Question about "It's Always Sunny in Philadelphia"
 

So I was watching "It's Always Sunny in Philadelphia" earlier tonight and decided to go to IMDB for a bit more info on the show when I came across this tidbit of info:

"The pilot was created by three Hollywood writers with a long list of credits. The first two episodes only cost $200 to make and were all shot on digital camera. It was then shopped around to different networks and was picked up by FX."

Anyone have any information about this, perhaps what cameras they used and any info behind the shooting process of the first two episodes?

http://us.imdb.com/title/tt0472954/trivia

I swear someone on this forum said it was a DVX100, or 100a. That his friend had borrowed it to shoot the show.

Yep, definitely DVX. Pilot was shot for $200 using a DVX100 and edited at home by Rob McElhenney, Charlie Day, and Glenn Howerton. The FX show was shot using DVX100a cameras for seasons 1 and 2.

Are they still using it? The show looks really nice for being on miniDV. It has this polish to it that I can't quite describe. . .how do they do that? It's not something so simple as lighting/composition/blah blah blah. . .some kind of color correction or post thing.

Josh, I read an article where the DP from that show said the key to making the DVX100 look so good was basically 1) warm cards 2) lighting the heck out of it, and 3) they upconvert to DigitBeta and edit/color process from there treating it like high-end video material.

Kinda makes sense; it's unbelievable how much sharper DV is with good light (and how flat & mushy it is without it).

That's a wicked show IMO. If I can relocate that source I'll post the link...

PS: Here's a link, looks like a press release from Panasonic after all:

http://www.digitalproducer.com/artic...e.jsp?id=45138

So I was right about the color correction, then.

Warm cards are simply light blue cards (or various shades of blue?) to warm the white balance when you WB on them, right?

The problem I have with this is, in my experience, it tends to "orange up" the entire image. Sometimes it seems like it looks good at the time, then you look at it later, and everything just seems too orange/red.

Seems like you could get more precise control by either lighting with CTO/an orange gel, or playing with the menu controls on the DVX.

Yeah, it can "orange up" things but if you use a white card that is just tinted blue, like the actual product (www.warmcards.com) it can be barely noticeable but nice, like on "It's Always Sunny...".

Same thing as white balancing with CTB over the lens, as my assistant/grip always likes to do. Less work than CTO on the lights, IMO. But it's like anything, it's about subtlety and not "holy orange!"

I'm a fan of repeatable accuracy, so using a well-calibrated warm card to WB every time is one way to get that. Whenever I tweak the DVX in-camera settings I get carried away and get into trouble...

Just to ditto Benjamin's comments about the warm cards. I recently picked up the set and have been really pleased with the result. The first time I used them the client just loved the look of the footage.

How bout that. . .


I used to white balance on light blue jeans, etc., and I always thought I liked the way it looked 'til later.

Does anyone know what they do to it post, specifically? I know it's some high end stuff that I'd never have access to, but is there some corollary in the low-end (i.e. Vegas) world? Specific effects, I mean.

Why is this necessary? Is it because most imported TV sets are shipped set for D65 to make them look brighter? Today they seem to be adjustable in most cases. Would pictures be deemed "warm" enough if consumers set their displays to D50?

It's more likely that some cameras (or maybe the DVX in particular) tend to shoot a slightly "cool" (i.e. the entire color spectrum shifted more toward blue than red) picture when white balanced normally. I know the XL1s sometimes did.

Haven't I heard tales of people making their own warm cards with pleasurable results? How's that work?

From what I've read the Da Vinci is pretty precise for CC & enhancement. Plug-ins like Magic Bullet's Look Suite might be considered an affordable way to process your video image.

Funny you mention the XL1s, I always thought Canons tended toward oversaturated, orange-red colors. You are right that the stock DVX WB settings seem a bit cool.

SOMETIMES it does tend to make a very warm white balance. Other times I've white balanced to what looks white, and come out with a bluer tint. I think the presets for sure tend toward uber warmth.

The Da Vinci is somewhere around $800k. Its main strength is that everything it does is real-time. Primary CC, secondary CC, and power windows (masking color corrections to specific areas; limited to geometric shapes). Other systems (i.e. Lustre) offer spline-based windows and more advanced features, but they don't quite have the same performance as a Da Vinci.

On the low end, you might look at Magic Bullet since it can do interesting things. It doesn't get as detailed as a Da Vinci in that you can't do secondary color correction or power windows / masking. However, you can look at what other people are doing with Magic Bullet and they are getting some interesting results with it.

Warm cards:
Most CRTs have a white point much blue-er than D65... in Japan, the standard is D9300 (or 9300K, not sure). At the higher color temperature, reds will appear de-saturated. Some/many TVs will oversaturate the reds (and maybe adjust the hue) to compensate for this.

What you should also consider is that your eye has its own auto white balance of sorts... so your eyes will white balance to whatever the TV set does anyways. The warm card will still have a small effect (i.e. the colors changing appearance as mentioned above) and the monitor may not have a linear response, so there's some subtle effects there too.

In the camera, colors will tend to blow out towards white. So if that happens, then your eyes will balance to the TV's white point and the image will appear like it has a tint on it.

Some NLEs can't fix color temperature problems perfectly (i.e. FCP) so shooting with a warm card may burn that into the image (and you can't quite get the same results as if you had shot normal WB).

2- There are potentially better ways to add a tint to the image. i.e. the Magic Bullet tinting looks interesting (and uses different math).

Thanks for the info Glenn. For those of us who don't know, what is the difference between primary and secondary color correction?

Thanks,
Ben

There are various definitions/interpretations for it.

One definition:
Primary corrections are ones that affect every pixel in the picture.
Secondary corrections are ones that affect only a certain portion of the picture... i.e. you isolate the change based on geometry or on hue/sat/luminance (or both).

Another:
Primary corrections are what you do first... it's your base grade. You'd do this on a scene-by-scene basis.
Secondary corrections are more detailed changes that you do on a shot-by-shot basis.

Glenn - yes , quite right it is 9300K - I was thinking of something else. And yes, D9300 means an approximation to 9300 K.

Benjamin - primary color correction means adjusting the whole image the same way (i.e. to correct color balance). Secondary color correction means changing the color of a portion of the picture. An example might be making a pink rose look more saturated while leaving the background untouched.

This isn't the same thing as the primary and secondary color correction plugins in Vegas, right? As far as I could tell, the secondary color corrector affected the whole luminance range in the image, whereas the primary affects shadows, mids, and highlights separately.

As for making a specific element in a shot (e.g. flower) look different from the rest, you could do that with multiple layers and a mask.

The math may be a little different than that in Vegas.... some color correctors do calculations in R'G'B' space and not Y'CbCr space (which is what Vegas does for its primary and secondary color correctors). The Da Vinci does ""YRGB"" which really means it can do corrections in either space.

2- Ok yes, the primary color corrector does affect a "range" of colors. To be more precise, secondary CC sometimes refer to affected colors based on hue and saturation (although those terms are ambiguous).

3- The D series illuminants are somewhat based on measurements of daylight... D9300 would be a continuation of that series, although I don't know if it exists and I don't know if it's the standard. 9300K is based on the Kelvin scale for black body radiators; objects when heated emit a certain spectrum of light; many light sources behave similarly to black body radiators.

So could one take the methods/effects applied on the DaVinci and get something comparable, if necessarily inferior, in one of the more affordable NLEs? I don't have magic bullet or After Effects, just the Vegas.

Yep, in Vegas you can do something comparable.

Hal,

If you go to iTunes they have a great little 20 min documentary about the creating of the pilot and how they sold it.

I'll give you a hint, though... which camera they used had nothing at all to do with their success.

Hey hey. . .


Back to the warm cards. . .


If you balance through a CTB gel, what strengths commonly give the best results? If you want that slight effect that a warm card would give you without reddening the poop out of the picture?

D9300 means "daylight 9300" which is a spectral distribution curve which has the variations characteristics of daylight (i.e. it is not smooth like the Plankian radiator's spectrum) but which generally follows the spectral distribution of a black body at 9300K. What the correspondence is to daylight I don't know for sure but a good guess is that an actual "average" daylight spectrum (from shade) was fit with Planckian curves and the temperature of the one that best fit was 9300 (I'm guessing the whole series is determined this way). OTOH a bunch of guys at the CIE may have sat around the table and "designed" it. It does exist but must be fairly new as I can't find the distribution in any of my reference materials which are at least 10 years old.

Well when it comes to white point, good information is unfortunately hard to find. Most sources incorrectly state that the standard reference white point for (non-Japan) TV is 6500K (it's really D65). Even Sony, who makes broadcast monitors, has gotten this confused! I have a broadcast monitor with a 6500K preset. The new monitors say D65.

1- I believe the daylight series illuminants are based off real data measurements taken in different locations around the world. I can't remember the details off the top of my head.

2- I think the Japanese studio video standard is 9300K, not D93. This is from reading Charles Poynton's book Digital Video and HDTV (which is generally very accurate).

A little playing with D65 shows that the Planckian curve which fits it best (rmse) corresponds to 5921K whereas the Planckian curve with a peak that matches D65's peak has a temperature of 6300K. Thus it appears that peak matching is more important than overall spectrum matching in labeling these daylight profiles with a color temperature but there is more to it than just peak matching.

Poynton does indeed refer to 9300K (and mentions that Japanese standards call for studio monitors to have a white point set for this). As his book was written in 2003 and the standard existed before that I am sure that D93 was not around at the time of the standard's adoption since the only references I can find to it are quite recent.

Poynton also mentions that 9300K results in a white 30% more blue than D65. That's so much that even I (color blind) can see it. If indeed Japanese manufacturers are shipping sets with 9300K (or D93) white points (and they appear to be) then this is going to result in a cold picture. My new set, while it doesn't allow selection of white point by color temperature, does allow the white point to be tweaked and I've adjusted it to look a lot warmer than when it came out of the box. My original thought in this thread was that if everyone adjusted his set for a more reasonable white point then warming gels, blue cards, CC filters, post operations etc. might not be necessary.

I think D65 corresponds to 6504K, although I don't know how that's calculated. You can think of D65 as greener than 6504K.

The following article "The White Paper" has some nice information on white point.
http://etconsult.com/papers/White%20...tt%20Cowan.pdf

A- In the North American market, many CRTs are shipped with a very high/blue white point. This makes the image appear brighter- presumably it sells more TVs. Maybe it stands out more on the store shelf, isn't overpowered by other light sources (other displays, store lighting), etc.

At the higher white point, your vision will mostly adapt to it. However, reds will appear de-saturated. Many TVs will compensate by over-saturating the reds. They may also change the hue angle of the decoded chroma too... although that might have something to do with the color of the red phosphors in the TV (I'm not sure there).

B- Some people evaluate warming cards on their camera's LCD or on a broadcast monitor... so the high color temp. may not necessarily be it. Or it might... people may be judging on things like a computer monitor (i.e. looking at pictures on a website). Although even in the best case scenario (D65), a point could be made to go slightly warmer. D50 is a little closer to natural daylight.

C- With extreme colors, our vision simply won't white balance to those colors. An example would be tungsten light, especially the lower temperature and dimmed types (i.e. nitelites). Streetlights is another example- the light will never look white. Another good example is "golden hour", the hour right before and after sunrise/sunset- the sunglight appears golden.

There is presumably some 'equilibrium' point where our eyes' natural white balance is at. This may be the ideal white point??

D- Some people just like some colors better than others. For example, my favorite color is blue.

E- It could also be the power of suggestion... i.e. the Monster cable effect. The power of suggestion can be very strong!

WRT to the "color temperature" of D65: Curve fitting is an art. D65 has a peak at 460 nm and a local null, at 430. If I fit from the null out to 780 nm I get 6507 K so maybe that's how they do it. OTOH if I fit from 435 - 780 I get 6603 K and fitting from 425 to 780 yields 6438 K. IOW if I play around with the portion of the curve to be fit, the weight given to various parts of the curve and so on I can get pretty much any number I want in the vicinity of 6500K. While this is instructive I'd really like to know how the CIE does it.

RE A: According to Poynton 9300K originated in the computer graphics industry where the reason for using it was additional screen brightness. With respect to Japan he mentions "cultural preferences". I think the answer is indeed that the sets look brighter in the store and at home and the eye largly adapts hence more sales. That's what drives many an engineering decision.

RE B: My question is to what should they adjust or should they adjust at all? We are often advised to do all our color correction using a studio monitor rather than the computer's monitor because of this very consideration. But how is a studio monitor set? Mine (Sony) has two choices: D65 and D93 (yes, it is labeled D93 so there is evidence that D93 exists - also mentioned several times in the "White Paper"). Then there's the question of the camera. What color space does it actually work in? I suppose each worker must choose the workflow that best suits his needs. The prevalent thinking seems to be to use a D65 monitor for making adjustments and then let the consumer worry about whether the picture is too blue. In this case then the judgement should be made on a D65 monitor - not the viewfinder of the camera or an outboard LCD display. Of course a guy who white balances per the camera manufacturer's instructions and finds the image too cold on a D65 display is perfectly justified in using warming techniques at the camera. This is an indication that the camera's balance is cold. And warming for artistic effect is, of course, a whole different ball game.

D50 is, of course, warmer than D65 but more closely resmbling daylight? I'd have to ask at what time of day, at what latitude, in which season at what relative humidity and so on. I recall, and this is a little hazy, that D50 was come up with back in the days that graphics arts people first got photomechanical reproduction capabilities (i.e. 4 color presses) good enough that the impression of the quality of an image was dependent on the light source it was viewed by. I believe that to this day evaluation booths for printed materials are set for D50. The thing that drove the decision was, if I remember correctly, the women's magazines which, you have to admit, have got fantastic photography. D50 was, again if memory serves, designed to simulate the mix of daylight and tungsten that the average housewife would encounter whilst reading Vogue at her kitchen table.

RE C: You can't accomodate unless there's something to accomodate to. This usually means something in the picture that you know is neutral. Snow is the perfect example. You know it's white so you set it white mentally even if it's blue as in shadow with a camera balanced for sunlight. Also the picture can't be too far off neutral on average. And this is a good thing because it allows photographers to capture the "golden hour" etc. Because of adaptation, the variability of natural and man made illuminants and the variability of human perception of color (remember all of color science is based on the "average 2 degree observer" ) I don't think there is an ideal white point. I might go for an area in CIE diagram inside which white points should lie but I can't see that there is one unique point.

RE D: Mine too.

RE E: Amen!

I believe the correct calculation is something like this:
First you'd figure out the x and y chromaticity co-ordinates of the curve. Look up the CIE color matching functions, it defines x y and z curves. You'd figure out the response in x y and z that D65 produces, then you normalize the results such that x + y + z = 1. This gives you the x and y co-ordinates.

*From here, I'm less sure. But you want to calculate the correlated color temperature.
You plot your x and y co-ordinates in Luv space (the *obsolete* Luv color space, not the newer one) and figure out which point on the Planckian curve is the closest (i.e. find the closest distance on a plot of everything in Luv space).

Hmm... I've seen a Sony monitor with the choices D65 and 9300K (I think so anyways).

And on a second closer reading, the White Paper does mention D-9300... and I consider the author to be a credible source. So I'm not really sure anymore.

I made the assumption that most of the time, D50 would be more representative than D65. This may be a bad assumption on my part.

Amen to that.

Turns out this is correct (search "correlated color temperature" in Google). Calculations are done in the 1960 Luv space by drawing the perpendicular to the Planckian curve through the u,v point having the coordinates of the distribution. Or one can draw the Planckian curve in x,y space with lines of constant correlated color temperature (not quite perpendicular in this space) and find the line closest to which the x,y coordinate of the spectrum lies (saves having to compute u and v). The locus of "D" white points is so close to the Planckian curve (p19 of the "White Paper") in xy space that you would probably incur little error by doing the comparison in xy space.

But sometimes Google / things you find on the Internet are wrong. :( Case in point: 6500K versus D65.

Anyways, I think we've really really gone off-topci from the original post. :D

Alas true but in this case Poynton confirms it (p223) and indeed running this algorithm against D65 returns 6504.2 K (published CCT 6504) and for illuminant C; 6775.4 K (published value 6774 K). I'm convinced.

Yes, a bit off topic but I needed to review this stuff for something else I'm doing related to the perceived color of beer and this discussion helped me do that. Thanks.