How do I generate images and videos in different styles?

Yes. You simply need to describe your style in plain language.

Select the model you would like to use. ChatGPT Image 2 is currently state of the art for most marketing and advertising creative use cases. The Nano Banana family from Google are all extremely capable as well.

Write your prompt, and be sure to include the medium you want to generate, e.g. a photo, an oil painting, sketch, cartoon, illustration or anything else.

Add details to refine your images, e.g. Add “visible brushstrokes with the style of Van Gogh”, or “charcoal drawing with monochrome shading”.

Reference images are often a more effective way to achieve a specific or consistent style. Models such as ChatGPT Image 2 and the Nano Banana family can pick up a style from one or a few reference images supplied alongside your prompt and apply it to new generations.

Simply upload or gather your references, then tell the model which qualities to carry over.

For example:

"Match the flat illustration style of the reference image, keeping the same palette and line weight."

This works wherever you generate images in Pencil, including agent chats, the Ads editor, Workflows and Sheets.

Style transfer means transferring the style from one of more images or videos to an existing composition. Using either method above, a descriptive prompt or a style reference image, the most capable models can apply a new style to existing content.

Supply an existing image or video alongside your prompt and describe the transformation you want, e.g.

"Restyle the image from attachment 1 in a watercolour illustration style, the same style as the example attached in reference 2, keeping the characters' poses and overall composition from image 1 completely intact and unchanged."

This is often more practical than regenerating from scratch, and support for it continues to expand across the newest image and video models.