Skip to main content
Prompts are how you tell Kive’s AI what to generate. Keep them short and clear. When you want a specific look, use studios, a style reference or train a style model.

​
Describing people in your images

You can describe people directly in your prompts without needing any special setup. Simply include details about the person’s appearance, age, clothing, or pose. For example:
  • “worn by 25 year old woman with long brown hair”
  • “male model in athletic wear”
  • “65 year old man with deep wrinkles wearing hiking gear”
  • “woman with short blonde hair in business attire”
This approach is free and gives you flexibility to change details like hairstyle, age, or clothing in each generation. When to use character models instead: Only use character models (with @modelname) when you need the exact same specific person to appear consistently across multiple generations. Character models require training with photos of one individual and are designed for consistency, not variety.

​
Build effective prompts

Think in four parts, then compress to one sentence: subject, context, style, and a technical hint. Example formula: [Subject], [Context], [Style], [Lighting/Composition detail]. Examples:
  • Modern armchair in a sunlit living room, mid‑century modern style, soft backlight, overhead angle.
  • 35 year old woman in business attire in a downtown office, cinematic style, 50 mm lens, soft rim light.
  • 65 year old weather-beaten man with deep wrinkles wearing hiking gear on a mountain trail, documentary style, natural light.

​
Use style references

Style references apply the visual aesthetic of one image to your new generation. To borrow a style from any image in your library or the Discover page, right‑click the image and choose Use style. The style will be inserted into your prompt as a link. Similarly, choose Use prompt to insert the original prompt from a Kive generation as a starting point. For uploaded photos, click Generate prompt to let Kive draft a detailed description of the image. Edit the generated text as needed, then add your own subject or scenario. Using style references from Discover

​
Advanced techniques

Multi-product shots: Tag multiple products or characters in a single prompt like @product1, @product2, @product3 or @character1, @character2 to create scenes with multiple trained models together. Studios and models: Combine product or character models (@modelname) with a style reference and a studio for precise control.
Style models cannot be combined with product or character models in the same prompt. You can use multiple product models together, multiple character models together, or product and character models together, but style models must be used alone.
Positive prompting: Focus on what you want rather than what you want to avoid. Instead of “no blurry text”, use “sharp text” or “crisp lettering”. Instead of “no dark lighting”, use “bright lighting” or “well-lit scene”. Describing the desired outcome produces better results than trying to exclude unwanted elements. Iterate: Generate drafts, review, and adjust your prompt. Save good prompts to boards for reuse.

​
Example prompts

Product photography: Minimalist skincare bottle on marble counter, mid‑century style, soft morning light, overhead angle. Portrait with character model: @ada in an office, cinematic style, 50 mm lens, soft rim light. Portrait with age specification: 40 year old dark-haired woman with wide-set eyes wearing athletic wear, outdoor setting, natural light. Cinematic scene: Foggy forest at dawn, golden light through trees, wide landscape shot.

​
See also