Adobe Photoshop Training with Dan Margulis

A client has a photo of 5 people, shot outdoors with harsh sunlight on the subjects creating bad shadows and highlights. She wondered if it's possible to fix the bad lighting, and since I've never been challenged with such a problem I thought I'd post and see if anyone has developed some techniques to soften such shadows and highlights.

Thanks in advance!

Jeremy Schultz

________________________________________________________________________

Difficult to say without seeing the image. It depends on the level of detail that exists in those areas. Post if possible and provide a link for viewing.

Les De Moss

___________________________________________________________________________

First try, in Photoshop CS, Image>Adjust>Shadow/Highlights. It requently works magic in situations like this.

Andrew.

___________________________________________________________________________

This kind of situation is probably best handled with the aid of channel masking. Make duplicate images and convert one to CMYK, the other to RGB. Look at the channels in both images. You're probably correcting in RGB, so you probably should look at a CMYK channel that is dark in the areas of the subject that contain shadow detail. Adjust a duplicate of this channel with Levels or Curves to darken the darks and lighten the lights. Now it's suitable for use as a mask. Make a selection from the modified channel and move it to your RGB file. Make the selection active. Duplicate the RGB layer and choose Layer>Add Layer Mask>Show Selection. Change the top layer's blending mode to Multiply and adjust the opacity to get more detail in the highlights (if there is any detail; if not, you're out of luck). Repeat this process to lighten the shadows, but this time use Screen mode.

What you want to do, essentially, is use channel masking to isolate the problem areas. To darken the washed out areas, you blend the layer with its duplicate, through a channel mask, using Multiply to darken the washed-out detail. To lighten the shadows, you do essentially the reverse by choosing Screen mode.

Depending on the quality of the image, this may or not work. But it's worth a try unless something better is suggested.

Let us know how you ultimately solve the problem, again presuming that there is enough detail in the image to work with. Sometimes you can create something from nothing, but it's probably not worth trying unless this is an extremely valuable photo.

Howard Smith

___________________________________________________________________________

Howard has some good advice , but as a rule of thumb I always try to use a Snapshot in conjunction with Curves for painting in image details before going to a mask. If the Snapshot does not work or I need more repeatability then I go to a channel mask.

My.02

Jim Rich

___________________________________________________________________________

I do this type of correction all the time. Post an FTP link for the group, or email me privately with a JPEG attachment.

J

___________________________________________________________________________

This question anticipates my September column, which introduces a very powerful method of dealing with this situation. Everybody is familiar with the basic scenario, which is that when confronted with something that's partially in bright sunlight and partially in shadow, the human visual system sees a far more balanced scene than the camera does.

We've had three good suggestions so far. First, the idea of using a channel mask works well sometimes but it often reduces contrast unnecessarily. A lot of what we perceive as highlight detail actually is not a highlight at all, but a quartertone or higher in the in the middle of a highlight area. So, the mask doesn't fully select it, and the true highlight will be darkened more than the detail it contains, flattening the overall look.

Second, painting in the detail via a snapshot or a layer mask works well provided there is a relatively small interest area to deal with. When there are several such areas in the image, it takes a long time and often looks unnatural.

Third, the Shadow/Highlight command of Photoshop CS is a big help. Properly used (which many people don't do), it's normally the easiest way of getting a reasonably good result, but it isn't as flexible as the following procedure, which takes longer but sometimes gives a much better result. I've been giving the method a workout and those taking the advanced course in November will get a full view of what it does.

1) Make a Luminosity selection, with Command-Option-~ or by command-clicking on the composite RGB icon in the channels palette. NOTE: as you get accustomed to this method, you will find when there are better options than this. For example, in this fleshtone image, making a selection based on the red channel (Command-Option-1) will probably work better, but a luminosity selection will work.

2) Select>Save Selection as a new channel (or, if you prefer, a new document).

3) Working on the new channel, sometimes a curve can be applied at this point, but it will work without it.

4) Working on the new channel, apply a BIG Gaussian blur--enough to wipe out all highlight detail. This is usually in the 10-20 pixel range. In faces, you should see only the outline of the head, and the eyes, everything else should be blurred out.

5) Invert the channel to make it a negative image.

6) Return to the composite RGB and make a duplicate layer.

7) On the top layer, Image>Apply Image>the New Channel>Overlay mode.

8) This will definitely add bite to the highlights. If you also like what it's doing to the shadows, you can now adjust the opacity of the top layer if need be, and you're done.

9) If you are happy with what's happening with the highlights but not with the shadows, change the top layer mode to Darken. Then, if a shadow adjustment is in order, you'll have to make a third layer and go through the process again, making sure that the blur in step 4 wipes out shadow detail, setting the third layer to Lighten, and altering its opacity as necessary.

10) After flattening the image, the highlights will now be full of contrast but probably a bit dark, and the shadows may be a bit light. So, apply a curve to restore full range.

Dan Margulis

___________________________________________________________________________

Dan,

Forgive my relative ignorance, but how does your method differ from simply duplicating the main image layer, desaturating/blurring/inverting the duplicate layer, and then blending with overlay?

I probably don't fully understand the significance of the luminosity selection and using Apply Image.

Enlightenment needed!

Bob Frost.

___________________________________________________________________________

Thanks for all your ideas. I tested Dan1s out and it works pretty well. Here1s one I played with during my trials:

(1) Correcting in RGB, I created a CMYK duplicate.

(2) In the RGB document, I used Apply Image->Black->Soft Light, 100%

(3) Then I discarded my CMYK doc and made another CMYK doc from the current RGB doc.

(4) In the RGB document, I used Apply Image->Magenta->Soft Light, 100%, inverted. Similar results to Dan1s technique, though I don1t have it here to compare....

In the end I wasn1t satisfied enough with any of these so the client gave me a different photo (shot under deep shade, which obviously has its own problems!) and was very understanding about it. I1m posting the light/dark photo in the photos section of the group. Some of these faces have nothing but blasted highlights and total shadow, and to get rid of that contrast completely and recreate professional lighting conditions is probably a larger task.

Jeremy Schultz ARTIST

2120 Grand Avenue, Suite 26

West Des Moines, Iowa 50265 USA

PHONE (515) 224-2025

CELL (515) 306-4348

WEBSITE http://jeremyschultz.com/

___________________________________________________________________________

Bob Frost writes,

The two do much the same thing but using a single channel is technically superior.

The problem with using all three channels for the overlay is not in the pure highlights and shadows, but in colored areas. Anyplace that at least one channel is >50% and at least one other is <50%, the color will be neutralized to some extent because the light channel(s) will darken and the dark one(s) lighten. There will also be a small loss of detail. Using a single channel as the overlay always moves all channels in the same direction and usually results in deeper, more pleasing colors, particularly in the near-shadows.

Also, you would need at least one extra layer for a three-channel method. The strength of Apply Image is that one can apply in Overlay mode but set the layer to some other mode (normally Darken or Lighten).

Plus, for those who get comfortable with the technique, it's often better to take a specific channel rather than just luminosity, and sometimes a curve is needed after the selection is saved. This would be more difficult in a three-channel overlay.

Dan Margulis

___________________________________________________________________________

Dan,

Many thanks for an excellent explanation.

Bob Frost.

___________________________________________________________________________

on 6/18/04 7:59 AM, Jeremy Schultz Artist wrote:

In the past, I've scanned a 35mm twice: setting up once to fix the 'over exposed' portions and setting up again for the 'in the shade' sections. I then combined the two in photoshop and got a really nice result. Of course this is not a photoshop solution per se without a re-scan. It did render a REALLY nice result though if you have the option to do a new scan. And of course a digital photo negates this workflow. :-)

Lanny Lathem

Bennett Graphics

___________________________________________________________________________

Not exectly, if RAW file is available. I use 2 renders from 1 RAW file with about 1.5eV differnce.

Best regards,

ib

___________________________________________________________________________

Iliah,

How are you blending the two files? I've tried using the Blend If sliders with moderate success. Is there a better way?

Bob Frost.

__________________________________________________________________________

I blend them using inverted/curved/blurred luma mask over lighter layer, the mask made from darker image.

http://www.pochtar.com/blending.jpg

Best regards,

ib