Cinque Terre: Results


Dan Margulis
 

I’ve posted the results of our Cinque Terre study as described below.

 

Reviewing: This is from the MIT study of 5,000 images. It is an image of a village in the Cinque Terre region of northeastern Italy. According to Roberto Tartaglione, who should know, this is a winter shot.

I have never used this in a class so I don’t know how difficult experienced people find this. My notes to the study, however, indicate that three of the five retouchers did a terrible job. The other two versions are in our gallery.

 

We have 22 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 #501 to #522. As with past studies, we also have a “par” version, #523. 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’m going to 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: Cinque Terre 
Because some of us would like a closer look at these, I also have zipped all 23 and uploaded a 54 mb file to our Files section,

 

Search for Cinque_terre_entries_061520.zip
 I look forward to your comments.

Dan Margulis

P.S. The next case study is announced today, look for a separate post.


Kenneth Harris
 

I actually prefer 502 to par. 

Ken Harris


Ronny Light
 

Overall, I prefer 509, except for some weirdness in the sky on the left. I like the color of the buildings and vegetation. I also like the cloud detail.

 

I like the par version, 523, although it seems a bit too contrasty in places.

 

After a lot of comparisons, those are my two favorites.

 

Are post processing adjustments to be global only?

 

 

Ronny

www.RonnyLightPhoto.com

5010 B Wilkerson Dr., Nashville, TN 37211

 

 

 


Dan Margulis
 



On Jun 15, 2020, at 9:35 AM, Ronny Light <Ronny@...> wrote:


Are post processing adjustments to be global only?

No, contestants can do whatever they want except that in these MIT studies the Unsharp Mask filter is off limits, and they can’t resize, crop, rotate, or do anything else that would make it impossible for the file to be used to make a par version.

Note, for example, that a lot of people went to some effort to make a better sky; the person responsible for #514, nothing daunted, just inserted another sky from some other picture.

Dan Margulis


Gerald Bakker
 

On Mon, Jun 15, 2020 at 03:59 PM, Dan Margulis wrote:
Note, for example, that a lot of people went to some effort to make a better sky; the person responsible for #514, nothing daunted, just inserted another sky from some other picture.
Haha, I was highly impressed by #514's sky, almost wrote something like "amazing what detail can be squeezed out of the original flatness, and it looks very convincing as well." Oh well... let's call it an uovo di Colombo. :-)

My own version is #503. This exercise proves that my taste, at least for this type of images, is very much on the conservative side. I find more than half the submissions too colorful. This includes the par version. Go too far (whether color- or contrast-wise) and you run the risk of producing a cheap postcard-look.

I think a good use of MMM is crucial to success for this image. Strong variation between the oranges, yellows and pinks of the houses makes the village look brilliant without needing much color boost. Some versions (e.g. #502, #518) i.m.o. fail in this respect: much color but not enough variation (both examples too much on the yellow side, at least the village part). Others, like #508, #517 and #520, do it very well.

Given all the discussion we've had on the processing of the DNG file... I wonder if there is a quality difference between versions that started from the DNG and those from the JPG? In theory a DNG would give more room for maneuvering, but I'd be interested if this is discernable in the final versions.
--
Gerald Bakker
http://geraldbakker.nl


Jan Hof
 

On Mon, Jun 15, 2020 at 08:09 AM, Gerald Bakker wrote:
I find more than half the submissions too colorful. This includes the par version. Go too far (whether color- or contrast-wise) and you run the risk of producing a cheap postcard-look.
This was also my impression when I saw the results. My own version is #522, and it is even more 'conservative' than Gerald Bakker's version. When I started my first strategy was to make the image as colorful as possible. But for me that didn't work out: I got rather depressing results. (At least IMHO). I was even wondering: why would people bother going to Cinque Terre :-).

So, my favorites ares: #503, #517 and #529.

-----

Jan Hof


jwlimages@...
 

Well, seeing my image (507) in context here offers a good dose of humble pie. I did a few versions*, and tried for a more colorful version than "normal" for me - following the instructions to make something for use in promoting tourism, I imagined the client reviewing versions & saying "more… more…" color intensity  ;-) But this morning it looks too candy-colored - the greens are still too fluorescent cyanish, and the sea & sky are a miserable failure. Oh well… That danged ole Par version wins again, I think.

 

* edit - I went back to my first working version - it's really pretty close to the par version, but I had rejected it as too uninteresting, HaHa!

 

Gerald, thank you for calling out the importance of MMM for this image. I used it in one version, then went with CB instead, maybe starting me down a wrong path.

 

Also, about starting with .dng's - yes, there are potentially large differences from the default & flat-rendered .jpg versions. One detail (maybe not critical) is that the default used ACR/LR's default capture sharpening, whereas the flat version .jpg (rendered with all settings zeroed) had none. Since this image was shot with an older camera (the original Canon 5D) at high ISO, the noise & softness of the flat .jpg create an appearance quite different from the default .jpg. If you start with the .dng & enable lens profile corrections, the geometry will be different from both rendered .jpg versions, as noted in the first Cinque Terre thread of course. Probably the biggest variable is which raw converter you use to process the .dng - Adobe (ACR or Lightroom), Capture One, and others (I used Iridient Developer) all produce very different color by default. Although you can vary settings to get similar results, I personally am finding that Adobe's color (Lightroom) looks a lot less appealing 'in my old age'. I do use LR a lot for my personal work, but I have made custom .dng profiles for my own cameras, which makes a big difference.

 

Anyway, it's great to see what everyone came up with, and I look forward to more comments & analysis. I have much to learn about using PPP tools.

 

John Lund

jwlimages.com


Harvey Nagai
 

"Here, let us assume that it is for promotion of tourism to the Cinque Terre
region of Italy."

Based on this premise none of the entries would compel me to visit this town
(including my own).

That speaks to the difficulty of this assignment rather than the efforts of
the group, of which many are very good, a few excellent.

If this was my personally-taken photograph, I would have tried to take it
a different way that spoke more of the moment, which according to the
camera metadata indicates it was a dark dull morning in late November.

Special mention to 505 who did go his/her own way, I find it the most
interesting one to look at.


Robert Wheeler
 

I have been trying out various methods for systematically evaluating images other than just perception- at-a-glance. My grasp of the images seems to be gradually benefiting from routinely looking at the histogram, manipulating a threshold layer, and clicking through the channels.

 

For color analysis, looking at the amount of “a” and “b” components in sky, greenery, faces, and neutral areas is gradually becoming familiar and useful. In addition, I have found it useful to make a temporary hue/saturation layer and then push each individual color separately to maximum saturation. I find the results often informative.

 

In the Cinque Terre image, sliding magenta to maximum saturation highlighted the cluster of flowers at the lower left. The flowers lend a bit of liveliness to the image if not lost in dullness. To my eye, entries 501, 502, 506, 514, 518, 519, and 520 have done ok keeping the flowers alive.

 

The original image had very low contrast and limited dynamic range. The threshold layer analysis showed a white gutter blown out, which is ok to leave without detail, with the next brightest areas in the sky/clouds where detail seems important. The shrubbery has dark areas that benefit from keeping detail. I like the images that improved brightness without making a color cast and without pushing contrast or sharpening too harshly (501, 503, 516, 517, 520, 522, 523PAR do this in my view). To me, the darker versions make the place seem less inviting, even if some are photographically interesting.

 

I made one version just using Lightroom sliders for corrections. I made another version using an approximation of the PPW approach via a number of Dan’s actions, with a few adjustments of layer opacity and final color tweaks to saturation and lightness of various colors. The LR version seemed fine but lacked some detail, while the PPW version had more detail but contrast was a little harsh and a bit of halo persisted along the cliff-water edge on the right. I enjoy my merged version better than either of the component versions. It ended up as number 520 in the results gallery.

 

Looking forward to hearing from others (and to learning more with the next challenge).

 

Robert Wheeler


john c.
 

Is everyone looking at these in Photoshop on high end color-managed calibrated displays? I’m surprised by some of the comments. Of course, there’s a case to be made for looking at them in a browser if social media usage is what this is supposed to be about.
 
John Castronovo


Robert Wheeler
 

I'm using a BenQ SW321C, regularly calibrated with i1DisplayPro colorimeter, to view the downloaded sRGB images in Photoshop. Have also had testing showing I do not have any of the common color vision deficiencies. Currently am actively engaged in upgrading my knowledge in hope of eventually being more worthy of this excellent equipment.


Thomas Hurd,MD
 

What a fun day! I was out with my wife all day looking at bathroom tile and countertops, just sneaking in some image review in the car on my iPad for relief. Funny thing happened. I was looking at various natural materials like granite and marble, and started to see the LAB values…(“I wouldn’t call it aqua, Deborah, but more green than magenta and more blue than yellow”) I also was trying to explain to my wife and the decorator that a yellow/brown tinted grout would bring out more blue in the tile.

It was really great to see everyone’s take on this image.

I didn’t think it was very easy. I took several tries at it, and after the first day’s efforts, I decided I had just lost control of the neutrals.  I particularly paid attention to the road on the left hillside as my primary neutral. Also the grillwork on the back of the structures in the foreground, and the porch on the left of the orange house in the right foreground. 

My impression was this was a late afternoon picture. I thought it was clever to look at the time from the camera, but this is probably a tourist shot, and they may not change their time setting. My evidence for an afternoon was the dark side of the hill on the right, the dark left side wall on the coral/orange house on the lower left hill. As I went through several variations of the color correction, I got some magenta/red in the clouds, and felt it went with the afternoon concept. I realize that not everyone took their image the same direction. I was still trying to imagine a tourism promotion shot of a fishing town, with empty roads as they bring their gear in for the early evening before they bed down early to get up before dawn to go at the sea once more.

Somewhere along the way to my entry 504 I held the neutrality of the road, but lost it in the grillwork, and the porch wound up a bit red. I also wound up with  purple in the right hillside, therein losing the green that came out so well in many of the other entries, and in some of my earlier tries. I liked the overall weight of my finish, but when I saw the first pink/purple flower in the left lower corner on several other entries, I knew I had veered off the correct path. I believe I concentrated too much on the reddish tint in the sky and let it bleed in to the wrong places. I don’t know how or when I missed the flowers. 

I felt as if I should have gotten a better result starting with the bigger hammer, but didn’t really understand its nuances. After watching the videos on the options in the panel Sunday after I submitted my entry, I found that using a green overlay layer did a pretty good job for a first step. Just as I write this I redid the picture in about 10 min starting with that and was pleased with the result. Without trying to force any specific look, I obtained a more balanced picture with the colors, and even kept the flowers purple.

As for how I thought the others did, 505 took a different mood. It looks to me like a stormy day, and the noise in the back looks a storm approaching the coast. 513 seems to approach the late afternoon mood I was looking for, but more successfully with more saturated (and more correct) colors on the hillside.

Best colors (*my personal favorites):

502 * slight yellow cast doesn’t bother me, as the colors are all in relative good proportions
503 * just great color all around, with a bias toward lighter contrast
506   stronger sharpening than the others but would probably print very well
507   good colors except the hillside is a bit too desaturated and missed the flowers, as mine did.
509 * good colors all around and held them through the darkening of the midtones and 3/4 tones. Also held all the areas I mentioned above stay very close to neutral 
510  *good colors more saturated than my taste, and I think the sky needs more definition.
[I found it challenging to get good contrast in the sky and not blow out the center]
513 The colors are true but too saturated for me. Great flowers
517 good colors, I would prefer the contrast be increased in the midtones
519 a lighter version of 517
521 good colors, low contrast
523 * Nicely done!

Other comments:

511 seems over sharpened with white haloes especially on the right side palm tree and hillside shrubbery
518 persistent yellow cast. But color blending at 50% on my mage improved mine.
516 overweighted in the lower tones to me, but kept the flowers looking good, and neutrals look good to the eye, but measure a bit warm

As to the comment by John about the monitors, I had a calibration of both my monitors a few weeks ago, and they are drifting apart. I actually did my final correction on my laptop, and it didn’t workout for me. I don’t think I can really blame the monitor, just the user.

Tom Hurd




On Jun 15, 2020, at 4:29 PM, Robert Wheeler <bwheeler350@...> wrote:

I'm using a BenQ SW321C, regularly calibrated with i1DisplayPro colorimeter, to view the downloaded sRGB images in Photoshop. Have also had testing showing I do not have any of the common color vision deficiencies. Currently am actively engaged in upgrading my knowledge in hope of eventually being more worthy of this excellent equipment.


Roberto Tartaglione
 

Dear All,
I share some thoughts about my submission, my job Is # 510…
No doubt the picture of Cinque Terre was shot in winter,
see the tangerine tree with fruits; in addition was shot at 1/160 f8 1600 asa: there was no many light in the scene,
The sky is hazy and although blue is not predominant in the scene, the perception we have is of a cold winter day.

I decide to not respect this “season” snapshot and to force the scene into a spring/summer day:
Making the mediterranean sea and the sky bluer, the greenery greener and the houses warmer.

If the sea was in a storm, it could have the sense to render the photo as a cold stormy winter day, but as I said before nothing but the tangerine suggest it.

The difficulty for me is about the reflection over the roofs that I find hard to control.
The scene lacks midtones: the greenery is too dark and parts of the houses are too light.

I found this exercise the hardest of the three, I’m not satisfied at all, despite all my efforts,
I feel that something is missing, maybe the absence of shadows that should be in a summer shot.

However, here is my workflow

  1. I started from the default version
  2. I added a new level, darker, processed by raw, with a mask I darkened the lighter part of the image, mainly the roofs.
  3. I checked for the color
  4. MMM
  5. And again MMM+ CB, masking to boost more the blue color
  6. I compared with a second version, basically with the same workflow but lighter in the green (I added Bigger Hammer to the Workflow) and with more contrast; made a mix of the two
  7. In the end, I added more color to the sea and partly to the sky.
  8. At the very end, I mixed with a third version, softer, made by the flat version and added more warmth with a curve.

I know it is not a fair way to move forward but this exercise was really hard for me and I proceeded by attempts
 I think I pushed too much the saturation and the warm tone of the image after all, but my first attempts, more “conservative” were even worse.


Roberto Tartaglione




Dan Margulis
 



On Jun 15, 2020, at 10:07 PM, Thomas Hurd,MD via groups.io <tomhurd@...> wrote:

I didn’t think it was very easy. I took several tries at it, and after the first day’s efforts, I decided I had just lost control of the neutrals.

There has been a certain amount of wailing and gnashing of teeth about the results here and sometimes it devolves into confusion.

I am all in favor of wallowing in sorrow and self-pity, provided you know what it is you’re sorry about. Is it a lack of technique? Or is it a failure of artistic version? Or possibly a little of both? The danger of not knowing is that you can decide you’re not as capable as you really are.

Here’s my hypothetical test for list members.

I hand you back the original files for Veiled Bride, Niagara Spray, and Cinque Terre. I also give you screen grabs of the par versions for each one. I explain that I seem to have lost my files (backup? what’s a backup?) and need you to recreate these par versions from scratch. Your artistic view of these par images is of no importance to me. Shut up and do the work.

Can you duplicate these three par images, starting from scratch?

I’d say that you’d have to be pretty good to get to the par version of Niagara, and that Veiled Bride wouldn’t be particularly easy either. But if you are told to duplicate the par version of Cinque Terre I don’t see why this is a big headache. 

So, be careful of saying “too hard” without defining further what that means.

Dan


Doug Schafer
 

I really enjoy not only working on and learning from my efforts; but also from viewing other submissions. I often see not only different approaches by others but some parts better and some errors to watch for in my own efforts. Great learning experience to improve, and to hear from "the expert, Dan" of his views on all the images.

Looking forward to the experiences with future images too. Every image has a "story" to tell and a use. Each "author" interprets and tells a different story. As a reader, some stories are better than others, but few are wrong. And, not having seen the scene in live real life, a little imagination goes a long way.

Doug Schafer


James Gray
 

I think I may be at a bit of disadvantage in trying to produce the best possible result in comparison to most of the people working on these case studies.  I have never taken somebody else's image and post processed it for pay or mostly as a favor to anyone.  I have post processed images taken by my wife and colorized an old image of her father (whom I never met).  I enter my own images into competitions and juried shows.  So, I tried to create a result of the Cinque Terre image that might impress a judge in one of those competitions.  That means an image with more drama.  The term used in the LAB book 2nd edition was "phantasmal".  I do not know if that is too strong of a term for what I did with Cinque Terre.  I was satisfied with my moody result.

James Gray


On Mon, Jun 15, 2020 at 5:27 AM Dan Margulis via groups.io <dmargulis=aol.com@groups.io> wrote:
I’ve posted the results of our Cinque Terre study as described below.

 

...

 

The files don’t have people’s names on them, and were random-generator numbered from #501 to #522. As with past studies, we also have a “par” version, #523. 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’m going to 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.

 

...

 

Search for Cinque_terre_entries_061520.zip
 I look forward to your comments.

Dan Margulis

P.S. The next case study is announced today, look for a separate post.


Robert Wheeler
 

First, I can attest that at least one other participant (me) also arrives without paid retouching experience. My enthusiast photography experience provides a good background for this advanced technical and aesthetic level where I still have quite a bit to learn. Second, I personally find it healthy to frame these image challenges as learning exercises rather than as contests to see whose version is "best." From that perspective, I find it a great advantage to benefit from a group that includes high levels of expertise and diverse perspectives. These exercises, and the group commentary, have greatly accelerated my learning, and I am deeply appreciative of the opportunity to participate.


James Gray
 

I figured there might be others who are serious photographers rather than paid retouchers.  I fully agree that these case studies are an opportunity to learn how to get better.  I was really thinking of the fact that I have never had the experience of a client telling me the result of my post processing was not what they wanted even after I thought I had done a good job.  I have had the experience of a judge telling my final image was not that great.  However, in those instances the judgement seemed to be more about the original image than just the post processing.  So, I do agree that these case studies are not primarily about who had the best image.  However, it is about what is better about the images that are picked for the par version and what are the specific flaws in the other versions.  Probably more important are the steps or procedures that worked well or that resulted in flaws.  I  think I would find it really difficult to start from scratch on the Cinque Terre image and try to come close to creating the par version as Dan suggested.  I have not actually tried to do that.  I actually like my version of Cinque Terre a bit better than the par version.  That is a lot more than I can say about Niagara Spray and Veiled Bride.  Disaster is probably too strong of a word for my Veiled Bride, but it was not something that I was pleased with.  Niagara was somewhat better, but I was not happy with my result.  No doubt I have learned a lot from the group commentary and Dan's individual evaluation of each image.  I think it is great that Dan is doing this.  At least I can work in RGB and LAB on these images.  When I took his course it was pretty hard for me as many of them had to be done in CMYK.  I really appreciate the opportunity.  I think I have finished the Colosseum, but am not sure what it is supposed to look like.  My goal has been to create an artistically interesting image.

James Gray 


On Fri, Jun 19, 2020 at 3:05 PM Robert Wheeler <bwheeler350@...> wrote:
First, I can attest that at least one other participant (me) also arrives without paid retouching experience. My enthusiast photography experience provides a good background for this advanced technical and aesthetic level where I still have quite a bit to learn. Second, I personally find it healthy to frame these image challenges as learning exercises rather than as contests to see whose version is "best." From that perspective, I find it a great advantage to benefit from a group that includes high levels of expertise and diverse perspectives. These exercises, and the group commentary, have greatly accelerated my learning, and I am deeply appreciative of the opportunity to participate.


Gerald Bakker
 

On Sat, Jun 20, 2020 at 06:24 AM, James Gray wrote:
I figured there might be others who are serious photographers rather than paid retouchers.
I am in the same position, and I also agree that these exercises are a unique opportunity to learn. When I submitted my corrections so far, I expected without a doubt to be one of the best. That's what you think, right, when you normally do everything alone. How could anyone ever do a better job than what I just accomplished?

When it comes to evaluating the other entrants, I see that someone who has taken a path that I beforehand considered unlikely to be successful, has lead to a wonderful version, albeit very different from my own. For me (I think) the greatest learning effect is on the artistic side: seeing the results of people starting from a different vision, e.g. about the mood they want to convey.
--
Gerald Bakker
http://geraldbakker.nl


Kenneth Harris
 

As far as I'm concerned, the interesting thing about pictures is how they show people's thoughts about what they think they're showing.  That includes capture and post.  Tools can be used to many different ends.   Usually they're not.  Which is good.  Imagine forever being in room where conversation is interesting.  

Ken Harris