← AI prompts · Trending · Ganesh Chaturthi

Ganpati saree portrait AI prompt

Two subjects in one frame, and the instruction every other Ganpati prompt leaves to chance: a deliberate visible gap between you and Bappa, and a gaze aimed precisely enough that your eyes actually meet the murti's eye across it. Copy this free ChatGPT & Gemini prompt below.

About this prompt

The sentence carrying the whole thing is place a beautifully decorated Ganpati Ji idol on the left side of the frame and me seated on the right side, with both faces clearly visible. It does three jobs at once. It declares a two-subject frame, which is unusual — most devotional prompts have one subject and a backdrop. It assigns each subject a side, which stops the model centring you and pushing the murti into the background: AI image tools have a well-known habit of demoting a second subject to a backdrop when nothing tells them where it belongs. And both faces clearly visible is the clause that forces the murti's eye to be rendered at all: with no visible eye there is nothing for your gaze to land on, and the emotional centre quietly collapses into a woman looking vaguely upward. The 2:3 crop, the head turn, the gap and the negative space are all corrections to the relationship this one line sets up, so that the picture reads as a moment between the two of you, not as a portrait that merely happens to have the murti in it.

Wearing a kurta instead of a saree? Three paragraphs change and nothing else does. Swap the garment line for a deep midnight-navy silk kurta with an antique-gold band collar and a wine-magenta stole; swap the accessories paragraph for a thin red tilak and a plain gold kada; swap the hair paragraph for short, neatly styled hair with your real hairline and beard kept intact. Leave the gaze paragraph exactly as written — it is the point of the image and it behaves the same on any face. The full swap is spelt out in Making it yours below.

The colour is the other deliberate decision. Everything in the scene is warm — marigolds, brass diyas, gold ornaments, a 2800–3500K key — and the saree is the one cool object in it. That contrast is why the prompt says do not make the entire image orange. It is also the cleanest line between this page and the traditional saree portrait, which is a solo, undated, committed-warm burnt-orange look with no murti anywhere in the frame. Same jewellery register, opposite colour contract, different job.

A word on how to treat the murti, because it matters more than the styling. The composition frames Bappa reverently: he is cropped by the left edge of the frame by design, and that crop is close: the prompt asks for the ornate crown, one expressive eye, the forehead markings, the cheek, the curved trunk, part of the raised blessing hand and the garlands. That is the partial view you get sitting right beside someone, not a diminished one — and he stays upright and elevated, never at foot level. It keeps you as the devotee beside him rather than a co-star: seated, chin slightly raised, quiet closed-mouth expression, never merged with or substituted for the deity's face. Nothing about it is ironic or branded, and it invents no ritual, temple or event. Two habits are worth adding after the image exists — regenerate if the murti has rendered badly, and say plainly that the picture is AI-generated when you post it. It is a festival photograph, not darshan, and it should never be passed off as a real photo taken at a real pandal. Ganesh Chaturthi 2026 falls on Monday, 14 September 2026, and the ten-day Ganeshotsav ends with visarjan on Anant Chaturdashi, Friday, 25 September 2026 — but nothing in the frame is dated, so the look holds up at Sankashti Chaturthi in any month of the year.

Know what a bad render actually is, because three of Bappa's correct features look like mistakes. He is Ekadanta, the one-tusked one: a single short, broken tusk beside a whole one is the icon, so a murti that comes back with two matching whole tusks is the render to regenerate, not the fix. He is Vakratunda, the curved-trunk one: the bent, sweeping curve is canonical exactly as the broken tusk is, and a trunk hanging straight is the thing that is wrong. And he is Chaturbhuja: four arms is the icon, not an anatomical error. The prompt's anatomy negatives — distorted hands, extra fingers, duplicate limbs — are aimed at your own body, not at his, so do not read them as licence to reject a correctly four-armed murti. What genuinely earns another generation is the crown melting into the garlands, the trunk dissolving or splitting mid-curve, fused or extra fingers on your hand, ornaments warping into each other, or a face where an eye has gone missing entirely.

How to use it

1. Tap Copy prompt above.
2. Open ChatGPT (image generation), Google Gemini or Nano Banana in image-to-image / reference mode.
3. Paste the prompt and attach one sharp, evenly lit, unfiltered, front-facing photo of your face. If you also attach a layout reference, label both: image 1 is my face, identity reference only; image 2 is a composition reference only, do not take any facial features from it.
4. If your tool exposes them — image-to-image apps that surface a denoise or strength slider — start around transformation strength 0.20–0.30 and, if there is an identity-strength control, set it high. ChatGPT, Gemini and Nano Banana do not expose these, so there you get the same effect by saying it in words: keep my exact face, relight and re-dress me, do not restyle my features. The beautify switch worth turning off is the one on the phone camera or gallery app that produced your reference photo, not anything in the image tool.
5. Generate at 2:3 vertical, then check three things before you post: the gap, the gaze, and whether the murti itself rendered cleanly.

The three ways this goes wrong

The gaze lands nowhere, or the pupils cross. AI image tools have a well-known habit of returning eyes that sit neutrally in the sockets, pointing wherever the head points rather than aiming at a target elsewhere in the frame — so at a 25–35 degree turn the result often reads as looking past the murti rather than at him. Asking for pupils deviated hard against the head turn is what manufactures the crossed, sclera-heavy look the prompt forbids further down. Let the neck do the work: raise the turn to approximately 40 degrees, and replace the pupil line with keep my pupils close to the centre of my eyes, shifted only slightly toward the inner corners, with no white of the eye showing on the inner side. Then repeat the geometry as the literal closing line — it is a short, self-contained instruction that is easy to lose in the middle of a long prompt and easy to find again at the end, and repeating it costs you nothing: final check: draw a straight line from my pupils to Ganpati Ji's visible eye; my eyes must be aimed along that line. If it still misses, make the target easier rather than the instruction louder — keep Ganpati Ji's visible eye higher in the frame than mine, but bring it nearer the vertical centre of the frame. He stays elevated, exactly as he should be, and your eyes travel up a gentler angle to reach him.

The gap closes and you end up pressed against the trunk. Six clauses in the prompt try to hold the distance and every one of them is relative or negative, which is the problem. Approximately 8–12% smaller than a tight close-up is the weakest sentence in the whole thing: there is no baseline close-up for the percentage to measure against, so it carries no information — and AI image tools have a well-known habit of falling back on the tight, overlapping two-shot whenever a distance is described only in relative terms. Give the gap an absolute size instead. Optional swap: the prompt above is left exactly as its author wrote it, but if the gap keeps collapsing, paste this sentence in place of the 8–12% line when you copy — my head should occupy no more than about 30% of the image height, and there must be a clear band of dark background between the tip of Ganpati Ji's trunk and my cheek, at least as wide as my own face. A gap measured in face-widths is renderable; not too close is not. Move no face touching the trunk up beside that sentence too, rather than leaving it stranded in the negative list hundreds of words below.

The navy turns brown and the rim light blows out. Warmth is the loudest signal in the prompt: ask for a warm scene and the navy saree usually comes back warmed along with everything else, which kills the cool-garment-in-a-warm-scene contrast the whole prompt is built on. Separate the two: the warm golden light falls on my skin, the flowers, the diyas and the murti; the saree must stay unmistakably cool deep blue — a clearly blue garment inside a warm scene; do not apply the warm colour grade to the saree or the blouse. The halo is the same instruction read too literally — one of the strongest cinematic elements becomes make it bright, and bright edge light at f/1.8 against a dark background bleeds into it. Change that to define the rim light as a thin bright edge on the hair and shoulder only; it must not bloom into the background or blow out the shoulder. Do not fix either by asking for less warmth overall, which flattens the marigolds and the diya flames and costs you the entire look.

Making it yours

• Men: swap three paragraphs, not the prompt. Garment line becomes a deep midnight-navy silk kurta with a fine antique-gold band collar and matching churidar, with a wine-magenta stole over one shoulder. Accessories become a thin red tilak on the forehead and a plain gold kada on the wrist; no earrings, no nath, no necklace, no bangles. Hair becomes short, dark, neatly styled hair with natural texture; keep my real hairline, beard and moustache exactly as in the uploaded photo. If the palm-on-cheek pose reads too soft, use both hands folded at chest level in a namaste, elbows relaxed and keep the head turn and eye line untouched.
• Mirroring the frame takes eight edits, not one. To put the murti on the right, flip the idol's side, your side, the crop edge, the farther to the right line, the pupil direction, the front-left key light and the camera-right rim light — then mirror the closing composition sentence too. Flip only the placement line and you hand the model contradictory signals, and it usually resolves them by putting the murti back on the left.
• 4:5 for a feed post instead of 2:3: change the ratio, then compensate for the lost height with keep Ganpati Ji's full ornate crown and my whole head inside the frame; crop the lower foreground rather than either face, and cut several blurred diyas down to one or two. If you pasted the optional 30%-of-frame-height rule above, keep it here especially — a shorter frame makes your head read larger and closes the gap again.
• No jewellery at all? Deleting the accessories paragraph is not enough — AI image tools have a well-known habit of putting ornate jewellery back into an Indian festival portrait when nothing forbids it, so a silent omission simply gets filled in again. Name each absence: no nath, no earrings, no necklace, no bangles and no bracelets; only a small deep-red bindi. Then widen a thin antique-gold border to a broad antique-gold border, because the bangles were the only thing echoing the murti's gold on your side of the frame.
• Make the seated pose actually read. The elbow rests on a knee the crop may not include, and an unsupported floating forearm is where distorted hands come from. Either widen the frame — frame me from mid-thigh upward so the knee supporting my elbow is visible — or remove the support: my forearm rests on the arm of a low seat just outside the bottom of the frame. Either way, say where the fingers go: my fingers curve along my jaw and cheek, below the eye and away from the lips.

Want the murti pressed right up against your cheek instead of an arm's length away? That is the opposite composition, and it has its own page — the Ganesh half-face split prompt, which fills the frame edge to edge with almost no background. For the full mandap scene, standing beneath a monumental idol rather than seated beside one, use the Ganpati Bappa standing portrait. Inviting people over for the darshan? The Ganpati invitation card prompt builds a 4:5 invite carrying your date, time and venue, and invitations usually go out one to two weeks ahead of the 14th. For the same saree-and-jewellery register without the murti — solo, undated and warm orange rather than navy — see the traditional saree portrait, and the Mahadev (Lord Shiva) prompt is the other devotional look in the library.

See every look in the Instagram trending AI photo prompts roundup, or browse the full AI photo prompts library. Not sure which crop suits you? Run the face shape finder, then read the face shapes guide.

Frequently asked questions.