What Generative Fill Can Do
This video was recorded when Generative Fill first arrived in the Photoshop beta in May 2023. It walks through the main uses of the tool with real photos: extending an image to a new aspect ratio, replacing objects with a text prompt, and building a complete scene on an empty canvas. The tool has improved since, but the workflow shown here is still how it is used today.
Use 1: Extend a vertical photo into a landscape
- Double-click the layer to unlock it.
- Take the Crop Tool and drag the canvas wider than the photo.
- With a selection tool, select the empty area on one side, including a thin strip of the original photo. Repeat on the other side.
- Click Generative Fill, leave the prompt empty, and click Generate.
Photoshop continues the sea, the sunset and the clouds so the join is invisible. Every generation gives three variations in the Properties panel, and the result sits on its own layer with a mask, so you can keep editing. In the second example, a photo with a shallow depth of field, the generated areas keep the same focus: sharp where the subject is, blurred further away.
Use 2: Replace an object with a prompt
In a photo of a girl holding a cat, select the cat with the Object Selection Tool, click Generative Fill, type “lion” and generate. The cat becomes a lion that matches the lighting and the pose. Type “dog” and you get a dog. In a cafe photo, the cup becomes a glass of coffee with the prompt “glass and coffee”, the salad becomes cereal, and a fork becomes a spoon. Each result has three options and you can switch between earlier generations at any time.
Use 3: Build a scene from nothing
Create a new HD document, select the whole canvas, and prompt “beach with sand and blue sky”. Then select smaller areas and add elements one by one: a boat, a beach house, trees. The first trees did not look right for a beach, so the prompt was changed to “coconut tree”, which fixed it. A second blank canvas became “astronaut walking on the moon” with “space station in the background”.
Honest notes from the video
- Results are not perfect. Some edges show cut-off areas, and the from-scratch images are not at the level of dedicated image generators at the time.
- It was a beta. The video expects better quality and more control in later releases, which is what happened.
- For photo editing tasks, extending and replacing, it was already very usable.
Conclusion
Generative Fill changes the routine parts of photo work: fixing aspect ratios, removing or swapping objects, and mocking up scenes. Always work with the three variations and the masked layer rather than accepting the first result, and be specific in prompts when the first attempt misses the context.
Links from the video
- Adobe Creative Cloud: https://www.adobe.com
- Join the members-only content: https://bit.ly/3lR3INj
- Premium tutorials on YouTube: https://bit.ly/3fEEvD4
- Premium tutorials on Udemy: http://bit.ly/2Cqmcgo
