1

I am trying to take a dramatic picture of a lit CFL with my 6-year-old Canon Powershot SD750 equipped with CHDK. It's pretty common simply to turn the exposure all the way down, but I wanted the whole image. As I don't have a tripod, I put my camera lens-up on my desk.

My first attempt was a 10-picture bracket (+/- 2EV, I think). It was incredibly noisy. I processed the image in Luminance HDR and then NDNoise and GIMP.

I did not like that very much. The aggresive noise reduction cut some detail, and the tone mapping made the bulb look really dusty and spider-webby. My next attempt was to blend 10 raw files (-5 to 4 EV) in TuFuse (JPG):

Then this went to Photivo for some dynamic range compression and local contrast enhancement (and some sort of sharpening, I think).

Then to NDNoise for noise reduction and edge enhancement.

And some minor tweaks in GIMP.

It fulfills all of my requirements, but I somehow expected the photo to be better. How can I make this image more interesting, and what are some general tips for photographing extreme dynamic ranges?

  • Are you trying this in the dark? Take the shots in normal indoors lighting or even in stronger light than that to reduce the extremely wide range. – Esa Paulasto Jul 3 '14 at 20:05
  • @Esa No, in slightly sub-office lighting in my bedroom at night. I'm not sure why the back ends up so dark. – Simon Kuang Jul 3 '14 at 20:06
  • 2
    The lightbulb is very bright, as you know. Your sub-office lighting really is quite dark compared to that bulb. In order to make the photo more interesting you should first decide what it is that you are trying to capture. Try to see your photo in your mind's eye before taking the first shot. – Esa Paulasto Jul 3 '14 at 20:12
  • @Esa Can you recommend some pathways to develop a similar shot? – Simon Kuang Jul 3 '14 at 20:17
  • 5
    I wonder if this photo is not that interesting because light bulbs are just not that interesting? You can expose a subject in as many different ways as you like, but you can't polish the lipstick on a pig, as the saying goes. – ElendilTheTall Jul 3 '14 at 21:05
19

In the immortal words of the late National Geographic photo editor Bob Gilka, "Kid, if you want to be a better photographer, you're going to have to stand in front of more interesting stuff."

That said, welcome to the sometimes not-so-wonderful world of the commercial/industrial photographer. As often as not, making a dramatic, exciting picture of something not unlike an ordinary light bulb in a relatively plain fixture (neither of which is in quite the sort of pristine shape you'd prefer) in such a way that it can be the eye-catching cover of a catalogue (well, eye-catching to people who are into that sort of thing, at least) is what puts food on the table. Hobbyists may scoff at the choice of subject matter, but then hobbyists always have a choice, don't they?

I suppose the first thing to do in a case like this (other than previsualizing) is to disabuse yourself of the notion that you can get it all in one shot without a lot of additional lighting and test exposures. (There was a time when that was the only real choice, and if you have the equipment it can still be the economical choice.) And by "all in one shot", I mean "all in one HDR sequence" as well. It's not that you'd need to shoot the sequence multiple times, necessarily, but that you'd need to develop it multiple times to get the best possible rendition of different parts of the image ignoring everything else in the picture. It's not like you're trying to compress the dynamic range of a scene with a long, smooth luminosity gradient (only the inside of the reflector bowl matches that description) — you have a picture with some very abrupt and radical jumps in luminosity between well-defined, distinct and easily-masked regions. Treating each of the regions as a separate problem to be solved will both save a lot of headaches and ultimately result in a better image. You have an editor that will let you combine the best elements of several images, so why not leverage its capabilities?

All of that work will get you a "reality" shot. And that may be what you want for personal or editorial work. If you're looking at something more along the lines of a commercial shot, there's still a lot of retouching to do. There always is. That spun aluminum reflector bowl is going to have scratches that run at an angle to what's there by design, and they are going to catch the light. Those scratches will be there in every single production example of the reflector, but they can't be there in the catalogue shot or people will think shoddy or used. The zinc chromate primer is uneven (it usually is; the surface is meant to be painted to match the decor). The bulb will have shadows from condensed mercury that you can't get around by cleaning the outside of the tube (although letting the bulb run for a long time before shooting can eliminate most of them). If there isn't dust here and there to deal with, there will be a prominent fingerprint that's somehow managed to etch itself into the surface somewhere. It's fiddly, painstaking work, but if you do it right, it will still look like you "have a really good camera" rather than like you created a 3D rendering.

  • 1
    Great answer! By the way, though, that's not aluminum: it's "frosted" plastic. – Simon Kuang Jul 4 '14 at 3:26
  • 1
    It's funny you should mention that last line, as as 3D renderings have become better and more realistic, they are increasingly being used instead of actual product photos! – Matt Grum Jul 4 '14 at 15:09
  • "Increasingly" is relative; for the most part it's still cheaper and faster to do the photography and retouch than to model (if no comprehensive model already exists), texture, light and render. People assume that they're looking at renderings when they're not, either because the retouching is just a little overboard (which is very often the case) or because they can't imagine how the photography was done. The bag o' tricks used in studio may be large-ish, but it's not big enough to include things like lighting tents/boxes and so forth that actually make things worse. – user28116 Jul 5 '14 at 4:59
  • 2
    The internet echo chamber doesn't help at all; people sharing their "reckons" just results in the equivalent of cargo cult science. There are only four things to know for product photography: reflection and refraction (and at a fairly elementary level at that), how to get adequate depth of field, and how to retouch. The rest is trial, error and patience. – user28116 Jul 5 '14 at 5:10
  • Funny thing is that if you go over the the blender SE, you will find that people are actually adding surface imperfections to renderings to make the more realistic and more like real photos! – 10 Replies Dec 14 '17 at 2:41
4

There is an article on strobist that goes over shooting a CFL bulb. The author forgoes HDR and the like and just uses speedlights. If you don't have access to strobes and are only using continuous lighting, then you can still balance the CFL with your other lighting; simply set your camera up for a longer exposure (stopping down the lens, low iso, and low shutter speed, e.g. 5 or 10 seconds). Start by figuring out what exposure time gives you the best results for lighting the hood of the light with your continuous lighting. Once you have that exposure time, move on to the bulb: turning it on then off for a different amount of time during each exposure. Eventually, you'll find a better balance between the two light sources.

If you're looking for inspiration outside of product photography check out the Neue Sachlichkeit photographers, particularly Blossfeldt.

1

The lightbulb looks overexposed, the dirt on the lightbulb which is so clearly visible in the first image is no longer visible in the final image. The whole point of HDR is to make all the details in both the extremely dark and light parts to become better visible.

0

The problem you are facing is that you want to take a circumstance with extremely high dynamic range (more than any digital camera sensor can handle) and make a photo with a more "normal" perceptible dynamic range. You have a few options here...

  1. Compress the dynamic range in the actual scene by adding lights. See @moorej's answer and related link to the Strobist article. This takes quite a bit of proficiency to do right, not to mention a bit of extra gear.

  2. Composite images taken with the bulb on vs off. In Photoshop (not sure what the terminology is in GIMP), you'd be using each image in a separate layer and using layer masks to mask in/out the parts of the image that you want from each frame.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.