Comparing Tools and Workflows - A Simple Image
jjbong
Registered Users Posts: 244 Major grins
This is the first image for our discussion. This is a picture of a bridge over the Rio Grande Gorge north of Taos, NM. Here is a JPEG I created directly from the RAW file. I only set the white balance in Camera Raw, using a WhiBal card as a reference:
I think this image is typical of many outdoor photos people take, and it has the typical problems. There is a definite cast, although it might not be obvious. It is pretty flat, and it lacks color.
Going back to Pathfinder's excellent dictum, it's not what I saw. I saw this incredible bridge over a deep gorge against an incredible background of mountains and sky. You don't get a sense of the depth of the gorge (this is a composition issue, that probably can't be fixed in post), but the background, sky and mountain, are not that impressive. The bridge is there, for sure, but not much else.
Anyway, I'll post my "post", and the steps to get there. Others are welcome to have at it. I'm sorry I couldn't find a way to post the RAW, but I'm happy to send the DNG to anyone who wants it and has a way to receive it (an email of 59 MB non fa strada, as the Italians say).
I think this image is typical of many outdoor photos people take, and it has the typical problems. There is a definite cast, although it might not be obvious. It is pretty flat, and it lacks color.
Going back to Pathfinder's excellent dictum, it's not what I saw. I saw this incredible bridge over a deep gorge against an incredible background of mountains and sky. You don't get a sense of the depth of the gorge (this is a composition issue, that probably can't be fixed in post), but the background, sky and mountain, are not that impressive. The bridge is there, for sure, but not much else.
Anyway, I'll post my "post", and the steps to get there. Others are welcome to have at it. I'm sorry I couldn't find a way to post the RAW, but I'm happy to send the DNG to anyone who wants it and has a way to receive it (an email of 59 MB non fa strada, as the Italians say).
John Bongiovanni
0
Comments
Moderator of the Technique Forum and Finishing School on Dgrin
I followed the 3 steps of PPW. I'm using PS CS3, so some of the procedures or short-cuts may be out of date.
One note - this is not the exact set of adjustments I would use for any given image. At each step in the PPW,
you need to evaluate the image and select the appropriate techniques. The ones I chose here were appropriate
for this image.
1. Correct for Color Cast
The foreground looks reasonable, but the sky is a little too green, and the mountain is quite blue.
Setting the second info palate to LAB, I scanned across the clouds. They should be neutral or blue, and what
I see is neutral to strong blue (0 to large negative values in the B channel), and neutral to slight green
(0 to -3 in the A channel).
I set up a curve adjustment layer and set a point in the green channel by ctl-click on a point in the clouds
with the largest negative in the A channel (the most green). I also set an info point there, and maybe a
few other points in the sky. I drag the point on the green curve down to get the A value to 0, without going
too red (positive in the A channel) anywhere else in the clouds. The curve adjustment layer is set to Color
mode, so it only affects color.
This is the curve, just a slight move in the Green channel.
And this is the result of that correction:
Slight but noticeable difference. Notice comparing the original that the open land between the bridge and
the mountains, along with the mountains themselves, are less green, and more realistic. I couldn't easily do anything
about the blue in the mountains.
2. Get the Best Contrast by Getting the Best B&W Image on a Luminosity Layer
First, see if there is any channel with better detail than the others. Here are the RBG channels:
The Red channel is clearly superior in the sky and the mountains, the Blue is clearly inferior everywhere,
and the Green is OK in the bridge and the foreground.
So I copy the background layer to a new layer. In this new layer, I do the following channel blends using
the Apply Image command multiple times, always blending the background layer into the copied layer:
- Replace the Blue channel with the Red Channel (normal mode, 100%)
- Darken the Green channel with the Red Channel (darken mode, 100%)
Then I do a curves adjustment to the copied channel (through Image->Adjustments->Curves) to bring out the
contrast in the key parts of the image. Since this is in luminosity mode, I only curve the Red and Green
channels, as the Blue contributes very little to contrast. The interesting detail is in the mid-tones, so
the curves are S-curves (I run the cursor over the areas of interest to determine the shape of the S in
each channel). In this case, I don't want to lose detail in the sky, so I keep the curves linear in the
high end of the curves:
At this point, I see whether Shadow/Highlight to the copied layer helps, and it does:
And here's what we have:
Significantly better. The only possible problem is that the channel blending changed the color of the shirt
of the person walking across the bridge. You see this as soon as it happens, and it's easy to correct, if
you care. In this case,I don't.
3. Boost the Color
Move the image into LAB, and do a simple linear curve inthe A and B channels, making sure that each curve
goes through (0,0), so that you preserve neutrals and only enhance, not change the colors. The only way to
do this accurately is with an action. Here is the curve:
Notice that the slope of the A curve is higher than that of the B curve. This curve will wildly enhance the
colors, so you use the Opacity to dial it back to something pleasing. I chose 20%.
At this point, I'm done except for sharpening:
I think I could do more in the sky and the background, possibly multiplying through a mask. But my guess
is that it would take a bit of effort to generate the right mask, and I'm not sure the image is worth it.
Some final comments. This may seem like more work than it is. My original pass (not carefull taking notes
and screenshots) was probably 3-5 minutes. Especially with actions, it's quite fast. What isn't fast is
the learning curve on the workflow. It does take a significant investment in time (at least, it did for
me), but I felt the results were worth the effort.
You have definitely increased the contrast and color saturation of the image. I cannot tell if the noise in the image has been altered as a result of the editing. I am not an anti-noise zealot, I even add it to images now and again. I have an image from the same vantage point I think. Edited as an Out of the camera jpg in 2003 from a 10D.
How large is the RAW file for this image? Can you ZIP your RAW file and email it to me if I sent you an email link that accepts larger files?
Moderator of the Technique Forum and Finishing School on Dgrin
This makes sense. I may have stumbled into a slightly better color here, as I noted the midground colors improved, also
I didn't look at noise at all.
Much better composition. You get a better sense of the depth of the gorge.
59MB. Send me a message.
What suggested this to me what that the image I had at this point was lacking both in weight (contrast) and color, compared to what I saw, and this is what the multiply technique can supply.
Thinking about it some more, I realized that I was just too lazy, as the experiments are easy to do. I started with the image after step 2, which is the contrast step:
Dan has developed a technique of multiplying through a mask, which can be done in LAB or RGB, and sometimes you have to manipulate the mask a bit. But it turned out that this image was pretty straight-forward. It starts with what Dan calls a false profile, which is basically assigning a profile with a different gamma to make the image interpreted as lighter than it is, to counteract the effects of the later multiply. I assigned it an RGB profile with a gamma of 1.7. I know this is a pretty terse description, and there may be some threads here that discuss the technique, but I can't find any. If there's interest, we can start a separate discussion.
I took the above image, duplicated the background layer and set it to multiply mode, and duplicated the copied layer:
Then I merged the top two layers, and assigned a layer mask. I'm in RGB. Using Apply Image to the mask for this merged layer, I looked for something that might work. What worked for the mask was the background RGB layer, inverted. I blurred it 30 px, which is the standard blur for this technique:
This is result:
And this is result after applying the LAB AB curve in the earlier post:
I think worth the few more minutes to do it.
Here we go.
Today, in the snail mail, I received a disc from John, with the dng file (2961.dng) for his image of Rio Grand Gorge Bridge....
I opened file in ACR v6.2 ( in Bridge ) via ctrl-R. Ctrl-O opens the file in Photoshop, but lets start in ACR in Bridge.
Here is the image as it appeared in ACR 6.2 without any adjustments
I am not going to include step by step screen shots, instead I rec you get an image of your own, and try to follow along with each step in ACR
There are 10 Panels across the upper right menu boxes in Adobe Camera Raw, including one with an iris aperture ( Basic ), a graph (Curves ), a pair of Triangles ( Sharpening ), etc.
Let's set the Camera Calibration to 2010 by clicking on the 8th panel ( the one to the right of the FX panel ), and under Process, select 2010 Current . The newer version of ACR has new camera profiles since 2003 versions, and these are worth the price of admission to CS5.
Return to the Basic panel #1.
Inspect the image - color does not look bad, with As Shot -> 5050, -1 tint I could live with this one but it is a bit blue.
Trying Auto for white balance -> 6300, +1 tint. Maybe some clouds in the sky-> much warmer tones.
It looks like this might have been a sunny day, so lets try Daylight 5500, +10 tint - definitely warmer than As Shot, but not as warm as Auto. Kind of feels southwest . Lets give Daylight a try.
Using the eye dropper on the sunlit beams under the bridge, -> too bright to to read for white balancing
At 300% viewing, I notice significant R/G CA bordering the windows of the white van, and the vertical bridge girders. The van is too bright to use for WB also, although I suspect it is white or off white.
For a start, I am going to choose Daylight at 5500, + 10 as my initial Custom WB.
Clicking on Auto versus Default on the Exposure settings box, is a modest improvement over the initial flat image from John's 5DMKII
But I think I can do better, so I will click on Default, and begin with the Exposure slider, while I hold down the option ( alt ) key and watch for blown highlights.
The histogram show some blown highlights with the white spike a the far right of the histogram, so I will begin by sliding the Exposure slider to the left a bit, to correct the over exposure - with the option key held down I immediately see that the bridge girders are over exposed, and are not fully captured until I cut exposure by over 1 stop. But by then the sky is getting too dark, so I will settle for about -1.0, which will leave some highlight loss in the brightest bridge girders under neath the bridge, but the Recovery slider will pull these back into range I think.
Now I will set the Black slider, watching the histogram as I do so, again holding down the option key, and moving the histogram to the left, until I am using the entire span of the histogram for the image data - lets use about 20. . Now with the Exposure and the Black slider, set, I will turn my attention to the Recovery slider and try to regain the blown highlights in the bridge girders underneath the bridge. It takes a fair amount of recovery, the sunlight is very bright down there, but a value of 85 for Recovery seems to have corrected most of the blown highlights from the sunlight with just a few specular reflections left. I will settle for a value of 85 which is high, but not that unusual in an image with strong sunlight and shade, like this one.
I played with the Fill slider, but eventually decided to leave it at 0 for now. Maybe I will slide the Brightness slider to the right just a notch to read +55,, to move the histogram a little more centered,.
Now we have a full range from black to white, with correction of the blown highlights, but the image still looks a bit flat to my eye. I will leave the Contrast sider at the default value of 25. I find I rarely change it.
Lets take a second at look at the color again. Using the eyedropper, the sunlit beams under the bridge read 250,250,250 a nice neutral white, and I think steel is frequently painted a light grey or white, which I believe is the case here. The top of the van now reads 237,237,237 - a light grey or off white, that looks good to me. The dark windows in the van vary a bit, but I can find areas of 31,30,35 which is a dark grey with maybe a tinge of blue, from the sky above. The jacket, of the fellow next to the van, reads 237,237,237, so I am reasonably convinced that the color balance is close.
Next I will add some Clarity, mostly because Jeffe Schewe says most images will benefit from a significant nudge in the Clarity slider, and I tend to agree. Lets try 25 to start. We will pass on Vibrance or Saturation for now. You can really see the R/G chromatic aberration on the borders of the van, now.
Lets move to the next panel, the Curves adjustment. The choices are Linear, Medium contrast, or Strong contrast, or Custom. My usual choice is Medium Contrast for most images with a broad tonal range, but each image must be evaluated on its own merits. Curves are easy to add later in PS if needed.
The third panel is Sharpening, and I think this image needs a bit. Move into 100% view, by clicking ctrl-+ until you are at 100%. At 200% it is easily seen that this image is not really perfectly sharp, but it was shot at f20 which a a less than stellar aperture for ultimate sharpness.. If you hold the option key down while adjusting the Amount slider, the image goes to B&W, but I usually do this slider in color mode without the Option key filip. I think we can easily go to 55 or so, without adding artifacts. I will leave the Radius at 1.0, and the Detail at 25 for now. I rarely change them for images of this size. The Masking slider creates a real time mask, as we slide the slider, that shows where there will be no sharpening by displaying black there when the option key is held down, and white where the sharpening will take effect. I like to have the sky black at least in most landscape images. I think a value of 65 works for me - I frequently find I am using between 45+ to 80 or so. Higher for high frequency images, for portraits where you want more limitations of amounts. There is no single right answer, but a range of acceptable amounts,
At 300%, you can see the noise and graininess of this image from a 5DMKII at ISO 320. You can also see some dust bunnies on the sensor, which we will fix in a minute.
For now lets deal with the noise. Raising the Luminance Noise slider to 10, begins to help, as does sliding the Color slider over towards 35-40. These values wil not kill all the noise, but certainly reduce it by 60% or more. You can see the results of your editing by clicking and unclicking the Preview button at the top upper right corner of the image box in ACR.
The next panel is the HSL grayscale panel which I will skip for now. This panel comes up when bringing Smart Objects back to ACR to create greyscale images for contrast control via blending multiple renderings, but we will skip that in this image for now. I will skip Split Toning for now too.
The next Panel we are going to, is the Lens Corrections panel, panel #6. Click on the Profile menu header.
Fortunately the EOS 24-105 is included among the lens with a lens profile in CS5. So we will click on Enable Lens Profile Corrections, leave Setup at the Default setting, and select the Canon EF 24-105 in the Lens Profile boxes. Examining the white van now reveals the R/G CA to be substantially moderated, almost gone, and we can see that the van must have been slowly moving as it is now clearly seen as not stationary by its blurred vertical borders. Usually I do not make any further adjustments and leave the Distortion, Chromatic Aberration, and Vignetting sliders at their default values, but in this image I think the van is rendered better with a setting for Chromatic Aberration of 140 or so. The residual red=green color edges fade further to grey.
The next panel is the Effects panel, for Grain and Post Crop Vignetting, but we will not need these for this image.
Now as I examine this image in ACR, versus, John's final image on dgrin, his image is significantly bluer than mine, but I am content with my numbers for the van and the white jacket of the man next to the van. The sunlit girders read 248,248,248 so they are still neutral also. I could increase the Curve to bring these values closer to 252,252,252, or thereabouts, depending on what your printer can demonstrate on paper. But 248,248,248 is fine for now.
John still has more contrast in the mountains than I have achieved, so I went back to the Curves panel and made some changes and will capture this final rendition as a screen capture with the graph caught as well. Here is the screen capture of my final stage in ACR. The bridge girders and the van now read around 252,252,252 so they are getting very white. Not blown with my printer and paper, but getting very close.
Here is the final image of the Raw converter with my final Curves graph captured
This then is the final image, without any processing in PS other than to convert directly to sRGB for display on Smugmug.. I meant to remove the dust bunnies in ACR, but forgot to go back to it, and thus, will tidy them up later in my pass through CS5.
After a pass through Photoshop, with a simple curve added to the sky and to the mountains, here is my final rendering
I have not done any pass through LAB, nor have I created a B&W image to blend with the original. Reading my steps in RAW conversion seems complex, but takes less than five minutes to do.
The original files of my images can be found here - http://pathfinder.smugmug.com/Other/Large-files/1789718_YdTAs#1117418761_Xh6pR
Moderator of the Technique Forum and Finishing School on Dgrin
The three things that jump out at me, comparing this with my final image:
1. sharpening, which I didn't do.
2. Yours is a lot more colorful. I didn't understand this, reading through your workflow, as there's no obvious color boost step. I'm guessing that you were just better at killing the cast, and mine was bluer than it should have been. The yellows in the mountains really come out in your version.
3. The contrast in the mountains is much better. Here, I'm not sure where that comes from. Perhaps you could comment on this.
Seeing this, I think I could go back and get something close to what you came up with. But the point of the exercise was to show what could be done outside of PS (largely) in the hands of someone sufficiently skilled. I don't think I can do significantly better. Perhaps someone else could...
Yes, I do sharpen in Raw for my Capture Sharpening. I also use output sharpening in Lightroom when I print. I rarely print from PS any longer, I just prefer LR3 that much better for print output, and i use the output sharpening in the Print Module as well.
1) I did local contrast enhancement in the file I uploaded to Smugmug with values of ~ 20 Amount, 50 Radius, 2 Threshold and Edit-> Fade to Luminosity Blend.
2) There is a lot more color available in the Vibrance and Saturation sliders which I specifically avoided so as to not play unfairly. Also the Hue/Saturation/Greyscale panel ( which I also did not use ) offers sliders to increase or decrease Saturation and Luminance and Hue in each of the major colors. This can be quite useful but is easy to overdo as well. I think part of the color is in the final curve which I just re-edited to display in my previous post. You can see the final curve I used in ACR now.
3) In the mountains in my final image, when I took the pass through PS, I selected the green hues with Select >Color Range > clicked on the greens in the mountain, and then did a curves adjustment layer with luminosity blending to bring out a little better contrast in the mountains.
4) I think the yellow in my mountains is because my image is not quite as blue - blue kills yellow and vice versa, just like magenta and green in LAB.
Here is a shot of Rio Grand Gorge I caught a while ago... 2004 with a 10D as a jpg - It was all I knew how to do 6 years ago
Let's do another image if you are willing.
Moderator of the Technique Forum and Finishing School on Dgrin
Yes. This one was fairly simple. I'll look for an post a more challenging one.
It is available for anyone who wishes to download the original file and give it their workflow and post here for comparison.
Please, if you do post an edited version here, post each of your editing steps as well, so that anyone can follow along and see precisely what you did, not just a vague outline
Moderator of the Technique Forum and Finishing School on Dgrin