Hi, how can we help you today?
Hi, how can we help you today?
Expand is Scenario's outpainting tool, and enables you to seamlessly extend your images while maintaining excellent consistency - whether for a scene/environment or a character. It's the tool to reach for when your generated images need "more space around them” - giving you more freedom to crop or adapt to different formats (mobile, tablet, etc).
When generating an image, the composition might not always turn out as planned. For example, the main subject might not be positioned exactly where you want it, or you might need more space or content around it. Expand is the perfect tool for these situations and is ideal for:
Extending backgrounds for immersive scenes (e.g., a platformer background endlessly).
Adding new elements to existing generated visuals (e.g. adding the lower body section of a generated character portrait).
Creating assets for different formats (e.g., mobile vs. tablet or PC).
Expanding part of a subject (e.g., adding the lower body to a generated character portrait).
Scaling assets to larger sizes without sacrificing quality or coherence.
You Can Access Expand - Two Ways:
From an Existing Image
Open an image in your Scenario library or generation screen (or even via the search menu), click the three-dot menu (ellipsis) in the top-right corner of the thumbnail, and select "Expand."
From the Main Scenario menu (to the left)
Navigate to the left sidebar (or click “m”). Under the "Image” section, click "Expand." Once the Expand screen is loaded, simply upload an image or pick one from your Scenario library to begin. You can also drag and drop an image to serve as an input, either from your computer or from the gallery to the left.
Once an image is loaded, three distinct areas appear in the image section:
Core Area: Stays unchanged, but acts as a reference for the added content (especially the style).
Chessboard Area: The empty space where new content will be created, adjustable to your preferred size.
Overlap Area: A highlighted mask between the two other areas, indicating where blending occurs on your original image. The content within this overlap area will be altered during the outpainting process.
Expand offers a variety of settings to shape your output, giving you control over the generation process and ensuring the added content stays consistent. Here’s how to set it up for the best results:
Step 1 Upload or Select Your Image
Start with a high-quality image (ideally 1024x1024 pixels or higher) for optimal detail retention.
Adjust Input Location
Decide where your original image sits within the expanded canvas; center, top/bottom or left/right. This frames your starting point and leaves the rest of the area for outpainting.
Set the Output Scale
Output Scale defines the size of the output image. As you adjust the output scale, the exact new dimensions of the output will appear live under the image. A larger scale (e.g., 200) creates more new content, while a smaller scale (e.g., 120) keeps extensions subtle. Preview the canvas to ensure it aligns with your goal. There are maximum output dimensions (2048x2048 for a square ratio, for instance). Scenario will automatically adjust settings to keep the output within the maximum dimensions.
Choose an Aspect Ratio
Select a predefined aspect ratio (e.g., 1:1, 16:9, 4:3) from the dropdown to match your project’s requirements.
Adjust Input Scale
Resize your original image within the expanded area using the input scale setting. Shrink it to emphasize new content or enlarge it to maintain its dominance in the final output. When modifying output scale, input scale will also automatically adjust.
Adjust the Overlap Setting (Important)
Adjust the overlap slider to control the size of the mask. A larger overlap (e.g., 30–50%) smooths transitions by blending more of the original image into the new content, reducing visible edges.
Inference Steps
20 is the recommended default. Inference steps balance speed and quality. For more detailed outputs, increase the steps, but it will take longer and consume more credits.
Seed (Optional)
Enter a specific seed value to replicate results across generations. This ensures consistency when iterating or sharing workflows. Leave it random for creative variation.
Write an Effective Prompt (Important)
You can craft a prompt to guide the outpainted content (that will be created in the chessboard area). You’ll find several options:
First, when loading a Scenario-generated image to Expand, the prompt will automatically load with the prompt used to generate the image (if any).
If you have uploaded an external image, you can turn the reference image into a prompt by clicking “Turn your reference image into a prompt” (blue link in the prompt box).
Or you can craft your own prompt, e.g., “a sprawling desert with distant mountains” or “a cyberpunk cityscape with neon lights.” Start simple, then add descriptors (colors, mood, objects) for precision. Focus on what matters and what needs to be generated in the area.
Once you’re ready, click "Generate" to create your expanded visual. Review the results in the content panel to the left, where you can zoom, compare (hit “c”), or select it for further edits.
Process: Select or upload your character image on the Expand page. Set the input location to the top, the new content will be generated to fill the surrounding empty areas, adjust the aspect ratio to 2:3 and scale the output to 120% with a 10% overlap.
Result: The expanded image now features a full body shot of the samurai in a vibrant, neon-lit street scene with additional background elements.
Process: Select or upload your scenic image on the Expand page. Set the input location to the center, the new content will be generated to fill the surrounding empty areas, adjust the aspect ratio to 1:2 and scale the output to 100% with a 10% overlap.
Result: The expanded image now features a wider view of the serene mountain landscape with two Buddha statues, enhanced by a vibrant sunset, prayer flags, and additional path details.
Process: Select or upload an image of your game asset on the Expand page. To create a vertical composition, adjust the aspect ratio to 9:16 and generate the image. Then, to produce a horizontal format, simply change the aspect ratio to 16:9 and generate again, allowing new content to fill the surrounding areas in each format.
Result: The expanded image now presents the superhero team in a vertical and horizontal layouts, showcasing a the cityscape background with all the building details, preserving the vibrant colors and poses of the characters
The expanded (outpainted) image might display some defects like visible edges at the boundary of the mask, or the expanded content might not meet your expectations. In these cases, you might want to adjust these elements to improve the outcome:
(i) Adjust Parameters, especially the Overlap Area
If transitions look jarring or seams are visible, increase the overlap setting (e.g., from 15% to 30%) to blend more of the original image into the new content. Experiment with output scale or input location if the expansion feels unbalanced.
(ii) Adjust the Prompt
If the new content doesn’t match your goal, you might need to refine your prompt. Make it longer and more detailed (e.g., “a dense jungle with tall vines, misty air, and glowing flowers”) for clearer guidance, or shorten it (e.g., “a quiet meadow”) to leave more creative freedom to the AI. Use Prompt Spark (sparkle icon) to rewrite it if needed.
After tweaking settings or the prompt, generate again. Compare results side-by-side in the content panel to identify what works best, and you can reuse the prompt or parameters from the three-dot menu as needed.
Expand integrates seamlessly into Scenario’s toolkit, allowing you to combine it with other features into versatile creative workflows like these (among others):
Improve Image Quality: After expanding an image, use "Enhance" to boost resolution and sharpen details. Use Style Images for even more refined results tailored to your style (link)
Refine Details: Click "Retouch" to launch the Scenario Canvas, where you can inpaint (add or remove) elements using specific custom-trained models.
Use as Reference (Image-to-Image): Apply your expanded image as a reference with a custom model to guide further generations.
Don’t forget to tag or save your expanded image to a Collection for quick access and smooth team collaboration.
Was this helpful?
Quentin