correcting amateur lighting...

wwjd wrote on 12/19/2014, 8:02 PM
Long story short, I make a ton of mistakes when I shoot. From forgetting to White Balance, to wrong lighting.

If anyone wants to play with Vegas to correct some colors, feel free. Here is the clip.
There is not much to the clip - just a focus test outtake - it's 4K MP4 from a GH4

https://copy.com/j0Swe9xEm9cDHd3S


The room is brown wood, the shirt is white, suit is very dark grey, tie is black, the screen in the background light blue. The is an LED on both sides set to 4200k (but they are not accurate), one with a BLUE GEL, the other a RED GEL. I was going for deep dark REDs and BLUEs and ended up with purples and pink/megenta/who knows what. How can I make the reds red and blues blue? I futzed around with color wheels, but don't know what I am doing, I guess. Feel free to post up ideas or tests.

To me, it is totally usable as is (views only know what you give them, unless you tell someone what was in your head), but I wouldn't mind getting closer to the colors I envisioned. If I use your correction idea, I'll be happy to credit you in the film, if ya want. Note: I do like hot, saturated, movie-style colors - like any other commercial trailer you find on YouTube.

Anyway, have some fun with it! Or not.

There are 3 other scenes in this location and lighting, so I will be matching them all up.

Comments

musicvid10 wrote on 12/19/2014, 9:26 PM
I think it's great; I don't know what you are after.
Raise the gamma by .05 - .10, and leave the rest alone. That's my instinct.
PeterDuke wrote on 12/19/2014, 9:31 PM
With red from one direction and blue from the other you are never going to get a natural look, nor was that the lighting director's intent. Leave as is.
farss wrote on 12/19/2014, 9:32 PM
Deep gels on white LEDs are not such a great idea.
I would have gone for Red and Blue LEDs which can be had pretty cheaply off eBay.
Just search for RGB PAR
Just be careful when you use them as it's easy to clip a channel, especially with the blue LEDs.

Bob.
wwjd wrote on 12/19/2014, 9:33 PM
Thanks for the input MusicVid.
I been messing with it for the last hour and, ya know, after playing with it a bit that is exactly what I came up with too. I used some hitfilm plugin that brightened it a pinch, and a grain layer that also bright it up just a little, and it seems fine. maybe i was smoking something and didn't know it. PLUS I think all the red an blue was making me THINK I was seeing a lot of purple.

we'll see if anyone else thinks of something.
wwjd wrote on 12/19/2014, 9:35 PM
I was the "lighting director" and know exactly JACK NULL about lighting. hahahaha! I did want lots of hard red and blue just like that, but sometimes it looks all purple and pink to me.
farss wrote on 12/20/2014, 12:34 AM
Red + blue = green. Purple isn't a colour, it's an artefact of human vision!

I've used red LEDs as an only light source, it looks better from the camera than the eye. Gel white lights and you get the reverse.

Bob.
PeterDuke wrote on 12/20/2014, 1:30 AM
"Red + blue = green. Purple isn't a colour, it's an artefact of human vision!"

Whoa! Every painter knows that mixing red and blue paints makes purple. That is subtractive synthesis where the primary colours are red, yellow and blue. The colours are created by absorption.

If you have coloured lights you have additive synthesis, where the primary colours are red, blue and green. Equal amounts of red plus blue makes magenta. A bit less red makes purple.
PeterDuke wrote on 12/20/2014, 3:20 AM
This thread has prompted me to read up on white LED lights. I found this site useful:

http://electronics.stackexchange.com/questions/125500/cool-white-led-bulbs-are-they-full-spectrum

Apparently most white LEDs, such as in my torch, are actually a blue LED with a yellow or red-yellow phosphor. The latter type produces a warmer light. White light made from a group of red, green and blue LEDs have a flatter spectrum, but are more expensive.

The spectrum of the former type is far from flat. You will see that there is a narrow peak in the blue region and a broader peak in the yellow or orange region. That is why using gel filters with these lights is not a good idea. It would be better to start off with red, blue and/or green LEDs in the first place.

A camera has three colour filters to separate the red, green and blue regions, analogous to a bank of three band pass filters in the electrical world. The characteristics were designed based on illumination by a light source having a more or less continuous spectrum. You can see that you are liable to get significant errors as a result, depending on the actual colours, if you use a LED light. Green objects, for instance, may be relatively poorly lit, because the colour falls between the two main peaks of the lamp.

I am not clear what you were aiming for. If you wanted strong reds and blues, you certainly got that in my opinion. How did you want green to come out? Black? If you increased the colour saturation you will suppress the magenta and purples.
farss wrote on 12/20/2014, 5:44 AM
[I]" Whoa! Every painter knows that mixing red and blue paints makes purple."[/I]

No argument there except purple / magenta isn't a colour. We see magenta even though the midpoint between blue and red is green however our eyes have green cones and they're seeing nothing. Faced with this conundrum our eyes / brain invents a colour we call purple or magenta.

]

I think you'd want to be careful using this blend of colours on skin.


Bob.
PeterDuke wrote on 12/20/2014, 7:28 AM
I see your point now, but I disagree with your definition of colour.

Colour is what we perceive when we see a single or mixture of wavelengths of visible light. Colour isn't restricted to single wavelengths, as occurs in a rainbow (e.g. there is no brown, actually unsaturated orange, in a rainbow), nor is it restricted to the three colours corresponding to the peak sensitivity of the cone cells in the human eye. The concept of colour preceded Newton's fiddling around with prisms or research into human eyes.

Colour is perceived as being somewhere in the triangle with the primary colours at its apexes, as depicted in the image in your post above. It is not like sound frequency which is unidirectional, ranging from low to high and with no wrap around.

Is hue different? Is magenta a real hue in your mind?

The Macquarie Dictionary defines colour as "the evaluation by the visual sense of that quality of light ... which is basically determined by its spectral composition ... Any colour may be expressed in terms of three factors: hue, chroma (purity or saturation) and brightness (or value). Generally the most obvious or striking feature of a colour is its hue, which gives it its name. The colour is qualified if necessary as pale, dark, dull, light, etc."
wwjd wrote on 12/20/2014, 10:00 AM
i dunno, just my inexperience and naiveity here. I am seeing purple in the rest of the clips (a ton of them) but it is probably in the parts where the red and blues are added together like on a back wall - not on a face left and right like this. And the lights get moved abit per scene for better coverage - which makes them darker/brighter depending on that subject.
It'll work out, just seemed like I did it wrong somehow. In the end, I'm having fun learning and trying things.

Thanks for all this input!
wwjd wrote on 12/20/2014, 10:02 AM
I discovered all that LED spectrum loss AFTER I spent on 3 large panels. Oh well. They will work for this and that. Cafefull white balance, correction and heavy grading like I like, I think they will work fine. Just not great.
Chienworks wrote on 12/20/2014, 11:21 AM
"Red + blue = green. Purple isn't a colour, it's an artefact of human vision!"

I think you mean magenta, not green. ;)

And yes, purple is a color! It's not a color we see directly, but it is a physical color. Every frequency of light is a real physical color, even those beyond infrared and ultraviolet. Our eyes only distinguish between reddish, greenish, and bluish, with mixtures of those three giving our brain the impression of all the billions of other colors in between and beyond.

Of course, color film and cameras are also made sensitive to reddish, greenish, and bluish, in an attempt to fool our brains into thinking our eyes are seeing full color vision.

I remember some philosophy/physiology discussions where we were debating whether the primary colors were the same for everyone and what would be the ramifications if they weren't. I think color printing, color photography, and color television prove pretty conclusively that for the majority of all homo sapiens, these are pretty close to spot on for everyone. If they weren't then the three-color artificial imitation wouldn't look like natural colors to anyone except those who's primary frequencies just happened to match closely. On the other hand, this might account for why there are a few folks who can't ever seem to get the color balance of their TVs quite right, or are apparently happy with it being off. It could be what they've got it set for is just the closest it will ever get for their eyes, and since they've never seen a screen show what is to them true colors, they don't know that it should be better.

And a bit more philosophical/physiological questions: isn't odd that the full spectrum we see just happens to come back around to the *SAME* purple/magenta-ish color at both ends? How marvelous that we end up with a full color circle that just starts again where it left off. And isn't it also amazing that it just happens to be pretty much exactly one octave of frequency? Why is that? Maybe those last two are related, or maybe it's just a happy accident.
Chienworks wrote on 12/20/2014, 11:25 AM
"We see magenta even though the midpoint between blue and red is green"

Fallacy there. Light frequency may indeed be a linear 1-dimensional array. However, since we have three different types of color sensing cones that gives us a 3-dimensional interpretation of color. On that 3-dimensional array, green is no where near between red and blue. It's a mistake to think that the linear origin of the color must have any impact on the brain's interpretation.
farss wrote on 12/20/2014, 4:13 PM
[I]" It's a mistake to think that the linear origin of the color must have any impact on the brain's interpretation. "[/I]

Never said it does.
What I did say (in effect) was that we cannot rely on how our eyes and brains interpret the real world to tell us what is real.
Magenta is a mixture of red and blue light or one could say it's the absence of green.

If one looks at the CIE chromaticity diagram you'll notice that the big curved part of it represents all the wavelengths of light. There's also a straight section that doesn't map to any wavelength of light, you need a mixture of wavelengths to get them and it can be quite difficult using just our eyes to work out what's going on with those colours. Evolution didn't bless us with spectrometers for eyes like it did for our ears :)

I'm far from certain how relevant this is to our world. I do know we used to use pure white light sources with gels which subtracted part of the spectrum to bounce light off objects that also subtract part of the spectrum. Today increasingly we're using light sources that are additive e.g. RGB LEDs, even worse they're pretty much monochromatic, something that gels never gave us. We're certainly facing some new challenges from all this. To plumb the depths of just how much we need to rethink things my approach is to go back to the basic physics.

Bob.
PeterDuke wrote on 12/20/2014, 5:13 PM
As I said before, the brain processes sound differently to light. The processes are not analogs of one another.

If the ear simultaneously receives a low frequency sound and a high frequency sound, we hear a low frequency sound and a high frequency sound. They don't get combined into an intermediate frequency sound. We place sound frequencies (or the fundamentals of complex sounds, even if missing) somewhere in a unidimensional space. If sounds are close together in frequency, we hear a beat, but that is something different again. In the early days when critical bands for sounds were first discovered, there was an attempt to see if vision had something similar. It doesn't.

The eye on the other hand usually has three types of cones, which defines a sort of triangular two-dimensional space, and colours (of a given brightness) are mapped somewhere in this space. People who are colour blind only have two types of cones. I read somewhere that the incidence of colour blindness was on the increase, so perhaps we are evolving towards a monochrome world! Colour correction would then be a thing of the past!

So far as LED light sources go, it would be better to avoid the blue plus phosphor type and stick to the RGB type. The latter is a better match to the sensors in a camera, but not such a good match to the cone sensors in the eye, which have the red and green closer together and the blue more distant. The rods sort of correspond to a luminance channel, with most sensitivity in the gap between the green and blue cones.

Even then, LED lights do not have virtual continuous colour spread as colours do in reality, so subtle colour changes will be lost.

Let's go back to limelight!.
PeterDuke wrote on 12/20/2014, 5:29 PM
"I am seeing purple in the rest of the clips (a ton of them) but it is probably in the parts where the red and blues are added together "

That is because red and blue added together does make purple or magenta. The only way to get only red or blue is to never mix them, which is very difficult to do with a flat back wall. If you don't want it to be purple, what colour did you want? You can't override the laws of nature and the processing that takes place in our vision.
PeterDuke wrote on 12/20/2014, 5:40 PM
As a point of trivia, we have seven colours in the rainbow because Newton thought there should be, analogous to the seven notes in a major scale in music. (Newton didn't always get it right!) He then got a friend to carve up the spectrum into seven distinct bands, because he didn't trust his own judgement. (Also not good, he should have got a panel of observers, not just one.)

I always thought that seven was chosen because of the special significance of seven in the Bible.

I could never distinguish seven major colours in a rainbow.
wwjd wrote on 12/20/2014, 6:00 PM
...where my "amateur" comes in... I wanted red and blue only or mostly but staring, running camera, setting up lights alone, I don't always see everything where I should, like a back wall being purple. :D
It's not a prob. For all anyone will know (except you guys) I WANTED a purple background :D
Chienworks wrote on 12/20/2014, 6:51 PM
"Today increasingly we're using light sources that are additive e.g. RGB LEDs, even worse they're pretty much monochromatic, something that gels never gave us."

I wonder how closely the three LED colors match the three frequencies that the camera thinks are R, G, and B. I suspect a LOT of LED manufacturers are only concerned with making their LEDs look like they are reddish, greenish, and bluish, and don't worry about them being the precise frequencies. This causes a lot of the same effect we used to get from old fluorescent lights and color film, where our eyes would adapt to the poor lighting, but the film would capture bronze skin, bleeding blues, and muted reds.

I will say Acer seems to have gotten it right with the LED backlight in my laptop, where the colors are just as good as the finest CRTs.
farss wrote on 12/20/2014, 7:18 PM
[I]"
I wonder how closely the three LED colors match the three frequencies that the camera thinks are R, G, and B. I suspect a LOT of LED manufacturers are only concerned with making their LEDs look like they are reddish, greenish, and bluish, and don't worry about them being the precise frequencies. This causes a lot of the same effect we used to get from old fluorescent lights and color film, where our eyes would adapt to the poor lighting, but the film would capture bronze skin, bleeding blues, and muted reds"[/I]

The RGB sensors in our cameras are not much different to the ones in our eyes. So if you shine the light from a RGB LED onto white and point your video camera at it, it will record white. The sensors in the camera don't record which R,G or B hit them, just how much of each part of the spectrum.

Where it gets messy is when you shine what appears to be white from the RGB LEDs onto something that isn't white. If it was something made from one of the pigments that only reflects a single wavelength of light and that wavelength is absent from your RGB LED then it will appear black. That's an extreme example, just to set the scene :)

The other issue here is LEDs produce no infrared light so the haemoglobin in the capillaries just under skin isn't reflecting visible light. That can give talent a pallid "zombie" appearance. Personally I'm very happy with my cheap Chinese Z96 lights. If it wasn't for them in many places I wouldn't have enough light or the talent would have deep shadows under their eyes.

If you want the best of the LED lights then the new "remote phosphor" ones are the go but they're pretty big and expensive. Even if I could afford them they're still too big for some of the places I have had to work in.

Back to the human eye. We're forgetting that our eyes also have rods which measure something akin to luma. Our cameras lack these, instead the luma signal is derived from a combination of the RGB values and that can be fooled. That makes auto-exposure even less reliable for stage shows that use RGB LEDs if they're outputting only blue. In my quest to avoid "blue blobs" I had to pull exposure back around 3 stops during one stage show.

Bob.