FLUX models read natural language and follow detailed descriptions closely, but they don’t support a negative prompt. Anything you want to avoid has to be expressed positively.
The generator writes a flowing description from your selections and converts common negatives into positive statements — for example “an authentic photograph, not a render” instead of “no CGI”.
Flux understands relationships between objects, so describe how things are arranged rather than listing them.
Instead of “no watermark”, describe a clean frame with no text. The generator does this automatically for common cases.
Lens, distance and depth of field strongly influence the look of Flux photos.
For editing an existing image, use a Flux Kontext model and say plainly what to keep and what to change.
Flux models don’t use negative prompts, so the generator folds important exclusions into the positive description instead.
Yes. The prompt is plain text, so it works in ComfyUI, Forge or any Flux interface.