← AI prompts · Trending · Collage

Candlelight photo collage AI prompt

Six frames in one 2:3 mosaic — candlelight, a red zari saree, antique-gold jhumkas and white flowers drifting past the lens. Copy this free ChatGPT & Gemini prompt, upload one photo, and your face stays the same in all six.

Six-frame candlelight photo collage made with the AI prompt — red zari saree, antique-gold jhumkas, candles and white flowers in warm amber light
Made with the prompt below — a seamless 2:3 six-frame mosaic in candlelight.
Use my uploaded portrait as the only facial identity reference. Preserve my exact identity throughout all six photographs: the same facial structure, eye shape, eyebrows, nose, lips, jawline, skin tone, hairline and natural facial proportions. Do not create six different women. Create a vertical 2:3 cinematic six-photo portrait collage, arranged as a seamless two-column staggered mosaic with no borders and no gaps. Both columns are equal in width. The left column contains three nearly equal-height images. The right column contains one large upper portrait, one narrow horizontal extreme close-up in the middle, and one lower portrait. Every image must have matching lighting, styling, color grading and facial identity. STYLING Style me as an elegant South Asian woman during an intimate candlelit festive evening. Dark black-brown hair gathered into a loose low bun with delicate natural flyaway strands framing the forehead, temples and cheeks. A tiny deep-red bindi between the eyebrows. Large antique-gold jhumka earrings with intricate detailing and tiny pearl drops. Deep crimson-red traditional sari or lehenga with dense gold zari embroidery, ornate floral and paisley motifs, rich woven borders and subtly reflective fabric. Add thin antique-gold bangles in the hand-touching-hair portrait. Makeup is refined and realistic: naturally luminous warm skin, visible pores and fine facial texture, defined natural eyebrows, brown-black kajal, delicate winged eyeliner, long separated lashes, warm copper-brown eyeshadow, soft peach blush, subtle cheekbone and nose highlights and muted rose-brown peach lipstick. Avoid heavy bridal makeup and avoid artificial beauty-filter skin. LIGHTING Illuminate the entire collage using realistic candlelight and warm tungsten light. Rich amber-gold color palette with burnt orange, deep brown, crimson red, cream and black. Soft directional candlelight from the side and slightly below, warm highlights across the eyes, cheeks, nose and lips, deep cinematic shadows, subtle golden rim light around the hair, glowing candle catchlights in the eyes, gentle highlight bloom, atmospheric halation and large circular golden bokeh. Dark intimate indoor background with no recognizable modern objects. Include tall ivory candles at different distances and small bouquets of white chrysanthemum or daisy-like flowers mixed with occasional muted red and orange blossoms. Some flowers should appear very close to the lens and become softly blurred. Candle flames should be bright, elongated and surrounded by delicate golden halos. PANEL 1, UPPER LEFT Intimate extreme facial close-up, my face filling the frame, head leaning gently toward a bouquet of white flowers, flowers partially covering the lower cheek and chin, eyes looking directly into the camera, subtle closed-mouth smile, loose hair strands across the forehead and cheek, sharp eyes with candle reflections, crop cutting part of the hair and chin. PANEL 2, UPPER RIGHT The largest frame, elegant three-quarter portrait, body turned away while looking back over the shoulder toward the camera, relaxed affectionate expression and subtle smile, red-and-gold embroidered sari draped across the shoulder and back, one ornate jhumka fully visible, several glowing candles creating vertical golden bokeh in the background. PANEL 3, MIDDLE LEFT Cinematic side profile facing left, eyes gently lowered, peaceful smile, loose strands near the cheek, detailed jhumka, bouquet of white and red flowers in the lower-left foreground, one tall candle glowing near the face, beautiful warm edge light outlining the nose, lips and jaw. PANEL 4, MIDDLE RIGHT Narrow horizontal extreme close-up of my eyes, eyebrows, bindi, upper cheeks and bridge of the nose, direct soulful gaze, one eye fully visible and the other partially cropped, crisp eyelashes, eyeliner and iris reflections, white flowers softly obscuring the lower nose and cheek, extremely shallow depth of field. PANEL 5, LOWER LEFT Warm chest-up portrait, head slightly tilted, one hand gently touching the hair beside the ear, gold bangles visible, looking slightly upward and away from the camera, soft natural smile, red-and-gold clothing visible, a large blurred candle flame glowing in the lower foreground, warm floral and candle bokeh behind me. PANEL 6, LOWER RIGHT Gentle three-quarter profile looking downward, quiet introspective expression, lowered eyes, detailed jhumka earring, loose hair strands outlined by golden light, white flowers beside the face, blurred candle bokeh behind me and the embroidered red garment visible near the lower edge. PHOTOGRAPHY Photograph every panel like a luxury cinematic editorial portrait using an 85mm lens, with selected close-ups appearing like a 100mm macro portrait lens, aperture around f/1.4 to f/1.8, extremely shallow depth of field, realistic optical bokeh, soft focus transitions, natural facial anatomy, high dynamic range, subtle film grain, gentle highlight halation, rich warm color grading and photorealistic detail. Maintain perfect facial consistency across all six frames. No borders, no text and no duplicated poses. NEGATIVE PROMPT different faces, identity drift, altered facial structure, face morphing, six different women, mismatched skin tones, duplicated photograph, repeated pose, asymmetrical eyes caused by generation errors, crossed eyes, distorted pupils, extra fingers, missing fingers, malformed hands, deformed earrings, floating jewelry, misplaced bindi, modern clothing, western dress, heavy bridal crown, excessive makeup, plastic skin, waxy face, overly smooth skin, pale grey color grade, cool blue lighting, harsh flash, flat lighting, cluttered room, visible furniture, text, logo, watermark, borders, frames, white gaps, cartoon, painting, illustration, CGI, oversaturated orange skin, crushed facial shadows, blurry eyes, low-resolution face. RECOMMENDED SETTINGS Aspect ratio: 2:3 Style: Photorealistic / cinematic editorial Identity or face-reference strength: High, approximately 75–90% Stylization: Low to medium Detail: High Depth of field: Very shallow Seed: Keep the same seed when generating panels separately

About this prompt

This is the most ambitious layout in the library: six frames in a two-column staggered mosaic at 2:3, all lit by candlelight. A red zari-embroidered saree, antique-gold jhumkas with pearl drops, a low bun with flyaway strands, and white chrysanthemums drifting close to the lens where they blur into the foreground.

What sells it is the light: warm candlelight from the side and slightly below, golden halos around the flames, catchlights in the eyes, and big circular bokeh behind. It's the festive-evening counterpart to the daylight collages.

Six frames is genuinely hard. The more panels you ask for, the more chances the model has to drift into a different face — and this prompt has the most of any in the library. Its own settings block has the answer: keep the same seed, set face-reference strength to 75–90%, and if the mosaic keeps failing, generate panels in pairs at a fixed seed and assemble them in any editor. That reliably beats fighting a six-panel render.

The six frames

1 · Upper left. Extreme close-up leaning into white flowers, eyes to camera, subtle closed-mouth smile.
2 · Upper right (largest). Three-quarter portrait looking back over the shoulder, saree across the back, candle bokeh behind.
3 · Middle left. Side profile facing left, eyes lowered, warm edge light along nose, lips and jaw.
4 · Middle right. Narrow macro of the eyes and bindi, one eye cropped, flowers obscuring the lower face.
5 · Lower left. Chest-up, hand touching hair beside the ear, bangles visible, a blurred flame in the foreground.
6 · Lower right. Three-quarter profile looking down, quiet expression, jhumka detailed, flowers beside the face.

How to use it

  1. Tap Copy prompt above.
  2. Open ChatGPT (image generation), Google Gemini or Nano Banana, paste it, and attach one clear, well-lit photo of your face.
  3. Generate at 2:3, with face-reference strength high and the seed fixed.
  4. Check panels 2 and 6 against each other — both are three-quarter views, so drift shows there first.

Tips for the best result

  • Build it in pairs if six fails. Two panels at a time at the same seed, then assemble. Far more reliable than re-rolling the whole mosaic.
  • Keep the room dark. Candlelight only works against deep shadow — if the background lifts, the atmosphere goes with it.
  • Watch the jhumkas. Ornate earrings are where models duplicate or float jewellery. Check each frame.
  • Don't let the skin go orange. Warm light plus warm grading tempts models to oversaturate. The negative prompt guards against it — re-run if it slips.
  • Swap the colour freely. Emerald, deep teal or mustard with gold zari all work; the lighting instructions are unchanged.

Fewer panels are easier to get right — try the 5-panel golden hour collage or the 3-panel harsh sunlight collage. For a single frame, see the traditional saree portrait or the candlelight close-up. Everything else is in the trending prompts roundup and the full library.

Frequently asked questions.

What is the candlelight photo collage prompt?
It is a copy-paste AI image prompt that creates one 2:3 image made of six seamless photographs in a two-column staggered mosaic. The frames run from an extreme close-up among white flowers, through a three-quarter portrait looking over the shoulder, a side profile, a macro eye crop, a chest-up portrait and a downward-looking profile — all lit by candlelight with a red zari saree and antique-gold jhumkas.
Why do I get six different faces?
Because each panel is effectively a separate generation, and the more panels you ask for the more the model drifts. Keep the seed fixed, set face-reference or identity strength to around 75 to 90 percent, and turn any beauty enhancement off. If it still drifts, generate two panels at a time at the same seed and assemble the mosaic in any photo editor — that is far more reliable than re-rolling all six.
What aspect ratio should I use?
Generate at 2:3, not 9:16. The two-column mosaic needs that slightly wider shape for both columns to read properly; a taller 9:16 frame squashes the panels. If your tool defaults to square or 4:5, set the ratio explicitly before generating.
Can I change the saree colour or use fewer frames?
Yes to both. Swap deep crimson for emerald, teal or mustard and keep the gold zari, and the lighting and posing instructions still work. You can also ask for four frames instead of six by dropping two panel descriptions — fewer panels usually means noticeably better face consistency.
Does it work in Gemini as well as ChatGPT?
Yes. It works in ChatGPT image mode, Google Gemini, Nano Banana and other identity-preserving tools. Paste the prompt, attach one clear photo of your face, and generate at 2:3. This is the most demanding layout in the library, so expect several attempts, or build it in pairs at a fixed seed.

Keep going

More trending AI photo prompts.

Copy-paste prompts blowing up on Instagram — all free, all keep your face.