# FlyAIgh

> FlyAIgh is a multi-model AI generation platform that gives creators access to every flagship AI video and image model — MiniMax H3, Sora 2, VEO 3.1, Seedance 2.0, Kling V3, Hailuo 2.3, Midjourney v7, Nano Banana Pro and more — from a single account. Users can generate videos and images, turn a one-line idea into a full storyboard with a screenplay and shot-by-shot video prompts (Director), build reusable workflows on an infinite canvas, and keep digital characters visually consistent across scenes.

FlyAIgh aggregates 36 leading AI video and image models behind a unified credit-based pricing system. It is built for content creators, designers, marketers, filmmakers, and anyone who needs to compare or chain outputs across multiple AI generation engines without paying per-provider subscriptions.

The platform is live at https://www.flyaigh.com and gives new accounts 25 free credits on signup, with no credit card required. Paid plans start at $11.99/month for 400 credits (Basic) and scale to $79.99/month for 3,600 credits (ULTRA).

## Core Pages

Core UI and main marketing pages support English, Simplified Chinese, Traditional Chinese, Spanish, French, German, Japanese, Korean, Russian, Brazilian Portuguese (pt-BR, /pt-BR) and Italian (it, /it). The language menu remembers the choice. Blog articles list only their actual translations; legal pages and model guides may remain English. There is no separate Portugal-Portuguese version.

- [Homepage](https://www.flyaigh.com): Product overview, hero video carousel of supported models, three pillars (Multi-Model / Canvas / Digital Characters), live prompt input.
- [Pricing](https://www.flyaigh.com/pricing): Six tiers — Free / Basic ($11.99/mo) / Basic Plus ($19.99/mo) / Basic Max ($31.99/mo) / Pro ($47.99/mo) / ULTRA ($79.99/mo). Every paid tier unlocks every model; the Free tier covers every image model plus the free-tier video models, with protected previews. Annual billing saves 20%. Every paid tier also includes clean downloads and MCP / API access; higher tiers add more credits at a lower price per credit, not extra models, resolutions or a faster queue.
- [Contact](https://www.flyaigh.com/contact): Support email agent@flyaigh.com, billing and partnership inquiries.

## Products

New purchases use Stripe. Where available, Alipay supports the Starter Pack and prepaid monthly/yearly plans without automatic renewal; cards support automatically renewing plans. Existing subscriptions stay with their original provider, including Creem. Available methods are shown at checkout. We do not store full card details. Prepaid plans expire at the end of the purchased month/year; unused plan credits expire and permanent credits remain. Active prepaid plans cannot be stacked or upgraded mid-term. Prices are USD; checkout shows applicable tax. Managed Payments restrictions must not be assumed to apply to standard Stripe Payments. Russian-issued cards, Mir and SBP remain unsupported.

- [Recast — AI Video Remake & Face Swap](https://www.flyaigh.com/recast): Remake STORM, APT and BAD video templates with your characters, or upload your own video plus a person photo to replace the whole person or only the face. Optional clean character-reference extraction; original sound kept by default on your uploaded video. Each job processes a selected window of up to 30 seconds. Seedance 2.5 (and, for person/face swap, Kling V4) accepts 30 seconds in a single pass; WAN 3.0, MiniMax H3 and Seedance 2.0 Pro / Fast (Flash) / Mini accept up to 15 seconds; Kling O3 (Kling V3 Omni) accepts up to 10 seconds. Switching models adjusts the selected duration. Users can choose a shorter section or automatic cutting, separate generation and joining, with a warning that segmented results may be less continuous than a single 30-second Seedance 2.5 or Kling V4 pass. Browsing is open to guests; video generation requires a prior purchase and credits, with the exact cost shown before generating. Direct person/face swap entry: https://www.flyaigh.com/recast?t=swap. Old /tools/recast links permanently redirect to /recast and preserve query parameters.
- [Music MV — AI Song & Music Video Maker](https://www.flyaigh.com/music): Describe an idea and AI writes the lyrics and makes two songs with Suno v6 (15 credits a run), or bring your own lyrics or MP3 / WAV. Then make a singing video (one person sings to camera, one camera or three rotating, up to 6 minutes) or a full MV (30–60 seconds of the song cut into singing, story and scene shots along the lyric timing; each shot can get its own first frame to review before filming, and one styling setting keeps the singer's outfit consistent). Singing shots are lip-synced with MiniMax H3 Enhanced; Seedance options are offered without a lip-sync guarantee. Your own song, photos, first frames and clips can replace any step; the song, SRT subtitles and the joined MP4 download separately. Video costs the model's per-second price (H3 Enhanced 480p: 2 credits/s, about 60 credits for 30 seconds), shown before generating; browsing is open to guests, and video generation requires a prior purchase.
- [Multi-Model Generation](https://www.flyaigh.com/generate): Access all video and image models from one interface. Supports text-to-video (t2v), image-to-video (i2v), first-last-frame video (firstlast), reference-to-video (r2v), text-to-image (t2i), and image-to-image edit (i2i).
- [Seedance 2.5 AI Video Generator](https://www.flyaigh.com/models/seedance-2.5): Dedicated Seedance 2.5 model guide and generator — single continuous takes from 4 to 30 seconds (the longest on FlyAIgh), native synchronized audio, up to 30 reference images, and 480p to 4K output.
- [MiniMax H3 AI Video Generator](https://www.flyaigh.com/models/minimax-h3): Dedicated H3 model guide and generator with 480p / 768p / 1080p and native 2K output, 5–15 second clips, stereo audio, first/last-frame control, and multimodal image, video, and audio references.
- [Video-to-Video AI](https://www.flyaigh.com/video-to-video-ai): Generate a new video from reference images, videos, and audio — restyle a clip, copy a camera move, keep a character consistent, or continue a shot. Powered by the Seedance and MiniMax H3 multimodal reference modes — up to 30 images on Seedance 2.5, or 9 images plus 3 videos and 3 audio on Seedance 2.0 / H3. A true regeneration guided by the references, not a frame-by-frame filter, so motion stays coherent.
- [Image to Video AI](https://www.flyaigh.com/image-to-video-ai): Upload one photo, describe the motion, and generate a video — almost every video model on FlyAIgh accepts a single reference image (VEO 3.1, Seedance 2.0, Kling V3, MiniMax H3, Hailuo 2.3, and more), each with its real per-model credit cost shown before generating. The basic single-photo sibling of the multi-reference video-to-video mode.
- [AI Animate Image](https://www.flyaigh.com/ai-animate-image): Animate any still image — portraits, old photos, products, artwork — into an MP4 (watermark-free on any paid account), on 19 animation-capable models with real per-clip pricing shown before you generate.
- [AI Wallpaper Generator](https://www.flyaigh.com/ai-wallpaper-generator): Generate custom phone and desktop wallpapers from a text description — image models, ratios for phone (9:16) and desktop (16:9), up to 4K on supporting models, names or text rendered into the design. Real per-image pricing shown before generating.
- [Free Storyboard Templates](https://www.flyaigh.com/storyboard-template): Three print-ready PDF storyboard templates — a classic 6-panel board, a 9-panel thumbnail grid, and a cinematic shot sheet carrying real film-grammar fields (shot size, camera angle, movement, cut type, audio). Available in A4 and US Letter, no email required, no watermark. FlyAIgh Director is the AI alternative: it fills in the same fields automatically from a one-line idea or script.
- [AI Storyboard Generator (Director)](https://www.flyaigh.com/ai-storyboard-generator): Turn one sentence into a complete storyboard — concept, visual style, a written screenplay, a consistent cast, and shot-by-shot video prompts. Six specialist agents run in sequence (concept → style → script → cast → storyboard → shot prompts). The screenplay pauses for the user to review, refine, and Lock before any downstream stage runs, so the user stays the writer. Each shot is compiled into a model-ready video prompt and can be routed to any supported video model. Planning is free; credits are spent only when generating images or video.
- [Canvas](https://www.flyaigh.com/canvas): Drag-and-drop visual workflows on an infinite canvas. Chain models together — generate a character on one model, animate it on another, polish prompts with AI — into a single reusable pipeline. Save flows as templates, run with one click.
- [Digital Characters](https://www.flyaigh.com/characters): Upload a few reference images, define a persona, and FlyAIgh keeps face, outfit and style locked across every video and image. Built on Kling V3 Omni, HappyHorse R2V, and Nano Banana Pro.
- [AI Image & Video Upscaler](https://www.flyaigh.com/ai-image-upscaler): Topaz-powered image and video upscaling — enlarge images up to 8x and video up to 4K while a model reconstructs real detail (not just stretching pixels). Five image models (Standard / High Detail / Recover / Art & CGI / Text & Graphics) and three video models (Standard / Faces / Generative), all on Topaz's official API. Usage-based pricing (per-megapixel for images, per-second for video), 6-credit minimum, credits auto-refunded on failure. First entry in FlyAIgh's TOOLS suite.
- [MCP Server for AI Agents](https://www.flyaigh.com/developers/mcp): Connect Claude Code, Codex, Cursor or any MCP client to FlyAIgh at https://www.flyaigh.com/api/mcp (Streamable HTTP, API key as a Bearer token). Eight tools: list_models, quote, generate_image, generate_video, get_job, list_my_generations, upload_file, get_balance. Same models and credit prices as the website, no API surcharge; requires a Basic subscription or above.

## Models Supported

### Video Models
- Seedance 2.5 (ByteDance): Flagship Seedance — single continuous takes from 4 up to 30 seconds (the longest on FlyAIgh), native audio, up to 30 image references in one generation (the richest on FlyAIgh), t2v/i2v/firstlast/r2v
- Seedance 2.0 Pro / Fast (ByteDance): Cinematic video with native audio and scene transitions, up to 4K (Pro), multimodal video-to-video (image + video + audio references), t2v/i2v/firstlast/r2v
- Seedance 2.0 Mini (ByteDance): Budget Seedance at half the Pro price — same multimodal reference stack (up to 9 images, 3 videos, 3 audio), native audio, up to 4K (upscaled), t2v/r2v only (no first/last-frame slots)
- FLUX 3 Video (Black Forest Labs): Text-to-video with synced audio and lip-sync, or keyframe-driven video — pin up to 10 stills on the timeline (1 opens the shot, 2 set first and last, 3-10 add evenly spaced beats) and the model animates between them; 5-20s, free native audio, t2v/i2v/firstlast/multi-keyframe
- Sora 2 (OpenAI): Official Sora 2 API access, t2v + i2v
- Sora 2 Pro (OpenAI): Higher quality variant, 720p / 1080p output
- VEO 3.1 Lite / Fast / Quality (Google): 8s clips up to 4K, three quality tiers
- Gemini Omni 1.1 Flash (Google): 3–10s clips in 16:9 or 9:16, 360p to native 4K, sound always included; t2v, i2v, first/last frame and reference-to-video with up to 10 images plus 3 reference videos of up to 3s each (reference videos are not billed)
- Kling V4 (Kuaishou): 3–30s single takes with stereo sound at 720p/1080p; t2v, i2v with optional last frame, and a reference mode with up to 10 images and 5 reference videos (30s total)
- Kling V3 (Kuaishou): t2v + i2v with first/last frame support, 720p/1080p, native sound with an on/off switch (no extra charge)
- Kling V3 Omni (Kuaishou): Unified text/image-to-video with multi-image reference syntax (up to 7 references, or 4 alongside a reference video); 4K cinema tier on every mode (4K does not accept a reference video)
- MiniMax H3 (MiniMax): 480p / 768p / 1080p and native 2K, 5–15s, t2v/i2v/firstlast/r2v with image, video, and audio references
- Hailuo 2.3 / Hailuo 2.3 Fast (MiniMax): SOTA instruction following with 15 camera commands
- HappyHorse 1.1 / 1.0 (Alibaba): Auto T2V/I2V/R2V/EDIT routing; 1.1 is the current default variant — sharper visuals, smoother motion, stronger cross-shot consistency (launch promo −40%)
- WAN 3.0 Video (Alibaba): Up to 30s in one take with native audio; multimodal references — up to 10 images, 5 videos and 5 audio clips, addressable from the prompt itself ("video 1", "image 2"); 480p/720p/1080p, t2v/i2v/firstlast/r2v
- WAN 2.7 Video (Alibaba): Multi-shot narrative video, t2v/i2v/firstlast/r2v
- Vidu Q4 Preview (ShengShu): Image-to-video and up to 12 reference images, 3–16s clips from 540p to 4K with native audio (no text-to-video)
- Vidu Q3 Pro (ShengShu): Full mode coverage — t2v/i2v/firstlast/reference
- Pixverse v6 / C1 (Pixverse): General-purpose t2v + an action/effects specialist (C1), first/last frame i2v
- Grok Imagine (xAI): Video generation with natural dynamics and audio

### Image Models
- GPT Image 2.5 Sunburst and Flare (OpenAI): OpenAI's newest image lines — Sunburst for fine detail, exact text and a Max quality tier, Flare for fast everyday edits; up to 16 reference images, t2i + i2i
- GPT Image 2 (OpenAI): Photorealistic image generation with superior text rendering, t2i + i2i
- Nano Banana 2.1 (Google): Newest Nano Banana — sharper detail and steadier multi-round edits, 1K/2K/4K, up to 14 reference images, 14 aspect ratios incl. 21:9 and 1:8, t2i + i2i
- Nano Banana 2 (Google): Advanced text-to-image generation
- Nano Banana Pro (Google): Flagship 4K image generation with image editing, up to 8 references
- WAN 2.7 Pro (Alibaba): Professional 4K image generation with thinking mode
- Midjourney v7 (Midjourney): Distinctive style, strong composition, draft/fast/turbo tiers
- Seedream 5.0 Lite (ByteDance): Versatile photorealistic image generation, 2K-3K output
- Seedream 5.0 Pro (ByteDance): Production-grade realism + precise image editing, up to 6 references, ultrawide 21:9
- Seedream 5.0 Pro Dola (ByteDance): Multi-reference editing variant — locks source structure while redrawing, up to 6 references, ultrawide 21:9
- Qwen Image 2.0 / Pro (Alibaba): Strong Chinese/English text rendering, unified t2i + edit
- Kling Image 3.0 / Omni (Kuaishou): Flagship t2i with strong subject consistency; Omni adds multi-image reference
- Z Image Turbo (Z.ai): Fast low-cost image generation, up to 1.5K

## Pricing

FlyAIgh uses a credit-based subscription system with 5 paid tiers plus a free tier. Every paid plan unlocks every model — no model gating between paid tiers. A free account can use every image model and the free-tier video models; its results are protected previews that cannot be downloaded.

New free image previews are medium-blurred with a centered FlyAIgh logo baked into the file. Any purchase, including the Starter Pack, unlocks the separate clear, logo-free originals. Existing previews are not automatically reprocessed; free video watermarks are unchanged.

Monthly plans receive the listed allocation each month. Annual plans receive 12× the listed monthly allocation immediately for the annual term. Unused subscription credits expire at the end of the applicable monthly or annual term and are replaced only after a successful renewal; grandfathered subscriptions retain their prior treatment.

- [Free Tier](https://www.flyaigh.com/pricing): Free to sign up, no credit card required. New accounts usually get 25 free credits, usable on every image model and on the free-tier video models (Grok Imagine and Hailuo 2.3 Fast); everything a free account makes is a protected preview that cannot be downloaded. Any purchase — the one-time $5.99 Starter Pack (200 credits that never expire) or any plan — unlocks every model and watermark-free downloads.
- [Starter Pack](https://www.flyaigh.com/pricing#starter-pack): US$5.99 once for 200 non-expiring credits, no automatic renewal. Available to never-paid accounts on the pricing page, in the homepage pricing section, on the in-app Billing page and via the Library image “Unlock clear original” button. Any purchase unlocks clean originals, including earlier work.
- [Basic Plan](https://www.flyaigh.com/pricing): $11.99/month or $9.59/month annual. 400 credits per monthly term, or 4,800 credits delivered up front per annual term. Unused subscription credits expire at renewal.
- [Basic Plus Plan](https://www.flyaigh.com/pricing): $19.99/month or $15.99/month annual. 720 credits/month with +8% volume bonus.
- [Basic Max Plan](https://www.flyaigh.com/pricing): $31.99/month or $25.59/month annual. 1,200 credits/month with +13% volume bonus.
- [Pro Plan](https://www.flyaigh.com/pricing): $47.99/month or $38.39/month annual. 2,000 credits/month with +25% volume bonus.
- [ULTRA Plan](https://www.flyaigh.com/pricing): $79.99/month or $63.99/month annual. 3,600 credits/month with +35% volume bonus.

**Real-money cost per usable clip**: Credits are worth about $0.03 each on the Basic plan ($11.99 / 400 credits), dropping to roughly $0.018 on ULTRA annual. So a short video clip costs about $0.33 (Hailuo Fast, 6s, 11 credits) to $2.04 (Sora 2 Pro, 4s, 68 credits) on Basic; most flagship 5-second clips land near $1.20–$1.70 (Kling V3 $1.20, Seedance 2.0 Pro $1.68), while Sora 2's 4-second clip is $0.72. Failed generations are auto-refunded, so the per-clip price is the cost per *usable* clip — you never pay for a broken render.

Credit costs by model (per 5-second video unless noted):
- VEO 3.1 Lite (8s fixed): 24 credits
- Hailuo 2.3 Fast (6s): 11 credits
- Grok Imagine (6s): 18 credits
- Hailuo 2.3 (6s): 18 credits
- Sora 2 (4s): 24 credits
- Pixverse v6 / Pixverse C1: 30 credits
- Seedance 2.0 Mini: 28 credits
- VEO 3.1 Fast (8s fixed): 68 credits with sound, 45 with sound off
- Seedance 2.0 Fast: 45 credits
- Kling V4: 64 credits (720p; 12.7 per second, about 19.4 at 1080p; with a reference video 7.4 / 11.3 per billed second = output + the longer of reference and output)
- Kling V3 / Kling V3 Omni / Vidu Q3 Pro: 40 credits (Omni 4K tier is 4.75x: 190 credits per 5s)
- Vidu Q4 Preview (720p): 30 credits (6/s; 540p 10, 1080p 37, 2K 60, 4K 120 per 5s; sound included)
- MiniMax H3: 14 credits (5s, 768p; was 21) — 480p 10 / 1080p 20 / native 2K 36
- Seedance 2.5: 77 credits (15.3/s at 720p; up to 30s takes)
- FLUX 3 Video: 57 credits (11.3/s at HD; a Draft tier costs about a third, FHD about 1.7x)
- Seedance 2.0 Pro: 56 credits (native 1080p / 4K tiers cost more)
- WAN 3.0 Video: 40 credits (7.9/s at 1080p, limited-time -30% off the 11.3/s list rate; 720p is half, 480p a quarter)
- WAN 2.7 Video: 45 credits
- VEO 3.1 Quality (8s fixed): 180 credits with sound, 90 with sound off
- Gemini Omni 1.1 Flash: 37 credits per 5s at 720p (7.3 credits/s; 360p x0.273, 1080p x1.545, 4K x3), sound included
- HappyHorse 1.1 (launch promo −40%): 33 credits
- HappyHorse 1.0: 75 credits
- Sora 2 Pro (4s, 720p): 68 credits

Image credit costs per image (headline price; scales with quality / resolution tier):
- Z Image Turbo: 1 credit
- GPT Image 2.5 (Sunburst / Flare, same price): 1–40 credits (token-billed by quality × resolution × aspect; low/1K from 1 credit, max/4K/1:1 40)
- GPT Image 2: 1–34 credits (flat by quality × resolution)
- Seedream 5.0 Lite: 3 credits
- Seedream 5.0 Pro: 5 credits (1K 3 / 2K 5)
- Seedream 5.0 Pro Dola: 5 credits (1K 3 / 2K 5)
- Qwen Image 2.0: 3 credits
- WAN 2.7 Pro / Qwen Image 2.0 Pro / Kling Image 3.0: 5 credits
- Midjourney v7: 6 credits; Midjourney v8.1 and v8.2 Preview: 6 (draft) / 8 (fast)
- Nano Banana 2.1: 2 credits (1K 2 / 2K 3 / 4K 7)
- Nano Banana 2: 4 credits
- Nano Banana Pro: 9 credits
- Kling Image 3.0 Omni: 8 credits

## Key Differentiators

- **Single account, every model**: No need for separate subscriptions to OpenAI Sora, Google VEO, Midjourney, etc.
- **Multi-provider failover**: When the primary provider has an outage, requests automatically route to an alternative (APIPod → APIMart → EVOLINK chain). Users get reliable generation even during provider incidents.
- **Visual workflow canvas**: Drag nodes, connect models, save reusable pipelines. Most AI tools force one generation at a time.
- **Character consistency**: Built-in tooling to maintain visual identity across generations, a problem most single-model tools cannot solve.
- **Shot-to-shot continuity (Director)**: The storyboard agent plans the cut between every shot — match cuts, J- and L-cuts, shared frames — and chains the last frame of one shot into the next, so generated shots splice into one continuous sequence instead of disconnected clips. Most AI video tools leave cross-cut continuity to the user.
- **Automatic refunds**: Failed generations automatically refund credits within minutes.
- **Billing-cycle refresh**: Subscription credits do not roll over. Monthly allocations expire and refresh at monthly renewal; annual allocations are delivered up front and expire at annual renewal. Grandfathered subscriptions retain their prior treatment.

## Legal

- [Terms of Service](https://www.flyaigh.com/terms)
- [Privacy Policy](https://www.flyaigh.com/privacy)
- [Refund Policy](https://www.flyaigh.com/refund-policy)
- [Acceptable Use Policy](https://www.flyaigh.com/acceptable-use)

## Character Replacement Guide

- English: https://www.flyaigh.com/blog/guide-replace-person-in-video-ai
- Chinese: https://www.flyaigh.com/zh/blog/guide-replace-person-in-video-ai
- Traditional Chinese: https://www.flyaigh.com/zh-TW/blog/guide-replace-person-in-video-ai
- Spanish: https://www.flyaigh.com/es/blog/guide-replace-person-in-video-ai
- French: https://www.flyaigh.com/fr/blog/guide-replace-person-in-video-ai
- German: https://www.flyaigh.com/de/blog/guide-replace-person-in-video-ai
- Japanese: https://www.flyaigh.com/ja/blog/guide-replace-person-in-video-ai
- Korean: https://www.flyaigh.com/ko/blog/guide-replace-person-in-video-ai
- Russian: https://www.flyaigh.com/ru/blog/guide-replace-person-in-video-ai
- Brazilian Portuguese: https://www.flyaigh.com/pt-BR/blog/guide-replace-person-in-video-ai
- Italian: https://www.flyaigh.com/it/blog/guide-replace-person-in-video-ai
- A practical workflow for replacing a video character using a photo and a source clip, with STORM and BAD comparisons, the difference between face-only and whole-person replacement, observed motion deviations, Recast duration limits, sound options and billing scope.
- The examples demonstrate whole-character and scene changes, not exact choreography or pixel-preserving face replacement. Browsing is free; video generation requires a paid account and credits.

## Optional

- [Full content reference](https://www.flyaigh.com/llms-full.txt): Long-form markdown dump of the full product, model catalog, FAQ, and policies for one-shot ingestion.
- [Sitemap](https://www.flyaigh.com/sitemap.xml): Machine-readable list of indexable pages.

## Make your character follow an existing dance

- English: https://www.flyaigh.com/blog/guide-make-character-dance-ai
- Chinese: https://www.flyaigh.com/zh/blog/guide-make-character-dance-ai
- Traditional Chinese: https://www.flyaigh.com/zh-TW/blog/guide-make-character-dance-ai
- Spanish: https://www.flyaigh.com/es/blog/guide-make-character-dance-ai
- French: https://www.flyaigh.com/fr/blog/guide-make-character-dance-ai
- German: https://www.flyaigh.com/de/blog/guide-make-character-dance-ai
- Japanese: https://www.flyaigh.com/ja/blog/guide-make-character-dance-ai
- Korean: https://www.flyaigh.com/ko/blog/guide-make-character-dance-ai
- Russian: https://www.flyaigh.com/ru/blog/guide-make-character-dance-ai
- Brazilian Portuguese: https://www.flyaigh.com/pt-BR/blog/guide-make-character-dance-ai
- Italian: https://www.flyaigh.com/it/blog/guide-make-character-dance-ai
- Real BAD mannequin → human office remake → tabby comparison, character image preparation, BAD template and own-video workflows, motion review, and music handling. Reference-guided generation does not guarantee exact choreography.

## Change a video background: method choices and a real comparison

- English: https://www.flyaigh.com/blog/guide-change-video-background-ai
- Chinese: https://www.flyaigh.com/zh/blog/guide-change-video-background-ai
- Traditional Chinese: https://www.flyaigh.com/zh-TW/blog/guide-change-video-background-ai
- Spanish: https://www.flyaigh.com/es/blog/guide-change-video-background-ai
- French: https://www.flyaigh.com/fr/blog/guide-change-video-background-ai
- German: https://www.flyaigh.com/de/blog/guide-change-video-background-ai
- Japanese: https://www.flyaigh.com/ja/blog/guide-change-video-background-ai
- Korean: https://www.flyaigh.com/ko/blog/guide-change-video-background-ai
- Russian: https://www.flyaigh.com/ru/blog/guide-change-video-background-ai
- Brazilian Portuguese: https://www.flyaigh.com/pt-BR/blog/guide-change-video-background-ai
- Italian: https://www.flyaigh.com/it/blog/guide-change-video-background-ai
- Foreground compositing versus reference-guided scene generation, with an office human input and separately generated office and rooftop cat variants. Covers character and scene references, BAD background options, motion, floor contact, lighting, sound and quotes. The case does not establish an unchanged original performer or exact motion.

## Make an outfit change video from one image

- en: https://www.flyaigh.com/blog/guide-make-outfit-change-video-ai
- zh: https://www.flyaigh.com/zh/blog/guide-make-outfit-change-video-ai
- zh-TW: https://www.flyaigh.com/zh-TW/blog/guide-make-outfit-change-video-ai
- es: https://www.flyaigh.com/es/blog/guide-make-outfit-change-video-ai
- fr: https://www.flyaigh.com/fr/blog/guide-make-outfit-change-video-ai
- de: https://www.flyaigh.com/de/blog/guide-make-outfit-change-video-ai
- ja: https://www.flyaigh.com/ja/blog/guide-make-outfit-change-video-ai
- ko: https://www.flyaigh.com/ko/blog/guide-make-outfit-change-video-ai
- ru: https://www.flyaigh.com/ru/blog/guide-make-outfit-change-video-ai
- pt-BR: https://www.flyaigh.com/pt-BR/blog/guide-make-outfit-change-video-ai
- it: https://www.flyaigh.com/it/blog/guide-make-outfit-change-video-ai
- A prepared SUMI illustration becomes a 16-second Vidu Q4 Preview costume clip. Includes first-frame input branches, the exact submitted English prompt with localized guidance, requested versus observed timing, and result-selection and editing limits. This is image-to-video, not proof of clothes-only editing of existing footage or exact garment matching.

## Replace two characters in a video

- en: https://www.flyaigh.com/blog/guide-replace-two-characters-in-video-ai
- zh: https://www.flyaigh.com/zh/blog/guide-replace-two-characters-in-video-ai
- zh-TW: https://www.flyaigh.com/zh-TW/blog/guide-replace-two-characters-in-video-ai
- es: https://www.flyaigh.com/es/blog/guide-replace-two-characters-in-video-ai
- fr: https://www.flyaigh.com/fr/blog/guide-replace-two-characters-in-video-ai
- de: https://www.flyaigh.com/de/blog/guide-replace-two-characters-in-video-ai
- ja: https://www.flyaigh.com/ja/blog/guide-replace-two-characters-in-video-ai
- ko: https://www.flyaigh.com/ko/blog/guide-replace-two-characters-in-video-ai
- ru: https://www.flyaigh.com/ru/blog/guide-replace-two-characters-in-video-ai
- pt-BR: https://www.flyaigh.com/pt-BR/blog/guide-replace-two-characters-in-video-ai
- it: https://www.flyaigh.com/it/blog/guide-replace-two-characters-in-video-ai
- Assign separate references in the APT duo template, replace one or both roles, and check each identity after every cut. Includes the full 14-second source and a historical WAN 3.0 result with a missed drummer replacement and inherited cat features; this is not a fully successful swap or a guarantee of stable identities.

## Character reference images from one image

Verified tabby input and generated front/back/face sheet, original prompt, design review and reference-versus-first-frame guidance. One historical image-edit example; no video consistency guarantee.
- [How to Prepare Character Reference Images for AI Video from One Image](https://www.flyaigh.com/blog/guide-prepare-character-reference-images-ai-video) (en)
- [只有一张角色图，如何准备 AI 视频参考图？](https://www.flyaigh.com/zh/blog/guide-prepare-character-reference-images-ai-video) (zh)
- [只有一張角色圖，如何準備 AI 影片參考圖？](https://www.flyaigh.com/zh-TW/blog/guide-prepare-character-reference-images-ai-video) (zh-TW)
- [Cómo preparar referencias de personaje para vídeo con IA a partir de una imagen](https://www.flyaigh.com/es/blog/guide-prepare-character-reference-images-ai-video) (es)
- [Préparer des références de personnage pour une vidéo IA à partir d’une image](https://www.flyaigh.com/fr/blog/guide-prepare-character-reference-images-ai-video) (fr)
- [Charakterreferenzen für KI-Videos aus einem einzigen Bild vorbereiten](https://www.flyaigh.com/de/blog/guide-prepare-character-reference-images-ai-video) (de)
- [キャラクター画像１枚からAI動画用の参照画像を準備する方法](https://www.flyaigh.com/ja/blog/guide-prepare-character-reference-images-ai-video) (ja)
- [캐릭터 이미지 한 장으로 AI 영상용 참조 이미지 준비하기](https://www.flyaigh.com/ko/blog/guide-prepare-character-reference-images-ai-video) (ko)
- [Как подготовить референсы персонажа для ИИ-видео из одного изображения](https://www.flyaigh.com/ru/blog/guide-prepare-character-reference-images-ai-video) (ru)
- [Como preparar referências de personagem para vídeo com IA a partir de uma imagem](https://www.flyaigh.com/pt-BR/blog/guide-prepare-character-reference-images-ai-video) (pt-BR)
- [Come preparare riferimenti del personaggio per video IA da una sola immagine](https://www.flyaigh.com/it/blog/guide-prepare-character-reference-images-ai-video) (it)

## AI video transition between two images

Start/end frame workflow, image selection, original and adaptable prompts, and checks for cuts or endpoint mismatch. Historical Seedance 2.0 case: two archived images and a full four-second result with a visible mid-clip viewpoint change. Provenance checked through linked storyboard/generation records and visual comparison; no retained provider request payload. Not a seamless-transition or success-rate claim.
- [How to Create an AI Video Transition Between Two Images](https://www.flyaigh.com/blog/guide-ai-video-transition-between-two-images) (en)
- [两张图片如何生成过渡视频？首尾帧与提示词实例](https://www.flyaigh.com/zh/blog/guide-ai-video-transition-between-two-images) (zh)
- [兩張圖片如何生成過渡影片？首尾幀與提示詞實例](https://www.flyaigh.com/zh-TW/blog/guide-ai-video-transition-between-two-images) (zh-TW)
- [Cómo crear una transición de vídeo con IA entre dos imágenes](https://www.flyaigh.com/es/blog/guide-ai-video-transition-between-two-images) (es)
- [Créer une transition vidéo IA entre deux images](https://www.flyaigh.com/fr/blog/guide-ai-video-transition-between-two-images) (fr)
- [KI-Videoübergang zwischen zwei Bildern erstellen](https://www.flyaigh.com/de/blog/guide-ai-video-transition-between-two-images) (de)
- [2枚の画像からAI動画を作るには？開始・終了フレームとプロンプト](https://www.flyaigh.com/ja/blog/guide-ai-video-transition-between-two-images) (ja)
- [두 이미지 사이를 연결하는 AI 영상 만들기: 시작·끝 프레임과 프롬프트](https://www.flyaigh.com/ko/blog/guide-ai-video-transition-between-two-images) (ko)
- [Как создать ИИ-переход между двумя изображениями](https://www.flyaigh.com/ru/blog/guide-ai-video-transition-between-two-images) (ru)
- [Como criar uma transição de vídeo com IA entre duas imagens](https://www.flyaigh.com/pt-BR/blog/guide-ai-video-transition-between-two-images) (pt-BR)
- [Come creare una transizione video AI tra due immagini](https://www.flyaigh.com/it/blog/guide-ai-video-transition-between-two-images) (it)

## Make a music video from your own song

Song and character-photo workflow for a short MV. Last Train Home case: source audio section 0:51–1:39, original character and shot images, and the complete 48-second film. Singing, story and scenery shot plan; audio-driven lip sync versus adding a soundtrack; wardrobe differences; current 30–60-second MV limit; separate SRT export. Full audiovisual review does not establish frame-accurate lip sync. Video generation requires a paid account and credits.
- [How to Make an AI Music Video from Your Own Song and a Character Photo](https://www.flyaigh.com/blog/guide-make-ai-music-video-from-your-own-song) (en)
- [已有一首歌，如何用角色照片做一段 AI 音乐 MV？](https://www.flyaigh.com/zh/blog/guide-make-ai-music-video-from-your-own-song) (zh)
- [如何用自己的歌與角色照片製作 AI 音樂 MV？](https://www.flyaigh.com/zh-TW/blog/guide-make-ai-music-video-from-your-own-song) (zh-TW)
- [Cómo crear un videoclip con IA usando tu canción y una foto](https://www.flyaigh.com/es/blog/guide-make-ai-music-video-from-your-own-song) (es)
- [Créer un clip musical IA avec sa propre chanson et une photo](https://www.flyaigh.com/fr/blog/guide-make-ai-music-video-from-your-own-song) (fr)
- [KI-Musikvideo mit eigenem Song erstellen: So geht’s mit einem Charakterfoto](https://www.flyaigh.com/de/blog/guide-make-ai-music-video-from-your-own-song) (de)
- [自分の曲とキャラクター写真でAIミュージックビデオを作る方法](https://www.flyaigh.com/ja/blog/guide-make-ai-music-video-from-your-own-song) (ja)
- [내 노래로 AI 뮤직비디오 만들기: 캐릭터 사진 한 장으로 시작하기](https://www.flyaigh.com/ko/blog/guide-make-ai-music-video-from-your-own-song) (ko)
- [Как создать клип на свою песню с ИИ и фото персонажа](https://www.flyaigh.com/ru/blog/guide-make-ai-music-video-from-your-own-song) (ru)
- [Como criar um videoclipe com IA usando sua música e uma foto](https://www.flyaigh.com/pt-BR/blog/guide-make-ai-music-video-from-your-own-song) (pt-BR)
- [Come creare un videoclip AI con la tua canzone e una foto](https://www.flyaigh.com/it/blog/guide-make-ai-music-video-from-your-own-song) (it)

## Make a photo sing your own song

Photo-and-audio singing tutorial. Original Mandarin song Deng Huo Man: exact camera A input, 27.75–37.875-second driving audio and full 10.125-second result; B/C inputs and complete 30.5-second three-camera film. Historical H3 Enhanced API test with FFmpeg assembly; current Singing video instructions are separate. Covers clear vocals, visible mouths, camera changes, lyric boundaries, failed-shot retries, and full-song limits. The example shows gaze and expression resets; review does not establish frame-accurate lip sync or quality in every language. Video generation requires a paid account and credits.
- [How to Make a Photo Sing Your Own Song](https://www.flyaigh.com/blog/guide-make-photo-sing-your-own-song) (en)
- [一张照片如何跟着自己的歌曲唱歌？](https://www.flyaigh.com/zh/blog/guide-make-photo-sing-your-own-song) (zh)
- [一張照片如何跟著自己的歌曲唱歌？](https://www.flyaigh.com/zh-TW/blog/guide-make-photo-sing-your-own-song) (zh-TW)
- [Cómo hacer cantar una foto con tu propia canción](https://www.flyaigh.com/es/blog/guide-make-photo-sing-your-own-song) (es)
- [Comment faire chanter une photo avec sa propre chanson](https://www.flyaigh.com/fr/blog/guide-make-photo-sing-your-own-song) (fr)
- [Foto mit eigenem Song singen lassen: So geht's](https://www.flyaigh.com/de/blog/guide-make-photo-sing-your-own-song) (de)
- [写真を歌わせる方法：自分の曲でリップシンク動画を作る](https://www.flyaigh.com/ja/blog/guide-make-photo-sing-your-own-song) (ja)
- [내 노래로 사진이 노래하게 만드는 방법](https://www.flyaigh.com/ko/blog/guide-make-photo-sing-your-own-song) (ko)
- [Как заставить фото петь вашу песню](https://www.flyaigh.com/ru/blog/guide-make-photo-sing-your-own-song) (ru)
- [Como fazer uma foto cantar com sua própria música](https://www.flyaigh.com/pt-BR/blog/guide-make-photo-sing-your-own-song) (pt-BR)
- [Come far cantare una foto con la propria canzone](https://www.flyaigh.com/it/blog/guide-make-photo-sing-your-own-song) (it)

## Make an AI product video from photos

Product-photo advertising workflow: select source photos, inspect generated references, check the ad plan, and review the full clip. Brewista case: the original photo directly produced the complete 20-second WAN 3.0 video, with music and voiceover added later. A separate next-day clean-reference experiment has a plan but no finished film. Includes original instructions, an adaptable brief, visible tag/lettering changes, and current sound/retry limits. Not an official brand ad or a guarantee of product fidelity. Video generation requires a paid account and credits.
- [How to Make an AI Product Video from Photos](https://www.flyaigh.com/blog/guide-make-ai-product-video-from-photos) (en)
- [只有产品照片，如何做一条 AI 产品广告短片？](https://www.flyaigh.com/zh/blog/guide-make-ai-product-video-from-photos) (zh)
- [只有產品照片，如何製作 AI 產品廣告短片？](https://www.flyaigh.com/zh-TW/blog/guide-make-ai-product-video-from-photos) (zh-TW)
- [Cómo crear un video de producto con IA a partir de fotos](https://www.flyaigh.com/es/blog/guide-make-ai-product-video-from-photos) (es)
- [Créer une vidéo de produit avec l’IA à partir de photos](https://www.flyaigh.com/fr/blog/guide-make-ai-product-video-from-photos) (fr)
- [Ein KI-Produktvideo aus Produktfotos erstellen](https://www.flyaigh.com/de/blog/guide-make-ai-product-video-from-photos) (de)
- [商品写真からAI商品紹介動画を作る方法](https://www.flyaigh.com/ja/blog/guide-make-ai-product-video-from-photos) (ja)
- [제품 사진으로 AI 제품 홍보 영상 만들기](https://www.flyaigh.com/ko/blog/guide-make-ai-product-video-from-photos) (ko)
- [Как создать рекламное видео товара из фотографий с ИИ](https://www.flyaigh.com/ru/blog/guide-make-ai-product-video-from-photos) (ru)
- [Como criar um vídeo de produto com IA a partir de fotos](https://www.flyaigh.com/pt-BR/blog/guide-make-ai-product-video-from-photos) (pt-BR)
- [Come creare un video di prodotto con l’IA a partire dalle foto](https://www.flyaigh.com/it/blog/guide-make-ai-product-video-from-photos) (it)