White Balance / Processing
3rdPlanetPhotography
Banned Posts: 920 Major grins
Need Advise.....
All of my photos seem orange to me. I white balance with a gray card and this is the result.
Equipment: 20d, Tamron 28-200 lens, 2 Alien Bees.
1. I set the camera to auto white balance.
2. I shoot a photo of the gray card.
3. I set the camera to Custom White balance (on top)
4. Then in the menu I select Custom WB and then select the gray card image.
Isn't this the proper way to set white balance? Do my photos look orange or red to you?
The parameter settins are all half in the menu if that helps.
Any Ideas or is the image what you'd expect?:dunno
oh yea... I'm also shooting RAW. Brought the photo in to picasa and did nothing to it but export it to JPG.
All of my photos seem orange to me. I white balance with a gray card and this is the result.
Equipment: 20d, Tamron 28-200 lens, 2 Alien Bees.
1. I set the camera to auto white balance.
2. I shoot a photo of the gray card.
3. I set the camera to Custom White balance (on top)
4. Then in the menu I select Custom WB and then select the gray card image.
Isn't this the proper way to set white balance? Do my photos look orange or red to you?
The parameter settins are all half in the menu if that helps.
Any Ideas or is the image what you'd expect?:dunno
oh yea... I'm also shooting RAW. Brought the photo in to picasa and did nothing to it but export it to JPG.
0
Comments
Also try using a white card and then set the white point using that, instead of grey.
-Fleetwood Mac
I read somewhere - maybe Andrew Rodney - that an 18% kodak gray card is fine for setting exposure, but that it is DEFINITELY NOT a neutral gray and should not be used for setting color balance.
Try using a white sheet of paper or a page from the back of one of Kelby's books that has a white, black and gray quarter page for setting white balance. See if that doesn't work better for you,
The other suggestion would be to chose a color balance setting other than AWB, but sun, shade, or tungsten whatever ( it does not matter ) when you shoot your white card. Then in ACR use the white card with the eye dropper to get the correct color balance for your images. The reason to avoid AWB is that it can vary from frame to frame, while choosing a setting will be consistent from frame to frame even if it is consistently wrong the same from frame to frame. You plan to correct this error in ACR once, not for each frame.
I think your image is too orange also - I am viewing with an Apple Cinema Display calibrated with a Spyder2Pro.
Moderator of the Technique Forum and Finishing School on Dgrin
I'd say the white balance is definitely off. If I sample a few points in Photoshop and look at the values in CMYK, I get C=0,M=27,Y=43 and C=10, M=38, Y=60. In both samples (one forehead, one on the arm), the Yellow is quite high and is partly why they look orange. I would expect the Yellow to be just a little higher than Magenta. In the first point, the Cyan is unusually low.
If I read this document about setting custom white balance on the 20d from a grey card, I'm not sure you are doing it right. I'd suggest reviewing the steps in that document or it says that the instructions are also on page 51 of the 20D manual.
FYI, when shooting RAW, I sometimes take a reference shot that includes a grey card and then use that photo to set the white balance on the rest of the shots in my RAW editor and don't worry about the white balance in the camera at all. In Adobe Camera RAW, I can just click the eyedropper on the grey card (which will make it neutral) and then copy the resulting temperature and tint settings to the rest of the shoot and my white balance is now set for all photos.
Homepage • Popular
JFriend's javascript customizations • Secrets for getting fast answers on Dgrin
Always include a link to your site when posting a question
OK, you don't want to use a gray card to WB in a raw converter but instead a white card! If you have say a Macbeth 24 patch color checker, you'd WB on the 2nd white square. Or something that's not a specular highlight but a white with some tone.
The WB on the camera plays no role in the raw data, its simply a recommendation. You're dealing with Grayscale data. While you can mess around with auto or custom WB, it will not affect the actual raw data at all. You're simply making an EXIF data recommendation that the converter may (or in some cases may not) see and use.
Again, the real issue here is probably using a gray card. Of course, gray or white card, it really has to be neutral!
Also, be sure you have a calibrated and profiled display. Its not impossible for a properly WB image to look off when it isn't (the display is).
Author "Color Management for Photographers"
http://www.digitaldog.net/
http://www.chrislaudermilkphoto.com/
WB, especially in ACR or LR is not the same as Gray Balance and there is a difference! Mostly due to the linear encoding of raw data and where the most tones (first stop of highlight data) fall. If you want direct quotes on this from Thomas Knoll who wrote ACR, I'd be happy to dig it up. The correct procedure is to use a non specular white for WB in both products. In a gamma encoded rendered image, you'd gray balance if the goal is to produce neutrality.
Author "Color Management for Photographers"
http://www.digitaldog.net/
All you need to know about PROPER WB in ACR/LR (see page 2 about the correct values to use).
Author "Color Management for Photographers"
http://www.digitaldog.net/
1. I white balanced with the "white" card versus the gray.
2. The camera color temperature is set to 5000k
That's it.. PP is the same, there is none. Just exported to JPG.
Would like to know your thoughts on the differences.
Perhaps just a little bright, but the skin tone color looks great to me and the CMYK numbers look like you would expect they should (19,43,48).
Homepage • Popular
JFriend's javascript customizations • Secrets for getting fast answers on Dgrin
Always include a link to your site when posting a question
Got to say, this skin tone, CMYK stuff is totally dependant on the CMYK color settings (profile) you want to use to translate numbers from RGB.
Load a CMYK setting for newsprint, then Japan 2002 Newsprint, then Euro ISO coated and tell me they aren’t all different.
RGB is a bit better but you have to define the RGB working space!
If you want to at least talk color numbers that are not ambiguous use LAB that everyone here loves to use. At least its a device independent color space, the color numbers are always defined.
Author "Color Management for Photographers"
http://www.digitaldog.net/
I'm using the techniques described here and they have worked quite well for me. They take much of the guess work out of using your eyes to judge it on screen. Are you saying there's a problem with this and that it doesn't work?
Homepage • Popular
JFriend's javascript customizations • Secrets for getting fast answers on Dgrin
Always include a link to your site when posting a question
I'm saying that depending on the color settings, the numbers can be significantly different because the numbers are calculated using a specific CMYK printing device defined by an ICC profile currently loaded in your color settings. You're viewing the numbers in RGB (maybe in a working space or output space) and each such space will produce a different set of resulting numbers depending on just that RGB color space. Try viewing the same skin tone image in sRGB, Adobe RGB and ProPhoto RGB. Then do the same and switch CMYK spaces. So you've got a big jumble of numbers based on a lot of variables, one of which probably isn't the device you're going to use.
Or you could have a profile of your display so the RGB numbers in your documents actually produce the correct color appearance for any user. I acually see the numbers appear as you saw them. IF the display provides a reliable preview of the RGB numbers (and I'll add, the RGB OR CMYK numbers for any device you have a profile), why all this numbers stuff? Or, use the profile for the final output device to provide the correct set of numbers. That's what profiles do. When you use that CMYK color space, you're asking for the correct CMYK values for that one printing condition.
Why view an RGB image, in a working space (realizing that users work with many different RGB working space) based on some CMYK device you're not printing to?
Here's another idea. Say once you find a set of RGB numbers in a working space that DO output to YOUR device that you like. You now save off a low rez version that you open when working with other similar docuemnts as a visual guide to color correct. You may have a set for various kinds of skin tone. In the current working space why not memorize (if you so desire) the current numbers for that RGB working space since using the correct output profile will traslate that into the correct RGB or CMYK numbers for any device you want?
Now you calibrate and profile the display such that the color numbers of ALL color spaces preview as you expect to see them (that's what color management in Photoshop is for). Why not use the correct previews AND the correct numbers based on the device you're actually going to send the numbers to?
This idea of using CMYK numbers or even ratio's was a decent one back in 1994. Now its not a big deal to actually see a simulation of the skin tone color appearance based on your output device in Photoshop. Better yet, the numbers are based on this device.
Author "Color Management for Photographers"
http://www.digitaldog.net/
I'm sorry, but I don't understand how what you wrote translates to a specific suggestion for how to adjust skin tone better than what I'm doing now. It sounds you think using CMYK ratios is dumb, but I don't understand what better technique you are proposing. Are you suggesting that I should just use my eyes on a calibrated display? I have no idea how to make RGB ratios work. While it seems like it should, in theory, be possible to use them, I can't find any simple rules that work like the CMYK rules.
In case I wasn't clear, I have Photoshop on a calibrated display, I'm lookng at an sRGB image and I've configured the info palette to show me CMYK values for a particular color sample(s) in the image. I find it extremely helpful to evaluate the M to Y ratio (Yellow never less than M) and the C to M ratio (C always a fraction of M, depending upon skin type and light conditions). My eyes will often tell me something is off, but these ratios tell me how far off and give me immediate ideas on what type of correction will work best and how much to apply. Since I've started using this technique (about a year ago), I'm a lot more accurate and faster at evaluating skin tone and making appropriate adjustments. I'm open to something better, but haven't seen or understood anything that works better yet.
Homepage • Popular
JFriend's javascript customizations • Secrets for getting fast answers on Dgrin
Always include a link to your site when posting a question
Lucky for me, Sarah is my fiancee and we can jump in the studio at any time
Yes. And I'm suggesting you use images that have output to your satisfaction as a numeric guide in the RGB working space you use, then you can output to ANY device you have a profile for. So you have both visual and numeric feedback based on any output device, not one you're not printing to.
Doesn't it make more sense to look at RGB numbers in one common color space (your working space) that can now be used for any output need?
We know that in a well behaved RGB working space, when R=G=B, you have a neutral. That's a ratio that works in all RGB working spaces (but not necessarily capture and output color spaces). Now the exact numbers for the same color appearance will be different in sRGB, Adobe RGB (1998) and ProPhoto RGB but they all equal out, so they are neutral. The other approach is to use some CMYK values for a neutral but that's solely based on a single output device, one you may or may not be using. And if you're using it, the next output device will produce different values. So I don't see the reason you'd use output color space numbers when they are device dependant and you might be using a dozen different devices (or the same device like an ink jet with a dozen different papers which would produce a dozen different necessary output values).
Not sure you even need to worry about numbers and ratios. How does the skin tone look when you view it? But, some folks feel kind of macho when they 'work by the numbers' so you could, using RGB documents of skin tone you know output correctly, begin to view them and teach yourself these ratio's and numbers.
Frankly, most by the numbers exercises is a lot of unnecessary work with the exception of knowing about highlight and shadow clipping values and neutrals. In RGB, that's really easy. We saw what makes a neutral in RGB working spaces. You probably have a pretty good idea about clipping for shadows and highlights by recognizing that 0/0/0 is a black hole and 255/255/255 is pure white. You can also use the calibrated display and Photoshop (ACR and LR) provide very useful clipping overlays for both saturation and tone clipping (Option/alt drag Exposure and Black sliders, what do you see? A white pixel clips to pure white, a black to pure black and colors show you color/saturation clipping).
But the numbers and ratio's are based on whatever CMYK color setting you have and wouldn't it be far more effective to simply look at the RGB numbers instead of numbers based on something you're not printing to?
The basic idea you want to use works in any color space you know the numbers for but the bottom line is, the profiles define these numbers.
Author "Color Management for Photographers"
http://www.digitaldog.net/
I'm sorry, but you've done a lot of writing to try to explain why the method I learned from Smugmug's page and have used quite successfully is a bad way to do things.
But, all I've been able to glean is that you think I'm doing it wrong and I'd be better off going by how it looks on screen than paying attention to any numbers. Or you want me to invent my own scheme with the RGB numbers when I've already told you I don't have an RGB scheme and don't know how to make one.
If that is, indeed, your conclusion then we can just disagree and be done with the conversation. If you have a specific scheme for looking at the numbers that will work in RGB, I'm happy to try it.
Homepage • Popular
JFriend's javascript customizations • Secrets for getting fast answers on Dgrin
Always include a link to your site when posting a question
And yes, we are done with the conversation.
Author "Color Management for Photographers"
http://www.digitaldog.net/
You implied it was bad when you said it was OK for 1994, meaning that we have since bypassed the method. And then again when you insisted that there are much better ways to do things. And finally when you said that working by the numbers was unnecessary and macho.
Here's what I understand your prescription to be. Get a fully calibrated system. Have a working space that mirrors, as closely as possible, the RGB parameters of your output device. Develop a series of samples, for different skin types in different lighting conditions, that you can use for purposes of comparison. Make corrections using the RGB numbers from those samples as a guide.
This might work. I don't see any reason why it wouldn't. But I also don't know for sure that it would end up getting better results that what they are now getting using CMYK numbers as their reference.
It also requires, as I understand it, that every edit be targetted to a specific output device. That's not practical for many people's work, where because of time constraints, people have to make corrections that will be acceptable for, e.g., delivery on a website and also for small prints (say 4x6 or maybe 8x10).
If you are advocating something that will require each person to put in a bunch of extra work, simply to get up to speed and have the requisite comparison skin tone samples, then I think it might be helpful to show, with some examples, how this more burdensome procedure will lead to better results.
But if the conversation is at an end, so be it.
Duffy
I didn't imply it. I did question it based on more modern approaches to handling matching image appearance on multiple devices.
Working by the numbers is far from mandatory and working visually is often a more effective way to handle image correction. Its certainly easier for users. Both used together is better, one alone is easier and more powerful. Numbers alone don't provide enough information to produce a specific color appearance (unless those numbers are in LAB).
No, I didn't say that about the working space. Its totally separate from the output space, hence the color space you would be far better off referencing as targeted color numbers.
IF you know the correct ratio or set of RGB numbers for a skin tone you prefer visually AND on output, all you have to do is know them for this one color space. Any other color space values will be calculated from this data. Its the foundation upon which every other color space will be based. Using the output color space is putting the cart before the horse unless, like the days of old, you only had one output device to worry about. This is the genesis of working by the numbers. But we don't have one device (some press in the next room) we have to target. We might have hundreds. Those resulting numbers will be defined with an output profile FROM the working space you have edited the RGB numbers to get a desired color appearance.
Unless you have some color blind users, you don't even need the numbers.
If you have documents that have a known and desired color appearance on output, you can now visually reference them while looking at the images that you think need correction. You have numbers that were used to produce the accepted color appearance in a single, source RGB working space you use all the time.
I would submit it would work better for those who don't want to mess with numbers and have a calibrated and profiled workflow. It will work no less with someone who, for whatever reason can't see the colors and has to work numerically.
Just the opposite. You're editing the master RGB image in your working space. Once you decide what output device you want to use, you can now soft proof it from the master. You can edit based on output conditions if you wish at this point but that's your call.
Actually when setup properly, it reduces work and as importantly, extra waste with output.
Author "Color Management for Photographers"
http://www.digitaldog.net/