How I Curved Your Shot

Gary Glass

In response to:

http://digitalgrin.com/showpost.php?p=575236&postcount=10

The first thing I did was decide what I wanted to achieve. I wanted more detail in the white water, and more black in the rocks.

I dragged over the white water areas and watched where the pointer showed up on my curve line. You make detail by making the curve steeper. So I grabbed a point around the middle and pulled it up, making the white end of the line steeper.

This worked fine for the highlights, and also darkened the blacks, but darkened them too much. I grabbed a second point in the black half of the line and pulled it down, giving up dark-midrange detail for the sake of not completely plugging the shadows.

Now it was clear that the whites were not white enough. (Also note from the histogram that there's no white point.) So I pulled in the white end of the line. That gave me more brilliance. It also over-steepened my original line, so I used my previous two handles to get the slope back where I wanted it.

I sat back and studied what I had for a minute. It wasn't quite right.

One of the things that works about your photo is the contrast highlights and shadows and also the contrasting motion and stillness. The still pools, especially on the right, had greyed out to the point where they were lifeless. I didn't have anything clear in mind on how to fix that, so I just grabbed my handles and started experimenting. A made a few small adjustments (and I mean small -- curves are touchy) to the handles I already had, and found the best compromise between plugging the blacks without losing the detail on the sea moss, and not flattening out the midlevels so much that they looked like mud.

Then I converted for smart sharpening and did a smart sharpen, played with settings and settled on amount 100%, radius 0.5. That gave me better detail over all, but especially in the rocks. Then I went back to my curves and tweaked them again, in light of the sharpening.

dups41

Thanks a lot for the explanation Gary, I really appreciate the time you have taken to describe your processing.

Ok, I have found a problem with my original inversion of the negative so here is take 2...

I have pieced together a calibration process to give linear output that I was hoping would reduce the need for major image manipulation after scanning. For the specific film, EI and processing, I have used a step wedge to generate a curve that inverts the scanned negative and gives linear output from -3 to +3 stops. The values outside of the seven stop range are compressed. I expected this would give a good starting point that would require a minimum of additional post processing.

For Ilford HP5 shot at EI 200, developed for 7:00 in HC-110 at 20C, and scanned raw on my Epson V700 this is the resulting inversion curve. This curve matches the scan of the wedge photo negative to an ideal seven stop step wedge created in photoshop.



A problem I introduced, however, was to meter the shadows at -2 stops, and then linearize -3 to +3 stops from the scan. This resulted in the shadows being too bright and the highlights compressed. The image on the negative was well within the linear range of the film but my inversion was adding an unintentional 1 stop offset and losing highlight detail due to the tonal compression.

Gary's curves magic restored the lost contrast in the highlights. My curving abilities, though, are far below this level and I had hoped that an accurate inversion of the negative would reduce the need for such drastic surgery. Also, the tonal detail that was lost in compression can probably not be completely recovered. After taking into account the 1 stop difference between metering and inverting the result looked much better to me than the previous conversion. This is how it looked with the only additional processing being a white point adjustment:



Now I could use a minor curve to get some extra contrast in the reflection bottom left and give some additional range to the highlights. While I have difficulty coming up with complex curves like the one that Gary put together above, the smaller adjustments seem more manageable for me. This is the additional curve:



And here is the net effect. Sure, the midtones look flat, but that is how they looked in the original scene - early morning and very overcast giving flat lighting.




This is not too far from the original version I posted in the whipping post, but the amount of post processing was greatly reduced. I had hacked away at the whipping post image for a long time to try and get a look I was pleased with. To me the version above has much improved highlight detail and darker shadows - problem areas in the original that were pointed out in the whipping post thread (link) and the post above. Here is a link to a larger version that has also had some sharpening applied.

So what do you think? Am I going backwards here?

Gary Glass

I think you're definitely headed in the right direction. It's a challenging shot. In your final version I think you've got the midtones and shadows in pretty good shape. I still think the highlights at top could use more range, but that's just me. I also would have sharpened it more than you did, but that partly depends on the size at which you expect to display it, among other factors.

By the way, there is on dgrin somewhere a nice tutorial on using curves.