shimokita
白黒
As far as I understand it, the different color temperatures that you set in a DSLR (Canon 5D mark ii for example) is accomplished through software (not a big leap of faith there). My question (if it makes sense): Is there an optimal color temperature setting for the 5Dii that gives the least (color temperature related) software intervention? Is it 5,500K or ?
I guess that I am just assuming (always risky) that the sensor/firmware/? is optimized in a similar way to film.
Casey
I guess that I am just assuming (always risky) that the sensor/firmware/? is optimized in a similar way to film.
Casey
segedi
RFicianado
If I understand your question correctly, AWB, auto white balance should give the best result in most conditions with minimal adjustment needed. Shooting in RAW allows you to change the color balance fairly easily in post-production. Even in post, there are the typical sunny, cloudy etc or manual adjustment of the color balance.
In tricky lighting situations, it is possible to take a photo of something that is white and then assign a custom color balance based on that photo to the subsequent photos. There are other white balancing tools as well, but I am not familiar with their usage.
I hope that helps.
In tricky lighting situations, it is possible to take a photo of something that is white and then assign a custom color balance based on that photo to the subsequent photos. There are other white balancing tools as well, but I am not familiar with their usage.
I hope that helps.
Chris101
summicronia
This question was investigated some years ago by Iliah Borg. You can google his solution, "UniWB" to see how it works.
lynnb
Veteran
If colour is critical (e.g. wedding dress) I shoot a Kodak gray card and set custom white balance accordingly.
shimokita
白黒
as expected, simplistic logic of the original question...
@segedi / @ lynnb: my question was more in the direction of the technical aspects of digital sensors vis-a-vis in-camera color temperature management. Sorry about my wording.
@Chris101: thanks for that suggestion. Much appreciate your comment/ suggestion. A google search lead me (among other places) to libraw.org and the following article...
White Balance in Digital Cameras: Problems
http://www.libraw.org/articles/white-balance-in-digital-cameras.html
So, now I have some reading to do... (Channel Noise and Raw Converters -and- Using Magenta Filter for Shooting With a dSLR Camera Under the Daylight) and at least for now and barring any tangents, I am headed in the right direction. I can understand a bit of how the author sees the relationship between a specific sensor and RGB channel processing - in theory at least...
I am also interested in using filters on a dSLR for color temperature adjustments under various conditions, the channel issues is a good starting point.
Casey
@segedi / @ lynnb: my question was more in the direction of the technical aspects of digital sensors vis-a-vis in-camera color temperature management. Sorry about my wording.
@Chris101: thanks for that suggestion. Much appreciate your comment/ suggestion. A google search lead me (among other places) to libraw.org and the following article...
White Balance in Digital Cameras: Problems
http://www.libraw.org/articles/white-balance-in-digital-cameras.html
So, now I have some reading to do... (Channel Noise and Raw Converters -and- Using Magenta Filter for Shooting With a dSLR Camera Under the Daylight) and at least for now and barring any tangents, I am headed in the right direction. I can understand a bit of how the author sees the relationship between a specific sensor and RGB channel processing - in theory at least...
I am also interested in using filters on a dSLR for color temperature adjustments under various conditions, the channel issues is a good starting point.
Casey
gavinlg
Veteran
On the 5d you can select a custom white balance with the 'K' option/icon. Then in the main menu set the 'k' white balance to your desired Kelvin temp - 5500K is good in most outdoors situations and tends to give natural colors I've found.
sol33
Established
Correct.
The raw sensor data has no native color temperature. Without software (either the camera firmware to render a jpeg or a raw developer like Adobe Lightroom) there is no color at all. So I don't think it makes sense to quantify software intervention. Color temperature is just a number which the software uses to calculate the colors.
If you are talking about manual intervention: You can let the software choose the "best" temperature and use automatic white balance in the camera or the raw developer. This will be ok in most cases. But sometimes (like with mixed light sources) it is a matter of taste and the software doesn't get it quite right. In such a case, if you correct it later in a raw developer, it doesn't matter what color temperature your camera was set to. The setting in your camera is used for generating jpegs and as a hint for a raw developer. It doesn't affect the output of a raw developer at all.
shimokita
白黒
@sol33: what you say makes sense
I guess my question-ing comes in at the point where if an argument can be made for using a magenta filter for daylight shooting -or- an 80A filter when shooting in a tungsten environment... there must be some gap between the reality/ promise of RAW and the various implementations/ of RAW converters. My take is that I am coming a little late to the technical discussions, and have some catching up to do - sorry about that.
If using a technique such as uniWB to profile your dSLR and then use the resulting three color histogram to recommend a color balancing(?) filter, then I would see that as a way to enhance the sensor/ firmware/ RAW converter/ system workflow in a similar way that was done with color film(s). Maybe I am misunderstanding some points...
I guess in the big picture I am talking incrementals, but it's somewhat interesting.
Casey
I guess my question-ing comes in at the point where if an argument can be made for using a magenta filter for daylight shooting -or- an 80A filter when shooting in a tungsten environment... there must be some gap between the reality/ promise of RAW and the various implementations/ of RAW converters. My take is that I am coming a little late to the technical discussions, and have some catching up to do - sorry about that.
If using a technique such as uniWB to profile your dSLR and then use the resulting three color histogram to recommend a color balancing(?) filter, then I would see that as a way to enhance the sensor/ firmware/ RAW converter/ system workflow in a similar way that was done with color film(s). Maybe I am misunderstanding some points...
I guess in the big picture I am talking incrementals, but it's somewhat interesting.
Casey
willie_901
Veteran
The raw data does not rely on any WB setting. For convenience the WB setting at the time you record the photograph is included in the raw file so your raw rendering software's initial display will be close the the WB either you or the camera choose. During post processing you can alter the WB as you see fit, or you can just use what was set in the camera, or you can let your rendering software automatically guess at the optimum WB.
If your scene had light from multiple sources, there may not be a single optimum set of WB parameters. For instance a room with tungsten light on one wall, a window with daylight and fluorescent light on another wall will not have a single WB setting.
If you use a color temperature lens filter, then you are eliminating data from the raw file. This does not make sense to me as you can produce the same effect without an optical filter by changing the post processing software's WB parameters.
If your scene had light from multiple sources, there may not be a single optimum set of WB parameters. For instance a room with tungsten light on one wall, a window with daylight and fluorescent light on another wall will not have a single WB setting.
If you use a color temperature lens filter, then you are eliminating data from the raw file. This does not make sense to me as you can produce the same effect without an optical filter by changing the post processing software's WB parameters.
Terry Christian
Established
I think it's important to take a step back and understand what white balance is, and is not. It is not an inherent bias in your camera's sensor.
Light, as we experience it in the real world, has color. When white light bounces off or passes through a colored object, it picks up the color of that object (through discarding the remaining colors). On an overcast day, the light is more blue. In a fluorescent-lit office, it may be more green. Incandescent (tungsten) light is more yellow. The color of the light on a scene affects how the tones of that scene appear to the eye and to the camera. Putting yellow-orange light on someone's face for a portrait is flattering, for instance; lighting someone with greenish-blue light is not.
Generally when you shoot with a digital camera, you want to record the scene neutrally. If the scene is lit with green fluorescent tubes, everything and everyone in the room will have a greenish cast, which you may or may not see. You want to lift this veil of green so you can see the scene as it actually is. The auto white balance setting on your camera may pick up on this and adjust for it accordingly, but it may not. You can set the camera's white balance setting to an appropriate preset, or you can shoot a grey card and tell your RAW capture software to remove the color cast that it sees on the gray card (some cameras can also do this at the time of capture, like gavinlg was saying above).
Now back to your original question: is there a setting that gives the least software intervention to the white balance? Yes, theoretically the one that is the closest to the white balance of the scene in front of you, which may change from moment to moment. So, it is best to use the Auto white balance setting in most situations to let the camera decide the best white balance. You can change it easily in your post-processing if you notice that anything looks a bit "off," like skin tones not looking quite natural.
Light, as we experience it in the real world, has color. When white light bounces off or passes through a colored object, it picks up the color of that object (through discarding the remaining colors). On an overcast day, the light is more blue. In a fluorescent-lit office, it may be more green. Incandescent (tungsten) light is more yellow. The color of the light on a scene affects how the tones of that scene appear to the eye and to the camera. Putting yellow-orange light on someone's face for a portrait is flattering, for instance; lighting someone with greenish-blue light is not.
Generally when you shoot with a digital camera, you want to record the scene neutrally. If the scene is lit with green fluorescent tubes, everything and everyone in the room will have a greenish cast, which you may or may not see. You want to lift this veil of green so you can see the scene as it actually is. The auto white balance setting on your camera may pick up on this and adjust for it accordingly, but it may not. You can set the camera's white balance setting to an appropriate preset, or you can shoot a grey card and tell your RAW capture software to remove the color cast that it sees on the gray card (some cameras can also do this at the time of capture, like gavinlg was saying above).
Now back to your original question: is there a setting that gives the least software intervention to the white balance? Yes, theoretically the one that is the closest to the white balance of the scene in front of you, which may change from moment to moment. So, it is best to use the Auto white balance setting in most situations to let the camera decide the best white balance. You can change it easily in your post-processing if you notice that anything looks a bit "off," like skin tones not looking quite natural.
sol33
Established
I missed the UniWB part in my first post. UniBW should be easy to set up for a Canon 5D Mark II. It's basically a better way to display RAW Histograms and fine tune exposure.
If you use color filters, you basically change the relative exposure of the three color channels. E.g. a 80A filter would decrease the exposure of the red and green channels much more than the blue channel. This way you can expose the three channels independently of each other and reduce noise somewhat.
In any case, no matter if you use UniBW and/or color filters, applying automatic white balance in the raw developer will lead to very similar (if not identical) images.
Personally I just bracket (and use HDR afterwards if needed) if I really feel unsure about the exposure. Changing color filters seems complicated. But I would be interested to see some results of this method.
shimokita
白黒
I haven't gotten as far as to implement uniWB and further still from interpreting the results ;-). So far I am starting to buy-into the use of color filters albeit for possibly only incremental gain... and yes I am interested to see actual test results... have to consider what would make for a good test example...
Casey
shimokita
白黒
@willie 901 and Terry... Thank you for the comments.
I was guessing that it would make sense to use a color filter that compensates for the variation in spectral sensitivity of the sensor rather than ask the RAW converter to provide a software solution. Likewise I was thinking that an 80A filter would similarly compensate for a tungsten lit environment... but I have to test it.
Ref: http://www.libraw.org/articles/white-balance-in-digital-cameras.html
Filter cost considerations aside... and for maybe only incremental results ;-)
Casey
I was guessing that it would make sense to use a color filter that compensates for the variation in spectral sensitivity of the sensor rather than ask the RAW converter to provide a software solution. Likewise I was thinking that an 80A filter would similarly compensate for a tungsten lit environment... but I have to test it.
Ref: http://www.libraw.org/articles/white-balance-in-digital-cameras.html
Filter cost considerations aside... and for maybe only incremental results ;-)
Casey
bensyverson
Well-known
You shouldn't worry about the native WB. If you're not careful, you will cause more problems than you solve.
The eye is most sensitive to luminance information, and luma is about 70% green. The eye is least sensitive to blue/yellow differences. Blue is the channel which gets manipulated the most by WB settings; in order to get to 3200°K, the blue channel needs to be boosted. But since the eye ignores that extra noise, it's rarely an issue.
If you try to compensate using a blue filter, you end up drastically cutting down the signal for the red and green channels as well. You can make up those 2-3 stops by using a longer shutter speed, but that's rarely an option indoors, unless you're shooting architecture. If you increase ISO to compensate for the filter, you're adding noise to all three channels instead of just one, so it's illogical.
Just forget you ever heard about native white balance, and become friendly with the Color noise reduction slider in Lightroom.
The eye is most sensitive to luminance information, and luma is about 70% green. The eye is least sensitive to blue/yellow differences. Blue is the channel which gets manipulated the most by WB settings; in order to get to 3200°K, the blue channel needs to be boosted. But since the eye ignores that extra noise, it's rarely an issue.
If you try to compensate using a blue filter, you end up drastically cutting down the signal for the red and green channels as well. You can make up those 2-3 stops by using a longer shutter speed, but that's rarely an option indoors, unless you're shooting architecture. If you increase ISO to compensate for the filter, you're adding noise to all three channels instead of just one, so it's illogical.
Just forget you ever heard about native white balance, and become friendly with the Color noise reduction slider in Lightroom.
Terry Christian
Established
Color filters were most often used with slide film, since slide film develops not into a negative, but a finished positive that is viewed directly by projection and not made into a print. So, getting the white balance correct in-camera was necessary. But with digital, putting a filter on your lens is possibly going to overcompensate for the lighting source and throw off your white balance in another direction, arguably one that your post-processing software might not be able to correct - like bensyverson was saying.
Also, keep in mind that every time you put a filter on your lens, you are also arguably degrading the image quality. So an image that you correct optically with a filter might prove to be inferior to an image that you correct with your RAW converter or Lightroom, since you could introduce problems like aberrations, fringing, etc., that you didn't have before.
Also, keep in mind that every time you put a filter on your lens, you are also arguably degrading the image quality. So an image that you correct optically with a filter might prove to be inferior to an image that you correct with your RAW converter or Lightroom, since you could introduce problems like aberrations, fringing, etc., that you didn't have before.
ColSebastianMoran
( IRL Richard Karash )
I think the most direct answer to your question is:
- For camera jpg, set to Auto-White Balance or set the WB for the lighting.
- For shooting RAW, it doesn't matter
- If you are going to adjust the WB in post-processing, then using RAW gives you the most "room" for adjustment and therefore the higher quality result.
The other question raised in the thread is "should I use a color filter so the light reaching the sensor is better balanced for the sensor sensitivity?" Maybe, but I would rather leave sensor tuning to the real engineers.
shimokita
白黒
@bensyverson: yes, I agree about the shutter speeds and ISO settings
@Terry: I remember sitting a watching hours of family slides... in-camera color correction was the way to go with slide film
@ColSebastianMoran: hahaha... those darn sensor engineers
Bottom line... I will do some testing, but not urgently and only after picking up some used filters ;-).
Thanks for the comments.
Casey
@Terry: I remember sitting a watching hours of family slides... in-camera color correction was the way to go with slide film
@ColSebastianMoran: hahaha... those darn sensor engineers
Bottom line... I will do some testing, but not urgently and only after picking up some used filters ;-).
Thanks for the comments.
Casey
Ranchu
Veteran
A cyan filter cleans up the color quite a bit under incandescents. It it worth the trouble? Yes, but I use film now, which looks normal.