• Image Generator — general-purpose creatives: product shots, social posts, illustrations, ads.
• YouTube Thumbnail — 1280×720 thumbnails plus matching titles, description & tags.
• AI Art / Illustration — stylised art, wallpapers, concept pieces.
• Pinterest Pin — tall 2:3 pins optimised for saves and click-through.
• Headshot Generator — professional profile photos from a selfie.
• Prompt Extractor — reverse-engineer a prompt from an image you like.

Every generation costs coins based on the Quality tier: Low ≈ 3 coins, Medium ≈ 12 coins, High ≈ 50 coins (exact price is always shown on the Generate button before you click). Failed generations are never charged.

1. Open Image Generator and describe your subject in one clear sentence.
2. Pick an Image Size that matches where the creative will be used (square for feed posts, landscape for banners, portrait for stories).
3. Leave Quality on Auto for your first tries — it balances cost and result.
4. Click Generate and review the output before spending more coins on a High-quality re-run.

Write prompts as: Subject + Setting + Lighting + Style + Composition

• Subject — the exact product, person or scene, with specific details (color, material, action).
• Setting — where it is: "on a marble countertop", "in a modern office", "against a plain studio backdrop".
• Lighting — "soft natural window light", "dramatic rim lighting", "bright even studio light" — lighting sells the mood.
• Style — photography vs illustration, and reference a look: "editorial photography", "flat vector design".
• Composition — "centered, negative space on the right for text", "close-up", "top-down flat lay".

Add unwanted elements to the Negative Prompt field: blurry, watermark, extra fingers, text artifacts, low contrast, cluttered background. This is the fastest fix when a generation looks "almost right" but noisy or low quality.

If you're not sure how to phrase an idea, type a rough sentence and click the Rewrite Prompt with AI icon above the prompt box. It expands your idea into a detailed, model-ready prompt — a fast way to learn what good prompts look like.

• One focal point. A single face, object or reaction — not five things competing for attention.
• Big emotion. Shock, curiosity, excitement — exaggerated expressions outperform neutral ones.
• High contrast. Subject should pop against the background so it reads at thumbnail size (as small as 120px wide in search results).
• 3–5 words max if you add on-image text — the title in the video panel already carries the rest.
• Curiosity gap. Show the "before" or the reaction, not the answer — the title + thumbnail together should make people need to click.

1. Enter your Video Topic / Title — the more specific, the better the AI understands the hook (e.g. "I tried the $1 vs $1000 burger" beats "food video").
2. Pick a Thumbnail Style that matches your niche: Gaming, Tech & Tutorial, Educational, Vlog/Lifestyle, Business/Finance, Entertainment, Motivational, or News/Review. Leave it on Auto if unsure.
3. Upload 1–4 Inspiration Thumbnails from creators in your niche whose style performs well — the AI matches composition and color energy, not the exact content.
4. Generate 2–3 variations and A/B test them, or swap the thumbnail after a few days if a video underperforms.
5. Turn on Apply Brand Guide → Overlay brand logo if you want consistent channel branding across every thumbnail.

Generate at Low quality first to check composition and readability at small size, then re-run your winning variation at High quality for the final upload. Keep a shortlist of 2–3 winning styles per niche in your Brand Library reference so future thumbnails stay visually consistent as your channel grows.

• Square (1024×1024) — Instagram/Facebook feed, general product shots.
• Landscape (1792×1024 / 1536×1024) — YouTube banners, website heroes, Facebook link ads.
• Portrait (1024×1792 / 1024×1536) — Stories, Reels, TikTok, Pinterest-style vertical ads.

Pick the size before you write the prompt — composition and negative-space direction should match where text/CTA overlays will go later.

Upload your actual product photo(s) under Reference Image (up to 8). On GPT-Image-1 this edits the real photo into a new scene; on other models it's used as a strong style hint. This keeps the product accurate instead of the AI inventing a lookalike — critical for ads where the product must match what's sold.

• Product Shot / Photorealistic — catalog, marketplace listings, retargeting ads.
• Lifestyle — top-of-funnel awareness ads, shows the product in use/context.
• Flat Design / Minimalist — app UI, SaaS, promo banners with text overlay.
• Cinematic / 3D Render — premium/tech products, launch announcements.

End your prompt with a composition instruction like "...with empty negative space on the left third for headline text". Ads with the product off-center and clean space for a CTA convert better than centered, edge-to-edge product shots — you (or your ad tool) need room to add the hook and button afterward.

• Always generate in the tall 2:3 portrait ratio — Pinterest crops anything else.
• Keep the top third clean for a bold headline overlay; that's what shows in the feed before someone taps.
• Bright, saturated colors and clear subjects outperform muted, busy scenes in the Pinterest feed.

Save your logo, brand colors and a reusable prompt template under Brand Library. Then on any generator, use Apply Brand Guide to auto-inject your brand's look and optionally overlay your logo — so a thumbnail, a pin and a product ad all feel like the same channel/brand without re-typing brand details every time.

• Low — fast, cheap drafts to test composition/idea before committing coins.
• Medium — everyday social posts and thumbnails.
• High — final assets for paid ads, print, or anything with heavy budget behind it.
• Auto — let FotoGen balance cost/quality when you're not sure.

Draft cheap, finalize expensive: iterate on Low, then re-run only your best prompt at High.

Generate multiple variations (2–4 images or 2–3 thumbnail variations) from the same prompt in one request, ship the top 2 to your ad platform or thumbnail slot, and let real click/conversion data pick the winner instead of guessing.

Found a high-performing ad or thumbnail from a competitor? Feed it into Prompt Extractor to get a usable prompt back, then swap in your own product/subject and brand details rather than starting from a blank page.

Save your best-performing prompts (from History) and reuse them as templates for new products or videos — a proven structure with a new subject is more reliable than reinventing the prompt each time.

• Too much in one image. One subject, one message — clutter reads as noise at feed/thumbnail size.
• Wrong aspect ratio for the platform. A square ad on Stories gets cropped and loses its focal point.
• No negative space for text/CTA. Plan composition for the overlay before generating, not after.
• Skipping the Low-quality draft. Spending High-quality coins on an untested idea wastes budget when the composition needed changing anyway.
• Inconsistent branding. Rotating logo placement/colors across posts makes a channel or store look untrustworthy — use Brand Library instead.

• Product ad: [product], on [surface], [lighting], product photography, centered with negative space on [side] for headline
• YouTube thumbnail topic: [specific hook/result], reacting with [exaggerated emotion], bold contrast, [niche] style
• Pinterest pin: [subject], bright saturated colors, flat lay / vertical composition, clean top third for headline
• Lifestyle shot: [product] being used by [person/context], [setting], natural lighting, candid, lifestyle photography