit engines are better on some scenes than others. And some tone mappers can be better on some scenes than others. This doesn't happen often. But it is a nice capability to have to get the desired image.
And don't forget Photoshop! Photoshop can add some layers to 32 bit files. Sometimes it is helpful to add an exposure
gradient or adjust the vibrancy or saturation before further tone mapping. Layer masks work as expected.
In short . . . it is a brave new world out there. We are just scratching the surface of High Dynamic Range.