A Few Questions (Metering, WB, CoC vs DoF)

[H4X1MA]

I know that these 3 things are a major part of taking good pictures, but for some reason my mind is not wrapping around them. I think that most of the articles I read are too technical. It's nice to know why something does what it does, but I need to know HOW to make it work.

White Balance: I think this is one of my main weak spots, and it may always be a weak spot since I am colorblind to greens and browns. I read through the tutorial on Cambridgeincolour.com but it still didn't say how you actually achieve a good balance (at least in a way that made sense to me). From what I gather you select something in the scene that is prodominent in the scene, then adjust your color slider to that color?

Metering: This is a complete mystery to me, even after reading about it. Can someone dumb it down and explain how it is physically done? What i gather, metering uses the highest white points and the darkest dark point to determine what the exposure should be. So if you spot meter a dark object with almost no white, then you will have a dark pic and vica versa. But what I think doesn't make sense or else I would think that manual camera modes would remove this option since exposure is more controlled by the shutter.

Circle of Confusion vs Depth of Focus: The article I read says that these are not the same, but it doesn't make sense on how they are not. CoC basically says that you lose focus after a certain point and objects start to get blurry, but isn't that the same thing that the DoF does? is CoC just a "scientific" explanation of what happens on a mostly non-visible level? What am I missing here.

 

[dxqcanada5]

I cannot tell you much about white balance.

Metering: B&H Photo - Introduction to Lightmeters - Part I
The reflective light metering in our current camera's are quiet sophisticated. There are programs built in to achieve an exposure based on an algorithm coded by the manufacturer.
Basically the meter provides you with a meter reading to render the metered area at 18% grey.
I you took a picture of a black card and a white card ... both should be rendered grey by a simple reflective light meter.

CoC and DoF are tied to eachother but not the same: Understanding Depth of Field in Photography

 

[pgriz]

Light "temperature" relates to the spectrum of light that a classic "black-body" would radiate at a specific temperature. If something is heated to, say 3,200K, it would be predominantly red with very little blue. As you increase the temperature, the spectrum of light has more and more contribution from the more energetic photons (towards the blue end of the spectrum), so by the time you get to 5,700K you get the spectrum which resembles that of the sun. As you increase the "temperature", you get more and more blue, with increasing contribution from ultraviolet, which we can't see.

The easiest way to adjust white balance is to include a pure white or pure grey object in the picture, and then have your post-processing software set its white balance based on that object being the standard.

Meters have been programmed to assume that 18% (or 12% in some cases) is the average reflectance if you average enough "normal" scenes. If your scene consists primarily of dark items, then the meter can be fooled because it will assume that there is just less light, and it will over-expose the dark items, trying to get the average exposue of 18%. If you have mostly bright or light-colored items in the scene, then the meter will similarly be fooled, but this time it will assume that there's too much light, and it will underexpose the light-colored stuff to make it come out to 18% grey. If you set your exposure based on "incident" light as opposed to reflected light, you'll get much more realistic tonalities, although when averaged, your photos will not necessary equal 18% grey.

Circle of confusion comes into play in terms of resolving power of the sensor or recording medium. Say you are focusing on a string of light arrayed along the same axis as your lens (ie, from your lens to infinity). Let's also assume that you're shooting slightly from the side of this line of lights so that you see a foreshortened line, but a line nevertheless. If you focus at a specific light (say at the 10' point), the lens will bring that light into perfect focus on the sensor. The light just ahead and just behind the "main" light will be slightly out of focus. That out of focus circle can be thought as the circle of confusion. The next lights ahead and behind the "main" lights will have larger circles. At some point, the circles will become larger than the sensor cell, and objects represented by these circles will start smearing themselves over more and more sensor pixels. So circle-of-confusion discussions depend on the resolving capacity of your sensor or medium.

 

[D-B-J]

Light "temperature" relates to the spectrum of light that a classic "black-body" would radiate at a specific temperature. If something is heated to, say 3,200K, it would be predominantly red with very little blue. As you increase the temperature, the spectrum of light has more and more contribution from the more energetic photons (towards the blue end of the spectrum), so by the time you get to 5,700K you get the spectrum which resembles that of the sun. As you increase the "temperature", you get more and more blue, with increasing contribution from ultraviolet, which we can't see.

Really? I never knew that!

 

[Dao]

Let me touch on the CoC vs DoF subject.

CoC and DoF are related. In fact, DoF is based on CoC.

First of all, in plan English, DoF is the range of distance that appeared to be sharp. But how can we define sharpness? That is where CoC comes into play. From Wikipedia, it defines CoC as "In photography, the circle of confusion diameter limit (“CoC&#8221 for the final image is often defined as the largest blur spot that will still be perceived by the human eye as a point."

In order to make sense, CoC conditions need to be defined. For example. if a tree is very sharp in your photo. But you enlarge it 1000x and look at it at a viewing distance of 1 feet, the tree is going to be blur, or say, out of focus. Therefore, it won't work without defining some of the variables.

<Ref: http://en.wikipedia.org/wiki/Circle_of_confusion>
- A normal vision person when viewing a paper at about 25cm distance is able to distinguish an image resolution of 5 line pairs per millimeter. And that is equivalent to a CoC of 0.2 mm. (Fact)
- When defining the CoC of a particular recording medium, we need to define the final output medium size, in this case, it is 8x11 format. So we are talking about viewing a printed photo in 8x11 format at around 25cm (a little less than 1 ft) viewing distance. (Defined)
- For 35mm film/full frame sensor, when it printed on a 8x11 paper, it is enlarged about 7 times. So the CoC of the 35mm film / Full Frame sensor is 0.2mm / 7 = 0.029mm (29 micron). For Canon aps-c format, it is 0.2mm / 7 / 1.6 = 0.018mm (18 micron) (Calculated)
- So when a tiny little light dot that hit the film / or full frame sensor with the dot size less than or equal to 0.029mm, it will appear to be sharp when print on a 8x11 paper.

Now, back to DoF. remember it is the range of distance that appeared to be sharp. And that sharpness is based on CoC. We defined CoC with a viewing distance, if we look at the photo 10 feet away, the DoF should change. (When you look at the billboard on the highway, it is sharp. But if you look at it at 3 feet away, everything seems out of focus). So when we calculate DoF, we need to know the recording medium (such as the sensor size). However, we assume other variables are fixed. (i.e. printed on 8x11 paper, view the photo at 25cm and has not additional cropping nor enlargement)


Now, when you calculating the DoF, make sure you input the correct recording medium. (35mm film, Full frame, 1.6x, 1.5x, 2x etc) And if you are planning to print it on smaller photo paper, the DoF should increase.

 

[H4X1MA]

Ok that's a lot of info to put through my brain sensor (har har) but let's start with WB since that one makes the most sense to me.

The following images were taken using Shutter mode yesterday around 11am. I'm still having trouble unstanding how you, as a person, know which WB to select when processing, let alone when shooting. Myself, knowing what the building actually looks like, I would go with the WB from White shot, but I still don't understand WHY I would be picking that shot. Is it because there is so much which in the image (clouds, marble, etc)?

Feel free to grab the original and do some processing. The last image is probably how I would process it, note that this was done in like 10 mins I would likely also B&W and sharpen the hell out of this pic to give it the "creepy feel"

Yellow Circles are where I pulled the WB from. Photoshop won't let me select pure white for some reason, which everyone but me probably knows lol

As Shot


WB From Grey


Wb From White


Processed

 

[Robin Usagani]

Let me touch on the CoC.

Dao likes Coc!

 

[clanthar]

Given your examples and the circles you've drawn I'm assuming you've set a grey or white point using the eyedroppers in a dialog like Curves or Levels as found in Photoshop or GIMP, etc.

I would not expect to white balance a photo using a white point like that. Especially since your original exposure has begun to blow out the highlights. Who's to say what you're clicking on in the photo that's white now because of overexposure was actually white in the original scene? Are clouds always white -- I know they're not.

In this specific photo you have two light sources that are different in color. The sun is shining on part of the photo and it appears to be well before sunset and after sunrise so the sunlight color is daylight. The building however which is the subject of the photo is lit by the open sky which is much bluer than the sunlight. So in this photo you have mixed light color.

You clicked on a point on the building I assume with a greypoint eyedropper. Again, the problem: was that spot in the original scene actually grey? What if it wasn't? So the eyedropper method has limited application. It can be helpful and it can go a long way to getting you there if you know you have a grey point.

Finally, this specific photo: You're highlights are blowing out from overexposure and yet the building is still too dark as it was underexposed. The dynamic range when you're photographing backlit architecture is usually prohibitive.

Joe

 

[H4X1MA]

Given your examples and the circles you've drawn I'm assuming you've set a grey or white point using the eyedroppers in a dialog like Curves or Levels as found in Photoshop or GIMP, etc.

I would not expect to white balance a photo using a white point like that. Especially since your original exposure has begun to blow out the highlights. Who's to say what you're clicking on in the photo that's white now because of overexposure was actually white in the original scene? Are clouds always white -- I know they're not.

In this specific photo you have two light sources that are different in color. The sun is shining on part of the photo and it appears to be well before sunset and after sunrise so the sunlight color is daylight. The building however which is the subject of the photo is lit by the open sky which is much bluer than the sunlight. So in this photo you have mixed light color.

You clicked on a point on the building I assume with a greypoint eyedropper. Again, the problem: was that spot in the original scene actually grey? What if it wasn't? So the eyedropper method has limited application. It can be helpful and it can go a long way to getting you there if you know you have a grey point.

Finally, this specific photo: You're highlights are blowing out from overexposure and yet the building is still too dark as it was underexposed. The dynamic range when you're photographing backlit architecture is usually prohibitive.

Joe

Why would it matter if it is white due to overex or because it is actually white. During processing I would think that white is white, and it really wouldn't matter. But now I think I am starting to grasp this.

Basically, if you were to take this same picture, and took a matte white color card with you and place it somewhere on the building, then what that card shows up as is actually white. If the card has a shadow overcasting it, it will appear slightly grey, but in reality the card is still white. So to get the correct WB you are actually trying to match what the white looks like, not what actual white is. For example, in this building you see lighter colored blocks which is actually white marble (it's in a marble quarry, go figure). So my white balanace really should be set, using the eye dropper, to the color of the lighter blocks on the building?

 

[DiskoJoe]

shoot in bw

 

[clanthar]

Basically, if you were to take this same picture, and took a matte white color card with you and place it somewhere on the building, then what that card shows up as is actually white. If the card has a shadow overcasting it, it will appear slightly grey, but in reality the card is still white. So to get the correct WB you are actually trying to match what the white looks like, not what actual white is. For example, in this building you see lighter colored blocks which is actually white marble (it's in a marble quarry, go figure). So my white balanace really should be set, using the eye dropper, to the color of the lighter blocks on the building?

The goal is to reproduce in the photo (on -screen or in print) the colors as they actually appeared in the real scene. For example in your sample-photo-camera-original there's a magenta/red color cast over the photo. The sky is purple rather than blue. The goal then would be to remove the color cast so that the sky in your photo was the color of the actual sky.

The color errors in our photos come from multiple sources. Your heading for this post is WB. Again the goal would be to record a matte white or grey card as colorless. Unfortunately the color of the light is always changing and the camera doesn't know that a grey card was grey and not greenish grey or redish grey. The camera can't measure the color of the light when it's taking a photo. With that variable unknown we get color errors. For example a typical digital camera set to auto WB and used to photograph a person in open shade will most certainly screw-up the color.

Back to the eyedroppers in editing software: When you use white or black or grey point eyedroppers, you're telling the software that the spot where you click was white, grey or black in the original scene. If that's not the case then you're simply introducing additional error. An option then would be to place an actual grey card in the photo and then use the grey point eyedropper to click on that card. Trouble is now you have a grey card in your photo. Many photographers have cameras that will allow them to set a custom WB. In this case you CAN use the camera to measure the color of the light. You take the grey card and place it in the light of the scene you want to photograph. With my camera I can then point the camera at the card and press a WB set button.

Back to the eyedroppers again. If you are going to try using them to WB a photo then use the grey one. Use the white and black ones to place white and black tonal points but not color.

Ultimately the eyedroppers are not the best route to go. They can help and I will use them as a short cut when I think I have an appropriate grey point, but in the end I make the call and manually adjust the color.

Joe

 

[H4X1MA]

The goal is to reproduce in the photo (on -screen or in print) the colors as they actually appeared in the real scene. For example in your sample-photo-camera-original there's a magenta/red color cast over the photo. The sky is purple rather than blue. The goal then would be to remove the color cast so that the sky in your photo was the color of the actual sky.

The color errors in our photos come from multiple sources. Your heading for this post is WB. Again the goal would be to record a matte white or grey card as colorless. Unfortunately the color of the light is always changing and the camera doesn't know that a grey card was grey and not greenish grey or redish grey. The camera can't measure the color of the light when it's taking a photo. With that variable unknown we get color errors. For example a typical digital camera set to auto WB and used to photograph a person in open shade will most certainly screw-up the color.

Back to the eyedroppers in editing software: When you use white or black or grey point eyedroppers, you're telling the software that the spot where you click was white, grey or black in the original scene. If that's not the case then you're simply introducing additional error. An option then would be to place an actual grey card in the photo and then use the grey point eyedropper to click on that card. Trouble is now you have a grey card in your photo. Many photographers have cameras that will allow them to set a custom WB. In this case you CAN use the camera to measure the color of the light. You take the grey card and place it in the light of the scene you want to photograph. With my camera I can then point the camera at the card and press a WB set button.

Back to the eyedroppers again. If you are going to try using them to WB a photo then use the grey one. Use the white and black ones to place white and black tonal points but not color.

Ultimately the eyedroppers are not the best route to go. They can help and I will use them as a short cut when I think I have an appropriate grey point, but in the end I make the call and manually adjust the color.

Joe

Hey Joe,

I really do appreciate your detailed response but I am still having trouble grasping this. I have also read a few articles online today and none of them just spell it out for you.

What I have come up with for a conclusion so far is: You see an object that you know is white, however the sun is casting light on the object which makes it a tint of yellow. In order to get the propper balance, while shooting, you will need to adjust the WB in your camera to the oposite of this yellowish color.

So say you have a yellowish tint on the known white object, you would adjust the camera WB to be slightly on the blue/purple side.
Say you have a blue tint on a known white object (blue sky reflections or water), you would set the WB on the camera to red/orange.

Does this sound correct?

I have a D3100 which doesn't seem like it has a lot of white balance options. I just took a white peice of paper, snapped a pic, told the cam to use that paper for WB, then took a pic in the room and everything ended up coming out with a yellow tint. My histogram showed a spike right in the center of the image for the paper pic, so I think that was right anyway.

 

[clanthar]

What I have come up with for a conclusion so far is: You see an object that you know is white, however the sun is casting light on the object which makes it a tint of yellow. In order to get the propper balance, while shooting, you will need to adjust the WB in your camera to the oposite of this yellowish color.

So say you have a yellowish tint on the known white object, you would adjust the camera WB to be slightly on the blue/purple side.
Say you have a blue tint on a known white object (blue sky reflections or water), you would set the WB on the camera to red/orange.

Does this sound correct?

I have a D3100 which doesn't seem like it has a lot of white balance options. I just took a white peice of paper, snapped a pic, told the cam to use that paper for WB, then took a pic in the room and everything ended up coming out with a yellow tint. My histogram showed a spike right in the center of the image for the paper pic, so I think that was right anyway.

I think you're considering the problem correctly, however fleshing out the details may be a bit more complicated. I think we have to get more specific about how you're working:

Let's back up some more: Outdoors during the day there's two lights and one of those is dominant. 1. the sun and 2. the sky. When an object is sunlit we call that daylight. We measure it in degrees Kelvin and although it fluctuates a bit with elevation and such it's right around 5500 degrees K. We can make film or set a digital camera to try and record white as white under daylight color and we do fairly well. When the weather turns overcast or we move into the shade on a sunny day the sky becomes stronger and the color temp goes up (bluer). To record white as still white we must adjust for this change in the light color. We can use a filter over the lens (light orange) or we can alter the software of the digital camera. The problem becomes substantially complicated when we either mix light sources and or begin to use the plethora of artificial light sources we've created that all produce different colors.

Film (back in the day) had a fixed color response. And when we used it with known light sources the task was fairly straightforward. With daylight film on a sunny day we got reasonably faithful color -- each film had it's unique biases, strengths and weakness. That same film on an overcast day and the photos came out blue so we'd attach an 81A filter to the camera. In open shade on a sunny day the photos camera out bluer still and we'd use an 81C filter to compensate.

Digital cameras complicate the process although for many users they do so unnecessarily. You have two paths to a finished photo through your camera. 1. You can record the sensor data and save a RAW file then open the RAW file using RAW conversion software and output an RGB processed photo or 2. You can record the sensor data and hand it over to the camera's image processor which will apply a sequence of computer algorithms to the sensor data and output an RGB processed photo.

Path 1: No need to worry about WB while taking the photos. The WB is set in the RAW conversion software after the fact. You typically have two WB controls in this case: the actual color temp in degrees K which you can set specifically to match the working condition and a secondary Tint control to compensate for color casts from artificial light sources. The advantage of this method is control in the hands of the photographer. The disadvantage is you've got to do the work yourself and be able to call the results.

Path 2: Software algorithms in the camera are going to try and WB and tone adjust the sensor data to a finished RGB photo. In this case the software is going to need all the help it can get and any info from the user concerning WB and tone will help. Therefore using the camera's WB setting is a positive step. Unfortunately this often doesn't help enough. This is OK for the consumer market -- the overwhelming majority of consumers would have never noticed the purple sky in the first photo you posted. The advantage to path 2 here is less time and work by the photographer. The disadvantage is a somewhat mediocre product.

Once you have your RGB photo you can continue to edit it as needed. If you followed path 1 then the editing process is usually some tweaks to the local tone response and possibly local color adjustment. If you followed path 2 then the editing process begins with fixing the mistakes that came out of the camera.

I don't know the details of how you're working, but I'm going to guess path 2. It may be that's the best way for you to keep working. It's the method embraced by the majority of people using cameras. In this case your first step is to use the WB presets on the camera. The next step is to get a WB target (not a white piece of paper): Michael Tapes Design - LensAlign - WhiBal - IJFR - WhiBal Certified Gray Card for White Balance This will help, but you will still suffer some grief from the camera's image processing software. Once automated software in the camera has missed the target with your photo, the same automated software in your computer (eyedroppers etc.) isn't really going to do any better.

Joe

 

[clanthar]

Here's an illustration that may help.

I was out taking photos yesterday. It was a bright sunny day with a blue sky. I took this photo below: My camera saved a RAW file and so I ignored the issue of white balance while taking the photo. In the screen shot you see below Adobe Camera Raw is allowing me to adjust the photo before final conversion to an RGB photo. In the top image the color temp is set to 5500 degrees K which is daylight. The photo was taken in open shade and as such it's major blue as would be expected -- the sky is blue after all.

The sensor in my camera has a fixed response to color just like film used to and over time I've gotten familiar with the performance of my camera's sensor. Knowing I had taken the photo in open shade I immediately set the color temp up over 7000 degrees K. With a little fussing I settled on 7150 which is in my experience with this camera pretty much what I'd expect. The blue color cast is gone and I'm ready to move the photo into Photoshop.

Joe

 

[H4X1MA]

I actually rely on imaged editing software heavily since I am new. I personally like a very bright vibrant photograph, and well since I don't know how to use the white balance I rarely get them lol. I would like to be able to do this with the camera though, editing images really strains my eyes lately so I try to do as little as possible. But I always shoot in RAW so that is a plus.

Lets see if I got any better this time. This is a picture from inside the building, but it may be a bad one to practice with. It seems a bit out of focus, and it looks like there is blue sky light coming in through the broken roof lighting the back wall. The lower back wall looks especially blue to me, it may be a reflection off of the water. The rest of the image doesn't seem to have the tint, so maybe this is one of those dual color images that are impossible to get right.

To me the front of the image looks about as real as I could get it. I tried using the brick mortar as a reference to what the true white should be since that is a typical color that is easy to remember as a reference. This time I didn't use the dropper, but the slider, and then typed numbers to get a closer value.

In this image I did need to adjust the Exposure in Camera Raw, the light coming in the roof was really blown out. -0.95 on both.

As Shot


4950k tint 0