← AI prompts · Trending · Cinematic portrait

Candlelight close-up portrait AI prompt

The moody extreme close-up everyone's posting — warm amber candlelight, one big creamy bokeh light, and loose hair strands falling across the face. Copy this free ChatGPT & Gemini prompt, upload your photo, and it keeps your own features.

Candlelight close-up cinematic portrait made with the AI prompt — warm amber light on one side of the face, blurred bokeh behind, loose hair strands across the cheek
Made with the prompt below — a cinematic 2:3 candlelit close-up that keeps your own face.
Recreate Image 1 as closely as possible in composition, camera angle, crop, lighting, atmosphere, hairstyle placement, facial expression, depth of field, and cinematic color grading, but replace the person entirely with the identity and facial features from Image 2. Identity is the highest priority. Preserve the exact recognizable likeness of the person in Image 2: same face shape, forehead, eyebrows, eye shape and spacing, nose structure, cheekbones, lips, jawline, skin tone, and natural facial proportions. Do not blend the two identities and do not retain any facial characteristics from Image 1. Image 1 must be used only as a visual reference for the photography. Create an intimate, extreme close-up cinematic portrait in a vertical 2:3 aspect ratio. Only the right side and central portion of the subject's face should be prominently visible, filling almost the entire frame. Position the clearly visible eye near the upper-middle-left area of the face. The opposite side of the face and foreground should be partially obscured by softly blurred dark hair. Reproduce the same subtle, relaxed expression: a very gentle natural smile, slightly parted lips, calm eyes, and an intimate candid mood. The subject should look directly toward the camera. Place several loose, fine strands of dark hair naturally across the forehead, eyebrow, eye area, cheek, and lips. The strands should feel spontaneous and realistic, not arranged or artificial. Keep the hair slightly messy, soft, and dimensional. Use warm, low-key, amber-orange lighting resembling candlelight or a dim tungsten lamp. Illuminate the visible eye, cheek, nose, and lips with a soft golden glow while maintaining rich shadows around the hair and edges of the face. Include a large, circular, heavily blurred warm bokeh light in the upper-left background, with smaller subtle golden highlights lower in the background. Use an extremely shallow depth of field. The visible eye and nearby eyelashes should be the sharpest point of the image, while the hair in the foreground, ear, background, and outer facial areas gradually fall into creamy blur. Add a realistic catchlight and detailed reflection in the visible eye. Simulate a professional full-frame camera with an 85mm portrait lens at approximately f/1.2–f/1.4, photographed from very close range. Use soft cinematic focus transitions, gentle highlight bloom, natural contrast, warm brown and amber tones, slightly muted saturation, subtle analog film grain, and smooth background bokeh. Maintain realistic human skin with natural pores, fine facial hair, tiny imperfections, soft under-eye texture, and believable highlights. Do not use plastic skin, beauty-filter smoothing, excessive makeup, or artificial sharpening. The final result should feel like a high-end cinematic film still: intimate, emotional, warm, mysterious, photorealistic, softly sensual without being explicit, and nearly identical to the reference photograph in framing and atmosphere — while unmistakably showing the person from Image 2. NEGATIVE PROMPT wrong identity, face blending, features from reference person, altered face shape, generic face, beautified face, plastic skin, excessive skin smoothing, waxy texture, heavy makeup, exaggerated smile, visible teeth expression change, closed eye, crossed eyes, asymmetrical eyes, distorted nose, deformed lips, extra facial features, duplicate hair strands, artificial hair, oversharpening, flat lighting, cool lighting, daylight, harsh flash, bright background, multiple bokeh lights, wide shot, full face centered, symmetrical composition, excessive blur on the eye, illustration, painting, anime, CGI, 3D render, low resolution, compression artifacts, text, watermark

About this prompt

This is a two-image, identity-preserving prompt. Unlike most prompts, it works from two attachments: the reference photo above sets the framing, lighting and mood (Image 1), and a clear photo of your face supplies the identity (Image 2). The wording repeatedly instructs the model to take nothing but photography from the reference and everything about the face from you — which is what stops the usual "blended stranger" result.

The look itself is a cinematic 85mm f/1.2 close-up: one eye tack-sharp, warm candlelight on the cheek, a big amber bokeh light behind, and loose hair strands drifting across the frame. It renders as a vertical 2:3, which suits both Instagram feed and Story crops.

How to use it

  1. Tap Copy prompt above.
  2. Open ChatGPT (image generation), Google Gemini or Nano Banana and paste it.
  3. Attach two images: the reference photo above as Image 1, and a clear photo of your own face as Image 2. Label them in the chat if your tool allows it.
  4. Generate at a 2:3 vertical ratio, and re-run once or twice for the sharpest eye and the closest likeness.

Upload your photo like this. For stronger facial accuracy, use a clear, unfiltered photo of yourself with your full face visible, neutral lighting, and no hair covering your eyes. The model can only preserve what it can clearly see — a dim, filtered or half-covered face is the single biggest cause of a poor likeness.

Tips for the best result

  • Don't ask for beautifying. The prompt deliberately keeps pores, fine facial hair and small imperfections — that texture is what makes it read as a photograph rather than a render.
  • One bokeh light only. If the background fills with multiple orbs, re-run — the negative prompt already guards against it.
  • Check the eye. The visible eye should be the sharpest point with a clear catchlight. If it comes out soft, regenerate rather than sharpening afterwards.
  • Keep the crop tight. If the model pulls back to a full centred face, remind it: extreme close-up, only the right side and centre of the face visible.

Want a different mood? Try the moody restaurant portrait or the red cinematic studio portrait. See every look in the Instagram trending AI photo prompts roundup, or browse the full AI photo prompts library.

Frequently asked questions.

What is the candlelight close-up portrait prompt?
It is a copy-paste AI image prompt for the trending cinematic close-up: an extreme crop of your face lit by warm amber candlelight, with one eye tack-sharp, a large creamy bokeh light in the background and loose hair strands falling across the cheek and lips. It renders as a vertical 2:3 image and keeps your own facial features rather than inventing a new face.
Why does this prompt need two images?
Because it separates style from identity. Image 1 is the reference photograph, used only for composition, crop, lighting and colour grading. Image 2 is a clear photo of you, used only for the face. The prompt explicitly tells the model not to blend the two, which is what prevents the common result where the output looks like a stranger who slightly resembles you.
How do I get the best facial likeness?
Upload a clear, unfiltered photo with your full face visible, neutral even lighting and no hair covering your eyes. Avoid heavy filters, sunglasses, strong side shadows and low-resolution crops. The model can only preserve features it can clearly see, so the quality of that one photo matters more than anything else in the prompt.
Does it work in Gemini as well as ChatGPT?
Yes. The prompt works in ChatGPT image mode, Google Gemini, Nano Banana and other identity-preserving image tools. Paste the prompt, attach both images, and generate at a 2:3 vertical ratio. Results vary slightly between tools, so re-running once or twice is normal.
Is this AI photo prompt free?
Yes, the prompt is free to copy and use. You only need an AI image tool such as ChatGPT, Google Gemini or Nano Banana, the reference image on this page, and one clear photo of yourself. There is nothing to buy.

Keep going

More trending AI photo prompts.

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