After a relative breather with the lion, many of you rated this one as the hardest of the set so far. I agree. A nasty lighting problem is compounded by the need to attend to many important details.
In common with Sunset on the Beach, and certainly with the Land of Pagodas that we’re now working on, this features a sun-and-shade-type situation where half the picture falls into shadows. We don’t need any further information to know that we will try to bring the two halves closer together, because the human visual system does a better job of that than a camera does.
Normally, though, we don’t try to absolutely equalize the two, as if both were shot under the same conditions. That’s what many of us, possibly the majority, tried to do. Some even tried to make the background darker than the foreground. I don’t condemn any of this but I personally prefer the choir to be somewhat darker than it is in the par version.
The alternative is to lighten the background while lightening the foreground, but again the character of the image works against us. This isn’t a typical portrait where the subject is the important part and we don’t need to concern ourselves overly with the quality of the background. We can’t afford to have the church seem washed out.
By and large, though, we got through these obstacles, in the sense that well over half the entries seem OK at first glance. That’s a high percentage compared to previous difficult images. But that was just the first hurdle.
When confronted with an image of this type (sun and shade, or similar) we’ve developed some good tools to bring the two halves closer together, at least as far as relative darkness is concerned. We usually don’t go as far as we did here, so we can often ignore another feature of this kind of image: since the two halves don’t share the same lighting they may not share the same color cast. Typically the light half is warmer and the dark half cooler.
We have special aggravating factors here. The reflections tell us that strong sunlight is hitting the background but, this being a church, the sunlight could be entering through stained glass, with unknown consequences for color. Plus, we rarely are confronted with critical color in both halves of the scene. As we saw in the lion exercise, the concept of “gold” occupies a fairly small range. That is very restrictive in the background, and the presence of fleshtones in the foregrounds is even more restrictive, especially since we have different ethnicities to work with.
Under these circumstances, we can’t simply correct color globally. Many people tried to correct for normal skintones, and introduced a nasty yellow or yellow-green cast into the background. Others knocked out much of the yellow in the background and wound up with overly pink faces.
That’s only the start of the face issues:
*They can’t be too pink, but they also can’t be seen as washed out.
*The shaping is critical; several people presented singers with no noses.
*Soft transitions are needed in the lighter hair as plugged shadows would be fatal.
*Because of shadows, adding too much contrast is likely to make some singers appear to have beards.
*The skin is full of unacceptable noise and artifacting.
Then there’s the other major issue. When we have factors that obviously can’t be reproduced accurately, like the sunset itself in one of our exercises, or the appeals to non-visual senses in the Bellagio (or Niagara Falls) exercises, we need to exaggerate them. We have such a case here. The gilded altarpiece is designed to provoke awe. We are not able to accurately represent its brilliance, so we have to find some other way to make it stand out from its surroundings.
With all these demands, it’s almost impossible to get everything right that needs to be. Choosing our best becomes a matter of which imperfections are the most acceptable. That guarantees a big vote for the par, which minimizes individual errors.
As always, the par has nothing obviously wrong, which is more than can be said for most of its parents. Accepting that as a given, I have to say I am not a big fan of this par, for reasons that may become clearer in my comments on individual images. We must remember that this is a “stupid” par, where each parent is given exactly equal weight. In this exercise, more nuanced blending would have gotten a better result.
Some other fine points:
*We generally did a good job holding detail in the red dresses, which is sometimes difficult when the color is so saturated.
*Several people realized that the necklaces needed to be emphasized by making them brighter and less pink.
*Some others realized that the large background painting could be selected easily and offered nice opportunities to cut the background cast and to emphasize the brightness of the gilding.
*The catchlights (specular highlights) in the background gilding were distracting and needed to be addressed.
*Some people remarked that they were trying to get more detail in the conductor’s black dress. I don’t understand why anyone would think this is important. This image is already very busy with things we *do* want people to look at.
*For the same reason, the audience and the foreground floor are better off darker than they are in the par.
One person found an elegant solution to the problem of too much activity in the picture (and simultaneously to the issue that it’s easy to make it overly colorful). As far as I can see everyone else should have done it, too. It will be discussed in my next post.
That discussion will have a lot of LAB technique, which happens to be important in this particular exercise.
The entries include four from my 2009 classes. By the standards of that time the group did well, but we’ve gotten better since then. So, as expected, they all fell into the top half of entries (after all, they were experts) but not quite to our top level. Incidentally, the group that was forced to work with the flat version did slightly better than those forced to start with the open one. Both groups had lots of trouble with the skintone, as we did. The flat team had slightly better contrast. This also was the general conclusion of all the images the classes worked on: for color it didn’t matter which version was the start point; for contrast the flat teams did somewhat better. Access to the raw image was irrelevant, although I could have made it relevant by including some where highlight detail needed to be recovered.
Anyhow, I’d say this is the group’s best performance so far, considering the difficulty.