Text-to-image generation creates a new image from a written description, the prompt. Well-known text-to-image systems include Stable Diffusion, DALL·E, Midjourney and Imagen; most are built on diffusion models.
Text-to-image vs redesigning your room
Text-to-image starts from nothing but words, so it invents the whole room: proportions, windows, view and light. That is ideal for inspiration (“a Japandi living room with a view of the sea”) and useless for knowing how your living room would look.
To redesign an existing space, tools use your photo as the starting point (img2img) and keep its structure with guidance like ControlNet. The prompt then only has to describe what changes.
Writing better prompts
Whatever the tool, the same principles apply: name the room, one style, real materials, the light, and what must not change. See prompt engineering and our prompt guide.