OK, so you have moving subjects and a huge dynamic range of light. How do you make an HDR?
N.B. The bottom triptych in the image shows the three versions of the exposure I split from the initial RAW file, which were used to blend the HDR. The two photos at the bottom of this page are comparisons of two different versions of the original, merged HDR made in Photomatix. See photo captions for details.
Learn photography with us in Tokyo:
Charles, one of my students, came out for a night photography lesson in Shibuya on Tuesday. We did various things as excercises with using a tripod, including some shots from the overhead walkways near Shibuya Station. The main issue there is that the walkways have some flex, and they bounce… making long-exposures like this a bit tricky. Over a 30-second exposure, the flex in the walkway can throw the focus right off. So, we found a slightly more stable section of the overhead area – near where the lift is situated.
HDRs in this situation, where there are lots of moving subjects, can be done by bracketing but you end up potentially with lots of horrible artifacts from where the different frames have recorded the objects in different places. We tried this, and the result was OK but would have taken a fair bit of editing to make great.
So, shoot a slightly over-exposed RAW file. Why? Well it’s better to have slightly blown highlights than shadows which are too dark, as the camera’s sensor is recording more data at the bright end of the spectrum. Lose the highlights a little and you can get them back with exposure-compensation when you process the RAW file. Lose the shadows, well they are gone. There simply isn’t any detail there to reclaim.
I opened my initial RAW file in Capture NX, used the exposure compensation slider to make two versions: one-stop over, one-stop underexposed. I save these as JPEGs in as high-rez as possible. Photomatix works quicker with JPEGs. I lso make a JPEG of the original file, as that is my ’0′ value file.
Then I open the three pics in Photomatix and blend them. Voila! HDR from a single RAW.
This method usually only works great my using three files; i.e. two at one-stop spacing from original. One-stop is best because if the difference in brightness between the files is too great, you will tend to get very lurid colours and extremes of brightness.
Charles shot some pics, I shot some. This one you see was a 20second exposure at f/20… which is way too stopped-down for this lens, as at that aperture distant objects get soft. But, this shot had the best traffic trails on it. f/11 is about the optimum level of dof and sharpness on this lens. But, when I was shooting at f/20 I had forgotten I had a neutral-density filter in my bag. Doh! 20seconds was a good length of espoxure, so when I had remembered about the filter, I shot a bunch more at f/11 and 20seconds. Much sharper but the traffic had thinned-out a fair bit so the trails weren’t as good.
At these sorts of junctions, you have to watch the traffic-lights and stay a while before you shoot, to see the flow of the traffic anf the timing of the lights. This was by far the best shot, as it has the trails from traffic turning in two directions, which really makes the composition pop.
So, what is the difference betwen splitting the RAW file and making three, which you then merge in Photomatix. Or, opening the original 14-bit RAW file direct in Photomatix and letting the program de-interlace, de-mosaic it and make a ‘pseudo-HDR’?
The split, three-file version [left] as compared with the de-interlaced version [right]. I think the splitting version wins it, very marginally, on luminosity and contrast. But they are very, very similar and it took way less time to make the one that used just the RAW file. Click each for larger versions.
Tokina 11-16mm f/2.8
One RAW file, split into three shots, 1-stop apart and then blended in Photomatix
Change to red and blue channels in Photoshop, to bring out the red of the car lights.