Remove objects from your videos with the Content-Aware Fill panel

Removing an unwanted object or area from a video can be a time-consuming and complex process. With the Content-Aware Fill feature, you can remove any unwanted objects such as mics, poles, and people from your video with a few simple steps. Powered by Adobe Sensei, this feature is temporally aware, so it automatically removes a selected area and analyzes frames over time to synthesize new pixels from other frames. Simply by drawing mask around an area, After Effects can instantly replace it with new image detail from other frames.

The tool gives you the option to help get the fill to blend seamlessly with the rest of the image. The Content-Aware Fill panel contains various options to help you remove unwanted objects and fill transparent areas.

To open the content aware fill panel, select Windows > Content Aware Fill.

• Lighting Correction: Enable this to handle dynamic lighting shifts in your footage. Incorporate lighting correction into fills to cleanly remove objects from footage where lighting changes from frame to frame. You can select from the three different correction strengths - Subtle, Moderate, and Strong. For best results, start with selecting Strong, in case it starts adding too much correction to the footage, go for Moderate or Subtle.
• Range: Choose whether to render the fill layer for only the work area or the entire duration of the composition. Setting it to work area restricts Content-Aware fill from pulling in content outside of the work area.
• Create Reference Frame: Creates a single-frame fill layer frame and opens it in Photoshop. Use reference frames to help Content-Aware Fill learn what the fill layer should look like. For example, if you have a video with complex background and Content-Aware Fill is unable to give you desired results, use the tools in Photoshop such as clone, and patch to create a better result on the reference frame. Once done, generate a new fill layer. Content-Aware Fill transfers the pixels from the reference frame to new frames in the fill layer. For some shots, you can create multiple reference frames at frames where the lighting or camera angle changes. Note that you can also use any-other process to create single-frame layers to guide content aware fill. It also respects content in those layers as a guide. This is a shortcut to do that.
• Generate Fill Layer: Creates a new fill layer. Analysis and rendering progress are displayed at the bottom of the panel. Content-Aware Fill prioritizes analyzing and rendering the frame under the Current-Time Indicator (CTI). While the fill is being rendered, you can move the CTI to a different frame to prioritize that frame, which can help you decide whether the results look correct before the entire fill layer is generated.

How to use Content-Aware Fill

The Content-Aware Fill panel includes multiple features to enable precise removal of unwanted objects from your vidoe. But all these features may not be useful in every situation.

Here is a summary of the steps you need to follow to use the Content-Aware Fill feature:

Lighting Correction in Content-Aware Fill

Handle lighting shifts in footage better with improved Content-Aware Fill. It helps you cleanly remove objects from footage where harsh lighting changes occur throughout your footage and get results that look more realistic without distracting artifacts. Without lighting correction enabled, Content-Aware Fill stays true to the reference frame and does not take into consideration any lighting changes.

Lighting Correction has three strengths:

• Subtle: Use when the lighting changes are not very dramatic. 
• Moderate: Use when lighting changes are moderately dynamic.
• Strong: Use when the footage has a heavy lighting changes. 

The old Content Aware Fill (when Lighting Correction is disabled) stays true to the reference frame and does not take into consideration lighting changes. It is more useful in footages with consistent lighting. Content-Aware Fill currently copies only local information from each image and then finds some global adjustments to fit this information into the target frame. When you use lighting correction on footage that contains lighting shifts such as shadows, highlights, lens flares, and auto exposure changes that are not correctly read, the color of the pixels copied into the hole are not the correct color, and the result do not look clean.

With this feature, you have the option to correct fill lighting inside of Content-Aware Fill like the actual fill layer that’s generated. This eliminates the need to generate several fill layers to correct it and is helpful for lighting and color changes. When lighting correciton is enabled, Content-Aware Fill gives back a fill layer which is seamlessly integrated with the rest of the footage. This Fill layer is lighter, but with the same quality.

This does not impact the speed of the performance than without lighting correction. The overall workflow is faster when working with footage with variable lighting because you don't need to take additional steps to clean up the results.

Work with Lighting Correction

• Start with analyzing the shot. See the lighting changes, the movement, and then decide what needs to be masked out.
• From all the Fill Methods, you don’t need to use Lighting correction for Edge Blend.
• Select Strong first. In case it starts adding too much correction to the footage, go for Moderate or Subtle.
• Once the settings are done, then click Generate Fill Layer.

Best practices to work with Lighting Correction

Situations where the Strong might not work well are footages with flickering lighting condition such as on grass or sandy shots that reflect a lot of light. The Strong can can color correct too harsh which can introduce more flickering. In those cases, try using Subtle or Moderate strength.

Examples

You can use the following examples to see how the different settings work in different situations:

Example 1: Footage with moving camera

This fill method is best used on footage with moving camera as it replaces the object entirely with what is behind it. It fills the transparent area by taking pixels from the current and surrounding frames, and outputs a seamless looking frame.

Example 2: Footage where the background is a flat surface without much detail

Example 3: Work with your footage in Photoshop

There can be footage where Content-Aware Fill alone may be unable to achieve the desired results. For example, a footage with water or areas where there is varied light and texture. In such situations, you can take advantage of the Adobe Photoshop tools such as Clone Stamp to finetune your footage.

In the above example, we remove the girl from the footage where the background has varied light and texture.

Content-Aware Fill settings

To open the Content-Aware Fill Settings dialog, click the menu icon (three bars) in the title bar of the Content-Aware Fill panel, and choose Content-Aware Fill Settings.

• Output Depth: The color depth, in bits per channel (8-, 16-, or 32-bpc), used to write the fill layer sequence files. By default Output Depth inherits the project's color depth.
• Output Location: Choose where After Effects After Effects saves the fill layer sequence files. By default the files are saved to a folder named Fills in the same folder as the project file.
• Path Type: Project Relative saves the files to a location relative to the project file. Absolute specifies the full path to the folder.
• Path: The path to the folder. For Absolute, this is the full path to the folder. For Project Relative, this specifies the path relative to the project file. A single dot (.) in the path represents the folder where the project file is located. Use two dots (..) to go up or down one level.
• Create Photoshop Reference Frame: Saves reference frames as a PSD file, and opens it in Photoshop. Enabled by default. If disabled, saves the reference frame as a PNG (8- or 16-bpc) or EXR (32-bpc) file and opens it in the system default application for those file types.
• Create Photoshop Sequence for Output: Saves fill layer sequences as PSD files. Disabled by default. If disabled, fill layer sequences are saved as PNG (8- or 16-bpc) or EXR (32-bpc) files.
• Auto-manage Unused Fill Footage: When a new fill layer is generated, you will be asked if you want to delete fill layer footage not being used in a composition. Disabled by default.

Delete Unused Fill Footage

To manually delete fill layer footage that is not being used by a composition, click on the menu icon (three bars) in the title bar of the Content-Aware Fill panel, and choose Delete Unused Fill Footage.

Important points and tips

• You can select an area in a footage using using any method that creates transparency, including:
  • Mask
  • Rotobrush
  • Eraser tool
  • Keying method such as Keylight method
  • Keyframing
• If you have 4K footage in an HD comp, use footage that is already scaled down via transcoding for better performance.
• Remove more complicated items in sections. For example, you can remove a person and their shadow separately more effectively than removing both at once.
• Smaller masks take less time to process over larger ones.
• The feature might give different results for a complex moving texture like water or areas where there is varied light and texture.
• If the mask is feathered, Content-Aware Fill blends in the filled region with the original content based on this alpha value (semi transparency). This is very useful when you use the panel, it causes a noticeable boundary between the filled region and the original content
• Generate Fill Layer renders sequences of images that are placed in the Timeline. These files may take up considerable space on the hard drives, depending on the kind of footage, and duration of your sequence.
• In certain cases, Content-Aware fill alone may be unable to remove object with precision. To improve results, create a reference frame and take it into Photoshop, and use the tools there to fine-tune the results. 
• At the time of shooting a footage, consider the background and the foreground of your composition to reduce editing time during post-production.