Candy AI Image Generator Guide Menu

Candy AI image prompts: tips for better results

A Candy AI image prompt should be specific: say who is in the picture, what they are doing and how they are posed. That is the substance of the three tips in Candy AI’s help center. This page restates them, adds general prompt-writing advice and gives ten example prompts. The examples are text to read: nothing is generated on this site.

Last checked:

What Candy AI recommends

Candy AI’s help center gives three tips. In our own words:

  1. Be descriptive and specific. Replace a bare noun with a described scene: who is there, how they stand and where they look.
  2. Say what the picture is for. Give the situation the image belongs to, such as a quiet evening at home, so the request has a context.
  3. Name the pose. If the pose matters, state it outright: standing, seated, leaning, looking over one shoulder.

The same article lists two limits: very complex prompts may fail, and exact consistency of hair color or clothing between images is not guaranteed.

General advice: six parts of a clear prompt

A prompt that covers six things leaves little to chance:

  • Subject: who is in the picture: an adult, with an age range, clothing and one distinguishing detail.
  • Pose: what the body is doing, the part Candy AI’s own tips single out.
  • Setting: where it happens, with one or two props.
  • Lighting: the source and its quality, such as window light or dusk.
  • Framing: distance and angle, such as close-up, waist-up or full-length.
  • Mood: one or two adjectives.

Write it as one sentence, most important part first: one scene, one action.

Anatomy of an image prompt: one example sentence split into six labelled parts, namely subject, pose, setting, lighting, framing and mood.
Fig. 1 The first example below, taken apart.

Ten example prompts

Each example describes an adult in an everyday or dressed-up setting, written for the generator; in a chat, open the request with “show me” or “send me”. Candy AI allows adult content for adult users within its Community Guidelines; these examples deliberately stay with ordinary scenes.

  1. Rainy café window A woman in her thirties in a mustard raincoat, seated with both hands around a cup by a rain-streaked café window, soft gray daylight, waist-up, calm and thoughtful.
  2. Rooftop at dusk A man in his forties in a navy overcoat leaning on a rooftop railing at dusk, city lights behind, warm low light, three-quarter view, relaxed.
  3. Hiking trail A woman in her late twenties in a red windbreaker pausing on a mountain trail, hand on her backpack strap, clear morning light, full-length, cheerful.
  4. Recording studio A woman in her thirties in headphones singing into a studio microphone, eyes closed, amber lamps behind, side close-up, focused.
  5. Formal gala A woman in her forties in an emerald gown standing on a marble staircase at a gala, hand on the banister, chandelier light, full-length, composed.
  6. Old bookshop A man in his thirties in a gray cardigan reading on a ladder in an old bookshop, warm lamp light, seen from below, quiet.
  7. Market stall A woman in her fifties in a linen shirt choosing tomatoes at a market stall, bright midday sun, waist-up, lively.
  8. Sunday kitchen A man in his thirties in a striped apron flipping a pancake in a sunny kitchen, morning window light, medium shot, playful.
  9. Gallery opening A woman in her thirties in a black blazer, arms folded before a large abstract painting, cool spotlights, over-the-shoulder view, thoughtful.
  10. Winter platform A woman in her thirties in a camel coat waiting on a snowy train platform at night, lamps glowing, full-length side view, wistful.
Try a prompt on Candy AI

Affiliate link. Adults 18+ only.

When a result misses: change one thing

Every new attempt uses tokens, so change one element at a time.

  • Wrong pose: put the pose first and name it plainly.
  • Cluttered or failed image: cut props and actions; Candy AI itself says very complex prompts may fail.
  • Wrong atmosphere: add the purpose or the mood.
  • Awkward crop: state the distance, such as waist-up.

To keep an ordinary chat from spending tokens on pictures meanwhile, see the toggle on the tokens and cost page.

Revision loop for an image prompt: write, generate, compare, change one thing, then repeat. A table pairs four common problems with the single change to try.
Fig. 2 One change per attempt shows which word mattered.

The consistency limit

Candy AI states that exact hair color or clothing may not stay identical across images. General practice is to repeat the details that matter in every prompt.

For a recurring character, design the look once in Character Studio; the caveat still applies. A picture you like can become a clip, as the image-to-video page explains, and the sharing rules apply before you post any result.

Sources

Candy AI help-center articles used on this page, read on 13 November 2026.