AI models on Craze
Every video, image, voice, music and sound model Craze can direct, with what each one makes and what it costs.
Every video, image, voice, music and sound model Craze can direct, with what each one makes and what it costs.
Starts at$0.006 per image
BiRefNet's high-quality background removal model. Remove backgrounds from photos while preserving fine detail in hair, fur, and complex edges. Available on every Craze plan.
BiRefNet uses bilateral reference learning to preserve fine details while removing backgrounds.
BiRefNet produces clean, accurate masks regardless of background complexity.
Upload an image and get a transparent PNG ready to use.
People, products, animals, plants, objects, and combinations all work well.
Craze creatively jams, researches and plans with you.
Craze generates anything imaginable. Video, audio, music, motion graphics and more.
Craze watches your footage and builds a final cut. With a real editor if you need it.
Every model here runs on every Craze plan, including Free, and Craze can use a different one for each shot.
| BiRefNetZheng Peng | GPT Image 2.5 SunburstOpenAI | Nano Banana 2.1Google | Nano Banana 2Google | |
|---|---|---|---|---|
| Makes | Image to image | Text + Image to image | Text + Image to image | Text + Image to image |
| Quality | — | Up to 1K | Up to 4K | Up to 4K |
| From | $0.006 per image | $0.135 per image | $0.036 per image | $0.071 per image |
| Task | Inputs | Resolution | Aspect ratios | Sound |
|---|---|---|---|---|
| Background removal | Image | Follows the source | Follows the input | None |
$0/ month
$19/ month
$59/ month
$199/ month
BiRefNet is a high-quality background removal model that isolates subjects while preserving fine detail. It's especially good at tricky elements like hair, fur, foliage, and complex edges that cause problems for simpler background removal tools. On Craze, BiRefNet is available on every plan, including Free.
BiRefNet uses bilateral reference learning to preserve fine details while removing backgrounds. Hair strands, fur texture, leaves, branches, and intricate edges come through cleanly. That precision matters for product photography, portraits, and any subject with complex boundaries. A portrait's hair stays detailed. Foliage stays crisp.
BiRefNet produces clean, accurate masks regardless of background complexity. Textured backgrounds, gradients, or busy scenes don't cause artifacts or edge bleeding. The mask is crisp and precise. A product against a patterned background isolates perfectly. A person against dappled light separates cleanly.
Upload an image and get a transparent PNG ready to use. No masking, no manual refinement needed. For most subjects, BiRefNet delivers clean results immediately. Most images process in seconds, making batch work feasible.
People, products, animals, plants, objects, and combinations all work well. BiRefNet adapts to whatever you throw at it without requiring parameter adjustment or specialized prompts. Same quality regardless of subject type.
You upload an image and Craze removes the background.
The process is fast. Most images process in seconds, making batch removal feasible.
BiRefNet is charged per image at $0.006 per image. The cost is the same regardless of image size, complexity, or subject type. No pricing axes. One cost per image. The cost calculator on this page shows your exact estimate.
every Craze plan, including Free include BiRefNet. Every paid plan has monthly credits, and users can remove backgrounds with available credits. It's available to everyone.
Remove backgrounds from product photos for web listings, catalogs, and e-commerce platforms. BiRefNet's fine detail preservation means product edges stay clean and professional. A watch edge stays sharp. Fabric texture shows clearly.
Remove backgrounds from photos for portfolios, corporate headshots, or social media profiles. Hair, skin edges, and fine detail all come through cleanly. A portrait stays detailed. Professional headshots look polished.
Isolate subjects for compositing, design mockups, or creative projects. Transparent backgrounds let you place subjects anywhere without edge artifacts or halos. Design work stays clean and professional.
Build compositions by isolating and combining multiple subjects. BiRefNet's edge quality means combined elements look natural without visible seams or blurry edges. Composited images look professional.
Generate lifestyle backgrounds or place products into custom scenes. Clean background removal makes compositing look polished. A product placed into a lifestyle scene looks like it belongs there.
Remove complex nature backgrounds while preserving detail in animals, plants, and foliage. Fine detail in feathers, fur, and leaves stays sharp. Nature photography stays detailed and authentic.
Background removal is often the start of a larger workflow. Remove a background with BiRefNet, then upscale the result with Topaz Upscale for a larger, higher-quality asset. Or remove the background, then composite the subject into a new background generated with FLUX.2 Pro or another image model. On Craze, each step is a separate operation but together they give you complete control over your final assets.
Both are high-quality background removal models on Craze. BiRefNet excels at fine detail preservation, especially for hair, fur, and complex edges, and is available on every Craze plan, including Free. BRIA RMBG 2.0 uses 256 transparency levels for smooth, semi-transparent edges and is available on Pro and above. Choose BiRefNet if you need to preserve fine detail. Choose BRIA RMBG 2.0 if you need smooth, natural edges for clean compositing. Both are fast and produce professional results.
Hand Craze your image and say what you want, and it runs BiRefNet for you. Every Craze plan, including Free, runs BiRefNet and every other media model, and Craze can use a different one for each job. Each image lands in your project, ready to place on the timeline of a video you export as MP4. Generations come out of your plan's monthly credits, from $0.006 per image, with top-ups when you need more.
E-commerce and product photography: Remove backgrounds from product photos for web listings, catalogs, and e-commerce platforms. Portrait and headshot retouching: Remove backgrounds from photos for portfolios, corporate headshots, or social media profiles. Design and graphic work: Isolate subjects for compositing, design mockups, or creative projects. Photo collage and editorial: Build compositions by isolating and combining multiple subjects.
BiRefNet is a high-quality background removal model that isolates subjects and creates transparent backgrounds. It's especially good at preserving fine details like hair, fur, leaves, and complex edges that often cause problems for other background removal tools.
You pay $0.006 per image for background removal. The cost calculator on this page estimates your cost.
every Craze plan, including Free. Every plan includes BiRefNet, even Free. Every paid plan has monthly credits, and background removals draw from them.
BiRefNet handles people, products, animals, and complex objects. It excels at fine details like hair, fur, foliage, and semi-transparent edges. It produces clean masks regardless of background complexity.
BiRefNet outputs a PNG with a transparent background. The subject is preserved on the transparent layer, ready to use in composites, e-commerce listings, or any context that needs a transparent PNG.
Yes. BiRefNet is built to handle fine details that other background removal tools struggle with. Hair, fur, wispy elements, and complex edges come through cleanly because the model uses bilateral reference learning to preserve detail.
Both are high-quality background removal models. BiRefNet is strong on fine detail preservation, especially for complex edges. Bria RMBG 2.0 uses 256 levels of transparency for smooth, semi-transparent edges. On Craze, both are available. Choose based on your subject type.
Remove backgrounds when you need transparent PNGs for compositing, e-commerce listings, design mockups, or when you want to isolate a subject for further editing or placement in another image.
