Last month, I presented a Culling exercise. I am uploading 2 from the set for your playing pleasure. If you are interested in another photo, I would be happy to add it here.
Story The van was at the shop, yet again, and ready to be picked up. The day was peaceful and sunny with short warm gusts. It was a 40 minute walk; on the way there, I came across a tree with interesting fruit. As usual, I didn’t have a camera on hand. This time I didn’t want to go home empty handed, so I snapped off a pod for safekeeping.
At home, I left the pod in the recycling bin at the porch, while I went inside to fetch the camera. The shadows were starting to change and if I were to get some good shots I would have to be quick. I also wanted to show how big the pod was in comparison with a regular snow or snap pea. I foraged the kitchen to find an edible that would do the trick. Yes, a kiwi would do.
The first half of the photos were on the deck on the side yard. Then I moved the subjects to the back yard, placing them on top the tarp covering the barbecue grill. In the latter case, they wouldn’t stay still, so I used the gardening hose and a broken branch to secure them.
I didn’t think too much about the staging. I didn’t have experience in that anyway and all I wanted was to demonstrate their difference in size and form. One thing that I did do was take multiple shots. The breeze and gusts were changing the light coming from the canopy so much that I want to give myself some insurance. The shots took a span of 20 minutes.
Nicely exposed image. The only tree with which i’m familiar that produces a seed pod is the Locus; we have one in our back yard, but its pods are longer.
I’m posting this one as a rawproc screenshot to show the filmic curve I’m using of late. The convention is to apply such with the parameters A=6.20, B=0.50, C=1.70, and D=0.06, and then follow it up with an S-curve to restore the contrast, but instead I’m inclined to modify the filmic curve itself to make the image look right, “One Curve To Rule Them All…” (OCTRTA). Hence the settings you see in the parameters window. Also included is the curve plot itself; note the decreased-slope toe, which crisps-up the black tones.
Here’s the toolchain commentary:
group - This is my standard proofing group, mostly settings from the camera data, applied in the order given below:
colorspace:camera,assign;
subtract:camera;
whitebalance:camera;
demosaic:ahd
blackwhitepoint:rgb,data - Scales the raw data to fit in the 0.0-1.0 floating point space.
tone:filmic,15.80,0.10,3.70,0.56,1.00,norm - OCTRTA! I’ve been messing with the ABCD coefficients to get a sense of what they do to particular images; right now, I have a sense of their behavior, but I really couldn’t coherently write it all out. Of note is that B and D affect the toe, A does a general lift.
crop:0.216783,0.220430,0.814046,0.817693 - crop of fruit and seed, get it?
Great to see more participants. So far each entry has its good and bad points. With respect to reality, I would say that @ggbutcher is closer with the kiwi and pod; @iarga the texture of the kiwi; and @kakashy the colour of the wood planks.
There is still more to explore in 1327: despite being pedestrian, there are quite a few subtleties in the scene. (Maybe it is my own limitation as a photographer. ) Also, I hope to see entries of 1335 at some point.
@age@iarga Would be wonderful if you shared more of your process.
Bonus
I had a Honey locust before it got infected and died.
This time I didn’t go for reality, but for an “over the top” edit.
I’ve been playing with frequency filters lately. This time I used: “Butterworth Bandpass”, “Sharpen [Whiten]”, “Freqy Pattern”. I put those results in gimp-layers for later use.
Then I edited the image in a “usual” way. “Mixer [PCA]” and “Curves”. Thereafter I used “TR’s Pixel Sharpener”, “GIMP Wavelet denoise” (Cb,Cr).
Then I combined the last results with the frequency edits (layer-processing in GIMP). Then a little bit “Pyramid Processing” and lots of “Tone Enhance”. Then on top of that G’MIC cli “deconvolve_richardsonlucy_blind” (by Jérôme Boulanger). This is really a great filter. Playing with the the autocorrelate threshold percentage between 20-35 and about 8-20 iterations gives nice results . I hope @Jerome_Boulanger will make a gui-filter in G’MIC for this.
Finally I used “LMS Adjustment”, for making this final result less painful for the eyes. I have to say that the original image has not a nice color harmony. I can’t change that without altering the colors completely. But then the result would be even more unrealistic.
by @afre.
) Also, I hope to see entries of 1335 at some point. 
