Turn a detailed brief into a portrait, a product idea or a new scene. Prepare your prompt and references here; Create opens Studio and starts generation once you are signed in and inputs are ready.
Name the subject, light and details to keep. Add up to 10 references for ordinary reference-led generation.
Check 1K or 2K and the credit cost. For transparency or a mask, open Studio directly and set those controls before creating.
Create on this page starts the task in Studio after sign-in and input readiness. Inspect the result, then download or refine it.
A fictional adult in a black satin dress, warm afternoon light and a quiet stone balcony: this direct Qwen Image 2.1 output follows one detailed text prompt. Describe the pose, fabric and camera distance together, then check hands, facial details and background consistency before sharing.
This Qwen sneaker comes from a PNG with real transparency; the smaller display copy preserves it. The checkerboard is the page background. In Studio, select Qwen Image 2.1 and enable Transparent background before creating. Describe one clear subject and inspect its edges.
Open Studio
In Studio, choose Qwen Image 2.1, add one reference and brush or upload a mask. White edits; black preserves. Use an opaque background and reference proportions. This single GPT Image 2.5 generation illustrates apple removal, not an actual Qwen edit.
Open StudioSwipe to explore related tools
Inputs, supported controls, credit cost, privacy, and recovery—before you start.
Generate from text or use references to guide a new image. SugarCube also provides transparent-background generation and brush or uploaded-mask editing in Studio. This page prepares ordinary text or reference generation; its Create action starts the task automatically in Studio.
The current price is 40 SugarCube credits for 1K or 70 credits for 2K per generation. Check the displayed cost before creating. Subscriptions and credit top-ups are available on the pricing page; this model is not promised as a free trial.
Ordinary generation supports up to 10 JPG, PNG or WebP references and a prompt of up to 5,000 characters. Give each reference a clear role. Mask editing requires exactly one reference plus a separate mask; the mask is not a second reference.
Open Studio directly, select Qwen Image 2.1 and enable Transparent background before generating. The sneaker shown here is an actual Qwen PNG with transparent pixels; the checkerboard is a display aid. Inspect fine edges and partially transparent areas before using a result in a design.
Open Studio, select Qwen Image 2.1 and add one reference. Paint the region or upload a mask matching its proportions: white for editing and black for preservation. Add a specific change prompt. The background must be opaque, and the reference determines the output proportions. Preservation and the final edit can vary.
The portrait and transparent sneaker are direct Qwen Image 2.1 outputs. The apple-removal diptych was generated in one GPT Image 2.5 request to illustrate the workflow. It is not an actual two-step edit, a Qwen result or a quality benchmark.
References remain in this browser while you prepare the landing-page draft. After Create, Studio starts processing once you are signed in and inputs are ready; processing sends inputs to the generation service. Results are available through authenticated owner-scoped access. Upload only images you have permission to use and review the privacy policy.
Failed generations are refunded automatically. A successful image uses credits even if you prefer a different variation. Review faces, hands, product details and generated text before publication; neither reference fidelity nor a perfect local edit is guaranteed.
Write a prompt above, or open Studio to configure transparency and mask editing.
Prepare my image
