You've been there. You have the perfect shot for your website header or a social post, but it’s just too narrow. Or maybe you need it to be a square for Instagram, but the original is a tight vertical crop. For years, the only way to fix this was a mix of the Clone Stamp tool, a lot of prayer, and maybe some creative blurring. It was tedious. Honestly, it usually looked like a smudge-filled disaster if you weren't a high-end retoucher.
But things changed fast.
✨ Don't miss: Getting the iTunes for Windows download right in 2026: Why it still exists and how to avoid the glitches
When Adobe dropped Generative Fill into Photoshop, the game for how to expand a picture in Photoshop shifted from manual labor to basically telling an AI what you want to see. It’s powered by Adobe Firefly, and while it feels like magic, there’s actually a specific workflow you need to follow if you don’t want the edges to look "off."
The Crop Tool is secretly your best friend
Most people think the Crop Tool is only for making things smaller. That's a mistake. If you want to know how to expand a picture in Photoshop, the Crop Tool is actually the starting line.
Open your image. Hit 'C' on your keyboard. Now, instead of dragging the corners inward, pull them out. You’ll see that checkerboard pattern—the transparency—bleeding out from your original image. This is your new canvas.
Before you commit, look at the top bar. You’ll see an option for Fill. Most pros used to set this to "Content-Aware Fill," which works okay for simple patterns like grass or sky. But if you want the "wow" factor, you’re looking for Generative Expand.
When you drag those handles out and hit Enter, Photoshop doesn't just stretch the pixels. It looks at the context of your photo—the lighting, the depth of field, the texture—and generates brand-new pixels to fill that empty space. It’s wild. One second you have a tight portrait; the next, you have a wide-angle shot with a realistic background that never actually existed.
Why Generative Expand beats the old ways
Content-Aware Fill was revolutionary in 2010. By 2026 standards? It's kind of primitive. Content-Aware works by sampling existing pixels and repeating them. It's great for removing a stray power line, but it’s terrible at "inventing" a new mountain range or continuing a complex brick wall.
✨ Don't miss: How to Reset Lenovo Smart Clock When It Stops Listening
Generative Expand is different because it understands objects. If your photo cuts off halfway through a coffee cup, Content-Aware might just smear the brown color. Generative Expand knows what a cup looks like and finishes the handle for you.
Mastering the Generative Fill workflow
Sometimes the automatic expand doesn't get it quite right. Maybe the perspective is slightly skewed, or the AI added a random bird where you didn't want one.
That’s when you go manual.
- Use the Rectangular Marquee Tool (M).
- Select the empty area you just created with the Crop Tool.
- Pro tip: Make sure your selection slightly overlaps your original image. About 20 to 50 pixels is the sweet spot. If you don't overlap, Photoshop won't know how to blend the new pixels with the old ones, and you'll end up with a visible seam.
- Click the Generative Fill button that pops up in the Contextual Task Bar.
- Leave the prompt box empty if you want Photoshop to just "guess" based on the surroundings. If you want something specific, like "snowy mountains" or "blurred city lights," type it in.
The cool thing here is the Variations panel. Photoshop gives you three versions of the expansion. Sometimes the first one is a hallucination—I’ve seen it turn a sidewalk into a river for no reason—but usually, one of the three is spot on.
Managing the resolution trap
Here is the thing nobody tells you about how to expand a picture in Photoshop using AI: resolution limits.
Adobe Firefly (at least in its current iterations) generates at a maximum of 1024x1024 pixels for a single generation. If you are trying to expand a massive 50-megapixel RAW file for a billboard, and you expand a huge section all at once, the new area might look a bit soft or blurry compared to your tack-sharp original.
To fix this, expand in smaller chunks. Instead of doing the whole left side at once, do it in sections. It’s a bit more work, but it keeps the pixel density high.
🔗 Read more: Dyson Purifier Humidify+Cool Formaldehyde: Is it actually worth your money?
When AI fails: The manual fallback
Sometimes you’re working on a document where you can't use AI—maybe for legal reasons, or maybe you're working offline. You still need to know how to expand a picture in Photoshop using the "old school" methods.
Content-Aware Scale is the forgotten middle child of Photoshop tools. Go to Edit > Content-Aware Scale. This tool is different from a regular transform because it tries to protect areas with lots of detail (like people) while stretching areas with low detail (like sky or water).
It’s not perfect. If you push it too far, people start to look like they’re in a funhouse mirror. To prevent this, you can actually "save" a selection of your subject first.
- Select your subject.
- Go to Select > Save Selection and name it "Subject."
- Then, when you use Content-Aware Scale, select "Subject" in the Protect dropdown menu in the top bar.
Now, when you drag the image wider, Photoshop will stretch the background but keep your subject’s proportions locked. It’s a lifesaver for quick layout changes.
The "Hidden" Technique: Outpainting with the Brush
If you're an artist or someone who likes a bit more control, you don't have to rely on the Crop Tool. You can use the Generative Brush.
Basically, you expand your canvas size (Image > Canvas Size), and then you literally paint over the transparency where you want the image to grow. This is particularly useful if you only want to expand a specific part of the image, like making a tree taller without changing the width of the whole photo.
Common pitfalls to avoid
I see people make the same three mistakes constantly.
First, they don't check their layers. Generative Fill creates a new "Generative Layer." It's non-destructive, which is great, but it can make your file size balloon if you have twenty different versions of an expanded background hidden in your layers panel. Flatten or convert to Smart Objects once you’re happy.
Second, lighting mismatches. If your original photo has a strong light source from the left, sometimes the AI gets confused and puts a shadow where there should be a highlight. You can fix this by using a clipped Curve adjustment layer over the generated area to manually match the exposure.
Third, the "AI Look." Sometimes the generated texture is just a little too smooth. To make it blend perfectly, I usually add a tiny bit of Gaussian Noise (Filter > Noise > Add Noise) to the generated layer. Just 1% or 2% is often enough to match the digital grain of the original camera sensor.
Putting it into practice
Getting a handle on how to expand a picture in Photoshop is really about experimenting with the balance between AI and manual tools.
If you're ready to try it right now, start with a simple landscape. Use the Crop Tool to pull the sides out by about 20%. Let Generative Expand do its thing. If it looks "mushy," undo it and try selecting smaller portions with the Marquee tool, overlapping the original image by a few dozen pixels.
Once you have the background expanded, zoom in to 100%. Look at the transition line. If you can see where the old photo ends and the new one begins, use a soft eraser or a layer mask on the generated layer to blend that edge.
Next steps for your workflow:
- Open a photo with a simple background (sky, sand, or a plain wall).
- Practice "protecting" a subject using the Save Selection method before using Content-Aware Scale.
- Compare the results of Content-Aware Scale versus Generative Expand to see which one maintains the textures better for that specific shot.
- Experiment with descriptive prompts in the Generative Fill box—instead of leaving it blank, try adding specific elements like "autumn leaves" to see how the AI integrates new colors into your existing lighting.
By mastering these layers of expansion, you're no longer limited by the frame you shot in the field. You can adapt any image to any platform, keeping the quality high and the "fake" look at bay.