Archive for the 'canidae' Category

but Adam, how do I fix white balance?

‍‍י״ג אייר ה׳ תש״ע - Tuesday, April 27th, 2010

A simple tutorial for a girl I know who asked me the question above – and I didn’t have time to answer cause there was so much to say and she was tired and overloaded and everything.   Time to answer.

I never have enough of that one.

Ok, so you get back from your photo shoot… and all your pictures look orange.

WTF?

In the old days, you bought film and made choices.  You would choose film speed (ASA/ISO, and perhaps DIN if you were a Nazi or are collecting Social Security), film size (35mm, 120, etc.) negative or slide, color or black and white.  If you chose color you had another choice – what type of light would you be shooting under?

Our brain makes things we know to be white look white under a range of lighting situations.  Film doesn’t have a brain. See, if you were outdoors, there was a film to make a pure white subject look white when lit by (midday) sunlight (5600K); if you were indoors, you could choose film balanced for Tungsten B lights (3200K) and even, and rarely, Tungsten A (3400K).  In an ideal world, our indoor lighting would be at the same color temperature-

-wait. I’m not going to get into the physics here, but you do need to know something about “color temperature” – and this is all you need to know, at least to start.  Those numbers above are in Kelvin (absolute temp= Celsius + 273.15). What heat has to do with any of this is not important.  You just need to remember the following five things:

  1. Sun = 5600K
  2. Cheap, powerful inefficient lights (Photofloods) = 3200K
  3. Expensive, powerful efficient lights (HMI) = 5600K
  4. The lower the number (temperature), the redder, and yes, confusingly, “warmer” the color
  5. The higher the number (temperature), the bluer, and yes, “cooler” the color

Now digital is awesome, because the sensor doesn’t actually see color, so we can make it up as we go along.  We can tell the sensor what to consider white, and then it figures out the rest of the palette from there.  Not only do we not have to worry about having the wrong, uncool film (pun mildly intended) in the camera, but we actually can balance for any temperature light source – not just the three options from the film days (and don’t ask about multiple light sources with different temps in the same shot.  That’s for later.)

But.

This is all well and good – important stuff to know – but your model still looks like a tangerine.  In all your shots.  And she’s already gone home or out to party with the band and do blow.  There is no reshoot.  You accidentally set the camera to shoot for something like daylight, but you were using cheap continuous bulbs from the hardware store and don’t even know what color temperature they are.
You are fucked.

Maybe not.

Now, the following steps will work with JPEG and RAW images, but it is far better to do this with RAW images.  There is no quality loss, and if you have to fix the exposure (because of human vision relating certain colors to luminosity) there is much more latitude and shadow detail to bring back.  For our workflow, we’re going to be using Adobe Lightroom (version 2.7 here), because we want to fix an entire photo shoot’s worth of images.  Photoshop is better suited for in depth correction of a single picture.  We’re going to fix all the shots perfectly – no estimation or guessing – in about one minute – assuming we took one simple precaution.

We’ll get to that.

Let’s have a look at our orange model.

Tangerine, Tangerine

Pretty, ain’t she?  But even she would agree that while warming filters are flattering, she is plenty orange enough without the help.  In fact, not only was this shot screwed, so was this:

This one too.

Hell, the prima donna  won’t even look at us until we have this taken care of.

Not looking at you.

So, let’s fix this.  See, we were careful and using a color balancing card in one test shot.  A 18% gray card would be fine, and another method could use a white card.  (Yes, I was going to do both as you might notice from the Lightroom screenshots, but it’s 3am now [edit: 4:50am])  Gray cards are traditionally used for establishing a exposure level for a section of the image with a reflective (often a spot) meter, however, as they are color neutral and fairly standardized (there are other mixes of gray, but 18% is the traditional choice),  you can use them as the basis for color balancing.

In Lightroom, this is very easy.  Below is a color calibration card set; there is also a pure gray only card, but this is fine – the colors are printed on a 18% gray background.  That’s all we’ll need.  This is a fancy card with at least four ways to do color balance.  Forget that.  This method will work on a $2.49 card.  (In other words, you need this and cost is no excuse.)

An ounce of prevention…

Ok, great you took this shot.  So you can fix this picture because you have a neutral tone that you know. “Wonderful,” you say, “but I need to fix all the shots.  Even the ones with different exposures, lighting angles, and with no pretty calibration card sitting the frame.  What now?” Well, here’s the thing.  Your camera was set to 5400K (warm daylight type fluorescent bulbs).  There was only one source of light in the room – an incandescent light of unknown color temperature.

Since there is only one type of light source, the difference between what your camera expected and what was there is exactly the same in every shot lit by the lamp.  If you needed to drop the color balance by 2,550K (which you will in this example), then all the shots need the same correction, regardless of how much light actually was reflected in the scene and regardless of the color of what reflected it (sorta, but again, that’s for later).

So to clarify – if your camera thinks the white is cool (a high number) and it’s warm (a low number), we’re going to be subtracting.  How did I know that I was subtracting 2,550?  And isn’t this is taking way longer than a minute to explain?

Well.  Now that you know the problem and you have taken the one precaution YOU WILL ALWAYS TAKE IN EVERY SHOOT, that is, YOU WILL ALWAYS HAVE A SHOT WITH THE MODEL HOLDING (AT LEAST) A GRAY CARD WITH EVERY LIGHTING COLOR CHANGE, the rest is very quick.

We’ve imported our shots into Lightroom and from the Library, we see the problem in all it’s ginger glory:

Red Read Red

Ok, so we see we have the important shot right there – a card with our preferred neutral color, 18% gray.  We click on it, and then go to Develop in the top right corner.  This will open up that image alone and give us more and finer tools to use.  When the mouse hovers over the image, it turns into a magnifying glass. We click on the part of the picture with the card to zoom in.  Now, on the right, you’ll see the Basic (very top) panel is open, set to Color (default), and there is a circle with an eyedropper.

Click on the eyedropper.

The tool we need

Now, that the mouse has turned into an eyedropper, we can move around and see the effects of picking various tones as the white balance point in the preview in the top left corner. We just want things “normal.” So, find the most even sample of the gray in the card that you can:

The place we need

And just click… and:

If it ain’t white, it ain’t right…

Now, that looks right.  But I said we would fix all of them in a minute, and even if you went slow, I should have 30 seconds left. More than I need.

While still in the Develop section on the picture we just fixed, right click on the picture (either the big one you just worked on or the small one in the filmstrip), and find Develop Settings/Copy Settings and click…

Where do we go…

Now we are going to take the important development change we made here (the white balance) and copy it.  Hit Check None at the bottom and then check the White Balance box (not suprisingly, the very first option, from an English reader’s perspective).  Click Copy.

Oh where do we go now…

Now that we have the change that we want, the rest is pretty obvious.  Go back to the Library View (top right corner) and select all the pictures that were from the late night photo shoot.  Don’t worry if you have the “fixed” picture in the selection.  We are going to paste a calibration number, not just “subtract” a number from all of the images (but that is the practical effect).

Where do we go…

Now we just do what we did to copy, but instead choose paste –  right click on any of the selected images and find Develop Settings/Paste Settings and click…

Oh where do we go now…

And just like that-

Sweet Child / Sweet Child O’Mine

Our model has her fluffy white hair back in every photograph… and is now ready for you to mess with her colors, but as you choose.

Addendum:  This is the most basic and simple method to get a decent working white balance after the fact.  There is always a gray card around, but it might not be perfectly neutral (though this is more common with the rise of digital cameras).  Color panels and white/gray/black card sets exist for a reason.  I use one of these. I have that with me at all times.  If I know I’m going to really be doing complicated color work with time to set up, this bigger and more versatile card is a lot more flexible (and my colormeter might come along).

A minute proposal.

‍‍י״ב כסלו ה׳ תשס״ח - Thursday, November 22nd, 2007

A month ago, in a review of the n810 on Maemo Apps, Jonathan Greene suggested that microb, the Gecko 1.9 “browser” for the Maemo platform, needs a branded name.  I wholeheartedly agree. In fact, I wasn’t sure whether to type microb or Microb or MicroB in the first sentence (the package name won out) as I have seen them all used.  I have a suggestion.

Fennec.

The littlest (red) fox around.  Actually, they are the smallest Canid around.  Hardy too, they survive throughout the Sahara and Arabian Peninsula, though are a popular target of hunting.  Drawing its water from vegetation, the fennec can survive without drinking for extended periods. They can be domesticated with some work, but that is not a task for those seeking the path of least resistance.

Sounds just like the mobile Linux experience.

Oh, and they are cute, too.