Solving the Dilemma of Optimal Lighting – Sky-Earth Stitching Technique

Written Beforehand

When we were still beginners and started reading photography textbooks, we would read something like this: “The golden hour for photography is the half hour before sunrise and the half hour after sunset.”

If we were just hanging out with a few friends on the roof, drinking iced Coke and chatting, it would be fine to head out before sunset.

But if we were going to shoot a rooftop photography piece, we would have to consider lighting, composition, and camera positioning. The best approach is to save excellent rooftop photography works, scout locations in advance, prepare thoroughly, and choose a day with amazing weather to set out.

This issue focuses on this dilemma in rooftop photography: the best time for the sky is half an hour before sunset, when the clouds start to be tinged with golden light; while the best time for the landscape is half an hour after sunset, when the city lights come on and the car trails take shape.

To perfectly capture the sky and the landscape in a single work, you need to composite them.

Here’s How You Can Do It

The two images above are rooftop photography works from Queenstown, New Zealand, last year. The camera position was on the Gondola viewing platform.

The former was taken at 7:30 pm, the latter at 8:30 pm. It is obvious that neither looks good on its own, and there will always be a part of the image that has a significant flaw, making it monotonous and lacking in color. Therefore, I chose to composite them, stitching together the most aesthetically pleasing parts of each image into a single image, resulting in the following:

(Photography by: Ding Yifeng)

Importing to the ACR Interface

Next, I’ll detail the process of my composite and some tips to keep in mind.

First, import the raw format of the two photos into ACR separately. Adjust the color temperature, tone, highlights, shadows, and other parameters of the photo in the ACR interface. The point I want to emphasize is that you don’t need to force the color temperature of the two photos you intend to composite to be consistent. This is a common mistake made by many beginners.

Stacking the Photos

Next, import the two photos into the stack. It’s crucial to note here that the two photos were taken an hour apart, and my camera position was on a busy viewing platform, so some slight tripod shake is inevitable. Therefore, please make sure to check “Auto Align”.

After importing, observe the photos and you’ll find that the top photo we want is just its lights and car trails, which are relatively bright. We can then use the luminosity mask technique to select the area, extracting the brighter part as a highlight channel. Here I’ll use the Starstail plugin to create a luminosity channel first.

In the channel interface, you can see that clicking H2 (Highlight 2) selects the ground lights and car trails.

Next, go back to the layer interface and create an inverse mask. At this point, you’ll see the entire night scene disappear. Since we only want the lights and car trails in the night scene, the next task is to carefully erase them with the protection of the channel selection.

The following two images show my process of erasing the mask. Using the channel, the highlights are automatically selected, and then the brush tool is used to paint on the mask.

There are also techniques for erasing the luminosity mask, mainly involving applying low opacity brush tools layer by layer, ensuring the mask effect has a gradual transition and avoiding distortion caused by abrupt changes in style.

Most importantly, with the protection of the channel selection, the brush tool will only affect the content within the highlight selection! This is the core significance of the luminosity mask, and you must try it yourself after watching the tutorial to feel the convenience of this technique.

Then merge the two layers, and you’ll see that the goal of having the best lighting for the sky and landscape appearing on the same screen has been initially achieved. Then comes the detail processing we are pursuing to improve the texture.

Here I still use Google’s Nik Collection filter system – Color Efex Pro filter plugin first.

Nik Collection is a suite of free filters developed by Google, the most commonly used ones are Efex Pro (color and brightness adjustment), Dfine 2 (noise reduction), Sharpener (sharpening), etc.

First is “Soft Contrast,” which is used to initially enhance the overall image quality. With simple settings, all three parameters are 5. Then there is no risk of the sky being broken, so you don’t have to worry about using masks. (The technique of avoiding and preventing color breaks will be discussed in future tutorials. For now, just remember the conclusion: large areas of plain skies are prone to color breaks. That’s all you need to remember. This photo obviously isn’t a plain sky, so no need to worry.)

Next, use “Tonal Contrast” to enhance the texture of key areas. This filter is easy to overdo, so the parameters are generally set to lower values first.

In this image, the key areas to enhance are the town’s buildings and the distant snow-capped mountains, so I use the filter’s automatic mask recognition feature to select them.

To further emphasize the focal point, there’s a handy filter called “Darken/Lighten Center.” Place the crosshair in the center on the town’s buildings, which can reshape the light and shadow, making the viewer’s gaze more layered and focused on the focal point.

I also observed that the details in the dark areas of the forest weren’t very clear. We could choose to use a curve to erase a mask to enhance this, but in the Efex interface, there’s a more useful filter called “Detail Extractor.” It ensures image quality while enhancing details.

The core of the Detail Extractor is to automatically recognize dark areas and restore details while maintaining image quality, which is more intelligent than using a curve mask to brighten up.

Then click “OK” and go back to the PS interface. You can see that at this point, the image quality has already undergone a qualitative improvement compared to the beginning. Landscape photography is like this, constantly accumulating small steps to achieve the final qualitative change.

Next, I want to enhance the colors of the evening sky and car trails, again using my favorite “Skylight” filter. After entering the Efex interface and turning it on, the effect is roughly like this, and then you go back to the PS interface.

Commonly used filters in Efex Pro, such as Soft Contrast and Tonal Contrast, adjust the image’s saturation, contrast, sharpness, etc., through built-in algorithms to achieve the purpose of enhancing image quality. The difference is that Soft Contrast is more gentle and can be applied globally, while Tonal Contrast is more forceful and generally only used carefully on the main subject of the image.

In the PS interface, you can see that the Skylight filter is applied to the entire image, which is not what we want. I only want it to affect the highlights.

So I once again use the Starstail plugin to create a mask so that the Skylight filter only affects the highlights. This way, the colors of our sky and car trails are enhanced, further highlighting the effect of cold and warm contrast.

Starstail plugin was developed by photographer “Half Island Snowman”, mainly for one-click completion of the tedious operation of creating highlight, shadow, and mid-tone selections. The foreign counterpart plugin TK is also good and can be tried out.

To improve the image’s transparency, I create a curve to increase the contrast.

The biggest concern with increasing contrast is overexposure of highlights and loss of detail in shadows, so I use the Starstail plugin to create a mid-tone mask so that the contrast curve doesn’t affect the highlights and shadows.

At this point, our image quality enhancement work is complete. The next step is the noise reduction and sharpening work that I do every time. I also use plugins to automatically complete these two operations, and the results are excellent.

Noise reduction uses Dfine 2.

Dfine 2’s algorithm is very mature, and unless it’s a Milky Way scene, there’s no need for additional manual adjustments.

Sharpening uses Sharpener.

Finally, we need to observe and remove any image flaws.

Since this is a distant shot of buildings, there are no wide-angle distortion issues. However, if we zoom in on the image, we can see that there are many dirty spots in the sky due to the camera’s CMOS being unclean, which need to be removed using the patch tool. This has no technical content, just carefulness and patience are required.

Next, I think the glass building in the forest is very distracting and want to eliminate it. Select the Lasso tool and select this area.

Then use “Fill” – “Content-Aware,” and you can replace the distracting building with the forest, thus avoiding distracting the audience’s attention from the main subject.

Then crop out the extra parts resulting from auto-alignment, and you can save and complete the output to get a satisfactory rooftop photography work.

Generally, we would also soften the clouds in the sky and the water surface in the landscape. This operation is more complex and difficult to grasp the degree of softening, so this article only introduces the concept without detailed elaboration.

Alright, have you learned about the post-processing of rooftop composite? If not, feel free to leave a comment. ~

Note: This article and image are from “Ding Yifeng.”