Re: Choir: Dan's comments
That’s an interesting question. First, I’m not saying that it’s bad, it’s likely the best one of the group, although personally I dislike it.
Having thought it over, I’d say that the underlying cause is that a par version minimizes poor technique, it averages out mistakes found in the individual parents and they aren’t as noticeable. The price is that it also averages out individual acts of unusual cleverness.
I surmise that the reason some might find this one unsatisfactory is the following: every par we’ve seen doesn’t contain obvious errors, such as excessive noise or clearly incorrect color. But until now, they’ve also had in common that there’s no immediately obvious way to improve them, in the sense that I point out and demonstrate ways to improve individual submissions in this thread.
This par, however, is different. Unlike the others, there *are* immediately obvious ways to make it better. Such as, blending #703 into it. Such as, selecting the painting and making it something more harmonious. Such as, making the altarpiece more spectacular against its background. Such as, darkening the foreground floor.
A few entrants can be found who saw the need for one or more of each of these moves. Some, though, have other issues that prevent them being a par parent. So these desirable moves fall into the category of “individual acts of unusual cleverness” that get averaged out in the par process.
Another way of looking at it: this image offers many more opportunities than usual for sharply different interpretation. In the lion image there were differences of opinion about color but I think we were all seeking the same tonal contrast. If we had converted every submission to grayscale there wouldn’t have been a whole lot of philosophical differences.
Even in something like Sunset on the Beach, which seems like it could have very different interpretations, it isn’t really so. We all knew we had to drastically lighten the woman and we all knew that pleasing color had to be added. It was only a question of degree.
To prove the point here: divide the picture into two parts: the choir, and everything else. We probably have a consensus on what the choir should look like. And indeed, few would object to the way it is presented in the par, whether in terms of the faces, the dresses, the necklaces, or the hair. It’s in the “everything else” category that we have real disagreements, disagreements that may not be resolved amicably by averaging.
In summary, in other case studies, I could generally have made marginally better pars if I could pick and choose how I was going to blend with each of the five parents, rather than the blunderbuss approach of 20% weight for each. Here, I’m pretty sure that a *much* better par could be produced by intelligent blending.