Blue flowers turn purple in a digital camera

[darekk]

when the color blue contains the addition of purple. For example common bugloss (Anchusa officinalis), also common chicory (Cichorium intybus) a little. I try to switch white balance adjustment from Auto: Ambience priority to Daylight and change hue and lightness of purple color. Flowers look more naturally, but not exactly like in a field:


purple HSL 0 0 0


purple HSL -10 0 -5, Canon Digital Photo Professional

See also:
https://www.google.com/search?client...nalis%22&udm=2 (some flowers are blue, others purple)
Purple is not spectral color (https://en.wikipedia.org/wiki/Line_of_purples) and perhaps this is the reason.

[DanK]

This is a complicated question.

First, what camera are you using, and how are you shooting (raw vs. JPEG)? Your second photo is labeled with Digital Photo Professional, which suggests that you are using some Canon camera. Are you shooting raw?

Almost none of the 2^14 color values captured by most Canon cameras are spectral colors, so I don't think that's the issue.

A lot goes into color rendering, including the lighting, the properties of the particular sensor, the electronics and programming in the camera, even the lens, and how the software you use processes the image. Automatic white balance varies in accuracy from camera to camera and from scene to scene, but it is always a guess.

I don't use DPP, but my understanding is that it uses the camera settings for its initial rendering of colors. So the programming in Canon's "picture styles" for your particular camera will influence the initial color balance, even if you are shooting raw. If you shoot raw, you should be able to tweak colors as you like in DPP. I almost never shoot JPEG, but my memory from many years ago when I did shoot JPEG is that color corrections in software don't always work well with JPEG images. (In fact, that was one reason I switched to raw.)

With my Canon cameras, I find that the most I need to do is shoot one image under the same lighting with a neutral card (I use a whiBal) and set WB and tint using that. With my newest camera, an R6 II, I don't often need to do even that. However, to get truly accurate color, one should create a profile for each camera and lens combination using a target like the ColorChecker. I have never needed to do that, but some people on this site have.

[darekk]

I shoot RAW and JPEG (two pictures simultaneously), mostly with Canon EOS 800D.
Someone answered in the iNaturalist forum, that this effect could be caused by infrared reflected by some flowers:
https://photobotanic.com/news/color-...geratum-affect
https://photo.stackexchange.com/ques...al-photography
They advice to change hue of one zone of the spectrum.
Maybe taking pictures during cloudy weather (infrared absorbed by clouds) would work too, if this is true for the sensor, not only the film ? Or even in shadow ?

[Geoff F]

That is a common problem which I have often encountered. Some plants, particularly intense blue colours, slightly change their colour when you bring them from outside daylight into a room. I have done that to try to get a certain reference point for my editing and found they appear different when placed beside my computer indoors.

Dan's suggestion of doing a Custom White Balance for your computer will help. Avoiding direct sunlight can often be an advantage. But I have never found any sure way to success except to struggle and keep going outside to check if I am getting close to my required colour answer.

[DanK]

To be clear, I'm not recommending setting a general custom white balance. What I'm suggesting is using a neutral card to get a neutral WB (blue-yellow) and tint (magenta-green) balance for a particular shoot, under particular lighting conditions.

One of the links Darius posted suggests using a Color Checker or Spyder Checker in the same way I use a neutral card. This is more work, as one has to import a profile rather than just use the WB took on an area of the test image. However, my guess is that this would be the most accurate way to do it.

I've never used DPP, despite having owned 5 (!) Canon DSLR and mirrorless cameras, but both Lightroom and Photoshop have tools for changing the hue of specific ranges of color. That might be enough.

[Manfred M]

I shoot RAW and JPEG (two pictures simultaneously), mostly with Canon EOS 800D.
Someone answered in the iNaturalist forum, that this effect could be caused by infrared reflected by some flowers:
https://photobotanic.com/news/color-...geratum-affect
https://photo.stackexchange.com/ques...al-photography
They advice to change hue of one zone of the spectrum.
Maybe taking pictures during cloudy weather (infrared absorbed by clouds) would work too, if this is true for the sensor, not only the film ? Or even in shadow ?

Be careful here, as the article deals with film, rather than digital photography. Films were sensitive to UV and in some cases IR. We frequently used UV filters on our lenses when shooting film to cut out this component UV component in outdoor photography. IR was less of an issue as films tended to be less sensitive to the near IR range.

Digital cameras all use both IR and UV cutoff filters on the sensor stack, so this is not really an issue any more, with the possible exception of very long exposures (measured in minutes).

This is unlikely the issue in these images.

[darekk]

Maybe infrared cut-off filter would block this ?

[darekk]

Be careful here, as the article deals with film, rather than digital photography.

I noticed that. But digital cameras can be used for infrared photography, so perhaps this depends on model ?

[Manfred M]

I noticed that. But digital cameras can be used for infrared photography, so perhaps this depends on model ?

In general, people that use digital cameras for IR photography have the IR filter removed from the sensor stack. People who want to try IR photography will sometimes place a filter on their camera lens to cut off visible light transmission, but allow IR. This is why I mentioned exposures measured in minutes. No cutoff filter is perfect, but in order to get any decent amount of IR (or UV) hitting the sensor, you need very long exposures when going this route.

We see this issue when using very dense ND filters (10+ stop) that cut of most visible light. Either the filter has a slight colour cast; Lee ND filters are known to have blue colour cast and B+W filters have a warm colour cast or there is a minor leak of UV or IR (or both) and that can effect the image. Again, in my experience, these exposures can be measured in many minutes.

[darekk]

So catching IR for example at 1/1000s is impossible, even such IR on the verge of the visible light ?

[Manfred M]

So catching IR for example at 1/1000s is impossible, even such IR on the verge of the visible light ?

Effectively yes. In fact, depending on the filter design, some of the far red wavelengths could be cut off as well. Unfortunately, the camera manufacturers do not share this type of data with us.

[darekk]

https://www.researchgate.net/publica...ligolectic_bee
Hannah Burger, Stefan Dötterl and Manfred Ayasse, Host-plant finding and recognition by visual and olfactory floral cues in an oligolectic bee, Functional Ecology 2010, 24, 1234–1240
p. 1238, Fig. 6: Mean spectral reflection of Echium vulgare and Anchusa officinalis petals (..)

They reflect red and perhaps infrared (in addition to blue), but the plot is cut at 700 nm.
However images of viper's bugloss (Echium vulgare) look well in my opinion, or at least not so badly like of common bugloss (Anchusa officinalis).
Common bugloss reflects less blue in comparison to red and is also darker. Maybe darker red is less visible for the eye ?
And this blue/purple color is sum of blue and red really, and there is difference between eye and camera sensor perception at least in case of common bugloss.

[tclune]

I don't know if this is the OP's problem, but I find that using an incident light meter can give more accurate exposure with red or blue flowers. Reflective metering often blows out the exposure with flowers. I'll sometimes adjust the EV, even with incident light metering, to avoid blowing a color channel. FWIW

[DanK]

My solution to that is to use reflective metering, but to set the camera to show the histogram separately for all three primary colors. In my experience, it’s most often the red that blows out in photographing flowers, but it obviously depends on the context. I then adjust exposure so that the rightmost histogram doesn’t clip.

This does require chimping once or twice, but it works very well.


Sent from my iPhone using Tapatalk