Seated in the Grass: Results


Dan Margulis
 

I’ve posted the results of the Seated in the Grass Study.

Reviewing: This is the final portrait in our series of case studies and is another example taken from the MIT retouching set. The original file is much better than the other two portraits we’ve worked on but there are still some challenges as well as some artistic calls, such as how much if at all facial imperfections should be downplayed, and what to do with the green background.

We have 27 entrants. When a person submitted two or versions, I chose the one I thought was better. Most people also submitted a list of their steps, thanks very much. I haven’t read these, because I’d rather get a sense of who was successful and who wasn’t before investigating why. 

The files don’t have people’s names on them, and were random-generator numbered from #1001 to #1027. As with past studies, we also have a “par” version, #1028. To get it, I chose what I thought looked like the five best entrants, and averaged them, each one weighted 20%. This often creates a version that is superior to most if not all of its parents.

I’ll have some things to say about this assortment, but as usual I’d like to open it up to group discussion first. What do the successful versions have in common? Meanwhile, if you’d like to know how your own version stacked up, download the par version and compare the two directly. Do you think you got the same kind of quality? If not, I hope you’ll find further discussion useful.

The folder is in the group Photos section, named Case Study: Seated in the Gras

I also have zipped all 29 files and uploaded a 36 mb file to our Files section,
Search for Seated-in-Grass_entries_072020.zip
If you are going to study these versions I strongly encourage you to download these files. Many of these entrants vary only in a minor way and it is hard to see the impact of a change without toggling back and forth between them.

I look forward to your comments.

Dan Margulis

P.S. The next case study is announced today, look for a separate post. We are now down to our final two case studies.


Robert Wheeler
 

For the Seated in the Grass challenge, I made a duplicate layer, applied a Camera Raw filter with the white balance dropper on a white area. Then applied green channel to the RGB in soft light mode with layer in luminosity mode. Shadows and highlights with PPW default settings was followed by the skin desaturation action. Then used the color boost action with end point layer curves adjusted with “Auto” and layer changed to luminosity mode. Finally moved back to sRGB and added a masked curves layer in luminosity mode to brighten her face. The result is #2011. I tried toning down the background several ways, was not happy with any, and finally decided to not be timid about bright color and enjoy the difference between bright green background and red hair. In comparison to other entries, I see that I missed an opportunity to tweak the face color to be a little warmer.

 

Sorting through the entries, I find wide variation in artistic choices. I learned that I tend to like the images with the subject fairly bright, that I like preservation of facial detail such as freckles more than the approach of smoothing the skin, and that I like warm skin tones better than cool ones. Others may choose differently. My top group includes 1002, 1005, 1006, 1011, 1021, and 1023 along with 1028 PAR.

 

With a mild vignette, blurred background, and smoothed skin, entry 1020 could easily be from a professional photo studio and many clients would like it well. I find the skin color of 1008 not pleasing. However, blending about 10% of that one into mine produces an interesting improvement. Many other images have aspects I like a lot along with aspects that do not work as well for my subjective taste, which would make it nearly impossible for me to generate a reliable rank-order of them.

 

It will be interesting to hear about preferences others have.

 

Robert Wheeler


Paco
 
Edited

Hi! Mine is 1005.

Kept the greens bright and shiny, the clothing was darkened and the highlights in them toned down and colored. Made a very dark version and painted/blended in the knee and clothing's highlights. Emphasized her hair color and eye color. The few red heads that I've met have had a skin color which is more R than Y so in CMYK values for the skin tone, R is aprox .03% more than Y. I'm guessing here; Dan will have an answer. Removed the darker freckles and corrected small patches of uneven skin texture on both sides. Removed R from the white in the eyes and corrected what little Y was in the teeth (what a beautiful smile!).

I hardly ever post my workflow because I feel that each one of us has their own particular way to solve whatever technical problems there are in an image. To start, I study the image and decide on what I want to accomplish before thinking about anything technical. Once I have a clear idea of where I want to get to, then I try and find the technique that will solve each specific problem. Most of the times things work out for me but like everyone else I crash and burn It is then that I try to learn how to not screw up again.

All the best!

Paco


Arthur Margolin
 

I'm getting to eagerly look forward to Mondays, the day that Dan posts all of the submissions, and the resultant nearly instantaneous recognition of where I went wrong with my version. Where else can you get such definitive feedback?
I saw immediately that in my version -- 1006 -- I rather single-mindedly targeted facial color and lost track of contour and modeling of the face.  This happened even though I created multiple versions and blended them. But, clearly, the multiple versions were constrained by my apparent lack of awareness of the narrowing of my attention. (It's also clear now that another pass of my final version through MMM would have benefited it.)
I find it somewhat difficult in this context to evaluate individual versions as free-standing entities. Rather, I'm more and more noticing, as Dan suggests, how particular characteristics, which might by themselves constitute "flaws", can be used to the advantage of other versions. For example, a blend with 1015 (which in fact looks fine on its own) markedly improves mine. 
(I also noticed that, inexplicably, I turned this lovely redhead into more of a brunette.)
As expected, the Par version looks like a very reasonable and balanced outcome of mutual assimilation.

This discussion, followed by Dan's comments on each version later in the week, makes this an exceptionally rewarding experience.

Best wishes to all,
Arthur

On Tue, Jul 21, 2020 at 6:02 AM Robert Wheeler <bwheeler350@...> wrote:

For the Seated in the Grass challenge, I made a duplicate layer, applied a Camera Raw filter with the white balance dropper on a white area. Then applied green channel to the RGB in soft light mode with layer in luminosity mode. Shadows and highlights with PPW default settings was followed by the skin desaturation action. Then used the color boost action with end point layer curves adjusted with “Auto” and layer changed to luminosity mode. Finally moved back to sRGB and added a masked curves layer in luminosity mode to brighten her face. The result is #2011. I tried toning down the background several ways, was not happy with any, and finally decided to not be timid about bright color and enjoy the difference between bright green background and red hair. In comparison to other entries, I see that I missed an opportunity to tweak the face color to be a little warmer.

 

Sorting through the entries, I find wide variation in artistic choices. I learned that I tend to like the images with the subject fairly bright, that I like preservation of facial detail such as freckles more than the approach of smoothing the skin, and that I like warm skin tones better than cool ones. Others may choose differently. My top group includes 1002, 1005, 1006, 1011, 1021, and 1023 along with 1028 PAR.

 

With a mild vignette, blurred background, and smoothed skin, entry 1020 could easily be from a professional photo studio and many clients would like it well. I find the skin color of 1008 not pleasing. However, blending about 10% of that one into mine produces an interesting improvement. Many other images have aspects I like a lot along with aspects that do not work as well for my subjective taste, which would make it nearly impossible for me to generate a reliable rank-order of them.

 

It will be interesting to hear about preferences others have.

 

Robert Wheeler


Dan Margulis
 

When I chose the ten exercises in this series it was no accident that this one came after Toast to Greece. That one started out with a lot of problems and contained a lot of traps for careless retouchers. You shouldn’t get into the mindset that all pictures are that difficult. This one certainly is not. The color is OK right out of the box, so all that is needed is attention to the basics, good artistic judgment, and a few of the tweaks that characterize good portrait work.

What are the possible traps? We want to add definition and depth to the face. In doing that we may make her freckling too pronounced. We may make the eyes too dark. And we may lose the redness of the hair. Most of us avoided these problems.

Additionally, we avoided two traps that I expected a few people to fall into. I was sure that somebody was going to try to turn this into a glamour shot, complete with plastic skin. And I thought that somebody would create unrealistic color in the blue jeans (no need to worry about the sweatshirt, as nobody knows what color it’s supposed to be). But, nobody was obliging enough to make these errors.

This shouldn’t take long, and many of you commented on the simplicity of your procedures. Some of you described very time-consuming workflows, which I don’t think are needed, but the results were good.

We also seemed to learn from errors in past studies. Not everyone can aspire to reach the par version. A more practical target would be #1003 or #1007, which are the best work from the MIT retouchers, who didn’t spend as much time on these exercises as we do. Were you able to do better?

It seems that almost everyone did. The main reason: after the discussion of the importance of range in the main objects in Cinque Terre and Colosseum, almost all of us planted a white point in the collar of the undershirt. The MIT retouchers did not, so their versions are flat and colorless compared to ours.

After our experience with the pattern in Panama and the noise in Toast to Greece, it’s good that we had a clear consensus on the freckles. They are a natural part of the skin and we should not try to eliminate them. OTOH we shouldn’t do anything to emphasize them. We generally accomplished this well.

We had many options for the background and here there was indeed a disagreement, as indicated in the private comments posted to me. Several people stated outright that they were deliberately making the grass far more saturated than it is in nature (as if that wasn’t obvious). They said they were doing this because it flattered the skintone more than a natural green would. But then a couple of others specifically said that we should be toning the background down (presumably because they thought it would distract from the foreground).

You can make your own judgment, but I favor the unnaturally saturated versions. Maybe this is the reason: I think the woman needs to be portrayed as healthy. I don’t like the ones where I consider her too pale. The danger of going the other way is that it’s easy to make skin that some consider too red or too colorful. Bright colors elsewhere, whether in the grass or the sweatshirt, help to fight that impression.

We did not do as well as expected in producing attractive eyes. I would have thought that after Veiled Bride this would have been more automatic. We also did not do a good job of attaching profiles to submissions—five of the entrants had tagged their versions as something other than sRGB, and three of these came in during the 24 hours before the deadline.

I’ll post comments on all versions soon. Meanwhile, we move on to our final two studies, which I don’t think are monstrously difficult.

Dan