About Background Removal
Abyssale's AI-powered background removal helps you isolate the main subject of an image by automatically removing everything around it, directly inside the Design Editor, on the layer you're already working on.
Note: Each background removal consumes AI credits. The exact cost depends on which model you choose and is shown next to the generate button before you launch, so you always know the cost up front.
Accessing Background Removal
Select the image layer you want to process
An AI toolbar appears at the bottom of the canvas, click Remove Background
The toolbar transforms into the Remove Background context, where you'll pick a model and any additional options before generating
Choosing a Model
Click the model dropdown to see all available background removal models. Each is listed with its relative AI credit cost and, where relevant, a badge such as New or Top Pick:
Model | Description |
Bria RMBG 2.0 (default) | High-precision background removal trained on licensed data, a strong, safe default for most images. |
Birefnet | Open-source segmentation using a bilateral reference architecture, a solid, freely-licensed alternative to the default model. |
Ideogram | Photorealistic generation with best-in-class handling of fine in-image detail, a good choice for tricky edges like hair or fur. |
PixelCut | An all-in-one AI editor built for e-commerce, with background removal as one of its core tools, a good fit for product photography. |
imageUtils | A straightforward, no-frills background removal option focused purely on removing the background from your layer. |
If your first result doesn't isolate the subject cleanly, especially around fine detail like hair, fur, or transparent materials, trying a different model is often faster than adjusting the same one repeatedly.
API note: via the API, remove_bg is supported on asynchronous generation. It still functions on synchronous generation but is deprecated there and not recommended, because background removal can exceed the 10-second synchronous limit. The model is passed as remove_bg_properties.model.
"Skip if already transparent" has no API property for now. It is a design-level setting: when the toggle is enabled on the layer, it applies to every generation of that design, including generations triggered via the API, and the API payload cannot set or override it.
Additional Options
Alongside the model choice, the Remove Background context gives you a couple of extra controls to fine-tune the result:
Trim Transparent Pixels
Optionally crop the image down to its subject after the background is removed:
Automatically crops away excess transparent space around the edges
Centers the remaining subject in the layer
Useful for keeping compact, efficient designs, especially when the subject occupies only a small part of the original image
Via the API, this option is remove_bg_properties.remove_bg_crop.
Skip if Already Transparent
Avoid spending credits on images that don't need background removal, useful when your source images mix already cut-out PNGs and regular photos, especially in batch generations:
When enabled, Abyssale first checks whether the image already contains a significant amount of transparency.
If it does, background removal is skipped and the image is used as-is. Only the check is charged: 1 AI credit.
If it doesn't, background removal runs normally. The cost is then the selected model's credit cost + 1 AI credit for the check.
Trim Transparent Pixels still applies either way: if both options are enabled, an already-transparent image is not re-processed but is still cropped down to its subject.
You can toggle this option from the Remove Background toolbar when generating, or later from the AI generation section of the Style panel, next to Trim Transparent Pixels.
Note: the transparency check itself always costs 1 AI credit when the option is on, even when removal runs afterwards. Turn it off if you know none of your images are transparent.
Manual Selection
For images where you want to guarantee a specific area is kept:
Draw a box directly on the image to indicate the area to preserve
Useful for complex images where the automatic detection might miss part of the subject, or include something you want removed
Note: Manual selection is only available inside the Design Editor, not via the API
Keeping Background Removal Consistent Across Variations
Once you're happy with a result, you can apply it automatically to every future variation of that image layer, for example, across multiple formats or a bulk generation, instead of repeating the process manually each time:
After generating, you'll be asked whether to save this as an AI action for future assets
You can also review or turn this on later without starting a new generation: with the layer selected, open the AI generation section in the Style panel on the right, where Remove Background appears as a toggle alongside your other AI settings (such as Auto Focus and Text to Image) for that layer
This is particularly useful for batch-processing multiple product images or keeping a consistent, clean look across a set of designs.
Best Practices
Image quality
Use high-resolution source images for the best edges
Look for good contrast between the subject and its background
Avoid busy or cluttered backgrounds where possible
Model and manual selection
Start with the default model, but try another (especially Birefnet or Ideogram's) if fine detail like hair or fur isn't isolating cleanly
Use manual selection for complex images where you need to guarantee a precise boundary
Batch processing
Turn on the Remove Background toggle in the Style panel when working across many variations
Check the first few results before running it across a large batch, and adjust your model choice if needed
Discover More AI Features
Abyssale Intelligence offers a range of AI-powered tools to enhance your design workflow.