{ illuma·blogspot·com }

Sunday, May 29, 2011

••◊ 5500k fluorescent lights...yeah right!

My previous post about the color shift of the Fader-ND neutral density filter got me thinking about another issue I've run into in the past.  I was setting up a shot last year after buying the Promaster 3-in-1 Cool Light shown below.  It's a simple soft box with four 85-watt fluorescent bulbs.  The bulbs *say* they are color balanced for 5500k, but in actuality the color balance of the camera doesn't come out at 5500k when lit with this light.  Sort of like the previous post, I wanted a short cut to white balancing if I didn't have a white balance card with me and the scene was dominantly lit by this light.


So...  I set up my Gretag White Balance Card and lit the card with the fluorescent light and read the white balance using the histogram on the back of the camera.  Because I didn't have a picture of the histogram for this blog post I pulled the footage into Premier/Photoshop and inserted the histogram.  You can see that blue is much less dominant than red or green.  As expected, green is dominant because this is a fluorescent bulb and that's what phosphorus does.  Even in the expensive professional Kinoflo lights, they have a nasty green spike in the spectral response.  Given all the downsides of tungstens and the bling-worthy prices of LEDs, I'll take the problems of fluorescent lights any day of the week.     


Starting with a white balance setting of 5500k I experimentally adjusted the white balance shift to (B2,M5) to adjust the white balance back to neutral.  As you can see in the last picture, the white balance card is now back to a neutral color.  So now when I'm in a pinch I know to set the white balance to 5500k and shift the white balance to (B2,M5).  No white balance card required.


There's always a downside to shortcuts, right?  Here's the deal with this method - it only works as a full system including the lens.  In a previous post I showed how different lenses from Canon render not only color but also exposure differently, especially if you use color temperature for white balance.  So if I was to shift to a Zeiss lens, for example, the white balance shift may be different.  It will also be different for a Tamron, Sigma, or non-L series Canon lens.  So this cook book white balance shift value doesn't always hold the same value.  The experiment has to be repeated on a lens by lens basis.  Pain for some, but my lens selection is fairly limited so no big deal here.

Monday, May 23, 2011

••◊ How to Feed a Giraffe... No, really.

Last Monday I had a chance to go on Safari at the San Diego Zoo Safari Park.  For a fee they take you out into the nature preserve with a park ranger on a flat bed truck and even let you feed the giraffes.  It's a simple morning breakfast of acacia leaves, but much better looking than their normal brown dry "dog food."  The park hosts 3500 animals with 4-5 tons of zoo-doo removed every day.  Below are some photos that prove once and for all that I am not a nature photographer, however it's always fun to share an experience through photography.


Baby giraffes have an incubation period of 14 months.  When they first come out their neck is all twisted up from being curled up inside momma.  It take a few months for an infant's neck to straighten out.


The park ranger explained that Rhinos form bachelorette groups and break off from the rest of the herd.  These two were part of a 3 female group known as the "plastics" - a reference to the movie "Mean Girls."  They rarely made contact with the rest of the herd.  When they rub their horns together it's a sign of friendship and bonding.


Being a gorilla has to be the most boring job in the park.  All there is to do all day long is people watch.  That job is for the birds.


Sunday, May 15, 2011

••◊ Neutral Density and the Fader-ND, part II

When I was writing my last blog post a question came to mind; "How do I quickly adjust for the Fader-ND's green tint if I'm already white balanced?"  A quick experiment was devised in my living room the next afternoon.  First we'll start with a simple setup that includes the camera in live view and a white balance card.  I did a custom white balance to the card, which is illuminated by simple afternoon daylight through the window.  I figured this was a reasonable test configuration since I typically use the Fader-ND outside in bright daylight conditions.  If you click on the photos you'll see that the histogram shows equivalent RGB values, indicating a good white balance. 


Next I put on the Fader-ND at the "Min" setting, which as we know from the last post means about -1.6 stops of "neutral" density.  Therefore I had to open the aperture to compensate.  I ideally I wanted to keep the exposure at middle gray since this is steepest part of the tone curve and the most likely place to show a color shift.  The first picture below shows the Fader-ND installed on the end of my 24-70 lens.



The surprise here is that we also see red shifting less than blue.  Although there is a slight attenuated shift in red, it appears that the greatest shift is with the attenuation of blue.  My goal here was to develop a quick rule of thumb for working on set, so the easiest solution when I'm not necessarily white balancing to a white balance card is to use the white balance shift/bracket function (WB SHIFT/BKT) built into the camera.  I started off with the white balance shift set to neutral (0,0).  When you enter this menu item it gives you a grid which represents white balance shifts in the typical green, blue, magenta, and red (don't know why they label it "A").


Through trial and error I found that shifting the white balance one step toward blue (i.e. one step away from red) and five steps toward magenta (i.e. 5 steps away from green) evened out the histogram.  Ah...back to normal.


Another aspect of this experiment involved seeing if the colors shift around when more neutral filtering is applied, but the color rendition seemed reasonably independent of the filter setting.  So now when I'm on set and need to make a quick change I know to shift the white balance one additional step toward blue and five steps toward magenta when I put on the Fader-ND filter.  That's a much cheaper fix than buying the Singh-Ray or the Tiffen filters.

Sunday, May 8, 2011

••◊ Neutral Density and the Fader-ND

After taking my camera outdoors and trying to film I quickly realized the need for neutral density filtering. On a sunny SoCal day you're lucky if you can stop way down to f/22 and get a picture correctly exposed. It's just too darn bright... and of course you really want to be shooting with the aperture open to f/5.6. A reasonable priced solution came in the way of the fader-ND product from Light Craft Workshop.


The filter is a two-part polarizer.  You screw in the inner polarizer onto the front of your lens.  The outer polarizer freely rotates to give you anywhere from 2 to 8 stops of light attenuation.  This was a life saver over the last few weekends of filming.  The producer/director wanted to go both indoors and outdoors, which mainly meant just screwing on the filter when we went outdoors and attenuating down the light to get the right exposure.  Below is a picture showing the filter at minimal, half, and maximum attenuation.  This solution isn't exactly a slam dunk because as you attenuate light there is a slight shift toward green.  You have to watch your white balance so the change isn't really noticeable.



  After using this tool quite a bit I was wondering why the filter only took a very slight turn to go from over/under exposed to correctly exposed.  It turns out the light attenuation isn't really linear (in camera stops).  It's more of a logarithmic curve.  I measured the actual attenuation by varying the ISO, shutter speed, and aperture on my lens while keeping a white card at 50 IRE (digital value of 128).  Below is a graph of the light attenuation with each graduated step of the filter.


As you can see the filter starts at about -1.6 stops of light attenuation (graph is inverted for readability).  The LCW documentation claims -2 stops, so with manufacturing margin it's doing what is says.  At the maximum filter setting the light is attenuated about -8.6 stops.  Again, the LCW documentation says about -8 stops, so they are right.  Most of the time when I'm outdoors I need 4-5 stops of attenuation, essentially where the graph's slope starts to get steep.  I guess that's why exposure setting is still a bit fidgety while using the filter.

The Singh-Ray Vari-ND is a higher end version of this product and doesn't seem to have the green shift according to others' reports.  For the budgets and products I work with the Fader-ND works fine.  Both the Vari-ND and Fader-ND are offered in multiple screw mount sizes so they fit many types/brands of SLR lenses.  In my case the screw-on filter mounts are all 77mm, so I only need one of these variable neutral density filters.  For lenses like the Zeiss CP.2's LCW offers a set of 4x4 matte box filters because the mount on those cinema lenses is just too large. I'm not even sure if those lenses have a screw mount.

Overall it's an essential piece of my camera kit now.  It allows me to work faster, better, and produce cinema-like images in just about all filming conditions using standard SLR lenses.

Sunday, May 1, 2011

••◊ Technicolor Picture Style for the 5D mark II, First Impressions

On Friday night I downloaded the new Technicolor picture style for the Canon 5D mark II.  Being familiar with picture style editing for the Canon SLR's I had some ideas on what to expect.  I typically use Faithful, Sharpness=0, Contrast=-2, and Saturation=-1, which gives me a fairly faithful color reproduction, but definitely leaves something to be desired as far as exposure latitude.  A custom picture style won't really extend the exposure latitude of the camera.  That's locked by the image pipe firmware and isnt' tweakable as far as I can tell.  However you can increase the amount of shadow detail by decreasing the contrast via a tone curve adjustment.  See my numerous previous posts about that topic for more information.

So...what can they adjust?  Technicolor can choose a color profile, for which the two viable selections are "Neutral" and "Faithful."  They can suggest sharpness, contrast, and saturation settings (Technicolor suggests 0,-4,-2).  They can also flatten the the tone curve to capture more gradations of highlight and shadow detail as well as tweaking the hue, saturation, and lightness.

In order to test this profile I adjusted the white balance to 5500K so the white balance remains consistent throughout the test.  I also metered the scene using my light meter to determine the correct exposure and used manual exposure mode on the camera to lock down any exposure differences due to the picture style.  Below are the comparisons between the Cinestyle picture style, two built-in pictures styles, and a custom-Faithful picture style I developed for better color accuracy (no tone curve adjustment).


When I first saw this profile I thought the color reproduction was better than the built-in profiles, but after inspecting it on a color calibrated monitor I have to say that the "faithful" and "custom" (based on Faithful) picture styles do much better at doing what their name claims.  I know the picture looks quite simple, but actually there is a lot going on here.  First, the top of the stairs basically goes into shadow so I had to adjust the tone curves of the three other profiles to match the shadow detail of the Cinestyle pictures style.  Noise was popping out all over the place on the built in profiles, whereas the noise wasn't that bad on the Cinestyle picture.  That might have been due to the under-saturated nature of Cinestyle.  However...there is a caveat.  If you look at the wall and under-ceiling on the left the Cinestyle picture has less difference in saturation between the wall (should be pink-ish) and the ceiling (eggshell white).  When I tried to color correct the Cinestyle saturation color gradients popped out everywhere and the ceiling essentially turned the same color as the wall.  The Faithful and Neutral picture styles had the correct amount of color difference.  So it does look like you make gains on shadow noise by under-saturating color reproduction, but lose on color correction in mid-tones.  The hues represented in these pictures are close to representative of skin tones, so I might expect similar results with people in the picture.  A quick vectorscope plot showed the wall near the skin tone line.

Another aspect of the technicolor profile is that they give you a tone curve to apply in post.  The question is mainly how to do this with the Adobe tools.  The first thing I did was to copy their data and plot the tone curve in Excel.  From the text file you might notice that they crush the shadows and clip the highlights a bit, which was unexpected for a profile that is supposed to help maintain better post processing.


The next step is to make a custom curve in Photoshop that reproduces the look-up-table (LUT) tone curve that Technicolor specifies.  You can save this curve out as a "*.acv" file and open it up in Adobe After Effects using the "Curves" effect.  Pretty easy, really...swear.

The last picture shows the difference between the Faithful (default tone curve) picture style and the Cinestyle with the tone curve effect applied with the Technicolor specified curve.  Faithful might win a little in shadow detail but this a tweaked curve so really it's not a choice of one being better than the other.  Rather, it's just an evaluation of Technicolor's tone curve using a standard reference for comparison.  Without the tone curve applied the Cinestyle picture style clearly has more noise-free detail at the top of the stair case.



So...does it super charge the 5D mark II and make it "better?"  I'm not sure I would say that, however knowing the pictures style's limitations this might be a useful tool to have in my back pocket.  The Technicolor picture style was really meant to allow higher end productions to put 5D footage though a standard color correction process along side with higher end digital (Red, Alexa) footage according to the Technicolor rep.  That's not likely to happen in my case, however when some tech tweaker-boy shows up on set claiming that this is the only way to go I can at least discuss the matter with some intelligence.