Now that I've got my MUSIC script up and running, I thought I would write a few thoughts down on Long Exposure (LE) capture.
LE is a photography genre that has wide appeal and is often associated with 'fine art' images, especially in black and white. For example Joel Tjintjelaar
LE is also a technique that gets linked with 'water and/or cloud' photography, ie where the movement in the water and the clouds becomes key part of the image.
LE photography therefore has a strong artistic dimension.
Of course, as we are undertaking LEs, by definition we are seeking to capture movement, eg in water or clouds; but not necessarily movement in everything, eg people or trees. Thus there is a balance we need to seek and that is best achieved by first having a vision of what you are seeking to achieve.
Technically, most photographers will create LEs using Neutral Density (ND) filters. Yes, we may get away with stopping down the aperture, but this should be avoided, in order to minimise diffraction softening in those parts of the image that are not moving.
As an aside, I believe, in time, NDs will become a 'thing of the past', as camera technology evolves. For example, Olympus already offers a Live Bulb feature, where one can 'watch' the LE evolve and stop it at a specific point. But for most, NDs remain the go to technology.
Some shutter control may be achieved by the use of circular polarising filters, but this will be limited to 1-2 stops, plus CPs risk introducing artifacts if we are using them with wide lenses.
Thus many photographers carry at least two to three NDs with them and sometimes need to stack these to achieve a specific LE time.
But NDs come at a cost: financially and technically.
The financial costs are simply a reflection of how much control (number of filters) you wish to have and how much money you wish to spend to offset the negative image quality associated with putting glass in front of your expensive lens, eg tints and image softening.
For example, the Formatt-Hitech “Firecrest Long Exposure Joel Tjintjelaar Signature Edition” NDs comes in two kits: Kit#1 comes with a ND 0.9 (3 stops), ND 2.4 (6 stops) and a ND 3.0 (10 stops); and kit#2 comes with a ND 3.0 (10 stops), ND 3.9 (13 stops) and a ND 4.8 (16 stops).
As for costs: think about $250 for kit#1 and over $300 for Kit#2.
In addition to a single ND's negative impact on image quality, many times photographers find themselves using two NDs to achieve a specific LE (shutter) time. In other words more image degradation.
Then there is the negative impact on the camera, especially the sensor. This is usually observed by seeing increasing noise as the camera remains working over a LE capture period.
Another problem we sometimes encounter with NDs is being able to balance the foreground exposure and the sky exposure. Yes, the dynamic range of modern cameras is incredible and getting wider, but, nevertheless, there are times when the sky exposure is radically different to the foreground's. We therefore may see more noise, especially in the shadows of a single (very long) LE ND shot of a high DR scene.
We can seek to offset the high DR threat in several ways.
First, by estimating a middle exposure and hope this covers the extremes; but this becomes problematic for very LE times, ie minutes of exposure. We thus risk blowing out the sky, or under exposing foreground elements.
We could use a graduated ND as well as an ND. But now we are back into the negative impact of putting too much glass in front of our lens.
In conclusion, although NDs are the usual way to achieve LEs, they come at a cost; which is why some like to have an alternative approach: namely multishot bracketing for a post-processed, simulated LE capture.
Up front let's state two obvious negatives about this approach.
First, there is the cost of additional post processing time. However, as this is likely to be an image you are creating for a reason, ie rather than a snap, I would suggest this cost is worth it. Plus, as mentioned in previous posts, if you are post processing in Photoshop, there are many tools/scripts to help you out, eg stacking and merging.
Second, there is the risk of creating a scene-specific simulation artifact. For example, if you are taking 10, 1 second exposures, ie to simulate a 10 LE exposure, and there is a repeating event, eg it could be repetitive wave action, you may see scene related artifacts; that you wouldn't see if using NDs. One may call these 'LE strobe effects'. Such artifacts may be reduced by simply shifting the base exposure to better match the natural frequencies in the scene, eg by using an ND to extend the base exposure, and, of course, being aware of the LE strobe effect.
These two 'negatives' aside, when I look at simulated LE photography I see nothing but positives:
- I don't need to purchase loads of expensive NDs, ie I mostly find a couple of NDs will cover me, say, a 3 stop and a 10 stop. I find these two will allow me to shift the base shutter into a more favourable zone for multishot LE bracketing, whatever the time of day;
- With or without an ND, I can create any LE simulation time, simply by taking the correct number of images;
- As a Canon Magic Lantern shooter, I can exploit Full Resolution Silent image capture, and extend the FRSP limitation, ie go beyond 15s;
- As I'm capturing n images of x seconds, to simulate an n.x LE shot, I'm not limited to one LE exposure in post. I could post process any number of (non-contiguous) images from my bracket set; thus creating 'any' LE I wish from the base exposure to an n.x LE;
- Related to the above, I can also decide to process any number of images for the foreground, and a different number of images for the sky and/or water. I thus have far greater control of the final image in post. As an example, I could use a single image for the static part of the scene, and paint in the LE parts, eg moving water, having processed these separately;
- Because I'm capturing a base exposure captured at a sensible shutter speed, I'm confident in setting a good exposure, with low diffraction softening, especially, like me, if you are also using an ETTR approach;
- As my shutter speed is not excessively long, I will minimise hot pixel problems;
- Finally, because I'm taking n images, the noise in the image will be reduced by sq.root(n).