Paul C. Buff, Inc. Technical Forum

Thank you again Paul/Luap for your contribution and, especially, to ventana for his hard work on color testing. The results from those tests are interesting and may prove very valuable; however, because of some of your findings of an upward direction of the color temperature (towrds the blue/cyan end of the spectrum) when going from bare head to fully diffused softbox, the accuracy of your data is thrown into some doubt. I'm not saying that it is inaccurate; I'm only noting that it is the exact opposite of what one should expect - that color temperature, from bare bulb to fully diffused softbox, should fall, not rise, because the most used diffusing cloths have a warming effect, not a cooling one. Perhaps the cloth used by PC Buff is different, perhaps not. We will have to get more data to intrepret this with any more accuracy.

Just a reminder here - I don't care very much what the exact color temp of these various flash unit/light modifiers are; what's really most important is by how much, if any, they differ from one another. That is where the real headaches occur in trying to run a controlled color shoot. All my lights on the product could be 4000 K or 7000K; it wouldn't matter at all. What matters is that one might be 4800 K and another might be 5400 K - that is trouble.

I have had to produce ridiculously color-controlled images for many clients - probably the most finickey of those have been auto interior prototypes of major auto design studios; for those clients I had to take measures that would make you laugh and cry at the same time. They brook no inaccuracy in the color of their textiles, rubber, plastic, metal, paint and dyes, when overall corporate approval is on the line. And it is hell, but a hell I've travelled through many a time, to try retouch an area into perfect color conformity which is receiving two or three different colors of light simultaneously and mixing at different rates over several different gradient patterns. Well, enough about me. I'm just mentioning this to indicate that I am well familiar with both the efforts to minimize these color differences and mitigate them when they happen anyway. If someone were to offer to pay me for figuring this all out, it would be nice, but, as it is, I have had to deal with it as just a small part of the whole production process.

Even the barest of reliable and accurate information on color uniformity and consistancy among one supplier's (Buff's) whole range of diffusing light modifiers (softboxes and PLM's) and their options would be of enormous value.

Regards

Dafrank,

How good are you at precise controlled color measurements? I am the only person in my company that can do this accurately and it's an immense job I don't have time to tackle right now. I would consider paying some one to do this, but there's not one in a thousand photographers with the technical skills to do this accurately, and there is little in the way of accurate test equipment on the market.

I have been extremely conscious of color response in my work during the past couple of years. This whole subject started coming to the fore front over that period. You would be surprised at the range of variation among essentially all pro photo products and I consider Paul C. Buff, Inc. among the best in this area. I buy a lot of "high end" products for comparison and could pass on a lot of horror stories if I chose to.

Luap/Paul,

To answer directly, I have had to become very good at it, albeit with MY OWN EQUIPMENT. I would say that I am definitely one of those one-in-a-thousand that could do it. I have had to develop ways to practically control color by trial and error that work for my equipment in certain environments or have had to create those environments from scratch.

As to using a meter for such work, while there may be a very esoteric model that could measure flash light output precisely enough, it is not the lack of precise metering that is the essential problem, but lack of a controlled environment that influences results. I own a Minolta II color meter with the Flash Color Receptor accessory, and, aside from the more critical environmental factors, for this kind of work it is only a crude and sometimes faulty tool. It is best used to roughly indicate the differences between one set-up and another, but not to very accurately measure any particular set-up with perfect precision. Besides, color temperature alone is only one measure of actual color, as you can easily see when you look at the results from the Minolta meter itself; two light sources can read as the identical color temperature in defgrees Kelvin, but also read as to require quite different CC filtering. So, practical measurements, as you somewhat already described, are what are most effective, customized to some sort of semi-official versions of particular camera sensors and raw software and camera profiles, or even, in the case of film, emulsion batches and chemical processing techniques. So, if a valid test were to be done, it would have to be specified that it was for a particular set of parameters, while indiviuals not using the exact same parameters would have to know that "their mileage might vary."

If you are implying, and I think you are, that you might entertain the notion of actually hiring me to develop a controlled test of your eqipment for these factors, I would probably say that I would agree to do so, if you should wish to do this, and if my time is available when you would like to have such testing conducted. Of course, many arrangements would have to be made, as well as planning for the exact location in which you would prefer to have such tests done.

In any case, if you would care to pursue this, let me know on the forum, and I'll contact you by phone and talk, or at least leave my email address for further discussion.

Regards,
David

For what it's worth, I just did some (reasonably) quick testing with an AB800 and Foldable Stripbox.

To measure the color temp, I used two methods:

• direct measurement from a Gretag Macbeth (now Xrite) spectrophotometer (Eye One Pro, if anyone's interested)
• shooting an image of a grey card (a Whibal card) and looking at the raw image data to see what the red and blue coefficients would need to be to balance with green, then comparing to the manufacturer's specifications to translate that as closely as I could to a range of color temperatures

When shooting bare bulb (no reflector/shield/modifier other than the AB800's baseplate), I saw a range of readings between 5828 and 5862 with the eyeOne. Adding the Foldable Stripbox lowered the readings to a range of 5583 - 5601. So, basically a difference of about -250 kelvin when adding the stripbox.

Translating the coefficients to a color temperature is not as exact a science as one would like, as the red and blue scales don't always line up nicely. For the bare-bulb test, I consistently came up with a red coefficient of about 1.655 and a blue one of 1.475. This would seem to match Nikon's ratings for a temperature somewhere between 5800 and 5950. So, the eyeOne measured value in the mid-5800s seems plausible.

Likewise, the coefficients for the softbox shot were 1.636 (red) and 1.551 (blue). This one gave a wider range (somewhere between 5400 and 5650), but still roughly coinciding with the eyeOne measurement.

These shots were done in a dark, neutral-colored room on a table covered with Savage super-white paper. I checked the paper with the eyeOne, and found it to have a Lab reading of 74.4/0/1.0 - so the paper isn't likely to have contributed any "coolness" to the readings. If anything, a very slightly warm tone instead.

Just wanted to add yet another data point to the discussion!

Regards -

I have replied privately with dafrank. Perhaps we can get some good independent testers involved in this.

I'm not a huge fan of X Rite. Their color cards are all over the place from patch to patch. I can't speak for their spectrophotometers. I've seen too many cases where this sort of equipment was sworn by and then seen terrible results in the real world.

As the final word, I use the same methods as most customers do . . . use the best pro grey card I can find (Kodak works best for me) and a d300 and a lot of RAW testing, coupled with evaluation of the images (greys in particular) VS known neutral on Mac monitor.

100 shooters can measure an Einstein of any other 5600°K light and give you 100 different readings between 5000° and 6200°. My four pro color meters vary by 500° from one to the other at the same position with the same light.

I hear you, Paul. I believe the eyeOne spectrophotometer to be a good tool (not to mention expensive), but it measures spectral power of light in 10nm steps. This may not be enough to give a precise reading for all light types (some light sources have significant spikes in energy levels smaller than a 10nm gap can capture). That's why I wanted to check the results in a secondary way that didn't include the eyeOne (in this case looking at the raw data). I was glad to see that both measurements seemed to indicate the same thing.

The number of variables in measurements of luminosity and color variance are immense, and I believe that most people underestimate the impact of environmental reflections/spill on the final image.

That notwithstanding, I've found that I'm still able to achieve consistent results when using the tools at hand in a consistent manner. Although I've never gone through the steps of attempting to measure my lights, it was nice to see some consistency when I did this recent exercise. And the results matched up quite well with my typical shooting results.

Most of my shooting is for relatively small products in a controlled environment, so my workflow includes shooting a reference shot with a grey card at the beginning of any sequence of final shots for a given setup. That reference shot then provides the basis for both color temp and relative exposure level from one shoot to the next. That's important for me, because the shots I did for a specific client two years ago with one product need to be relatively coherent with the new shots I'm doing for them today. In other words, if the new label's slightly darker and more orange now, it needs to be slightly darker and more orange when viewing the images side-by-side, even though they were shot years apart. If I simply process the shots without a common reference, they'd potentially be all over the place.

So for me (and I assume dafrank as well), the actual color isn't quite as important as the consistency and predictability from light to light and shot to shot. If it's consistent, then we can use a good process/workflow to insure that we get what we need. I have every reason to believe that you are striving for that as well. Until someone can prove otherwise, I'm content from what I've read from you and others that you have achieved that goal with Einstein. Hopefully others can step in with outside data to support that.