simple technical photo tips

If you want to take a technical photo, a photo that is useful for product or tutorial purposes, there are a few very simple techniques that will improve your results.

One: digital cameras need a lot of light.

Photosensors essentially build up an electrical charge when hit by photons. The fewer the photons, the worse the signal/noise ratio. You want the most photons you can arrange, so plain old transistor heat noise is held at a minimum by comparison.

Also, if your camera can be set to overexpose or underexpose shots, and also show you a histogram of pixel values after your shots, then you should use these features together. Expose low enough that brightly-lit white objects aren't getting "clipped" (pixel values mashed to pure white with no detail), but expose as high as possible otherwise. This again improves the signal-to-noise ratio. Then you can adjust to taste in (#5) below.

Two: put your camera on a tripod.

Electronics guys should be aware of how LONG a few milliseconds really is, so "I can hold this thing still for 1/60second" should sound completely silly if you think about it.

Three: the larger the source of light, the softer the shadows.

Not many people have a 2x1 meter photo studio diffuser strobe lamp in their house. However, most people have a nice white ceiling with a matte surface. Bouncing a strobe light or even a halogen desk lamp off a large white ceiling or large white cardboard reflector does wonders for softening the shadows.

Four: make sure it's in focus.

I can't tell you the number of times I've seen people hold a digicam up to a tiny device from about two feet away. On their 2 inch preview screen, it seems sharp enough, but when you LOOK at it, the image is totally out of focus. The camera manual will tell you the closest distance the lens is capable of focusing.

Five: post-processing is essential.

The goal of the exposure-judging circuitry in a camera is to ensure that the AVERAGE PIXEL VALUE is MEDIUM GRAY. That's it. That's all there is to it. There are some specialized definition of "average" on better cameras, but every camera since 1960s with an electronic exposure meter has aimed for that.

If you take a photo of a snowy scene, or like my product shots, a photo of a white background, that will end up looking dingy gray in the end. This is what the camera ASSUMES is the right thing to do, but you know better. You want white things like snow or backgrounds to look like white.

Most photo editing software will let you adjust the "levels" or tone "curves" of the image. You should adjust the curves so that the brightest parts of actually-white objects have the maximum value, and the dimmest parts of actually-black objects have the minimum value. This usually increases your scene contrast, which also boosts the color saturation.

Released (cc) Creative Commons Attribution Only
Ed Halley
[ e d @ h a l l e y . c c ]



For the gear-heads out there who want more information about my camera, settings, etc...

These were taken with a Canon 5D (mk I) camera, using a 550 EX Speedlight offcamera. An ST-E2 controller was used to trigger the Speedlight, but I kept the flash in "manual" mode to avoid underexposing the white backgrounds. My lens was the very common and versatile Canon 28-135 IS USM, at the 135mm end of the zoom. The shutter was set to a fixed f/16 aperture (for maximum depth of field) and 1/125sec (since the flash determined the exposure).

The tripod's not interesting, but I have a Gitzo Explorer that lets me swing the camera overhead or back a little bit for angle shots with minimum hassle. The tripod is important to avoid hand-shaking vibrations, and also makes it MUCH easier to do multi-shot composites.

I made a very simple stage for these shots using a plain white foamcore panel (found at any art store or even Wal*Mart), about 18" x 24". I aimed the strobe upward and bounced it back down to the scene with a second plain white foamcore panel. The hero parts were laid out and wires arranged to avoid overlapping in confusing ways. Tutorial shots were shot straight down (so you can read the Arduino pinout labels) but product shots were at more interesting angles). I try not to forget to dust the scene because any lint, grit or hair just means more retouching.

I used RAW format to ensure I could adjust exposure with 12 bit data. Editing curves in the usual 8 bit JPEGs is possible but leads to graininess, especially when you do multiple adjustments. I used Bibble Labs RAW toolset to achieve the basic exposure adjustments and rough cropping. I balance out the color temperature so white looks white (not flash blue and not tungsten orange). I pull in the contrast so the foamcore background is mostly pure white, without sacrificing detail in the actual parts. Bibble also offers Noise Ninja and dead pixel corrections.

From Bibble, I save JPEG and do any retouching in the GIMP (Gnu Image Manipulation Program). As of this writing, GIMP is still limited to 8 bit color channels, but I don't need to adjust exposure curves so this generally doesn't matter. I use the corrective direction to rotate parts if I want them to look level, then crop everything down more. I clone out any remaining lint or plastic defects that I don't like.

In the cases where I want to show a light being lit, this is hard to do (and even harder to do convincingly) with a white background. I take two shots, one with flash and one without flash. I lay the images atop each other as separate layers, make one of them semitransparent to get them to line up accurately. Then I add a layer mask and try to allow just enough contrast from the lit-bulb version to combine with the flash-white version so it looks somewhat faithful to the way our eyes perceive color and light.

Anyway, not trying to distract the Arduino topics, but people were starting to ask about specifics. Instead of throwing other threads off-topic, I opened this one. Thanks for the interest.

Nice work and excellent information. Thank you VERY much.
:slight_smile:

Yes, indeed. Thanks very much, halley. It's nice to see all this information collected in a single place.

Mikal

Thanks for the info. If you're going to post beautiful pictures like those, people ARE going to want to know how you take them!

One question - you mention adjusting so white is white in post processing. Can't you also do that at exposure time using either manual exposure or the +/- "adjustments" available. Sorta the old "zone system" principles; if you're scene is whiter than 18% gray, you expose for longer than your meter says so that it actually LOOKS white. (I thought your description of the histogram feature would accomplish the same thing, but apparently not?)

Ok, two questions: what about fixed lighting vs flash? I don't think I trust most of teh digital cameras to do the right things with the flash. Or not the things I want it to do, anyway, such as reacting appropriately to apreture or shutter priority automatic exposure.

Your first question is mostly true. You can overexpose the whole scene until the white background is truly saturated to pixel white (the term is "blown").

If I were shooting a larger product, I could have a separate flash that is dedicated to the background alone, and I'd pump it up so it blows out the background with two stops (four times) the light intensity of the rest of the product. And yes, the histogram on the camera would show strong blast right up against the right edge, showing a lot of pure and almost pure white overexposed area.

On a small product, one way to do that would be an elaborate setup where I put the device on a bit of plexiglas, so it could float over the background far enough to separate out the lighting. But then I lose the shadows on the white area too.

If there's white stuff in the shot, and you can't separate the background with dedicated illumination, then you get a lot better control in the computer than in the camera. It's really hard to see what areas will really overexpose and what won't, on the camera's little screen, even if it has those blinky warning pixels. Also, I could combine two shots: one adjusted brighter for background, one adjusted less aggressively for the product.

As for your second question: not trusting your camera to do the right thing. I agree. But flash is no different from fixed lighting, other than having a lot faster exposure times (less chance for vibrations to blur the shot), and less heat (fixed lighting is usually halogen or incandescent).

The camera only has a few metrics it can use to make decisions. The camera's only intent with regards to exposure is to ensure that the average pixel is middle gray. This extends to the flash: if you let the camera control the flash, it will see a flood of white and throttle back the flash power so as NOT to blow out any part of the image. My camera does this by taking a tiny unseen test shot with a lower-power pre-flash. If doing a product or studio shoot, I assume full control over the flash (and all other camera settings). Manual mode everywhere. Another benefit of turning this off is there's no preflash, so I can use additional "dumb" light-triggered strobes instead of expensive "smart" IR-protocol-driven strobes.

There's an article in the latest Make magazine about taking good technical photos. The technique shown there involves placing the object an a piece of glass, so that no shadow is cast below the object. Also, this makes the background out of focus (it's further away), so that texture and small blemishes don't show. I haven't tried it yet, but it's on my list of things to do over the holidays!