Snow never neutral?

Viking

Its winter where I love. Very, very Cold! We have snow everywhere. And everytime I take a photo on a snowy landscape, and open the RAW file in ACR its always-always-always have a blue color cast. I use the neutral pipet on the snow and the picture gets Neutral. And it looks terrible - The snow is white, but the other whites spaces and other color become a little bit of.

So, my question. Is snow pure white or does the blue sky make the snow reflect some blue and then create a blue cast. Picture of snow looks more realistic if they have little blue color cast. And no, Im not color blind! And the snow I have photographed have not been urinated by dogs or cats or something like that. ;-)

Best Regards
// Johan

rutt

I thought about this a lot last winter. See: http://dgrin.com/showthread.php?t=6366

In short, I think snow is rarely neutral and it really depends on whether it's in sun or shadow and what color the sky is. If you do neutralize a sunlit patch of snow in ACR, it will actually look yellow (even though it's really neutral). What I do these days, is neutralize the lightest patch of snow (if there isn't a better neutral point) and then push the B curve a touch toward blue so that the sunlit snow measures roughly -1 and the shadowed snow is even bluer.

pathfinder

If you shoot in RAW, why not choose sunlight or shade as the White Balance in ACR? That should set the white balance on the spectrum the image was shot in.

I agree with rutt - snow is a reflector and reflects the ambient lighting - white, blue, or even orange in the late afternoon sun. But at high noon on a sunny day,it should be near white.

Viking

pathfinder wrote:

If you shoot in RAW, why not choose sunlight or shade as the White Balance in ACR? That should set the white balance on the spectrum the image was shot in.

Becous I like custom white balance. I dont always get the result I want with the sunlight, tungsten and the rest. But mostly.

Thanks for all replys.

rutt

Snow is really ever so slightly blue, even in yellow sunlight. You will lose this if you overexpose it in an effort to make it come out white instead of gray. I like to shoot snow scenes with no exposure compensation. It comes out gray, but I fix with with a the L curve as shown in the thread I cited above.



This is high noon, sunny day. The sunlit snow is very slightly blue. The shadowed snow is bluer. I didn't do anything special to make it bluer; I slightly steepened the blue endpoint of the B curve to get the sunlit snow to be just slithly blue and then the shadow snow was bluer all by itself. Looks very natural to me. And see how I've both made the snow look the right color and not blown it out, as I would have it I'd used exposure compensation of some sort.

dandill

rutt wrote:

Snow is really ever so slightly blue, even in yellow sunlight. ...

The reason is, the vibrations of the atoms in the ice crystals of the snow absorb mostly infrared light but also a small amount of deep red visible light, making the visible light that is reflected slightly blue.

Bob Bell

dandill wrote:

The reason is, the vibrations of the atoms in the ice crystals of the snow absorb mostly infrared light but also a small amount of deep red visible light, making the visible light that is reflected slightly blue.

To prove dandil's statement take a look at icebergs which are just heaps of compressed snow. The inner parts are blue, maybe slightly green. Snow and sand are very hard to capture because of the dynamic range needed for the subtle color changes. I have seen a few amazing shots taken from 1Ds that are right on.

I think if you were shooting from the ground and had a different background that would of helped. The better beamer turns your flash into an oval shaped flash that is larger than that bird so I am guessing the flash equally lights the snow and the bird making it very 2D. stand in your house and flash a wall to see the shape it makes.

pathfinder

: John, I just returned from Steamboat last week and have a few shots in the sunlit and shaded snow. I processed these straight in RAW as shot - the sunlit scene received +1.33 EC and the shaded scene received +0.33EC.All shot with a 5D and a 24-105 L plus a B&W polarizer. ~12,100 feet:):

Here is the sunlit scene - my son descending - the sunlit snow reads pretty white but is not blown - three different points in the snow 244,239,235 222,217,213 234,227,223 and right on the snow on his boot reads 252,251,247
[imgl]http://pathfinder.smugmug.com/photos/52030665-L.jpg[/imgl]

Here he is again in the shade - The sunlit snow read 241,239,238 or thereabouts, the shaded snow reads 44,76,117 in the depth of the rut
[imgr]http://pathfinder.smugmug.com/photos/52031045-L.jpg[/imgr]

Is this snow too red, white, or what?? Would you prefer more blue?

I love the Digital Color Meter Program on the MAC!!:):

rutt

pathfinder wrote:

Here he is again in the shade - The sunlit snow read 241,239,238 or thereabouts, the shaded snow reads 44,76,117 in the depth of the rut

Is this snow too red, white, or what?? Would you prefer more blue?

Close enough. I think the important thing is that it's more blue in the shadows than in the sun. I like it a little blue even in the sunlight, but I've been known to have mood swings on this particular topic. But it has to shade to blue or it really looks wrong.

If you use no EC when you shoot, the sunlit snow might come out a little more blue. Then if you use the L curve to lighten it up in LAB, the impossible color effect probably will keep it ever so slightly blue. Does that make sense?

pathfinder

rutt wrote:

Close enough. I think the important thing is that it's more blue in the shadows than in the sun. I like it a little blue even in the sunlight, but I've been known to have mood swings on this particular topic. But it has to shade to blue or it really looks wrong.

If you use no EC when you shoot, the sunlit snow might come out a little more blue. Then if you use the L curve to lighten it up in LAB, the impossible color effect probably will keep it ever so slightly blue. Does that make sense?

I understand that by underexposing snow, you do not blow highlights, and that you can reset the tonality in RAW processing or in LAB post ARC.

But, if you underexpose an image, by definition aren't you increasing the noise level in the image - even if just a bit? Underexposed images always contain more color noise than properly exposed images - don't they??

Impossible colors?? Underexposed snow is grey, not white, and not really blue - now if it is shaded then yes it is blue, and expanding the L probably does "average" the impossible colors in to create the blue you are desiring. At 12,000 of course, the light is "more blue" ( actinic maybe ) than at lower levels, but when I was skiing in the sunlight - I just thought of it as white.

Believe it or not, John, but this entire conversation was going through my mind as I shot these images, as I knew this topic would come up. Hence, the polarizer, and the + EC as I described.

I liked the shot in the shade with the sun in the tree top - Do you think the colors are believeable on your monitor? Or would you prefer more blue?

colourbox

rutt wrote:

I think the important thing is that it's more blue in the shadows than in the sun.

That is exactly what we were taught in painting class back in school, with regards to color behavior. A sunlit neutral object should be on the warm side, and the shadow cast by a light source is always complementary to the color of the light source, in this case cool, or blue. That means snow wouldn't really be white in either light or shade.

pathfinder

I had to stop and think about my post processing steps - These went through the LAB portrait routine you presented...

Import from RAW - find thresholds - dup - apply blue or green channel as a luminosity bleand - merge - dup - overlay mode with the curves you supplied and the blend to taste and then finally sharpen twice. I like that workflow for a lot of images.

rutt

pathfinder wrote:

I liked the shot in the shade with the sun in the tree top - Do you think the colors are believeable on your monitor? Or would you prefer more blue?

I knew you'd pin me down. I like that shot and the colors look belivable. If it were mine, I'd nudge the sunlit snow just a shade toward blue, but it's yours and I like it fine the way it is. Next year I might change my mind.

As I said, the only important thing is to get the shaded snow a bit blue. And don't overexpose it to get it white. Expose your subjects properly and get the snow right in post.

pathfinder

colourbox wrote:

That is exactly what we were taught in painting class back in school, with regards to color behavior. A sunlit neutral object should be on the warm side, and the shadow cast by a light source is always complementary to the color of the light source, in this case cool, or blue. That means snow wouldn't really be white in either light or shade.

Thanks Colourbox, I told John that snow could be orange or red in the late afternoon sun ( I know I have seen that!!) In the images posted...

Just slightly warm in the the sun - + in the RED channel - slightly blue in the shade - Just like my points I gave the data for.

pathfinder

rutt wrote:

I knew you'd pin me down. I like that shot and the colors look belivable. If it were mine, I'd nudge the sunlit snow just a shade toward blue, but it's yours and I like it fine the way it is. Next year I might change my mind.

As I said, the only important thing is to get the shaded snow a bit blue. And don't overexpose it to get it white. Expose your subjects properly and get the snow right in post.

The sunlit snow in the first scene, MIGHT tolerate just a few points more blue - It seems ever so slightly too grey to my eye here, but then I would be blowing the snow in his boots unless I masked them out.

rutt

pathfinder wrote:

I like that workflow for a lot of images.

Me, too. With variations (MFM instead of the overlay curves), it's what I used for the ballet shots.

pathfinder

I used the same workflow to create this image of a plowed field and a rather grey stand of timber in the river bottom.

The first image is the straight from RAW jpg - the second is via a luminosity blend and MFM blending. The second matched my mind's eye much better. Shot with a 5D and a 24-105 L

[imgl]http://pathfinder.smugmug.com/photos/52513889-L.jpg[/imgl]


[imgr]http://pathfinder.smugmug.com/photos/52513635-L.jpg[/imgr]

rutt

The field looks great. How did you make the luminosity blend layer? Did you use the red channel for the sky? If not try it and you might get better clouds.

pathfinder

Actually I used the red channel for the sky and masked it out for the timber, and then used a blue or green ( I forget )luminosity mask for the timber and foreground and masked out the sky. Best of both worlds that way. I liked how it brought out the reds in the tops of the timber.

There are just so many variations you can do with this workflow!

rutt

pathfinder wrote:

Actually I used the red channel for the sky and masked it out for the timber, and then used a blue or green ( I forget )luminosity mask for the timber and foreground and masked out the sky. Best of both worlds that way. I liked how it brought out the reds in the tops of the timber.

There are just so many variations you can do with this workflow!

Usually, you can just use Dan's favorite trick: the blend-if sliders and you don't have to make a mask at all. The sky has will be B negative. That way, if you do have to make a mask, it can be very sloppy.

I think this workflow is the real point of the book. Get any cast out in RGB. Make a luminosity blend layer for contrast and detail. Do something to enhance color and tweak the L curve a bit. Sharpen twice. Clean up the blacks in CMYK, perhaps. Sounds easy, doesn't it?

pathfinder

You're right of course that Blend IF sliders would also work -

I still don't think in that mode although I discussed it in my review of chapter 7 or so. Masks I am real comfortable with - but I'll keep that Blend IF trick in mind next time, and give it a whirl again.

rutt

pathfinder wrote:

You're right of course that Blend IF sliders would also work

Not only faster, but generally better when they do work because they use just the property you want, in this case the color of the sky, to identify the area to blend. When they don't work, I find that starting with one of the color LAB channels as in Ch 9 gets there almost as fast and again better than anything I can draw by hand.

pathfinder

I actaully think I made my mask by drawing a linear gradient, rather than with a brush, for the sky. Maybe that is why the clouds weren't more dramatic - hmmmm....

Something I have been playing with - using gradients to creates 'masks' rather than selections. Sometimes it works - soemtimes not, Just another tool:):

rutt

pathfinder wrote:

I actaully think I made my mask by drawing a linear gradient, rather than with a brush, for the sky. Maybe that is why the clouds weren't more dramatic - hmmmm....

Something I have been playing with - using gradients to creates 'masks' rather than selections. Sometimes it works - soemtimes not, Just another tool:):

When there is a color clue, as with blue skys, you just can't beat using one of the color LAB channels as a starting point. That way you are using the very property of the image that you want to enhance to determine how to enhance it. I suggest a careful rereading of Ch 9 and some practice with its techniques. I found this one of the best parts of the LAB book. In fact, my improved ability to correct image elements independently sparked the questions in this thread.