Content-Aware Fill has long been one of those features in Photoshop that feels like magic. It’s perfect for cloning or fixing larger areas of an image without having to re-sample over and over with the clone stamp or healing brush, or to define the source area using the “patch” tool. But this tool sometimes grabs the wrong source material and creates epic failures. So simplicity is great until is doesn’t work. Adobe gets this and they gave us a lot more control in Photoshop CC 2019 with a new version of Content Aware Fill.
The new version is a huge improvement. The biggest change is that you can easily specify which parts of the image can or cannot be used as source material. This eliminates the biggest problem, where random objects show up in the fill. Another critical change is that you get a live preview as you make changes to both the source material and areas targeted for fill. There has always been a bit of unpredictability in the tool, and these live previews allow you to more quickly explore different settings to find exactly what you need. And finally, there are several more controls to allow you to specify whether or how source material is manipulated to deal with specific issues, such as a repetitive pattern disappearing into the distance.
Both the old and new version of the tool are available in Photoshop CC 2019. To access the old tool, you can click <shift>-<delete>, <shift>-<F5>, or go to Edit/Fill and set the contents dropdown to “content aware”. The new tool is in a different place and has no keyboard shortcut by default (though you can customize your keyboard shortcuts to use it if you prefer). Just go to Edit / Content-Aware Fill (not simply “Fill”). If either of these items are grayed out in the menu, make sure that you have selected a normal pixel layer (Smart Objects are not supported) and that you already have an active selection (which you can change once you enter the new tool, but you will need an active selection to use it).
The workflow for Content Aware Fill is a bit iterative, but generally works looks this:
- Create a hard-edged selection with the lasso tool (<L>) and go to Edit/Content-Aware Fill.
- Click the circular reset arrow for “fill settings”, as the default settings are usually the best place to start.
- Refine the green sampling areas with the brush (accessed by clicking <B> or the icon in the toolbar at left). If you do not see any green, make sure “show sampling areas” is check and click the circular reset arrow next to that checkbox. Use the minus option in the tool options at top left (or hold the key to use the opposite mode) to see a minus cursor and paint out any areas of green that should not be used for Content-Aware Fill. You can add additional areas with the brush in plus mode if remove too much or need to include something outside the original green area. Note that the preview will update after any change to the green areas.
- You may refine the lasso selection as needed with the lasso tool in the toolbar or by clicking <L>. This is of course helpful if your original selection wasn’t perfect, but also serves another purpose. Making even very tiny changes to the selection can dramatically change the fill preview. So try making very small changes to experiment with different fills, even just a pixel far removed from any problem areas. Think of it as a way to provoke random changes in the results. This is a great way to improve the quality of the fill after you have refined the green areas.
- If you need to zoom or move around, you can use the icons in the left toolbar or the standard <Z> to zoom and <space> temporary hand tool. Note that checking the “scrubby zoom” option in the zoom tool settings allows you to click and drag to zoom smoothly in or out of the image window on the left, but that the preview of the fill in the right window unfortunately ignores this setting. The easiest way to zoom in and out of the preview window is to use the zoom slider at the bottom of the preview.
- Once you’ve done everything above, you may wish to experiment with different options in the “fill settings”. They generally behave in a way consistent with their naming, but the results can fairly unpredictable, so experimentation is the best approach.
- Color Adaptation specifies how much the source material can be changed in tone or color to fit its new surroundings.
- Rotation Adaptation specifies how much the source material can be twisted in the fill. Unlike color adaptation, it doesn’t do a great job at matching its surroundings, which makes the results highly unpredictable (ie, its much more about rotation than adaptation). But it can be very useful if you need to change orientation, such as in rocky textures that may look like too much of a repeating pattern without some variation.
- Scale allows the source material to be resized, which can be helpful for patterns that change size, such as a brick wall shot on an angle.
- Mirror allows the source material to be flipped, which is great for sampling from something which is symmetrical and opposite to the area you need to fill. This is good for things like architecture and reflections.
- Before clicking “ok” choose which output you’d like: new layer (great for working non-destructively with minimal increase in file size), current layer (best avoided unless you feel very confident in the results), or duplicate layer (helpful if you want to keep more source material for cloning later, but this generally causes a wasteful increase in file size and you can always clone from the original layer).
You may find that you get a great, but not perfect result. That’s fine. Once you’ve finished with Content-Aware Fill, you can then use the spot healing or clone tools to further improve the results. The goal is to have Content-Aware Fill do about 90% of the work for you, which should ultimately give you faster and better results.