Prompting Process

Prompting is in essence, describing to the diffusion model what you want to generate. Let's say you want to generate a portrait photo of a handsome man with long brown hair at a beach, wearing white shorts, with the shoreline and the sea in the background.

To generate a good result, you need to prompt like you're explaining the scene with details and more. Let's observe this through an example:

Prompts:

a handsome man standing at a beach, long brown hair, wearing white shorts, shoreline and the sea in the background

Prompts:

a handsome man standing at a beach, long brown hair, wearing white shorts, shoreline and the sea in the background, (RAW photo)++, portrait, (high detailed skin)+, 8k uhd, DSLR, soft lighting, high quality, film grain, Fujifilm XT3, depth of field, bokeh

As evident from the example, although both images managed to capture the key elements of our idea, the one on the right captured it with greater detail and resemblance. This is due to the addition of a greater amount of details. Such as: type of image, lightning details, type of lens, and the addition of depth of field and bokeh.

Although these detail and quality prompts vary from image type to image type (i.e., using lens prompts when generating photographs whereas using artist names when generating illustrations), it is always better to include more details in your prompt. We will further examine different styles and quality prompts in the coming sections.