Sogni AIInstall

Summary

Edit an image using a chosen Qwen Image Edit model by applying a text prompt describing the desired modifications to one or more context images.

Inputs

• imageEditModelSearch — Optional text to filter available Qwen Image Edit models by name, updating the model dropdown.
• imageEditModelId (required) — Identifier of the Qwen Image Edit model to use for editing the image, e.g., 'standard' for detailed or 'lightning' for faster results.
• imageEditPrompt (required) — Textual description of the edit to apply to the input images, e.g., 'Change the background to a beach sunset, make the sky orange'.
• contextImage1Property (required) — Name of the binary property containing the primary input image to edit.
• contextImage2Property — Name of the binary property containing a second context image, if applicable.
• contextImage3Property — Name of the binary property containing a third context image, if applicable.
• imageEditNetwork — Choice of network type for generation, 'fast' for quicker output or 'relaxed' for slower but cheaper processing.
• imageEditAdditionalFields.generationSettings.negativePrompt — Text describing undesired features to exclude from the generated image, e.g., 'blurry, low quality'.
• imageEditAdditionalFields.generationSettings.stylePrompt — Style instructions to influence the visual style of the edited image, e.g., 'photorealistic, cinematic lighting'.
• imageEditAdditionalFields.generationSettings.numberOfMedia — Number of edited images to generate, from 1 to 10.
• imageEditAdditionalFields.generationSettings.steps — Number of inference steps to perform; defaults depend on model (typically 20 for standard, 4 for lightning).
• imageEditAdditionalFields.generationSettings.guidance — Degree to which the generation should adhere to the prompt (guidance scale); defaults to model's default (4.0 standard, 1.0 lightning).
• imageEditAdditionalFields.generationSettings.seed — Random seed for reproducible results; omit for random.
• imageEditAdditionalFields.output.downloadImages — Whether to download the resulting images as binary data (recommended to avoid URL expiry). Defaults to true.
• imageEditAdditionalFields.output.outputFormat — Output format for images, either PNG or JPG. Defaults to PNG.
• imageEditAdditionalFields.output.sizePreset — ID of a size preset for output image dimensions (e.g., 'square_hd'); leave empty for default size.
• imageEditAdditionalFields.output.width — Custom width in pixels when size preset is 'custom', range 256-2048.
• imageEditAdditionalFields.output.height — Custom height in pixels when size preset is 'custom', range 256-2048.
• imageEditAdditionalFields.advanced.tokenType — Token type to use for payment, options are 'spark' or 'sogni'; defaults to 'spark'.
• imageEditAdditionalFields.advanced.timeout — Maximum time to wait for the edit operation in milliseconds; defaults to 60000 for 'fast' network or 600000 for 'relaxed'.

Output shape

a list of edited images returned as binary data properties or URLs depending on configuration

The number of returned images corresponds to the 'Number of Images' parameter. Images can be downloaded directly as binary data to avoid the 24-hour URL expiry, or provided as URLs. Output images follow specified format and size settings. If multiple context images are provided, the model uses them as input context for the edit.

Links

• n8n community nodes installation guide