DPReview.com is closing April 10th - Find out more

Pink sky - why?

Started Mar 23, 2022 | Discussions
Dunlin Senior Member • Posts: 2,611
Pink sky - why?

Could someone please explain this to me:

If I take a RAW file from my canon EOS 600d, and very under-expose it in software (LRc) the sky has a tendency to turn pink/purple - is this normal?

Below is an example. The first shot is the correct exposure (save for the sky), and the second is very underexposed. In the second photo (on my monitor at least) the sky turns partly purple after adjustment in LR.

First photo - correct exposure.

Second photo - under-exposed with purple sky.

Could it be because I need to calibrate my monitor?

Thanks.

-- hide signature --

Jethro B.

 Dunlin's gear list:Dunlin's gear list
Canon PowerShot SX410 IS Olympus OM-D E-M10 II Canon EF 35-80mm f/4.0-5.6 III Olympus M.Zuiko Digital ED 40-150mm F4-5.6 R Gimp +6 more
Canon EOS 600D (EOS Rebel T3i / EOS Kiss X5)
If you believe there are incorrect tags, please send us this post using our feedback form.
Tom Axford Forum Pro • Posts: 10,076
Re: Pink sky - why?
1

I suspect it may be a characteristic of your camera rather than your monitor.  The other possibility is that the sky was actually pink (or your white balance was incorrect), but that was not obvious when over exposed.

I have never noticed the sky turn pink with shots from my recent cameras (Olympus and Panasonic), but some older models sometimes showed colour casts in such circumstances.

Here is an example in which I moved the Exposure slider in LR to -3.0.  The sky is still the usual colour, although very dark.

sybersitizen Forum Pro • Posts: 24,306
Re: Pink sky - why?
10

Jethro B-UK wrote:

If I take a RAW file from my canon EOS 600d, and very under-expose it in software (LRc) the sky has a tendency to turn pink/purple - is this normal?

Most of the pixel values in your original sky are not pure white. Some areas have more red, some have more blue. Adjusting the 'exposure' (really just shifting pixel values) emphasizes those originally minor differences, making them much more prominent.

Could it be because I need to calibrate my monitor?

No, that's not the reason.

OrigamiCactus Contributing Member • Posts: 667
Re: Pink sky - why?
4

Jethro B-UK wrote:

Could someone please explain this to me:

If I take a RAW file from my canon EOS 600d, and very under-expose it in software (LRc) the sky has a tendency to turn pink/purple - is this normal?

Below is an example. The first shot is the correct exposure (save for the sky), and the second is very underexposed. In the second photo (on my monitor at least) the sky turns partly purple after adjustment in LR.

First photo - correct exposure.

Could it be because I need to calibrate my monitor?

Thanks.

Absolutely nothing to do with the monitor, the sky is already pink on your "correct" photo.

And because most probably the color channels are saturated, when you bring the exposure down there is no info left there, so it just darkens the already pink sky, making it more visibly pink. You can see that the sky looks low-res and blocky, because there just wasn't any information left after the overexposure.

Also generally digital cameras do better shadow recovery than highlight recovery, so ETTR would be the preferred method to boost dynamic range of the image, but you have done the opposite of that. Expose the sky correctly, always shoot at base iso!!!, and then bring the rest of the image up.

And a further tip on your "correct exposure", the exposure is pretty wrong imo.

Shoot F8, after that diffraction starts to soften your image, on your APS-C camera.

Whenever possible please shoot base iso, especially on an older camera like that, if you want to do any post processing.

I don't see any part in your image that would have benefited from the 1/1000 shutter, 1/500 would have sufficed.

But what can you do to save the current photo? You can try the WB adjustment, and play around with "tint", the green-pink slider.

 OrigamiCactus's gear list:OrigamiCactus's gear list
Fujifilm X-Pro3 Fujifilm X-T4 Fujifilm XF 18-55mm F2.8-4 R LM OIS Fujifilm XF 23mm F2 R WR Fujifilm 50mm F2 R WR +1 more
OP Dunlin Senior Member • Posts: 2,611
Re: Pink sky - why?

OrigamiCactus wrote:

Jethro B-UK wrote:

Could someone please explain this to me:

If I take a RAW file from my canon EOS 600d, and very under-expose it in software (LRc) the sky has a tendency to turn pink/purple - is this normal?

Below is an example. The first shot is the correct exposure (save for the sky), and the second is very underexposed. In the second photo (on my monitor at least) the sky turns partly purple after adjustment in LR.

First photo - correct exposure.

Could it be because I need to calibrate my monitor?

Thanks.

Absolutely nothing to do with the monitor, the sky is already pink on your "correct" photo.

OK.

And because most probably the color channels are saturated, when you bring the exposure down there is no info left there, so it just darkens the already pink sky, making it more visibly pink. You can see that the sky looks low-res and blocky, because there just wasn't any information left after the overexposure.

Also generally digital cameras do better shadow recovery than highlight recovery, so ETTR would be the preferred method to boost dynamic range of the image, but you have done the opposite of that. Expose the sky correctly, always shoot at base iso!!!, and then bring the rest of the image up.

Thanks for the tip, will do.

And a further tip on your "correct exposure", the exposure is pretty wrong imo.

Shoot F8, after that diffraction starts to soften your image, on your APS-C camera.

I usually shoot at f/8 anyway.

Whenever possible please shoot base iso, especially on an older camera like that, if you want to do any post processing.

Of course.

I don't see any part in your image that would have benefited from the 1/1000 shutter, 1/500 would have sufficed.

I used a fast shutter speed to avoid blur with my 100-300mm telephoto.

But what can you do to save the current photo? You can try the WB adjustment, and play around with "tint", the green-pink slider.

Thanks for the tip.

-- hide signature --

Jethro B.

 Dunlin's gear list:Dunlin's gear list
Canon PowerShot SX410 IS Olympus OM-D E-M10 II Canon EF 35-80mm f/4.0-5.6 III Olympus M.Zuiko Digital ED 40-150mm F4-5.6 R Gimp +6 more
OP Dunlin Senior Member • Posts: 2,611
Re: Pink sky - why?

Tom Axford wrote:

I suspect it may be a characteristic of your camera rather than your monitor.

Oh dear, I hope not.

The other possibility is that the sky was actually pink (or your white balance was incorrect), but that was not obvious when over exposed.

I have never noticed the sky turn pink with shots from my recent cameras (Olympus and Panasonic), but some older models sometimes showed colour casts in such circumstances.

What brand (s) was the older camera?

Here is an example in which I moved the Exposure slider in LR to -3.0. The sky is still the usual colour, although very dark.

-- hide signature --

Jethro B.

 Dunlin's gear list:Dunlin's gear list
Canon PowerShot SX410 IS Olympus OM-D E-M10 II Canon EF 35-80mm f/4.0-5.6 III Olympus M.Zuiko Digital ED 40-150mm F4-5.6 R Gimp +6 more
OP Dunlin Senior Member • Posts: 2,611
Re: Pink sky - why?

sybersitizen wrote:

Jethro B-UK wrote:

If I take a RAW file from my canon EOS 600d, and very under-expose it in software (LRc) the sky has a tendency to turn pink/purple - is this normal?

Most of the pixel values in your original sky are not pure white. Some areas have more red, some have more blue.

Why? Is there a way to fix/avoid this?

Adjusting the 'exposure' (really just shifting pixel values) emphasizes those originally minor differences, making them much more prominent.

Right.

Could it be because I need to calibrate my monitor?

No, that's not the reason.

Good to know.

-- hide signature --

Jethro B.

 Dunlin's gear list:Dunlin's gear list
Canon PowerShot SX410 IS Olympus OM-D E-M10 II Canon EF 35-80mm f/4.0-5.6 III Olympus M.Zuiko Digital ED 40-150mm F4-5.6 R Gimp +6 more
OrigamiCactus Contributing Member • Posts: 667
Re: Pink sky - why?
1

Jethro B-UK wrote:

OrigamiCactus wrote:

Jethro B-UK wrote:

Could someone please explain this to me:

If I take a RAW file from my canon EOS 600d, and very under-expose it in software (LRc) the sky has a tendency to turn pink/purple - is this normal?

Below is an example. The first shot is the correct exposure (save for the sky), and the second is very underexposed. In the second photo (on my monitor at least) the sky turns partly purple after adjustment in LR.

First photo - correct exposure.

Could it be because I need to calibrate my monitor?

Thanks.

Absolutely nothing to do with the monitor, the sky is already pink on your "correct" photo.

OK.

And because most probably the color channels are saturated, when you bring the exposure down there is no info left there, so it just darkens the already pink sky, making it more visibly pink. You can see that the sky looks low-res and blocky, because there just wasn't any information left after the overexposure.

Also generally digital cameras do better shadow recovery than highlight recovery, so ETTR would be the preferred method to boost dynamic range of the image, but you have done the opposite of that. Expose the sky correctly, always shoot at base iso!!!, and then bring the rest of the image up.

Thanks for the tip, will do.

And a further tip on your "correct exposure", the exposure is pretty wrong imo.

Shoot F8, after that diffraction starts to soften your image, on your APS-C camera.

I usually shoot at f/8 anyway.

Whenever possible please shoot base iso, especially on an older camera like that, if you want to do any post processing.

Of course.

I don't see any part in your image that would have benefited from the 1/1000 shutter, 1/500 would have sufficed.

I used a fast shutter speed to avoid blur with my 100-300mm telephoto.

But what can you do to save the current photo? You can try the WB adjustment, and play around with "tint", the green-pink slider.

Thanks for the tip.

Ah okay, I take it that your 100-300mm is not stabilized?

In this case yeah, sometimes there is nothing you can do and 1/1000 is needed.

It is just that especially on older CMOS sensors, the dynamic range diminishes rapidly when deviating from base iso, especially on an old Canon made sensor.

But with ETTR at base iso you can get faster shutter speeds, because technically you are "underexposing", but you will bring the exposure back up in post for the underexposed parts, leaving the highlights as they are.

Even with a really old sensor, you should be able to get an image with decent dynamic range with this method, and if all else fails, you can always take multiple exposures and combine them into HDR.

 OrigamiCactus's gear list:OrigamiCactus's gear list
Fujifilm X-Pro3 Fujifilm X-T4 Fujifilm XF 18-55mm F2.8-4 R LM OIS Fujifilm XF 23mm F2 R WR Fujifilm 50mm F2 R WR +1 more
tony field Forum Pro • Posts: 13,389
Re: Pink sky - why?
1

Jethro B-UK wrote:

Could someone please explain this to me:

If I take a RAW file from my canon EOS 600d, and very under-expose it in software (LRc) the sky has a tendency to turn pink/purple - is this normal?

Below is an example. The first shot is the correct exposure (save for the sky), and the second is very underexposed. In the second photo (on my monitor at least) the sky turns partly purple after adjustment in LR.

First photo - correct exposure.

Second photo - under-exposed with purple sky.

Could it be because I need to calibrate my monitor?

Thanks.

The problem IMHO is that the picture was taken in hazy conditions which can  cause colour variations depending upon what is causing the haze.

-- hide signature --

Charles Darwin: "ignorance more frequently begets confidence than does knowledge."
tony

OP Dunlin Senior Member • Posts: 2,611
Re: Pink sky - why?

Please could somebody see what they could do with the RAW file?

Here is the DxO linear DNG I used (dropbox):

https://www.dropbox.com/s/lx6o933dbakx2to/20220314-IMG_6303.dng?dl=0

(I've still got the original .CR2 file if needed.)

-- hide signature --

Jethro B.

 Dunlin's gear list:Dunlin's gear list
Canon PowerShot SX410 IS Olympus OM-D E-M10 II Canon EF 35-80mm f/4.0-5.6 III Olympus M.Zuiko Digital ED 40-150mm F4-5.6 R Gimp +6 more
sybersitizen Forum Pro • Posts: 24,306
Re: Pink sky - why?

Jethro B-UK wrote:

sybersitizen wrote:

Jethro B-UK wrote:

If I take a RAW file from my canon EOS 600d, and very under-expose it in software (LRc) the sky has a tendency to turn pink/purple - is this normal?

Most of the pixel values in your original sky are not pure white. Some areas have more red, some have more blue.

Why?

Maybe your software is doing something odd.

Is there a way to fix/avoid this?

Depends on what the cause was.

Post the RAW file somewhere so people here can download it and process it with a variety of software.

Adjusting the 'exposure' (really just shifting pixel values) emphasizes those originally minor differences, making them much more prominent.

Right.

OP Dunlin Senior Member • Posts: 2,611
Re: Pink sky - why?

OrigamiCactus wrote:

Jethro B-UK wrote:

OrigamiCactus wrote:

Jethro B-UK wrote:

Could someone please explain this to me:

If I take a RAW file from my canon EOS 600d, and very under-expose it in software (LRc) the sky has a tendency to turn pink/purple - is this normal?

Below is an example. The first shot is the correct exposure (save for the sky), and the second is very underexposed. In the second photo (on my monitor at least) the sky turns partly purple after adjustment in LR.

First photo - correct exposure.

Could it be because I need to calibrate my monitor?

Thanks.

Absolutely nothing to do with the monitor, the sky is already pink on your "correct" photo.

OK.

And because most probably the color channels are saturated, when you bring the exposure down there is no info left there, so it just darkens the already pink sky, making it more visibly pink. You can see that the sky looks low-res and blocky, because there just wasn't any information left after the overexposure.

Also generally digital cameras do better shadow recovery than highlight recovery, so ETTR would be the preferred method to boost dynamic range of the image, but you have done the opposite of that. Expose the sky correctly, always shoot at base iso!!!, and then bring the rest of the image up.

Thanks for the tip, will do.

And a further tip on your "correct exposure", the exposure is pretty wrong imo.

Shoot F8, after that diffraction starts to soften your image, on your APS-C camera.

I usually shoot at f/8 anyway.

Whenever possible please shoot base iso, especially on an older camera like that, if you want to do any post processing.

Of course.

I don't see any part in your image that would have benefited from the 1/1000 shutter, 1/500 would have sufficed.

I used a fast shutter speed to avoid blur with my 100-300mm telephoto.

But what can you do to save the current photo? You can try the WB adjustment, and play around with "tint", the green-pink slider.

Thanks for the tip.

Ah okay, I take it that your 100-300mm is not stabilized?

Correct.

In this case yeah, sometimes there is nothing you can do and 1/1000 is needed.

It is just that especially on older CMOS sensors, the dynamic range diminishes rapidly when deviating from base iso, especially on an old Canon made sensor.

But with ETTR at base iso you can get faster shutter speeds, because technically you are "underexposing", but you will bring the exposure back up in post for the underexposed parts, leaving the highlights as they are.

Even with a really old sensor, you should be able to get an image with decent dynamic range with this method, and if all else fails, you can always take multiple exposures and combine them into HDR.

-- hide signature --

Jethro B.

 Dunlin's gear list:Dunlin's gear list
Canon PowerShot SX410 IS Olympus OM-D E-M10 II Canon EF 35-80mm f/4.0-5.6 III Olympus M.Zuiko Digital ED 40-150mm F4-5.6 R Gimp +6 more
OP Dunlin Senior Member • Posts: 2,611
Re: Pink sky - why?

tony field wrote:

Jethro B-UK wrote:

Could someone please explain this to me:

If I take a RAW file from my canon EOS 600d, and very under-expose it in software (LRc) the sky has a tendency to turn pink/purple - is this normal?

Below is an example. The first shot is the correct exposure (save for the sky), and the second is very underexposed. In the second photo (on my monitor at least) the sky turns partly purple after adjustment in LR.

First photo - correct exposure.

Second photo - under-exposed with purple sky.

Could it be because I need to calibrate my monitor?

Thanks.

The problem IMHO is that the picture was taken in hazy conditions which can cause colour variations depending upon what is causing the haze.

This isn't the first time I've noticed it.

-- hide signature --

Jethro B.

 Dunlin's gear list:Dunlin's gear list
Canon PowerShot SX410 IS Olympus OM-D E-M10 II Canon EF 35-80mm f/4.0-5.6 III Olympus M.Zuiko Digital ED 40-150mm F4-5.6 R Gimp +6 more
bclaff Forum Pro • Posts: 13,929
Re: Pink sky - why?
1

Jethro B-UK wrote:

Could someone please explain this to me:

If I take a RAW file from my canon EOS 600d, and very under-expose it in software (LRc) the sky has a tendency to turn pink/purple - is this normal?

...

It is normal to get a hue shift.

It can be fixed using curves or levels.

As for an explanation ...

Color comes from the ratio of red to green and blue to green values.

Two things happen when you "reduce" the exposure.

First, all the values get smaller but since they are integers the ratios start getting off.
For example 42 / 14 = 3 but if you divide by 5 you would get (42/5)=8 1(4/5)=3 and 8/2 is 2 2/3 not 3

Second, and usually more serious, is that the values don't actually start at the same "zero". This is called BlackLevel. Let's assume BlackLevel is supposed to be 64.
In the above example we have (106-64) / (78-64) = 42/14 = 3
But what is BlackLevel is reported as 64 in the Exif but is actually 62 for one channel and 63 for the other.
Now we have (106-63)/(78-62)=43/16=2.69
And dividing by 8 (43/8)=5 16/8 = 2 so 5/2 = 2.5
Even if you can't follow the math just remember the smaller the values the more inaccurate the ratios and more sensitive they are to small errors up/down.

-- hide signature --

Bill ( Your trusted source for independent sensor data at PhotonsToPhotos )

knickerhawk Veteran Member • Posts: 7,615
Re: Pink sky - why?
2

Jethro B-UK wrote:

Please could somebody see what they could do with the RAW file?

Here is the DxO linear DNG I used (dropbox):

https://www.dropbox.com/s/lx6o933dbakx2to/20220314-IMG_6303.dng?dl=0

(I've still got the original .CR2 file if needed.)

Please post the original CR2 to dropbox. It's hard to eliminate variables possibly added by DXO's conversion to a DNG.  The DNG read in Rawdigger looks like the green channel was blown in the sky, which wouldn't be surprising. The original CR2 will be more definitive. Thanks.

bclaff Forum Pro • Posts: 13,929
Re: Pink sky - why?

knickerhawk wrote:

Jethro B-UK wrote:

Please could somebody see what they could do with the RAW file?

Here is the DxO linear DNG I used (dropbox):

https://www.dropbox.com/s/lx6o933dbakx2to/20220314-IMG_6303.dng?dl=0

(I've still got the original .CR2 file if needed.)

Please post the original CR2 to dropbox. It's hard to eliminate variables possibly added by DXO's conversion to a DNG. The DNG read in Rawdigger looks like the green channel was blown in the sky, which wouldn't be surprising. The original CR2 will be more definitive. Thanks.

I see no clipping but the Exif in that file shows no BlackLevel  (which will be treated as zero) but BlackLevel for the 600D is 2048 so that's an issue.

-- hide signature --

Bill ( Your trusted source for independent sensor data at PhotonsToPhotos )

knickerhawk Veteran Member • Posts: 7,615
Re: Pink sky - why?

bclaff wrote:

knickerhawk wrote:

Jethro B-UK wrote:

Please could somebody see what they could do with the RAW file?

Here is the DxO linear DNG I used (dropbox):

https://www.dropbox.com/s/lx6o933dbakx2to/20220314-IMG_6303.dng?dl=0

(I've still got the original .CR2 file if needed.)

Please post the original CR2 to dropbox. It's hard to eliminate variables possibly added by DXO's conversion to a DNG. The DNG read in Rawdigger looks like the green channel was blown in the sky, which wouldn't be surprising. The original CR2 will be more definitive. Thanks.

I see no clipping

Rawdigger doesn't report any clipping, but as I said it "looks like green channel was blown in the sky". For instance, here's the raw histogram for a selection of the sky:

Something's not right, which is why I asked for the CR2 instead of relying on the linear DNG that DXO produced.

but the Exif in that file shows no BlackLevel (which will be treated as zero) but BlackLevel for the 600D is 2048 so that's an issue.

The missing blacklevel may be symptomatic of DXO mucking up the DNG. Regardless, we're not worried about problems at the shadow end and color shifts in the darks. The problem in this file is in the highlights.

OP Dunlin Senior Member • Posts: 2,611
Re: Pink sky - why?

Here is the requested .CR2 file:

https://www.dropbox.com/s/2otyueo0t56z0qh/IMG_6303.CR2?dl=0

-- hide signature --

Jethro B.

 Dunlin's gear list:Dunlin's gear list
Canon PowerShot SX410 IS Olympus OM-D E-M10 II Canon EF 35-80mm f/4.0-5.6 III Olympus M.Zuiko Digital ED 40-150mm F4-5.6 R Gimp +6 more
bclaff Forum Pro • Posts: 13,929
Re: Pink sky - why?
1

Jethro B-UK wrote:

Here is the requested .CR2 file:

https://www.dropbox.com/s/2otyueo0t56z0qh/IMG_6303.CR2?dl=0

Looks fine to me (-4 stops in Photoshop)

Something strange in your workflow.
I think that zero BlackLevel is the primary issue.

-- hide signature --

Bill ( Your trusted source for independent sensor data at PhotonsToPhotos )

knickerhawk Veteran Member • Posts: 7,615
DXO appears to be the culprit
6

Jethro B-UK wrote:

Here is the requested .CR2 file:

https://www.dropbox.com/s/2otyueo0t56z0qh/IMG_6303.CR2?dl=0

Thanks for posting the original raw. This confirms that the green channel is indeed blown in the sky. Here's the raw histogram from the sky portion of the scene:

As you can see, the red and blue channels are intact but the green is blown. This isn't unusual for overcast skies. However, clipping only the green channel in the sky isn't normally a problem for LR/ACR. The Adobe engine is usually quite able to "recover" the blown highlights in the green channel in the sky. It does this by applying the intact red and blue channels to the missing green channel highlights. Color shifts are also not normally a problem here. I've taken many, many overexposed sky shots when I'm aggressively ETTR'ing with my cameras (mostly Olympus) and color shlfits aren't an issue.

Where things seem to have gone off the rails is with the DXO conversion to a linear DNG. As BIll noted, the blacklevel is wrong. I'm also seeing a big change in the default WB and also a weird truncation of the data is shown in Rawdigger (which is what led me to ask for the original CR2). When the DNG is processed in ACR, the color shift is obviously a big problem. When, however, the original CR2 is processed in ACR, there isn't any color shift in the sky. It looks like DXO is the real culprit here. I have DXO but normally only use it for lowlight scenes that require DeepPrime. Sorry, I can't be of too much help with why it's mishandling this particular image.

EDIT: Just checked in DXO. If you export with only the noise and optical corrections, the file is fine. If you export with all corrections, you're baking in the clipped green channel in a way that ACR can't overcome.

Keyboard shortcuts:
FForum MMy threads