Digital Darkroom Assignment for the Week: 4/16-4/23

cletus

Color to Black and White

This assignemt is designed to complement fish's current photo assignment.

Use the photo editing software of your choice to convert a color image to black and white. The point of this challenge is to find a method of going from color to black and white that you feel comfortable with and gets you consistent results. If you already have a favorite method for converting to black and white, try to discover a new method!

You are encouraged to post not only your final image, but also your original color image and a description of your color to black and white conversion method.

Advanced Spin: Feel free to add toning or other special effects to your image. The only requirement is that the entire image must be monochromatic. No mixing color and black and white in the final image... We'll save that for another day!

cletus

I haven't had a chance to go out and shoot fish's new assignment, but here are some example conversions. To provide a little help for those that might need it, I've provided a vauge description of the method for each of the conversions.

Here is the original:


#1) Change the mode


#2) Map the gradient


#3) Adjust the hue & saturation twice


#4) Mix the channels

lynnma

cletus wrote:

I haven't had a chance to go out and shoot fish's new assignment, but here are some example conversions. To provide a little help for those that might need it, I've provided a vauge description of the method for each of the conversions.

Here is the original:


#1) Change the mode


#2) Map the gradient


#3) Adjust the hue & saturation twice


#4) Mix the channels

Cool! it may take a day or so to get a good response to this great new challenge as we have quite a lot of work to do... or at least I have...:D looking forward to seeing some great black and whites..

lynnma

Heres my original with a little glamour blur...

lynnma

heres my black and white...

Cranked up the green first.. then adjustment layer-channel mix. played with the levels.. sharpened and dodged the eyes a bit.

cletus

lynnma wrote:

heres my black and white...

Cranked up the green first.. then adjustment layer-channel mix. played with the levels.. sharpened and dodged the eyes a bit.

Very, very nice lynn

rutt

lynnma wrote:

heres my black and white...

Cranked up the green first.. then adjustment layer-channel mix. played with the levels.. sharpened and dodged the eyes a bit.

You lost a lot of detail when you converted to B+W. Did you mean to? Pam's face is flat and blown out in the B+W version, but there is lots of detail in the color.

What do you think about the following version?



This was very quick and dirty. The red channel had very little facial detai. Both the green and blue channel were better, but the green channel looked better for the jacket. I applied the green channel to the red channel before desaturating and then sharpened the result. Probably an even better result is possible. I didn't explore the otehr color spaces or think very hard about it.

I did remember that Dan Margulis' "Professional Photoshop" has a chapter, "Friend and Foe In Black and White" devoted to converting color to B+W. Dan goes into a lot of detail about how to get the best possible results and I didn't have time to read very deeply. But quick and dirty, the problem with a naive conversion is that although the face has color contrast. it doesn't have luminosity contrast. Each L pixel gets a value equal to 70% of its green component, 30% of its red component, and 10% of its blue component. In this case, most of the cheek highlight is very bright in green channel and blown all the way in the red channel. So red isn't adding anything. The green channel of the face is also bright, but it has detail. By substituting it for the red channel before the conversion, we avoid the loss of detail.

As I said, Dan goes into a lot of detail. I think this chapter would be good reading for people who want to make great B+W images from color ones.

rutt

Reading Dan a little more deeply -- The red channel is a foe in this picture because it blows out the detail in the other two channels. Dan's quick and dirty rule of thumb with faces - discard the other two channels and just use the green channel, assuming the color original is has good contrast and color. In effect I did this by duplicating the green into the red. Blue only counds for 10% in the conversion, so leaving it alone didn't really matter.

pathfinder

rutt wrote:

Reading Dan a little more deeply -- The red channel is a foe in this picture because it blows out the detail in the other two channels. Dan's quick and dirty rule of thumb with faces - discard the other two channels and just use the green channel, assuming the color original is has good contrast and color. In effect I did this by duplicating the green into the red. Blue only counds for 10% in the conversion, so leaving it alone didn't really matter.

Isn't much of the noise in the blue channel also? I know when I try to photograph old B&Ws with a digital camera, frequently the blue channel has a lot more noise than the red or green channels.

cletus

pathfinder wrote:

Isn't much of the noise in the blue channel also? I know when I try to photograph old B&Ws with a digital camera, frequently the blue channel has a lot more noise than the red or green channels.

Another reason to stay away from the red channel when dealing with portraits is that it tends to empasize blemishes.

rutt

pathfinder wrote:

Isn't much of the noise in the blue channel also? I know when I try to photograph old B&Ws with a digital camera, frequently the blue channel has a lot more noise than the red or green channels.

This may be true, but it doesn't matter because blue counts for so little in the conversion.

pathfinder

rutt wrote:

You lost a lot of detail when you converted to B+W. Did you mean to? Pam's face is flat and blown out in the B+W version, but there is lots of detail in the color.

What do you think about the following version?



This was very quick and dirty. The red channel had very little facial detai. Both the green and blue channel were better, but the green channel looked better for the jacket. I applied the green channel to the red channel before desaturating and then sharpened the result. Probably an even better result is possible. I didn't explore the otehr color spaces or think very hard about it.

I did remember that Dan Margulis' "Professional Photoshop" has a chapter, "Friend and Foe In Black and White" devoted to converting color to B+W. Dan goes into a lot of detail about how to get the best possible results and I didn't have time to read very deeply. But quick and dirty, the problem with a naive conversion is that although the face has color contrast. it doesn't have luminosity contrast. Each L pixel gets a value equal to 70% of its green component, 30% of its red component, and 10% of its blue component. In this case, most of the cheek highlight is very bright in green channel and blown all the way in the red channel. So red isn't adding anything. The green channel of the face is also bright, but it has detail. By substituting it for the red channel before the conversion, we avoid the loss of detail.

As I said, Dan goes into a lot of detail. I think this chapter would be good reading for people who want to make great B+W images from color ones.

Rutt - I like your version of the face better than Lynn's - sorry Lynn - you definitely have more detail in the face, but your jacket is TOO bright - I prefer the darker tone of the jacket as in Lynn's picture ( Hooray for Lynn!!)

I think the big bright area in the lower right distracts the eye from the center of interest - the face. That is why I would burn in the jacket to lower its tonality. But I kind of like darker borders to keep the eye from wandering.... wandering eyes will always get you in trouble.

dkapp

cletus wrote:

Color to Black and White

This assignemt is designed to complement fish's current photo assignment.

Use the photo editing software of your choice to convert a color image to black and white. The point of this challenge is to find a method of going from color to black and white that you feel comfortable with and gets you consistent results. If you already have a favorite method for converting to black and white, try to discover a new method!

You are encouraged to post not only your final image, but also your original color image and a description of your color to black and white conversion method.

Advanced Spin: Feel free to add toning or other special effects to your image. The only requirement is that the entire image must be monochromatic. No mixing color and black and white in the final image... We'll save that for another day!

I was pretty excited when I saw the topic. I've gotten more into B&W over the last few weeks, and my favorite photos seem to have a little color in them.

Hopefully I will be getting a Nikon D70 tonight & I will be able to get out this weekend and get some photos for the two new challenges.

If you want to see my most recent B&W work w/ a splash of color, it can be found here:
http://dkapp.smugmug.com/gallery/97307

I can't wait to see what people come up with.

Dave

hutchman

Cletus,

This is tough! I am out in a field I know nothing about with this one. I have tried several conversions and have not been happy with them at all. I took a pic of a bald tree against the sky one day thinking there might be picture in it somewhere. After playing with the image, I decided that it pretty much sucked. Then I tried to convert it for this project. It did not come out too bad, but I would like some critique.

The original.........





The transformation...........






I changed the image to grayscale.
Opened an adjustment layer for levels until I achieved the desired look.
Opend another adjustment layer for brightness and contrast (really bumped the contrast).
I then sharpened the original with unsharp mask - 0 threshold, 2 pixel radius, and 70%.

The unsharp mask seemed to really make it jump.

I kink of like it, but I don't really know what it "is supposed" to look like. Enquiring minds want to know.

Hutch

rutt

pathfinder wrote:

Rutt - I like your version of the face better than Lynn's - sorry Lynn - you definitely have more detail in the face, but your jacket is TOO bright - I prefer the darker tone of the jacket as in Lynn's picture ( Hooray for Lynn!!)

I think the big bright area in the lower right distracts the eye from the center of interest - the face. That is why I would burn in the jacket to lower its tonality. But I kind of like darker borders to keep the eye from wandering.... wandering eyes will always get you in trouble.

No need to burn, just use better color theory. This time I understood what was going on a little better:

1. Apply image red channel blend mode darken to green channel
2. Apply image blue channel blend mode darken to green channel
3. Keep only the green channel
4. Sharpen

Deacon

Flower B&W
Playing around,









Deacon

rutt

Color theory in B+W conversion
Here is a very simple color to b+w excercise.

Consider this image:



Here's what happens when PS converts it to grayscale:



Not very nice, eh? What's going on? Although the colors of the stripes are very different, I chose them carefully to have identical luminosity, using photoshop's luminosity formula: L = 60% G + 30% R + 10% B.

So, in this case, we have to do something before we convert to grayscale. Once we convert, the information we need to distinguish the strips is gone and there really is no good way to recover it.

Perhaps we can use one of the channels for the b+w image?

Red:



Green:



Blue:



So no single channel offers us a nice contrast between the colors. Notice that the blue strip has faded to white in the blue channel, but the red and green have not faded to white in their channels. This is because I needed to choose a green and a red with the same luminosity as 255 blue in order to make this example work. But that means (according to the formula) I needed to make the red strip 3 times darker than the blue stripe and the green stripe 2 times darker than the red stripe.

I used the channel mixer to produce a b+w version that distinguishes the colors:



This blend was 126% R, 92% G, 4% B. Other blends can get the colors into different luminosity orders. Here is 92% R, 34% G, 64% B:



This was just a little study to see if I could get my feet on the ground about the basic theory behind color->b+w conversion. Hope someone else found it interesting. I'm working my way through the chapter "Friend and Foe in Black and White" in Professional Photoshop Dan Margulis. Dan has lots more detail and lots more practical information as well. But there's only so much I can absorb at a time.

cmr164

The original shot was pre-race as the runner read through her paperwork.
ISO Speed: 400
Aperture: f8.0
Shutter: 1/500
Focal Length (mm): 400

Note the long focal length. As usual click the image for the larger version.




Now converted in 'xv' but dropping saturation on RGB channels



And then after listening to the chatter here, taking only the G channel and reducing the saturation.

rutt

Why black and white?
I'd like to add another parameter to this assignment. Discuss the pros and cons of converting the image to black and white. What was gained and what was lost? Is the image actually better in black and white than it was in color? If so, why?

In the good old days, black and white was cheaper than color, and in fact sometimes color just wasn't possible at all. But today, even newspapers can print color photos and black and white prints can even be more expensive then color. So the reason for b+w has to be aesthestic. Let's try to understand that as part of the assignment.

cletus

hutchman wrote:

I changed the image to grayscale.
Opened an adjustment layer for levels until I achieved the desired look.
Opend another adjustment layer for brightness and contrast (really bumped the contrast).
I then sharpened the original with unsharp mask - 0 threshold, 2 pixel radius, and 70%.

The unsharp mask seemed to really make it jump.

I kink of like it, but I don't really know what it "is supposed" to look like. Enquiring minds want to know.

Hutch

Hutch, I think it looks pretty good! I tried working with your orignal, but the JPEG artifacts keep making things look funky.

Here is something you might want to try. This is the technique I used in conversion #3 in my earlier example. If you like this technique, be sure to thank zero-zero for bringing it to dgrin!

• Create a Hue/Saturation adjustment layer, but don't make any adjustments right now.
• Change the blending mode of your Hue/Saturation layer to Color

• Add a second Hue/Saturation adjustment layer above the first. Set Saturation to -100

• Now go back to your first Hue/Saturation layer and start playing around with the sliders. The Hue slider will make the biggest impact. The Saturation and Lightness sliders can then be used to fine tune things.

I hope this helps!

pathfinder

rutt wrote:

No need to burn, just use better color theory. This time I understood what was going on a little better:

1. Apply image red channel blend mode darken to green channel
2. Apply image blue channel blend mode darken to green channel
3. Keep only the green channel
4. Sharpen

Like you I would liked to have the full file rather than the 800 pixel size here, but I just did a little dodging and burning - not too sophisticated - and then a duplicate layer with Gaussian blur at 2, and then overlay blending mode - How does it look?

cletus

rutt wrote:

I'd like to add another parameter to this assignment. Discuss the pros and cons of converting the image to black and white. What was gained and what was lost? Is the image actually better in black and white than it was in color? If so, why?

In the good old days, black and white was cheaper than color, and in fact sometimes color just wasn't possible at all. But today, even newspapers can print color photos and black and white prints can even be more expensive then color. So the reason for b+w has to be aesthestic. Let's try to understand that as part of the assignment.

Great idea rutt!!!

In this case, B&W helps to reinforce the mood of the image. The color image shows a stark, desolate place. The change to black and white helps to make those feelings even stronger.

cletus

cmr164 wrote:

The original shot was pre-race as the runner read through her paperwork.
ISO Speed: 400
Aperture: f8.0
Shutter: 1/500
Focal Length (mm): 400

Note the long focal length. As usual click the image for the larger version.




Now converted in 'xv' but dropping saturation on RGB channels



And then after listening to the chatter here, taking only the G channel and reducing the saturation.

Great shot Charles. I think this is one of the best you've posted here! The conversions look really good too, but I think it might work best in color

lynnma

I can see this is going to be a very interesting forum... some great input and stuff to learn.. looks like I'm gonna have to print out the whole darn thing Part of my problem with Pam was that I gaussian blured before I converted.... flattened everything out too much.. some nice changes guys, thanks

rutt

pathfinder wrote:

Like you I would liked to have the full file rather than the 800 pixel size here, but I just did a little dodging and burning - not too sophisticated - and then a duplicate layer with Gaussian blur at 2, and then overlay blending mode - How does it look?

To my eye, I got a little better detail around that highlighted cheek. You did get the jacket darker. You can get that same effect with blending. I stopped short because I didn't want to lose detail in the shadow around her throat. But now that you went there, I have to admit it works.

rutt

I thought I'd try out Eric's most recent image to test the understanding I'm beginning to develop.

Here is Eric's conversion:



and here is mine:



I think I got better detail in the sky, deeper color in the ground, and better contrast all around. What to do you guys think?

1. Apply Image -> red channel to blue channel, darken mode, 100%. Now we have the best of the sky (from red) and everything else.
2. Duplicate Image, convert to grayscale
3. Apply Image -> red channel from 1 to duplicate
4. Curves -> steepen a little at both ends (we dont care if we loose a little detail in the window in favor of improved contrast. There really is no white point, so we can tighten up a little in the highlights.
5. Sharpen

The whole point of 2&3 was just to discard the red and green channels and move the red channel to greyscale. There must be a more elegant way to do this. Anyone know?

rutt

I thought I'd take a stab at the steeple; it seems like a good excercise because of the detail in the brick work.

Here is my result:




The blue channel had beautiful brick detail, which a naive conversion would lose because the red channel overwhelms it (our foe). On the other hand, the clouds live in the red channel.

Careful sharpening is going to be improtant to this image. We want the brick pattern to come out, but we don't want moire patterns or loss of detail due to light halos.

1. Apply image -> red into blue, darken, 100%
2. Discard red, green channels, convert red to grayscale
3. Create 2 duplicate layers
4. Blending options for top layer -> lighten, 50% opacity. Sharpen this layer conservatively for light halos.
5. Blending options for middle layer -> darken, 100% opacity. Sharpen this layer aggressively for dark halos.

The trick for separate light and dark sharpening is due to Dan Margulis' book.

cletus

rutt wrote:

The blue channel had beautiful brick detail, which a naive conversion would lose because the red channel overwhelms it (our foe). On the other hand, the coulds live in the red channel.

Careful sharpening is going to be improtant to this image. We want the brick pattern to come out, but we don't want moire patterns or loss of detail due to light halos.

1. Apply image -> red into blue, darken, 100%
2. Discard red, green channels, convert red to grayscale
3. Create 2 duplicate layers
4. Blending options for top layer -> lighten, 50% opacity. Sharpen this layer conservatively for light halos.
5. Blending options for middle layer -> darken, 100% opacity. Sharpen this layer aggressively for dark halos.

The trick for separate light and dark sharpening is due to Dan Margulis' book.

rutt,



Your conversion looks really good!


Although I haven't done it yet, I think this image would be a perfect application of masking. The edge between the foreground and background is very well defined, which should make coming up with a mask fairly straigt forward. Also because the foreground/background boundray is almost entirely straight edges, it should be fairly easy to make a vector mask.


Once the mask is made, the foreground and background could be optimized independent of each other.

lynnma

I started again from scratch with this... any improvement??? critique away, it's all a learning curve...

What would you have done..

lynnma

I'm not sure who said what where now but I think it was Pathfinder who said this "

1. Apply image red channel blend mode darken to green channel
2. Apply image blue channel blend mode darken to green channel
3. Keep only the green channel
4. Sharpen"

Can you take me to square one pleasea senor... I'm adobe challenged it seems.. are you using adjustemt layers - channel mixer - I think not and I can't seem to figure out where to start. must be the alchohol......
...I think it might have been Rutt rereading.. methinks I must order Dan Whatshisnames book immediately if not sooner...

rutt

cletus wrote:

rutt,

Although I haven't done it yet, I think this image would be a perfect application of masking. The edge between the foreground and background is very well defined, which should make coming up with a mask fairly straigt forward. Also because the foreground/background boundray is almost entirely straight edges, it should be fairly easy to make a vector mask.


Once the mask is made, the foreground and background could be optimized independent of each other.

Because the edge is so clear and because the values are so different, no mask is really required. That's the beauty of plate blending. You can use the blend if option to decide which layer is chosen. The result looks much more natural than any selection or other local enhancement, IMHO. In this case, I basically achieved this by blending in the red channel darken-only. It didn't have any effect on the building which was already darker than the in the red.