I have Bill Raven's picture profiles loaded into my EX-1 and noticed that one of the settings under the pictures profiles menu is for "white" (I'm assuming white balance). I'm hoping someone can help clarify things for me. Raven's has the "white" setting in each of his profiles set to "5600k Preset". I've always kept the camera's default WB preset at 5600k, however, I hardly ever use it, as I typically do a custom white balance every time I shoot, saving it to the A and/or B WB custom presets. My question is, what is the relationship between the WB setting in the picture profiles menu and the camera's WB settings.

Because Raven's PPs have a WB 5600k preset, does it mean the PPs will work when the camera's WB setting is on the 5600k preset? Or do I need to go into the picture profiles and reset the White setting each time I do a custom camera WB? For example, I just ran a test: I did a custom WB and stored it to WB-A. The WB was 3600k. Then I went into the PP settings and changed the "white" setting from "5600k preset" to "3600k preset". I noticed that the camera's WB preset was effectively changed in the process and the default 5600k WB Preset changed to 3600k. Oddly, however, when I toggled back and forth between the camera's WB preset and the custom WB-A setting--which are BOTH set at 3600k--there was a slight, but perceptible change in color temperature between the two. Since both are the same color temperature, I assumed there wouldn't be a difference when toggling back and forth. Perhaps it's because one setting is displaying 3600k with the PP settings, and the other is displaying 3600 without the PP settings. Would someone be kind enough to shed some light on what I am seeing here. I will greatly appreciate it.

Thanks so much for any assistance.

Cheers,

-Emile Wamsteker

Perhaps it's because one setting is displaying 3600k with the PP settings, and the other is displaying 3600 without the PP settings.


What are the PP settings?

Not knowing the PP settings, and presuming they are affecting color downstream of the base white balance temperature read by the camera, that is the logical explanation me thinks.

Hi Max,

These are the PP settings

Matrix ...............on
Select................hisat
Level..................0
Phase.................-5
R-G...................75
R-B...................0
G-R...................-18
G-B...................-32
B-R...................-27
B-G...................13
Gamma Level.............. 0
Gamma Select.............CINE1
Black..........................-12
Black Gamma..............0

Hi Emile...

Just read your post. White Balance is another way of "normalizing" the camera sensor for the "color temperature" of the light that's illuminating your picture. Very generally, White Balance is around 5600 deg Kelvin for bright sunlight and 3200 deg Kelvin for indoor incandescent lighting. An actual scene can vary from these numbers depending on the light source or light mix.

When one white balances a camera, they are basically making the color interpretation by the camera as neutral as possible. In order for the Picture Profiles I posted to work properly, the camera MUST be white balanced for the light source. Otherwise the PP's are worthless.

Hi Emile and Bill,

According to the post Emile did execute WB to the scene. Then changed the Preset temperature to match (numerically) to what the camera read. When switching between the scene WB and the Preset a shift is observed.

If you can post frame grabs of the scene WB and Preset Emile that would help. Barring that I'd go with temperature shift by numbers not being as accurate as the same number derived from a scene WB. Even though mathematically 3600 equals 3600. Doesn't sound like it makes sense I know. For a cohesive answer to that I'd want to first know Sony's process for programming in Gamma and Red Blue channel response in correlation with the kelvin choices when changing the Preset.

Hi Emille

The PP settings give you the ability to specify the preset colour temperature, I guess because you might want it to be appropriate for where you predominantly use the camera. If it lives in a studio, 3200 would make more sense than the 5600 or higher for outdoors use. You could have some PPs that you use in daylight and others in studio, so might want a different preset WB.

However, it is only the preset, and is only used if the switch is set to preset. If you do a white balance (or use ATW) the preset value makes no difference.

N

First I want to thank you all for helping me on this.

I am still not entirely clear, but making progress.

It may help for you to know how I understand WB and tell you how I have my WB workflow set up.

The EX-1 has a WB toggle switch on the side with three settings: PRST, A, and B. I know you know this, but please bear with me, I just want to make sure we're all on the same page.

PRST is the factory preset which comes shipped at 3200k. A and B are programmable for custom WBs. Actually "B" comes shipped in Auto Trace WB, but it can be configured as memory, which I have done.

My setup: I changed PRST from the factory preset of 3200k to 5600k (for run 'n gun situations outdoors when I don't have time to do a custom WB). I programmed "A" to 3200k (for run 'n gun situations under tungsten when I don't have time to do a custom WB). "B" is for custom WBs only. Because I try to custom WB as much as possible, I am mostly toggled to the B setting.

So, when I do a custom balance (toggled and saved) to B memory, are my PPs pulling those WB numbers? I ask this question because when you go into the PPs menu [PICTURE PROFILES>SET>WHITE>PRESET WHITE: (color temperature)] it appears that it is set up to pull whatever number is associated with the PRST toggle--in this case 5600k. If it is pulling 5600k, and I am toggle to "B"--which is, say 3600k--then there is a problem.

More clarification will be greatly appreciated.

Emilie....

Several comments:
firstly, the EX1 should use whatever WB setting you have selected with the switch, regardless. A= custom preset, B= custom preset, PRST=3200 or whatever you've set in the PP menu.

Having said this, I believe there is some evidence that the WB derived from pressing the WB Button, on the front of the EX1, is not exactly correct. IOW, if you see a difference between an embedded preset of 3200 and a "calculated" 3200, I would assume the in-camera preset is correct and the "calculated" 3200 derived from pressing the WB button, is not correct.

Just as a side note: I hope you're taking your WB measurements from an evenly illuminated white or grey card. It's very important that the white/grey card you use to set the WB is illuminated by the overall illumination source and not influenced by a colored reflection off of a nearby specular source like a blue wall, red brick wall, or whatever.

Hi Bill, and Max,

That's perfect. All right, so I'm good on the PP white setting. WB values go into PP based on my external toggle switch position.

About the difference between the calculated and embedded preset? The camera is programmed to show the color temperature in 100k increments. It does not break it down into single unit values. Could it be that the true reading is somewhere in between? For example, in this situation, I did a white balance and the number that came up was 3600k, which I saved to "A" memory. Is it possible, for example, that the true calculated reading was 3649, but a NUMERIC value of 3600k was shown because it needs to round up or down the numeric display value to the nearest 100k? However, the IMAGE itself is displayed on the monitor using the under-the-hood true value of 3649, and this is why there is a difference between the embedded 3600k image (which is a true 3600k) and the calculated 3600k image, which is really 3649 with a displayed numeric value of 3600k? I should mention the shift is very slight, but perceptible. If this might be the case, could a 49k shift even be seen?

Hi Bill, and Max,

That's perfect. All right, so I'm good on the PP white setting. WB values go into PP based on my external toggle switch position.

About the difference between the calculated and embedded preset? The camera is programmed to show the color temperature in 100k increments. It does not break it down into single unit values. Could it be that the true reading is somewhere in between? For example, in this situation, I did a white balance and the number that came up was 3600k, which I saved to "A" memory. Is it possible, for example, that the true calculated reading was 3649, but a NUMERIC value of 3600k was shown because it needs to round up or down the numeric display value to the nearest 100k? However, the IMAGE itself is displayed on the monitor using the under-the-hood true value of 3649, and this is why there is a difference between the embedded 3600k image (which is a true 3600k) and the calculated 3600k image, which is really 3649 with a displayed numeric value of 3600k? I should mention the shift is very slight, but perceptible. If this might be the case, could a 49k shift even be seen?


Hi Emile and Bill and Nick,

It's a good question I think.

Can't speak for what is happening inside an EX camera until I run it by an EX engineer although traditionally that's how it works. What the temperature meter in your camera reads is accurate to its photocell but the readout is rounded off to the nearest 100k even if the actual temperature is in between the 100k space. As I understand it, this by reason of 100k jumps having been determined to approach the limit of perceptual color changes and being easier to work with when matching temperatures with other cameras and lighting.

You said it's a slight difference and that is making sense. That's why I asked if you could post frame grabs since I don't know how good your eyes are.

Emilie....


Having said this, I believe there is some evidence that the WB derived from pressing the WB Button, on the front of the EX1, is not exactly correct. IOW, if you see a difference between an embedded preset of 3200 and a "calculated" 3200, I would assume the in-camera preset is correct and the "calculated" 3200 derived from pressing the WB button, is not correct.


I'm assuming the opposite Bill. Until I know Sony's engineering steps that determined what for example "3600k" is supposed to be when programming the Preset position, between custom and preset WB I'd call Preset inaccurate. Sadly there is no accepted industry standard for the photocell designs going into devices when it comes to reading temperatures. Take 5 cameras and WB on the same card and you will get 5 different readings. This is also the case with color meters.

Max,

I will post some frame grabs tomorrow. I guess I can post them directly to this thread? I'm new to this forum so I'm not sure. Looks like I can.

Thanks for the feedback.

Cheers,

-Emile Wamsteker

FYR
http://www.dvinfo.net/conf/sony-xdcam-ex-cinealta/234446-strange-wb-results.html

Hey Bill,

Haven't read that thread yet, but I will. Thanks.

Max,

Attached are two files. Both were created with all the camera settings in manual mode. No AWB, no auto Iris, no nothing. I was shooting through a Letus adapter with a Canon 35mm 2.0 wide open. I didn't manipulate the images in any way. I transferred them to FCP via XDCAM and then exported them as still images from FCP.

One file, pictured to the left, titled EMBEDDEDPRST3600, represents a frame grab from footage when shooting toggled to PRST (at 3600k). Just to remind you, the PRST setting was originally set at 5600k, but I switched it to, 3600k through the PP menu [under PICTURE PROFILES>SET>WHITE>PRESET WHITE: 3600] before shooting the clip. The other image, pictured to the right, represents a frame grab from footage when shooting toggled to the WB "A" setting which custom white balanced to 3600k.

The difference is actually more pronounced than I could make out on the camera's monitor. Look at the shadows in the embedded preset version--they shift to cyan.

Anyway, I don't know if I'm splitting hairs here. I am mainly concerned with knowing how the camera "thinks" so that I can control it and therefore can get predictable results.

What is the light source?

The preset white balance settings assume a pure full spectrum light source while the manual white balance will to some degree compensate for light sources that are not full spectrum or have peaks in some colours. For example light reflected by some materials or paints may not be full spectrum as some wavelengths may be absorbed by the material. Normally you would expect a manual white balance to give you the purest whites for the given lighting while the preset will give you a close approximation.

One advantage of using a fixed preset is that it can make grading easier as you have a consistent setting to start with. It is also useful for multi-camera shoots to help get all cameras to match when they can't all white balance together at the same time.

Hi Allister,

Thanks.

The light source is window light on an overcast day. I guess what we are trying to figure out is why two separate 3600k readings give different results. One of the readings was created and saved to the WB "A" memory, and the other is the camera's embedded preset of 3600k. I suspect the difference lay in the rounding up and down of numbers by increments of 100k. While the camera may display a numeric value of 3600k, the true value (which could be 3649, or some variation thereof) is reflected in the actual image.

As much as we want to be scientific about this, the true tolerances of WB and PPs are not as tight as most of us would like, and only a Sony engineer would be able to explain the discrepancy I've found here. That said, I've learned a couple of valuable things out of this: take care to white balance your camera properly and choose the picture profile most appropriate to the given scene. Further, keep your head about you and realize it ain't gonna be perfect but that you should be close to the desired result. In the end, it will be refined in Post anyway. The idea is to try to come as close in-camera as possible to minimize the amount of time spent tweaking in FCP later.

Thanks again, everyone! Especially Bill for generously sharing the picture profiles he built.

Cheers,

-Emile Wamsteker

You guys are over thinking this whole business.

When the camera gives you a color temp estimate of what your white balance is, it is only that - an estimate not an dedicated expensive color meter. Over the years I have seen these numbers vary wildly. They are just estimates from a cheap camera circuit designed to help you have a relative idea of what your color temp is.

The presets are probably more accurate but again they aren't designed to be perfect color meters. The main thing is they are a different circuit so it would be plain amazing if they actually matched. Don't expect 2 cameras to match perfectly and don't expect these numbers to match other cameras. Use a decent monitor if you care about your color and get used to what your numbers actually mean.

Leonard...
I agree.

The difference is actually more pronounced than I could make out on the camera's monitor. Look at the shadows in the embedded preset version--they shift to cyan.

Anyway, I don't know if I'm splitting hairs here. I am mainly concerned with knowing how the camera "thinks" so that I can control it and therefore can get predictable results.

Well Emile, after seeing the pictures you are certainly not splitting hairs. That is a big difference. It is NOT a difference in between the 100k space. But from seeing them I'm glad to say it looks like there is nothing wrong with your camera.

For an overcast sky through a window the Preset is looking as it should that being more neutral. And the Auto WB is looking proper that being more warm. When you approach 4k especially, and for outdoors, preset will produce a picture more neutral than an Auto WB.

The order of accuracy as I think you know is such:
Preset
Auto WB
Manual WB

*******

General Comment:
Auto WB is pressing the camera's WB button, many
refer to this as Manual WB. This is a misnomer.

What is being increasingly referred to as "Auto WB" with
these sub $10k cameras is more aptly termed
Auto Tracing WB.

"Manual WB" in a professional camera infers manual
balancing of the encoder channels by an engineer.

*******

The preset may look cyan in comparison to the Auto WB here but it is actually more neutral given the light source. I was matching two broadcast cameras (Sony) today and had a similar difference under tungsten. The same experiment under tungsten will show a similar effect with green replacing cyan. Further below...

Do consider any color balance derived from any camera's Preset and Auto WB calibrations is created by the manufacturers inclination towards what looks pleasing for their cameras. An Ikegami Preset will be different from a Sony, from a Panasonic from a JVC. What you are seeing in Preset is the manufacturer's textbook rule of what that temperature should look like, and sometimes this will look ugly because as Alistair alludes that textbook rule doesn't control the lighting in front of the camera. Of course, the other way around hence Auto WB.

One temperature Preset is one thing. Preset temperature programming is another and a relatively unique feature. Thus the question stands on how Sony determined the color balance for the temperature steps. How were the incremental values of gain determined for the R and B channels?


...a fixed preset... it is also useful for multi-camera shoots to help get all cameras to match when they can't all white balance together at the same time.

If it is the last resort yes. I'm confident you know they will not match perfectly especially different cameras. More especially from different manufacturers.


You guys are over thinking this whole business.

When the camera gives you a color temp estimate of what your white balance is, it is only that - an estimate not an dedicated expensive color meter. Over the years I have seen these numbers vary wildly. They are just estimates from a cheap camera circuit designed to help you have a relative idea of what your color temp is.

The presets are probably more accurate but again they aren't designed to be perfect color meters. The main thing is they are a different circuit so it would be plain amazing if they actually matched. Don't expect 2 cameras to match perfectly and don't expect these numbers to match other cameras. Use a decent monitor if you care about your color and get used to what your numbers actually mean.

Not with you on that Leonard.

One would be wholly justified in expecting your camera to give you temperature readings accurate to it photocell within a range of +/-200k. Beyond this range is NOT an area of acceptable estimation. It is without argument an absolute performance requirement, and any professional camera not performing up to it is technically defective.

Essentially the working solution to temperature readings as they affect color balance variation is familiarity with equipment behavior. Specifically this is by determining numerical offsets. Every piece of gear with a temperature meter inside it will behave differently than the next except for all other things being equal, two identical units.

Leonard, if you light using a "dedicated expensive color meter" the accuracy of the meter is irrelevant vs. your camera. As I've said in the previous posts and other threads you will be dealing with two different photocell designs and deciding on temperature with your meter will not produce that precise color balance in your camera. The only relevance is the offset between your camera and the meter.

In my opinion the video monitor is in a way the blight of video. In essence, a handicap instrument than an instrument for accuracy. There is a wide span of indeterminate variation between phosphor batches and LCD backlights. On the other hand the variation between color meters and cameras is determinable.

FYI, yes you can make two EX cameras match as well as any other professional camera. This is done first through the maintenance menu, not through picture profiles.

FYI, yes you can make two EX cameras match as well as any other professional camera. This is done first through the maintenance menu, not through picture profiles.

Sony is mute on this whole subject because they provide NO information what so ever on the maintenance menu and the most basic of window dressing to make practical use of their picture profiles. Please do share to enlighten the rest of us mere mortals ;-)

"One would be wholly justified in expecting your camera to give you temperature readings accurate to it photocell within a range of +/-200k. Beyond this range is NOT an area of acceptable estimation. It is without argument an absolute performance requirement, and any professional camera not performing up to it is technically defective."

Max, Its nice for you to set those standards, I've just never seen that kind of accuracy in cameras and I don't think the manufacturers care about it all that much. Just my guess. I remember when the D30 cameras first came out , 3200 degree lights would read as 2200. we got used to it. Sony cameras often used to balance green as a matter of course so we got used to carrying green gel to cheat it back.

Presets look different both because of color temp and the colorimetry and matrixes of the individual cameras. Lots involved.

I would never use a color meter to set my video color balance, just trying to say that I doubt whether manufacturers care that much about this system being accurate.
After 25 years shooting, I find a trustworthy monitor and waveform to be the best way to set exposure and color and don't personally know anyone who prefers any other method. That's why in a well crewed professional situation you usually have a good monitor, a paint box and a tech.

Of course once you know your camera then you can trust how it white balances or what the presets look like - until you've seen the evidence though its a crap shoot.

Hi Leonard,

My point is if your camera read tungsten as 22k that does not make your camera inaccurate. You said carry green gel so I'll assume this was ENG with varying light sources. As long as it does not read 22k one day and 25k the next. The green issue aside, 22k is accurate to the photocell receptor in that camera so it can be considered tungsten balance for that camera. Now if you're matching another camera which reads the same light as 32k, you know that means 22k in your D30 so you won't raise your D30. Or if you want to cool the image you won't light or balance at 35k you'll raise to 25k.

I haven't seen D30s, D35s, D50s or D55s balance that far off. Not saying you're wrong but I haven't seen it. If AWB yielded so much green that is not normal. Some red or blue is typical of the SD Sonys but not green.

I understand what you mean. Not saying don't use monitors but on the scale of accuracy monitors don't score high in my book. Waveforms are an evolved incarnation of a meter and obviously in the top of that scale. It deserves a look back to note how the film industry developed and matured in this regard. It wasn't with the use of monitors but meters. As you know not having WYSIWIG required DPs to know their cameras and stock intimately, creating the image with care. I think video has been spoiled in that we place unconditional trust in the monitor. So the history of video is wrought with ugly lighting and bad levels. A waveform as you included can solve all that of course.

Meters are highly useful, they shorten lighting time immensely. When I know the sensitivity of a video camera and can match my light meter I no longer need to walk back and forth to the monitor or the waveform while lighting. I basically use them only for a final check this way. Especially useful when lighting something in the distance.

As far as color meters, many many uses. You can pre-balance 1 or 10 cameras to a scene when you've metered the lighting beforehand and you don't have to be on location. On location hook up, turn the cameras on and everything is matched. You can match different temperature sources without a monitor or waveform. You can tune LEDs to not cause false color errors without having a camera up. It goes on and on.

Many video guys are unaware of why or how metering for video can be used. Of course being me I blame this on the security blanket of the video monitor. But if you use a waveform then we can be friends.

Sony is mute on this whole subject because they provide NO information what so ever on the maintenance menu and the most basic of window dressing to make practical use of their picture profiles. Please do share to enlighten the rest of us mere mortals ;-)

Hello Barry,

I have to agree with Sony on being mute on the maintenance menu. It is very simple to mess up your camera with the amount of controls available in there which frankly I was surprised they included, but that's part of the reason why my company bought one.

Please understand I don't want anyone to mess up their camera. To describe it what you'd be doing is what the camera does when you press the AWB button and ABB function. Except you will be doing this manually. Adjusting gain on the RGB channels for White and Black. After this you'd then go into the PPs to match all the gammas, then you will be lining up the matrix.

I can't say how strongly I recommend to stay away from this unless you fully understand what you are doing to the camera. A vectorscope is a necessity, a waveform is preferrable and a test chart with color chips is needed. If you have to do this then have an engineer or DIT perform it. If something goes wrong (which it shouldn't) you at least have some recourse now as they will be responsible for the work.

If you want to get as close to this as possible then perform AWB not using a white card or warm card but using a greyscale. As always the only test charts, be they white cards, warm cards or anything else, I recommend is DSC Labs.

Of course you can always decide to learn camera engineering too. Read, talk with engineers, maybe take a course. There are "engineers" I know who follow some strange processes so let's say it's good part "art" also. I'm always learning myself.