Week 2: Exercise 1

[leeharper_admin]

Here is my Merry-Go-Round correction.

I've been watching and reading the class materials (the videos are great by the way  ;)), so I initially adjusted the image in RGB according to the screenshots in the notes. I used this correction as a baseline for my own.

I felt that the baseline correction had a bit too much contrast, so I created a slightly flatter RGB version (I just adjusted the master curve), and compared the two. The more aggressive curves were much better for colour, so I decided to duplicate my version and take it back into CM so that I could use the Lab Saturation slider.

The saturation slider boosted the blues and greens too much, but I found that by backing the slider off until the greens and blues looked better left the rest of the colours looking too weak. Consequently I adjusted the A and B curves manually, to control the saturation boost.

I liked the RGB version with the Lab saturation adjustments, but decided to check HSB and CMYK as well. CMYK was a disaster! I'm not sure whether it was just my ineptness, or whether the nature of the image made it unsuitable for a CMYK correction - in any case, I quickly gave up on that approach. I liked my HSB correction much more, and whilst I was in CM I thought that it looked better than the RGB version. Upon comparision I decided that I only preferred the luminosity of the HSB version - and upon close observation, I only liked the HSB version in the brighter-half of the image. Therefore my final version is mostly the RGB correction (with Lab saturation adjustments), with the HSB correction being applied (in Luminosity mode) through a (blurred) RGB composite layer mask.

I have put together a small comparision image so you can see the effect that this HSB addition has (I think that the Reds and Blues look better for it - and it seems slightly more three-dimensional to me).

I have attached the relevant curve files, and comparision images.

Let me know what you think can be improved about my attempt. Also, any words you might have about why I had no luck with CMYK would be really interesting (I know that we will get to this eventually, so feel free to hold back on detail if you want to).

All the best,
Lee.

[ggroess]

Lee,
I have to be careful here because you really have opened up this image.  I think for the number of times I have seen people struggle with this that have you certainly have pulled much more out that I would expect. Having said that and please take it as a complement...I have to say that I think it is too open...not by much but by enough that It has made me want to comment.

OK...All that said

If you apply the curve I have attached and look at a few specific areas of the image I think you will see that you have lost something important here...

1) The necks of the horses;  The information is there but they lack shape.  I started to take this into CMYK to just add some K to the image but like you I found the CMYK to be tough... more on that later...I really ended up using the next best thing for shape and tonality...the L channel in LAB.  If you look at the hue clocks I have set you will see changes are modest but yet the effect is strong. 

2) The deep shadow on the top of the carousel...This comes down for me as a "significant details" problem.  What I mean by that is it is sometimes good to have areas of the image with little or low detail.  the goal for this image was to open the shadows and you certainly did do that but if you expand the area you allow to be darker by say 2-6 points... you can retain the shape of the image and keep some of the visual interest and contrast;  It also helps the overall color and tone of the image.

3) Lastly the skin tone on the child is too red.  If you lift the very end of the A curve  in LAB you can push that skin tone into a better spot without messing up the color too much.  You can also fix this in RGB.

Overall you made a vast improvement on the image; I get the feeling you were not entirely happy with some of the choices you had to make.  This image is a good example of compromise there is only so much you can squeeze out of it at the resolution we have to work with...

Greg

[ggroess]

And now for the CMYK...

I think the quarter tones are the hardest part of this image.  To begin with the image needs to be opened up more than I would do using just curves alone.  To make faster work of this I copied the background and switched the layer mode to screen.  I did this 3 more times before I went into the CM process.  I could of also used the shadow highlight adjustment in PS but sometimes I just whack an image with a hammer....

Then I opened in CM and used CMYK as shown.  As you can see the K channel is really the problem it requires the most adjustment...I think I even went too far in my example.  The color channels are a straight shift of the shadow ends of the curve to add some color contrast and saturation.  Each channel was adjusted to the same point then individual channels were adjusted.  I did not check a neutral and did this by eye...I spent less time on this than I did writing this response...I would probably take the image back into CM and do a simple LAB pass to fix any problems at this point.

So...to take this all the way back to your original question...it can be adjusted in CMYK but you need to be in a better starting point before you go to the curves.

Greg

[leeharper_admin]

Thank you for your replies Greg, they were really useful. Also, the curve files you attached are much appreciated  :)

My computer has been transcoding a DVD all day, so I've yet to try your suggestions (hopefully I will get a chance in the next hour or so) - but I'm very glad you made them. I'm here to learn as much as I can, and won't take any criticism badly. I'm dealing with the twin problems of not being very good at writing curves, and also I'm spending a fairly long time on each image (so I frequently reach points in my corrections where I can't see the wood for the trees). It's tremendously useful to have a fresh (and discerning) pair of eyes critiquing what I'm doing. Don't hold back - and don't worry about causing offence  ;)

Hopefully I will be able to log into Boot Camp within the hour, and I'll take a look at your curves.

Cheers,
Lee.

[leeharper_admin]

I hate transcoding DVDs!  ;)

I have just had a chance to explore the curve files that you uploaded. Thanks so much for uploading them - the CMYK curve particularly was really instructive. The whole adjustment was giving me lots of 'Ah Ha!' moments  ;) I'd never have guessed that steepening the CMY curves would boost saturation (as I said before, I don't really get CMYK yet), and the K curve looked really odd before I started to play with it.

Working with your K curve was a really great learning experience. The shape looked so strange to me until I started to move the points individually, to see where they affected the image. I love how pulling the lower endpoint to the right makes the blue and white curtain look. I had to switch between the K and All Channels view to see what was going on, and can see now that the K channel was contaminating the rich blues - and so, clipping those values to white in the K increases saturation, which was a real revelation.

It's funny, both you and Greg M screened this image to get it into better shape before you moved into CurveMeister. Having seen you both take that approach it now seems so obvious to me, but at the time I was working on my version it didn't occur to me once. That has been (so far) the real value of CM 101 - seeing and reading about the approaches other people take.

I'm going to work on some of the other exercises now, but I just wanted to thank you for setting off all of those lightbulbs!

Cheers,
Lee.

[ggroess]

As I am sure you know...light bulbs being turned on are one the best rewards for your efforts...Thanks!

I am planning an interactive session were I screen share and we talk via Skype. 

What Saturday works better for you?? 
July 10th or July 17th?? 

I would work out the time as well but need a date first...I'll be posting a notice on the board.  But I wanted you to get thinking about it...
Greg

[leeharper_admin]

The 10th might be a little better for me, but I'm sure I could do the other date if that works out better for Boyd.

As for the time, I'm in the UK - we are on GMT (Summertime) here at the moment. As I am posting this it's 20:12 here.

Looking forward to chatting  :)

Lee.