Which is better, using “In camera white balance PRE” or using third party software (such as RawTherapee) for setting the white balance?
(I forgot to mention, this is assuming everybody already knows a white balance value is not required for a raw image file to be processed, and can be readily provided later within software.)
I hear a lot about software being better with most image rasterizing versus using a camera’s limited image rasterizing abilities, however white balance seems like a very non-trivial task and allowing the camera to use a custom white balance seems fine to me.
Some benefits I realize using setting the white balance within the camera prior to taking a series of photos:
The white balance is already set correctly, so JPEG’s and/or embedded JPEG’s within RAW image files (eg. Nikon’s NEF raw files) are already white balance calibrated correctly. (NOTE: JPEG’s are likely always best taken setting a custom white balance in camera, as a JPEG has issues with correcting white balance afterwards.)
One less step when using a raw image editor, correcting the camera’s image auto white balance using a Color Checker white balance card, along with having to remember to take the additional photo of the Color Checker during the photo session!
The in camera preview better reflects the final corrected image color tones, as the in camera image is already white balance calibrated.
Funny, Stack Exchange has a similar question posted, yet all the answers seem to get side-tracked and fail to mention which is best.
I’m going to take a guess at the final answer or best suggestion and state, when time permits, it is probably best to always take along a Color Checker or White Balance card, and configure the camera to use a custom white balance for each photo session. My only guess for using third party software to perform this task on raw image files, software does a better job for some unknown reason. But for the time being, I’ll assume this is a simple task easily left for the camera to perform!
One benefit I just thought of, taking a separate Color Checker or white balance photo/image, a better record is kept of the white balance conditions during the photo session. For example, also taking a photo/image of the Color Checker color tones, can easily adjust tones for artistic reasons.
If you shoot raw, and you should, then the white balance from the file doesn’t really matter. If the file white balance is correct, use it. If not, correct it.
A color chcker or gray card (way cheaper), is a good solution!
If your question is whether it is better to set a good white balance in the camera than to shoot with a white balance which is wildly off, then the answer is yes, it is better to have a good white balance in-camera even when shooting raw.
In addition to the reasons you mentioned, some steps involved in processing a raw file which take place before the white balance the user set in the program takes effect can perform better on correctly white-balanced data [1]. Since these steps lie before the white balance as set by the user in the program is applied, then how should the program get a good white balance? Either it doesn’t (multipliers [1,1,1]), or it assumes daylight, or it sets the white balance based on values from the image metadata, or it performs auto-white balance. If the program uses white balance values from image metadata, then the white balance you set in-camera matters to some degree.
The histogram you see on your camera’s display is based on the embedded JPEG, it does not show the real raw data. As such, the white balance again makes a difference to some degree, as the user adjusts the exposure and other parameters based on the histogram or preview - as might the camera’s algorithms.
Approaches such as UniWB [2][3] aim to take advantage of the fact that the in-camera white balance setting ends up affecting the raw file in one way or another.
In my quest for a batch workflow to produce proofs, I’ve found having a measured WB in the metadata really helps to get a better-looking proof. My cameras support collecting such a measurement and inserting it in the metadata of subsequent shots, and I’m finding this really works well in indoors-natural-light situations. I still just let the camera do daylight auto WB for grab-n-go exterior shooting.
Either way, I take a look at the proof and decide if WB needs further attention. And, what I usually do for that is to inspect the histogram, looking for shifts in the RGB channels, especially at the high end. Then, if I see a discernable shift, I just scooch the camera multipliers to line things up, and that usually makes the picture look better. Very unscientific, but it works…
Great, thanks for confirming my thoughts upon the easiest workflow method!
I think Morgan stated, “some steps involved in processing a raw file which take place before the white balance the user set in the program takes effect can perform better on correctly white-balanced data [1]” was one of my area of concerns/questions.
In other words, during digital cameras minimal processing with rasterizing RAW images/exposures, better more favorable results are seen with a good preset white balance.
I’m finding it’s best to always ensure to revert the camera to AUTO WB after using a preset, and trying to remember to retake the test white balance photo again just prior to the next photo session.
This is really tricky to remember to do, but the previous step does help to avert using an older incompatible white balance setting during the next photo session. (With the auto setting being more easily fixed, else becomes extremely important to take a photo of the color checker white balance card.)