Choose between a new image and a photo edit
The AI image generator is a workspace for still images. Start with text when you want a scene that does not need to preserve a particular photograph. Use a reference image when the subject, arrangement or visual details in that photo matter to your request. Available modes depend on the model selected in the controls above. Read the active mode before adding files: a written description alone does not make a specific person or product a reliable reference.
This page serves a different purpose from the Burning Bridges party effect. It helps you prepare a portrait, explore a background or develop a visual idea without requesting animation. If your goal is the cup-in-hand performance, return to the party tool on the homepage when it becomes available. If your goal is movement with your own direction, use the general video workspace. Creating a still image does not automatically create or start either video task.
Write a prompt that can be evaluated
A useful prompt describes the subject, setting, composition, light and changes you want. Give the model a clear main task before adding visual style. For example: “An adult wearing a plain navy shirt, waist-up portrait in a warmly lit lounge, face looking toward the camera, both hands visible, natural photographic detail.” This is a prompt-writing example, not a sample generated by the site or a promise that every detail will appear correctly.
For an edit, distinguish what should change from what should remain. You might ask to replace a busy background with a simple interior while preserving the subject’s face, clothing and pose. Avoid conflicting instructions such as a close-up face with the entire body visible. If you need a particular layout, describe the position of the subject and leave space where a title or other graphic will later be added. Short, concrete instructions make it easier to identify which part of a result needs revision.
Prepare a suitable reference photo
Use a clear image that you have permission to upload and process. For a portrait, choose one adult subject with a visible face and enough separation from the background. Heavy blur, strong filters, severe shadows and objects across the face make the starting information harder to interpret. A full-body task needs visible clothing and limbs; a tight headshot cannot reliably supply those details. The generator may infer missing information instead of recovering what was outside the frame.
Check the upload control for its current accepted formats, file-size limit and number of references. Selecting a file in the active uploader sends it for processing and storage before you press Generate. Remove unrelated private information from the image before selecting it. Do not upload identity documents, financial records, intimate imagery or a photograph of someone who has not agreed to this use. More references are not automatically better if they show different people or contradictory appearances.
Check settings and credits before generating
Select only from the models and settings actually shown in the workspace. Aspect ratio affects the shape of the result, so choose the intended destination early: a tall composition can suit a phone screen, while a wider composition leaves room for a landscape scene. A ratio change can also change framing. It is not simply a guarantee that the same image will be stretched or cropped without other changes.
Review the displayed credit requirement after adjusting the settings. Generation costs can differ by model, mode and output options. See Pricing for monthly plan prices and credits. Unused subscription credits expire at the end of each billing period and do not roll over. Cancel before renewal to stop future charges. An available credit balance and permission to use a model are distinct from a claim that generation is free. If the catalog or cost cannot be loaded, resolve that state before submitting a task.
Review an image before reusing it
Look beyond the overall style. Examine eyes, teeth, fingers, jewellery, repeated objects, clothing edges and any lettering. Compare the output with the reference if preserving identity or a product detail matters. The generated image may change those details even when the prompt asks to keep them. Treat an attractive result as something to inspect, not evidence that every requested property was preserved.
When preparing an image for later animation, check that the person is large enough in the frame and the intended moving body parts are visible. An unusual pose, obscured hand or cropped cup can make the next task harder to control. Keep the version you evaluated rather than substituting a different image after reviewing it. A successful still image does not establish how a video model will animate that image. Review the full video separately if you use it in another workflow.
Use history and recover from interruptions
After submission, follow the task status. A queued or processing state is not a finished image. If the browser loses its connection, check your creation history before sending the same request again. A network error in the page does not establish that the provider stopped working. Repeating a submission may create a separate task with a separate credit charge.
If a task fails, keep its reference and any visible error message. Compare the task status and balance before assuming credits have been restored. Technical credit restoration and a refund to a payment method are separate processes. For support, provide the task reference, approximate time and a concise explanation of the problem. Never send a password, a full card number or unrelated personal documents.
Download and publish deliberately
Open the finished result and use its available download action. Check the saved file at the size you plan to use rather than judging only the small preview. Keep a local copy of work you need: history and hosted media are not promised as permanent archives. A completed download also gives you a stable version to compare when you request an edit or raise a quality concern.
Before sharing, consider the rights in your photo, any visible marks and the person’s likeness. Generated output does not automatically give you permission to imply an endorsement or misrepresent a real event. No blanket commercial licence is promised here. Label synthetic portrayals clearly, review the acceptable-use rules, and obtain any permissions required for your intended use. Use the privacy contact for questions about an account, uploaded material or deletion.
Burning Bridges party tool · AI video workspace · Subscription pricing · Support and privacy contact · Acceptable use