Playing With Black And White Conversion (Specifically Greg Gormans)

imax

Thanks Andy for the information that is on Greg Gormans site. Also thanks for the information on Petteri's Pontifications All great stuff. I tried several this AM and I liked this one the best. Let me know what you think!


Original

Grayscale

Using Greg Gorman Black And White Conversion Techniques

Thanks for taking the time to look. Have a great day.

Joe

JimM

I like the conversion, but as for the image, I really want to see the eyes open.

imax

JimM wrote:

I like the conversion, but as for the image, I really want to see the eyes open.

So How Are These For Eyes

THe Original

And The Conversion

Thanks

Flyinggina

Let's see. I liked the first b&w conversion on the first photo (the Greg Gorman is too dark - at least on my screen), the open eyes in the second photo (the conversion is good too) but best of all the hair in the first. What looks best when you make a print?

Virginia

DavidTO

Part of Gorman's workflow is to use curves to control the density of the image. It looks like this is something that you should play with, as all of your conversions are too dense.

thegreenegg

Here's my first try using the instructions. Couldn't find them all but it helped. Thanks Andy .

Orginal:


Greyscale:


Greg's Black and White:


Critiques as always welcome.
Ashley

Jnichols

thegreenegg wrote:

Here's my first try using the instructions. Couldn't find them all but it helped. Thanks Andy .

Orginal:


Greyscale:


Greg's Black and White:


Critiques as always welcome.
Ashley

I'm so glad you posted greyscale and then the Gorman method. What a difference. I have to find that link!

rutt

Greg Gorman's method worked better than just using convert to gray scale, but did it really work better than just taking the green channel?

Here is that green channel, straight from the color version you posted:



This is a much better starting point than the first step of Gorman's recipe, just taking the L channel from a direct conversion to LAB. Why? Because the red channel sucks and contributes to the L channel in the conversion. Look at the channels in RGB before you convert and you will see.

From this starting point, you can follow the rest of Gorman's recipe and I think you might get a better result. Instead I just used a slight S curve and a little USM. I suppose if she hates her freckles, she might like the Gorman result better. But if we want to show more facial depth and detail, we have to prefer my version:

DavidTO

rutt wrote:

Greg Gorman's method worked better than just using convert to gray scale, but did it really work better than just taking the green channel?

I prefer the Gorman, in this case. The skin has a very nice glow to it. The green makes her skin too dark and inaccessible.

Owen

rutt

DavidTO wrote:

I prefer the Gorman, in this case. The skin has a very nice glow to it. The green makes her skin too dark and inaccessible.

Jeez, then just mix a little red into the green before converting to B&W. Or start with the green and follow Gorman's method. Or use a little different curve.

The point is there is a decision to make before throwing away the color info and Greg's method completely punts by just using the default photoshop rgb->b&w blend, 60G + 30R + 10B.

DavidTO

rutt wrote:

Jeez, then just mix a little red into the green before converting to B&W. Or start with the green and follow Gorman's method. Or use a little different curve.

The point is there is a decision to make before throwing away the color info and Greg's method completely punts by just using the default photoshop rgb->b&w blend, 60G + 30R + 10B.

Well, for me the point is that Gorman's method is a great, easy recipe that produces very nice results for portraits. I wouldn't use it on every shot, like I think it's the wrong choice on the shot of the cops, but for skin, it works very nicely. It's easy, fun and pleasing. And his recipe uses the L curve, not RGB.

And you are right...it's invaluable to know how to use the Channel Mixer well. That's something that I have a lot of learning to do. But I also don't see the need to overcomplicate something that works quite well.

rutt

It's not the whole Gorman recipe that I'm questioning, just the idea of mindlessly accepting the L channel after conversion to LAB as the starting point. Photoshop has this little formula that it uses to do that. The brightness of each pixel is determined as 60% of the green channel, 30% of the red channel, and 10% of the blue channel. Accepting that without question is the same as a B&W film photographer deciding never to use different film or filters. You give up a dimension of creativity. In the case of the girl with the freckles, deciding how much red vs green to use is deciding how much the freckles are going to to contrast against the skin, what shade her hair will be relative to her face, how much the shadows under her eyes are going to show. Those are all decisions which can be made at the time the channels are blended into B&W and not easily afterwards. Play with this image by putting each channel on a separate layer and changing their relative opacity (hint, keep green at 100% on the bottom.) Try using the blending options for the red layer to reduce or emphasize the highlight on her nose.

AFTER you do that, then the rest of Gorman's recipe may work really well to get the tonal range, contast, and sharpness you want. But I don't agree with blindly using Photoshop's built in formula for the initial B&W conversion.

In the case of the girl with the freckles, it may be that my final version turned out too harsh. But look at the green channel before I started to mess with it and compare to the naive L channel in LAB. For most faces, the red channel only obscures detail. If this is what we want, fine. But it's a decision that shouldn't just be left up to some stupid program.

leebase

I like this one (buxom woman) better than your first (hair boy) as I felt that one was too dark.

Lee

DavidTO

rutt wrote:

AFTER you do that, then the rest of Gorman's recipe may work really well to get the tonal range, contast, and sharpness you want. But I don't agree with blindly using Photoshop's built in formula for the initial B&W conversion.

See, I don't think the recipe works if you take the green channel, for instance. The L (as you know) has a very different gamma than any of the RGB channels, and I think that if you use R,G, or B you've broken the recipe from the start, and you are heading down another path. I think that the recipe is dependent on that L gamma to work.

thegreenegg

Rutt,
I see what you are doing.
I still do like Greg's option better.
It helped make the picture light, like I wanted, and still left the freckles. Another thing was that I did not want the contrast between the skin on the hand and the face to be that different.
I will keep your ideas in mind for other headshots that I have to do in the future. It may work better for others than this one.
Thanks,
Ashley

edgework

If the point of this operation is to attempt to create a B/W representation of the information in the original color, then it's clear that no method is going to cut it without some carefully drawn curves to help out with contrast.

Here are the elements in the original that need to be captured in a B/W conversion:

1. There is a complicated light effect under her eyes. Between the shadows and her cheekbones a marked lightening occurs. This is both lighter and more saturated than the tones just beneath, and considerably more so than the shadows above.

2. Something similar is happening along the ridge of her nose. The side shadows fade away to a cleaner flesh tone, and then, right along the center is a distinct highlight.

3. The freckles are separated from the background as much by color as value. They're not a lot darker; somewhat, true, but the they are also a markedly different color.

4. Her hands have a lot of subtle detail in them which is helped by tonal differences, the kind of thing that gets crushed in a BW conversion.

5. The shadow along her left shoulder (HER left) contains faint but distinct hair detail. It's an open question how much of this could or should be brought out in B/W, but there is no question that care should be taken not to plug up that shadow. in the same area, her shirt, though deep in shadow, is clearly pink against the green of her sweater.

I agree with Rutt that there's no particular advantage to be gained by starting with the L channel from LAB, certainly no more than starting with a straight Grayscale conversion.

My problem with the green channel is that, by definition, it will strengthen any red areas. Often this helps the lips, but here, I think it makes them too pronounced. There are some red bands coming off her nose on either side, running down towards her mouth, which become blotchy spots in BW. The variation in her cheeks becomes much more pronounced in the Green channel, yielding more blotchiness. And the subtle light play around her eyes vanishes in the Green plate altogether. And, as the overall tone is heavier, a curve that lightens the lower end will only make the variation that much more pronounced.

I played with various setting using the Gorman recipe and it seemed to blow past a lot of the subtlties that I think are important. Since a lot of the variation in this face is tonal in nature, the L channel comes off as pretty flat, and all the steps after seem to be an effort to engineer back into the image what never needed to be taken out in the first place.



The three samples I included here all had a contrast cure applied as the final step. Each curve was different but all were basically "S" type shapes, with the major move in the highlights. The idea was to try to bring all three into close proximity, so that the differences in detail could be accurately assessed.

Looking at the different channels, I liked the density of the Green, but the Red contained much more variation in lighting across her face, even though it was overall to washed out and lacking in detail. So I split the difference. The Blue channel on its own is a disaster, but to keep from being discriminatory, I ended up using R=45%, G=45% and B=10%. For my money, this does the best job in the overall appearance of her face, though, compared with the color original, the contrast is a slightly harsher, but this seems to be a reasonable tradeoff for variation in the original that is primarily a difference in tone. In the shadow by her shoulder, both the ribs of her sweater and the distinction between sweater and pink top are in evidence.

Gorman comes off worst in her face, but there is slightly more detail visible in her hands. The shadow at her neck is overwhelming. I could probably find a curve that enhanced the detail in more than one area, but why bother, when good results are just a click away?

The big surprise for me was how good a result came out of a straight Photoshop conversion to Grayscale. It was helped by a contrast curve, but it could be argued that its overall effect is superior to the channel mixer version in that the freckles are not as much in evidence and, therefore, the feeling of heaviness that plagues all the other versions is absent.

wholenewlight

I'll toss my examples into the mix:

Original (with camera raw adjustments):


PS Grayscale:

Link to Large Version

Gorman Method:

Link to Large Version

Finished Product using Gorman (with a little more curves, toning, dodge & burn):

GoBlue!

John -

I love that image. Very well done!

Jim

pathfinder

I learned Greg Gorman's technique last winter from the Epson On Line Print Academy and find it useful in B&W conversions for portraits and moody, dark B&W prints. It certainly is not always the best technique and I suspect Mr Gorman would agree with that statement. I don't think it is as useful for landscapes or technical images as for portraits, precisely because of its usefullness for portraits.

Rutts comments about the detail present as a result of judicious choosing of the channels with the Channel mixing technique is very pertinent and accurate. Why should we blithely accept 30R+60G+10B as the only greyscale conversion, indeede!!??

Well,.... Frequently B&W conversions are precisely to diminsh the detail present in an image - look at a lot of the old dramatic B&W portraits with harsh shadows et al. Or the old grainy Tri-X images. Lack of detail sometimes increases drama and piques the viewer's eye, doesn't it?

Also, we tend to speak of G Gorman's technique as if there is a single, simple recipe, whereas I think of it as a series of several choices ending finally in whether I want to blend back in color or monochrome to add contrast and drama. I think I may try some channel mixer blends before moving on the the inverted luminosity blending in the GG techniqe due to Rutt's suggestions.

And Edgewater has done an excellent job of demonstrating that there really are no simple formulas - each image is different and the goals of the photgrapher are not always the same each time - so sometimes, even a plain old grey scale conversion works. Just like it life, it is good to have several different tools at hand. Just like in any highly skilled occupation - beginners know one technique or two - pros know several different ways of skinning a cat

david_h

imax wrote:

So How Are These For Eyes

Very nice pair of eyes.
I couldn't resist a quick BW jobbie myself. Since I often have to process a lot of pictures at one time, I have quite a few actions for different BW copnversion styles.
Here's an example that used just one keystroke from your colour original.....



This action actually uses the Lightness chanel. plays with curves a bit and a bit of levels with a pinch of grain for good measure.