立即开始
GPT Image 1.5 Edit balances performance and quality with transparent background support and input fidelity control. Supports multi-image editing (up to 16 images) and streaming output. Default quality is low when omitted.
输入
Description of the desired edits, supports up to 32,000 characters
Background transparency. 'transparent': transparent background (requires png or webp format), 'opaque': solid background, 'auto': model decides
auto
Input image URL(s) to edit. Direct HTTPS URLs as string array. Supports up to 16 images for GPT Image models
提示：可拖拽文件、从剪贴板粘贴（Ctrl/Cmd+V），或提供 URL。
Control fidelity to original input image. 'high': preserve original details, 'low': more creative freedom
input_fidelity
Output quality level. 'low': fast draft (default), 'medium': balanced, 'high': best quality
low
Output image size. Preset resolutions only
1024x1024
结果
暂无结果
运行模型后，结果将在这里显示。
