Flash versus continuous lighting are we at the crossroads?.

[rjlittlefield]

Here is a link to a thread on the Fredmiranda Lighting & Studio Techniques forum by "gintasr," who made a portrait light out of four-foot fluorescent bulbs with electronic ballasts. Having an interest in DIY lighting projects, a statement of his stuck with me: "For a ballast, you MUST make sure it is ELECTRONIC, otherwise you will be getting flicker and your exposure will vary frame to frame. The electronic ballast will let you use any shutter speed without a hint of flicker." I have not tested his assertion. If correct, it must not apply to CFLs--or at least not to all CFLs.

Right, so I just now pointed my DSLR at a new shop-light fixture with electronic ballast and fairly new tubes. Here's the data, the two most extreme frames from a batch of 20, again with the 1/320 second exposure:



I checked again after a 40 minute warmup -- no change.

It's a great question where this variability is coming from. Using an oscilloscope I checked the electronic drive on the tubes and looked at the light output with a simple phototransistor detector. The light output shows a definite slight sawtooth variation at 120 Hz. The electronic drive is strongly dominated by a 44.9 KHz complex waveform, but there's a small 120 Hz sawtooth superimposed on that too. I'm guessing that what we're seeing is the output of a 44.9 KHz switcher whose input comes from a less than perfectly filtered full wave rectified power supply.

In any case, this performance is clearly much better than the old type ballast, but also clearly not perfect. I'm sure it would be fine for portrait photography so what the fellow wrote on the Fredmiranda forum is correct for his audience.

For the stacking that we do, the situation is less clear. Small amounts of frame-to-frame brightness difference can be corrected nicely by the software, but the color shifts don't go away. In a PMax output, I suspect they would be invisibly smoothed, but in a DMap output especially with OOF background there's some potential for contouring.

Is the CFL you photographed dimmable?

It's labeled "not for use in dimmable luminaires".

--Rik

[rjlittlefield]

Just because somebody might ask... (Who, me?)

Here's the same test done against a cloudy sky.



Presumably the variation here is a bit of difference in the actual exposure time, again nominal 1/320 second.

By the way, these were all with the lens wide open so no diaphragm variation.

--Rik

[Chris S.]

Just because somebody might ask... (Who, me?)

Funny, Rik--I was just reading your prior post, and thinking that variations in shutter speed and diaphragm still needed to be eliminated to really nail this, when in came the notification that another post had been added to the thread.

Nicely nailed.

Presumably the variation here is a bit of difference in the actual exposure time, again nominal 1/320 second.

Can we indeed presume this, or could this amount of variation be produced by the sensor (and associated electronics) itself? I've never thought to check. Also not sure we care, as this variation is much less than the one you demonstrated for the fluorescent lights.

--Chris

[johan]

It's a great question where this variability is coming from. Using an oscilloscope I checked the electronic drive on the tubes and looked at the light output with a simple phototransistor detector. The light output shows a definite slight sawtooth variation at 120 Hz.

If I understand this correctly, you're saying there is a 120 Hz sawtooth. Ignoring the 44.9 KHz complex waveform, would an exposure at 1/120s then not make for optimum uniformity outputs?

[Chris S.]

in other situations (especially with light-weight, dark-colored subjects in regimes requiring a lot of light), flash can induce violent subject motion during even the shortest exposures, and can be an irrecoverable cause of motion problems. Some have described this situation as “rare”—and it is probably so for many photographers—but for others of us, it is a very common problem. Certain subjects simply cannot be imaged with flash

Fwiw, I use flash to get around vibration problems. The biggest limitations I've come across are mechanical ones, of positioning, so I use all sorts of homemade things like barn doors to focus the light where I need it. I'm curious about this quoted section though. What would be some examples of this?

Johan,

Craig, as usual, came up with exactly the right link very quickly.

Here is another post in which I described movement of carbon-smudged glass-filter fibers. While perhaps not the most interesting photographic subject, it seemed a clear demonstration of the mechanics of flash-induced movement, because neighboring fibers without the black smudges did not move, while the smudged fibers did. In another post, Charlie Krebs listed flash-perturbable examples of wider interest to this community: ". . .a subject such as a butterfly wing, midge antenna, or other low mass, unsecured subject. . . ." And here, Rik Littlefield demonstrated the effect with moth debris.

Ironic, isn't it? Flash can be either cure or cause of motion problems in photomacrography.

--Chris

[rjlittlefield]

If I understand this correctly, you're saying there is a 120 Hz sawtooth. Ignoring the 44.9 KHz complex waveform, would an exposure at 1/120s then not make for optimum uniformity outputs?

Sure, if you capture exactly one cycle then almost all the variation will go away.

Setting the shutter to capture exactly one full cycle (or multiples thereof) is powerful medicine for overcoming flicker.

For example, here is our nemesis the old ballast revisited at 1/250 second:



and here is the very same old ballast shot at 1/60 second -- definitely a lot better!



One of the posters in the thread that Chris linked to noted that: "Now if someone does want to use magnetic ballasts (T12) I think they can get around the flicker problem by shooting at 1/125th of a second or slower (not my preferred method)." The part about "or slower" is not quite sufficient; for example 1/80 would still flicker because it would catch all of one half-cycle but only part of the next one.

I used 1/60 for this demo to avoid possible problems with asymmetry in how each bulb lights up on the two half-cycles.

Of course 1/120 and 1/60 are for US standard power at 60 Hz. It would be 1/100 and 1/50 in many other places.

--Rik

[elf]

How about controllable RGB LEDs? http://dx.com/p/45w-300x5050-smd-led-rg ... -5m-139647

Would one or two of these put out enough light?

[johan]

Thanks guys, that's interesting. It'll be a long time before I do anything like 40x, my setup can stretch to 20x but beyond that I really just don't have the right workspace.

[Len Willan]

I have noticed that a GIGAmacro Professional Photography System uses Continuous Lighting and this may be of interest.

The Complete System that is shown in the link below uses Ultra Bright LED with Color Calibrated Lighting with Active Cooling
http://gigamacro.com/wp-content/uploads ... m_1000.jpg
They say,” Our systems cost only $20,000 to 25,000 USD. We custom configure each system to meet specific needs; costs can vary depending on the camera and lens options selected”
For an example of its results:
Boenasa nigrorosea (Lepidoptera: Arctiidae: Lithosiinae) – Rosy folded-wing footman moth from Dominican Republic.

Specimens and identifications provided by Dr. John E. Rawlins and his staff from the research collections of the Section of Invertebrate Zoology, Carnegie Museum of Natural History.
This 364 megapixel macro photograph was created from over 10,000 individual photographs.
Click and drag in the photograph below to explore the specimen.
http://gigamacro.com/2011/rosy-folded-w ... hotograph/
Are Ultra Bright LED with Color Calibrated Lighting with Active Cooling another option , compared to modern wireless flashes for use and relative cost and use
Len Willan

[rjlittlefield]

I have noticed that a GIGAmacro Professional Photography System uses Continuous Lighting and this may be of interest.

I don't know any details of the lighting system used by GIGAmacro. It looks like a lot of thought has gone into the system so I'd be inclined to expect good lighting there.

Are Ultra Bright LED with Color Calibrated Lighting with Active Cooling another option, compared to modern wireless flashes for use and relative cost and use

Well, as written that's a marketing bullet with lots of buzzwords in it. Aside from "LED", there's a lot left to the imagination.

But in any case, my answer to your question would be "Sure." As always, the devil is in the details of implementation and in what you want to optimize for.

LEDs and flashes have very different characteristics. Important fundamental differences are spectrum and maximum instantaneous power. Also important -- though entirely in the details -- is effective repeatability from exposure to exposure.

For spectra, the "white" LEDs that are actually RGB arrays can have awful spectra consisting almost entirely of narrow spikes. No problem, exclude those. That leaves integral white LEDs that consist of a mix of phosphors excited by a blue LED. Those spectra are more lumpy than incandescent, but arguably a lot smoother than electronic flash. (See for example the "spectral outputs" graph at http://en.wikipedia.org/wiki/Flashtube.) Personally, I think there's no clear advantage either way here.

For maximum instantaneous power, flash wins hands down. That's what lets it freeze vibration so well.

For repeatability, it all depends on the details. A well regulated LED driver should be completely stable. A not so well regulated LED driver is likely to have the same sort of 120 Hz ripple that we talked about earlier with the electronic ballast fluorescent. I just now checked my DIY LED system based on a BuckPuck driver and a cheap power supply, and to no surprise I found on it a substantial amount of 120 Hz ripple from the power supply, in addition to the expected (and unimportant) 200 KHz switching component from the BuckPuck. But I've seen substantial variability in flashes too, especially when they're operated at low power and thus have to switch on and off very quickly. We've talked about that problem before, HERE.

If I were designing from scratch, I probably would pursue a route similar to what I see in the GIGAmacro setup.

--Rik

[rjlittlefield]

Santa brought me some JANSJÖ lamps for Christmas, so of course I could not resist looking them with some instrumentation. Quick summary is that they definitely flicker, but much too fast to matter: around 60 KHz = 15 microseconds per cycle. I could not detect any variation at longer periods. For practical purposes they are stable continuous illumination.

I received both pedestal and clip-on styles (HERE versus HERE). Interestingly, they are different colors! I haven't measured the color temperature, but of the two I've checked the pedestal unit is much "warmer" than the clip-on.

--Rik

[Len Willan]

Thank you to Riks Santa ,
The procession moves on , looking forward to more on continous lighting,
At least what you see is what you get !

[rjlittlefield]

The procession moves on , looking forward to more on continous lighting

For your amusement, here is a type of lighting that you probably do not want to use except at the Christmas season.

Not only do they have strong color casts, but they flicker like crazy at AC power frequency!

At least what you see is what you get !

Indeed. I generally have these things lighted when I put them on the tree. The combination of motion and flicker produces some really interesting visual effects. It's best to stay away from the eggnog while doing this!



--Rik