Beach at Sunset: Dan's comments

Dan Margulis

When confronted with an image of the sun-and-shade type, where half is strongly lit and the other not, normal procedure is to attempt to bring the two halves closer together. The reason is that humans adjust easily to two types of lighting in the same scene whereas cameras do not. They therefore present us with more contrast between the two areas than we would see ourselves. For PPW adherents, the normal way of closing the gap between the two is with a Hammer action if we are strongly interested in picking up more highlight and shadow detail, or a false profile plus multiplication through a layer mask if we are not.

Decent cameras are bad enough at such scenes, but at least they don’t try to outsmart the photographer. Smartphones and tablets do. They assume, probably rightly so, that their typical user is a fool, and consequently they build in a lot of tweaks that make for better images if that supposition happens to be true. Greens will get greener and blues bluer but fleshtones won’t, for example.  Unfortunately, what works well for typical photographs can do badly on weirdos like this one. What seems to have happened is that the iPad decided that the color variation in the sky and water made it the most important part of the image, enough so that it would pay to blow out the lighter portion of the sky and darken the beach drastically in order to give the favored area more snap. To overcome that, we need a Hammer and a false profile, and very likely need to select each half and work on them separately as well.

The sun is low in the sky, and it will be going down soon, but it hasn’t done so yet. Forget the iPad’s interpretation. The beach and the woman are getting a lot of light, so the original is way, way too dark in the entire beach area. How do we know?

a) The sea is getting plenty of sunlight, as seen by the powerful colored reflections that extend all the way to the water’s edge. But somehow the sunlight doesn’t extend into the beach?

b) The gravelly beach surface isn’t very reflective, but human skin is. Reflections are coming off the woman’s hair, nose, cheek, knees and hands, and the top of the camera as well. This strong lighting cannot mysteriously avoid lighting up the beach as well.

c) The boats, and the village in the background: nobody has their lights on! How can this be, if the scene is dark? (The few lights that are on look to be of the type that are timer-controlled, and would go on earlier to account for days shorter than those of mid-July.)

d) Look at the par, or better yet, at Paco’s #417 (which is lighter). These two convincingly portray daylight conditions on the beach. If the beach were truly in nighttime conditions or even dusk such a correction would be impossible. Don’t believe it? Our next case study, the Bellagio, is a night shot. Try to make it a daytime shot and see how far you get!

Now that we know that the two sides need to come closer together, what are we trying for. I disagree with Hector when he says,
So, I focus on making the woman in this picture as bright as possible (especially if they look very dark).

You'd be surprised, they might not even notice the sunset or the ocean.

That might be true of a photo of me at that age, or of my parents/grandparents, because not so many exist that we can afford to be selective. But here we have a child of our century. She probably has a thousand better photos of herself. If nothing was interesting about the sunset, why should she save this one at all, let alone ask for a corrected version? 

I don’t necessarily disagree with the people who used the word spectacular to describe what was wanted in the sky, but it doesn’t have to be spectacular to be interesting. A lot of the entries are what I’d describe as over the top in the sky, but who’s to say that isn’t what’s wanted?

Certainly the sky won’t say it, because the sky isn’t the client. The woman is, and whatever her opinion of the sky may be, she will insist that she herself look good enough, meaning much, much better than the original. Could her friends even recognize her in that one, considering she isn’t facing the camera?

What does it take to make a decent picture out of this mess? Not much, IMHO.

1) The sky/sea combination. Anything from interesting to lurid is acceptable. Any color combination is acceptable. It’s almost all personal taste.

2) The woman. A lot of this is personal taste, too, but there are some mandatory features, easily seen in Hector’s #401 and Paco’s #417:
*Healthy skin color, not too gray or too jaundiced.
*Distinctly blond feel to the lighter hair.
*A feeling of softness to the hair, not a sudden jump from lightness to black.
*Denim jeans and gray shirt approximating their real colors.
*Something done to minimize the noise in the fleshtone.
*Good shaping of the skin, taking account of the reflections.

Considering that there are only two priorities and just about anything will satisfy the first, you would think this would be a fairly easy exercise. But I could count the entrants who meet both requirements on my fingers.

I became so irritated after receiving around half a dozen consecutive poor entries with descriptions of lengthy workflows that I said to myself, “I could make an acceptable version of this bleeping thing in one minute.” So I did. Here are the steps, admittedly it would take longer without the PPW panel. The † symbol means that ordinarily I would consider some modification of the settings but with one minute there was no time for thinking, so I made the “normal” choice.

Click Velvet Hammer
Flatten and Click False Profile 1.4†
Add Background RGB as layer mask†
Gaussian Blur 30 pixels†
Move to LAB
Click Skin Desaturation
1-second lasso selection of woman†
Click MMM+CB
Adjust downward MMM Color and Color Boost.

That’s it. No sharpening, no channel blending, no curves correction, no retouching. The result is #407. Is it in the same league as our best entries? No. Does it have an interesting sky and an acceptable woman? Yes. Consequently it’s better than a healthy majority of our entrants.

Hopefully not yours, though. I’ll post some individual comments in the next day or two.


>On Feb 25, 2021, at 6:08 AM, <> wrote:

>What does it take to make a decent picture out of this mess? Not much, IMHO.
1) The sky/sea combination. Anything from interesting to lurid is acceptable. Any color combination is acceptable. It’s almost all personal taste.
2) The woman. A lot of this is personal taste, too, but there are some mandatory features, easily seen in Hector’s #401 and Paco’s #417:
*Healthy skin color, not too gray or too jaundiced.
*Distinctly blond feel to the lighter hair.
*A feeling of softness to the hair, not a sudden jump from lightness to black.
*Denim jeans and gray shirt approximating their real colors.
*Something done to minimize the noise in the fleshtone.
*Good shaping of the skin, taking account of the reflections.
>Considering that there are only two priorities and just about anything will satisfy the first, you would think this would be a fairly easy exercise. But I could count the entrants who meet both requirements on my fingers.

For this challenge, I decided not to read the details of the "brief" and just follow my guts to what is ( still obviously) a tourist snapshot. The many versions clearly indicate how subjective our personal decisions get in the way!!

I only wanted to comment that, in my own personal opinion, the PAR version resembles pretty much what all "modern" smartphones, like the iPhone 12 or Samsung 21 will do with such a situation: an extreme HDR-ized version of reality.

As a photographer, I have to question an overall uniform luminosity across the entire image, when the direction of the light is obvious,  and there should be a normal light fall-off, even when viewing the scene with our eyes. This false flatness in the PAR is unrealistic.  The old iPad instantly detected a silhouette-like situation, the user did not properly tap on the subject to expose her properly, although that choice may have burnt the background, so the image is a compromise between the landscape and the main subject, the blonde girl in her vacation.

This approach has been widely accepted in current architectural photography trends, where so many shooters and retouchers have gone so far into the ultra-low contrast, HDR imaging techniques, that the photos actually look now almost identical to the CGI versions, and this, I think, will lead to the eventual replacement of photographers for good, given CGI versions are capable of rendering even weaker shadows and less contrast than digital capture. Why hire a shooter when the computer performs better?
I have this perennial discussion with a friend who is both a photographer and an amazing CGI creator. Take a moment to view some of his fantastic work and by all means, assume that around 90% and more of the content of each image you will see is computer-generated.

At this point, he is taking pictures of the locations (the ones that exist in real life), but only to have a better rendition of the lighting, which is something he claims he can not yet fully generate in the computer with as much "life" as real light. I insist that lighting is the one thing that even experts in CGI still have to respect, so why are we, common mortals,  going to willingly break that rule, just because we can??

So, back to our girl at the beach, the new Night Modes in the smartphones would probably render this image even brighter than the PAR version, pleasing the consumers even more, while removing any sense of realism to the actual situation. It is a personal choice - one more time- to choose how far we prefer artificial over reality if the goal is achieved.

Just to show my point I am attaching a very small version of a picture I took at midnight December 31st, or should I say, past midnight on January the 1st, with the night mode shot on an iPhone 12 Pro Max. Just as it was captured, no adjustments applied.
Totally surreal look, with no lighting other than the moon ( was not even full).

If the Bellagio image were shot with any of these smartphones in Night Mode, it could certainly be adjusted to look like daytime, but how realistic that is??


Jorge Parra

Dan Margulis

401 Chosen for the par version, but I did not want to. I discussed the reasoning in my other post. Hector Davila stated that he believed the picture is all about the woman, in which case this is perhaps the best entry. Most everyone else believes that it is also about the sunset. I prefer some of the schmaltzed-up exaggerated colors of some of the other versions, but would accept almost anything, provided it isn’t as bland as this. The problem is that any potential substitute either has a much worse woman or a background that, if better than this one, is too blue or isn’t particularly exciting.

402 This is a very sensitive handling of the sky. But if the water is this bright, the beach can’t be this dark. And even if the scene is lightened the woman is too gray.

403 This has a similar sky effect to #402 but a more appropriate darkness distribution and a nice portrayal of the shore trees. The difficulty is a magenta cast throughout.

Purplish skies are not a bad idea, certainly better than blue clouds. But the light source that’s making the sea an appealing shade of purple isn’t shining on the woman’s back, so her shirt shouldn’t measure as magenta. One solution would be to run a tiny color gradient from the left, using the complementary of the purple. I’d prefer the simple method of a curves adjustment layer, click the black eyedropper tool, and use it to designate the darkest part of the camera as a black point. Then change layer mode to Color, and it becomes the best version we have where the woman is portrayed as this dark. Additional comments on this version at #406 and #419.

404 Chosen for the par version. This person, like Hector in #401 and many others, divided the image into two halves and worked on each individually.  Granted that approach, when working on the left half curves should make the boat lighter, as it is the natural highlight of that side.

405 A reasonable version for those looking for a relatively dark foreground. John Furnes has commented elsewhere on why he thinks this is the desirable approach.

406 David Remington was looking for an overall effect. He made several moves through gradients, filled in the holes in sky, and did some retouching of the woman. In retrospect, he posted

I went with the violet purple color palette of the sunset sky, accentuated it and used it throughout the image. I wanted a uniform color scheme. Now it seems a little too cool and the contrast is too harsh. I like some of the softer takes. If I had shot this scene myself I would say I over lit it. Too much fill on the woman. I agree with those who think a bit darker would be more natural. Also, I went with a silhouette for the coastline. I appreciate how open it is in Paco's version [#417]. I would go somewhere in the middle.

I agree that having excessive blue in this scene is not desirable and these comments could apply to several other entries. It’s one thing to say we don’t want blue but it leaves open the question of what we do want. Fortunately, we have a couple of extreme versions to play with—#403, if you’d like to move toward purple, or #412 for a more golden look. Blending either into this one 50-50 would be an improvement. Furthermore, if you apply #412 to this version at 50% opacity, then toggle back and forth, #406 as submitted seems positively icy.  

407 This is the version I prepared in one minute, sans sharpening, blending, color correction or retouching, as described in my earlier post.

408 Doug Schafer, like others, was disturbed by the holes in the sky. He selected them and added a warmer hue. Then, however, he reduced overall brightness on the theory that the sky should match the darkness of the woman.

409 John Gillespie thought that the spectacular nature of the scene called for a lot of handwork in the sky. What happened in the water is perhaps more typical of Las Vegas than Greece, but it also could be just what the client wants. However, if the water is so brilliantly colored by the sunlight, the woman can’t possibly be this dark.

410 Gerald Bakker used a 1.4 gamma false profile routine to get an agreeable presentation of the woman. He also concluded that the original had too much blue, so he used a Hue adjustment to move it toward purple/red. In retrospect, he finds there is not enough color. I think there’s also a slight mismatch in feeling between the two sides: the beach area is warm (note that the woman’s hair is more auburn than in other versions) while the background gives a yellowish impression. Slight overlay blending of the B channel into the A can add warmth quickly. As submitted, this version would be substantially improved if blended 50-50 with #412, which moves the woman’s hair toward blond and adds warmth to the background. Additional comments on this version at #416 and #428.

411 This person was having a busy week and could only spend five minutes on this exercise. So he did it quickly in Lightroom, concluded he had overdone the lightening of the woman, blended some of the original back in, selected the sea and sky, and added contrast. It came out rather well, I think.

412 Chosen for the par version. This is mine, and it looks to me like the day is about five degrees hotter than in anyone else’s. That’s fine with me for July in southern Greece. The woman was adjusted with a false profile and the Velvet Hammer action, but I also worked on both halves of the scene separately. Like others, I was bothered by the blown-out areas in the sky. Not feeling like exposing my painting skills, I took the lazy way out: I found another sunset image and threw it into this sky at a low opacity. Also I blended the B channel into the A, Overlay mode, for added warmth.

Note also that this is a pretty good “auxiliary” image. Even if you don’t want to go this far with the golden look, this one probably improves most of our images if blended in at 15% or so.

413 This one gives a frosty appearance. The woman’s shirt and her camera both measure B-negative, blue.

414 If you’re looking for a darker treatment of the beach, this one came out nicely. After applying a false profile, the person blended with the red channel to ligthten the skin more than the rest of the foreground. He also ran Velvet Hammer, but cancelled it because he felt it was making the beach too crunchy. 

415 An interesting impressionistic background but the foreground is much too dark.

416 This person tried treating it as two separate images, didn’t like the result, gave up, and did the correction mostly through masks in a different program. The result is reasonable but seems too cold and lacking in color generally. The notes mention adding cyan to the foreground, presumably to achieve the properly desaturated fleshtone. Whether this was the cause or not, the shirt, the camera, and the whole beach are decidedly B-negative, blue.

This version reminds me of #410: both nice contrast, both short on color generally, and both with a cast, blue here and yellow in #410. Since blue and yellow are complementaries it follows that a 50-50 blend of the two will be better than either parent.

417 Chosen for the par version, and IMHO the best individual entry by quite a bit. Paco Márquez explains:

Duplicated the background layer. Assigned Apple RGB 1.0 gamma, multiply layer mode with a blurred RGB mask.

From then on, using the R channel, made masks for sea, sky and woman, used Curves, Sel/Col, Hue/Sat, Shadow/Highlight and Dodge and Burn.

Started color correction with a point on the white of the boat on the left.

The pebbles in front of her take on a reddish cast from the sky so I left it in because removing it made it too unreal.

Darkened the top sky a bit to keep the attention on her.

The sea and sky colors might be too exaggerated but in her mind's eye that's how she saw them.

Her skin I felt had a great tan so I kept it on the warm side.

Did what you've been suggesting and made extreme versions which I later merged into each other and I think that's a great idea.

The result is an excellent woman, convincing color, and a sky/water combination that, if not “spectacular”, at least is quite interesting. I’m not really inclined to quibble about anything, as I like this version better than the par. My only suggestion is that the blues could stand to be a little darker, this would make the reflections in the water more convincing. So, Apply Image of the par to this one, but check Mask=Blue.

Additional comments on this version are at #406, #434, and #437.

418 The introduction of large amounts of yellow into the lightest areas of the sky is interesting but something more orange would likely have been a better choice. The darkness of the beach is apparently deliberate, since the person started off with a false profile at .75 gamma, which would have allowed him to brighten the woman much more than this. What was described is a very complicated workflow—so complicated,  in fact, that it didn’t leave any time to run Auto Tone at the end. It wouldn’t have made the version better but it would have suggested that Auto Contrast would. Without a full range, there’s a limit on how good an entry can be.

419 Robert Wheeler wanted to make the sunlight spectacular. I can buy that there is a rainbow of reflections in the water, but the bright yellow sun seems to quarrel with the blueness of the rest of the sky. He says he wants it warmer and suggests blending 50% of the par into it. That certainly improves things but I would repeat the suggestion that I made in #406, which also has a sky that seems overly blue: it’s better to move toward magenta (like #403) or toward golden (like #412). I’ve tried the 50% blend of each into this version, as well as 50% of the par, and I like the par least of the three for this particular blend, which is instructive.

420 This image turned into a grayscale for reasons that aren’t apparent from the description.

421 This person used an inverted luminosity mask to lighten the image, this may account for its washed out appearance. A false profile procedure would likely get better results.

422 Ronny Light probably makes himself unpopular by saying “This is my favorite photo of recent case studies.” He used a simple workflow with some special attention to the sky. The result is satisfactory, although I find the blueness overpowering. See also a comment about this version at #430.

423 John Castronovo has posted about his interpretation, and from that and his written notes I surmise that he feels the darkness of the woman is about right. In the sky, he properly warmed up the scene by shifting away from blue and toward purple. The sky still is full of distracting holes. For that reason, if you want a purplish version, I’d prefer #403 in spite of John’s good treatment of the woman. But for a result better than either, try blending #403 into this one in Darken mode. That clearly illustrates the value of getting some warmth into those holes in the sky.

424 Here’s what I was talking about in my first post when I warned against having too abrupt a transition from blond to black in the woman’s hair. Also, parts of this sky measure as too green.

425 This one comes not from us but from a student of Matthew Croxton. Like me, he stole a sky from elsewhere and used it to beef up what was found in the original.

426 James Gray created a nice rainbow effect in the water, but the left side has a yellow cast, measurable in the shirt and in the shorts. Blueness of clouds fights with the water

427 The Grinch, having just recovered from stealing Christmas, opines as follows:
This reminds me why I haven’t taken a sunset (or sunrise) picture in 15 years.  People who do deserve what they get, absent heroic measures.

Another to add to the collection of interesting right halves with unsatisfactory left ones.

428 Nice shimmering reflections in the water, foreground too dark. Frustrated, in the other thread Kent Sutorius says

I don’t understand how you decide on how light the foreground should be in the photograph

Most rules about color correction don’t involve what should be, rather they involve what should not be. The woman’s light hair should not be even slightly green, or excessively red. The beach should not be blue or green.

We have rules like this because experience has shown that they are not matters of personal taste. Rather, if they are violated the version will almost unanimously be rejected in favor of an otherwise similar image that honors them.

This can be done to some extent with the color of flesh, but not its overall darkness, as what might be true under one lighting condition would not be for another. For example, some would likely say that the skin in #401 is too light, while others would say that #410 it is too dark. However, there would be differences of opinion, for which reason I’d say that they’re both acceptable.

Skin significantly lighter than #401 or significantly darker than #410, on the other hand, is likely to get a near-unanimous thumbs down. So it would be appropriate to say that doing either of these things is wrong—in the context of this particular image. But that doesn’t help us much with the next one.

401 Christophe Potworowski deliberately maintained a reddish-purplish look, but despite some efforts to lighten the foreground with LAB curves, conluded “It’s still a lousy image.”

401 Chosen for the par version. We finally find one to satisfy those looking for a relatively dark beach. It’s got everything we need: A convincing woman, nice coloring in the sky, a cheerful and colorful overall look. To see how important that is, compare this one to #422, which is just as colorful and also similar in weight. But #422 emphasizes blues. I say that #430 is sunnier and happier. Edward Bateman describes his workflow:

I wanted to try an experiment with myself. I made 3 versions… each made on separate days without looking at previous versions

Version 1 started with the Bigger Hammer using the Blue channel. Then curves for some color adjusting. Then an Adjustment Layer adding 19 points of Vibrance using a darkened mask from the blue channel to limit the Vibrance to the sky. Then a HighPass/Overlay set to soft light at 75%.

Version 2: I tried to do it all using just the RAW tools.

Version 3 started with curves for some color correction. Then Bigger Hammer using the Blue channel.  Then HighPass (26) blended with Soft Light and excluding the woman.

I was surprised at how different each version was. That really gave me something to think about - and to compare. So then I  blending my versions. I placed one version on top and adjusted the opacity to find the balance of the images I liked the best. Version 1 and 2 were blended at 60%.  I repeated that with Version 3 with 50% opacity.  Then I used Shadows/Highlights with DM default setting at 50% strength to the flattened version.

A crazy way to work… but I learned a lot.  Curious to see how others’ versions look… there was a lot of room for interpretation on this one. I felt that the woman would want to look tan (based on season and clothing) … and made a bit redder by the warm sunset light.
A multi-version approach is most effective when, as here and to a lesser extent in Concert on the Beach, there is a lot of doubt as to what you’re trying to achieve. One approach that you guessed would be silly may be unexpectedly good, and the one you had the most hope for may turn out to be garbage.

431 Another good sky coupled with an overly dark woman. The person described about 10 minutes of work, mostly S/H and Bigger Hammer.

432 The strong purple cast in this one comes from the person’s belief that the clouds represented an incoming storm. To that, a reminder that “Red sky at night=sailor’s delight. Red sky in the morning=sailor’s warning.” There are atmospheric reasons why this is true, and they've been known for a long time:

Jesus said, “When in evening, ye say, it will be fair weather: For the sky is red. And in the morning, it will be foul weather today; for the sky is red and lowering.” Matthew XVI: 2-3

433 The notes describe a deliberate attempt to darken the left side, so as to make the sky/water combination more dramatic. That it does, and we might have considered such a move if the beach did not contain an important human being.

434 Here is one of the nicest backgrounds, full of interesting color without being as oppressive as many of the other efforts. The beach area is OK but this could be substantially improved by treating this as two separate images, as many of us did. With everything selected except the sky and sea, establish a white point in the boat on the shore. Do this, and it becomes a reasonable choice for best version for those who feel that #417 is slightly too light. See another comment about this version at #437.

435 An outright night shot.

436 You may recall the story of this Greek family reunion, which I told in the “A Toast to Greece” exercise in the 2020 set. My wife attended, but I was not keen on being one of only a couple of people out of 30 who didn’t speak Greek, so I stayed home. My punishment was that my wife volunteered me in absentia to color-correct all the photos taken by these people with their various devices over two weeks. She told them that I was very fast and could probably knock off the whole batch of two thousand or so in a couple of hours. Furthermore, they would all be masterpieces, since I am supposed to be the best in the world at doing this.

I need not involve you in the marital difficulties that ensued, but the result was a “compromise” where I agreed to look at all of them and correct not more than a hundred. I wasn’t limiting myself to three minutes as in the MIT study, but I was in no mood for time-consuming adventures.

This, then, is what I did in 2018. I note that the water/sky combination is very close to that of the par, except that this one does a better job of adding yellow to the lightish areas and is therefore superior. The woman’s hair also came out well IMHO, and I also like that the gravelly beach is not as attention-grabbing sharp as it is in the par. The shape of the woman’s skin doesn’t measure up to the par, however. 

437 Jim Sanderson, posting to the list, agrees that a light interpretation is correct because the sun has not yet set. However, his notes on procedures are a warning

I found this one to be the most difficult for me so far.  If there's anything I've learned (so far) from this exercise it's never to use an Ipad to take pictures at sunset on the beach.  I kept running into problems with negative A values sneaking up in unexpected areas.  I won't list all my steps as I wandered from adjustment layer to adjustment layer using various and sundry masks.  Thanks ever so much for the lesson in anger management as I tried (and succeeded) in not throwing my coffee cup through my expensive color critical monitor on several occasions.  

When frustration sets in it helps to look at individual components. Colorwise, this has a greenish feel. Also most of us preferred more color overall. It’s important to note that when the fleshtone of a light-haired person gets this light, more color is necessary.
To get a striking demonstration, grab hold of Paco’s #417, which happens to be about the same darkness as this one, and also #434, where the woman is considerably darker. Put #417 on top of each of the others in Color mode. With #437, that’s a big improvement as more pinkness flows into the flesh. With #434 it’s a failure, because skin that dark can’t support as much color as in Paco’s version.

In terms of contrast this also leaves something to be desired, as can be seen by switching the Paco Color layer to Luminosity mode. The brief explanation is that in #437 there seems to have been a subconscious decision that detail in the background hills is more important than in the woman. Lightening things enough to bring out details in the hills shortchanges the range of the skin. So if we were to decide that detail in those distant hills is the most important thing in the image, then sure, this one beats Paco’s. But nobody is going to decide that.

438 A plug-in was used here to add shape to the clouds, after filling in the blown-out areas. The background seems good, but the woman is unconvincing because the transition from light to dark in her hair is clunky.

439 The par version.

Bill Theis

I am very very happy that I didn't get comments that the woman's skin was "swarthy" this time, only unconvincing due to clunky hair.  I can live with that.  I was convinced that she was backlit and that we are seeing that dark side with no reason to expect it being artificially illuminated by a big thunderhead in a clear sky.  I did dodge the upper arm and leg since I felt the light should be coming from her front and perhaps above.  All in all a very good exercise and I agree with Dan's choices for par which make for a great comparison along with all the other versions from everyone.  Thanks.

Christophe Potworowski

Thanks Dan for the comments. By “lousy image” I was referring to the tendency in the image to break up under the influence of different adjustments, for example the skin of the woman on the beach or the sky. On the other hand, the the term may refer to the quality of my take (# 429)on the image. Either way, I’m grateful to be taking part in these series of exercises.