--- name: ad description: "Reference guide for AI advertising — a product (or the person's own) in three formats from one product sheet: a UGC video with the recurring AI character, a 15 s product commercial, static creatives with the offer text. Source experiments 9.10.2026 on the fictional OSKA sparkling cold brew: GPT Image 2.5 product sheet → MiniMax H3 Max (UGC 14 s, spot 15 s) or Seedance 2.5 720p for the final spot; historical ElevenLabs voice and Music settings, distinct from the author's preference to avoid ElevenLabs Sound Effects. This is not FlashBoards policy. Points to ad-* sub-skills. Use when the user says ad, advert, commercial, UGC ad, product video, packshot, banner, creative, реклама, рекламный ролик, UGC, баннер, креатив. Two external routes: the H3 draft method of this skill, and the quality route — Seedance 2.5 reference-to-video; the source preferred Seedance for quality, but the side-by-side measurement was a product commercial, not every format." --- These creative references describe source experiments with external tools. Use your own supported account and tools; this download installs no automation. Settings and timings are experimental observations, not current FlashBoards availability. Use only media and voices you have permission to use. # ad · reference workflow > **Source author's sound preference (9.10.2026), not FlashBoards policy:** avoid ElevenLabs Sound Effects (`/v1/sound-generation`, `fal-ai/elevenlabs/sound-effects`) for effects, room tone, ambience and jingles; the author found them strange and louder than the voices and preferred ElevenLabs for voices (TTS v4, Speech-to-Speech, Audio Isolation). Historical demonstrations below and in related guides still include ElevenLabs Music beds and an earlier Sound Effects ambience test; those are separate experiments, not a universal voices-only rule. For your own mix, use permissioned film sound (line-free gaps, Demucs without vocals), the video model's native sound or supported video-to-audio tools; keep effects under the voice. Measured on source experiments (9.10.2026, OSKA — a fictional sparkling cold brew can, the demo character Mara Vale as the creator). Article with every prompt and picture: https://flashboards.yaroflasher.com/learn/formats-and-exhibition/ai-video-ad/ ## 🏆 Source experiment pick 1. **The product is a picture, never words** — one product sheet (4 sides, the label close, the can in condensation) goes into every image and video → `ad-product-sheet`. 2. **UGC** — the character's voice (ElevenLabs v4 with tags) + a start frame with the product (GPT Image 2.5) + H3 Max reference-to-video 9:16 with beats tied to word times; room tone and a product sound mixed in → `ad-ugc`. 3. **Commercial** — 3 keyframes with the product (macro, the action, the end card with the slogan) + ElevenLabs music → H3 Max (draft) or **Seedance 2.5 720p** (final: closer cuts, finer liquids, its own synced sound effects) → `ad-spot`. 4. **Statics** — GPT Image 2.5 Flare medium draws the product, the scene and the offer text in one pass: 1:1 feed, 9:16 story with the character, 16:9 banner → `ad-static`. ## Two external routes: draft or quality | Route | Model | When | |---|---|---| | **Draft** — the method of this skill | MiniMax H3 Max / H3 Max Turbo, 480P | drafts, many takes, a tight budget | | 🏆 **Quality** | **Seedance 2.5** reference-to-video (`bytedance/seedance-2.5/reference-to-video`, `task: reference`): the same pictures as `image_urls` (@Image1…), the same audio as `audio_urls` (@Audio1…), 4–30 s a call, native synced sound, 480p / 720p / 1080p | the final cut, when the look matters | **Source observation (9.10.2026): Seedance 2.5 was preferred for quality; H3 remains the draft route.** Measured side by side on a 15 s product commercial (`ad-spot`): Seedance 2.5 at 720p — cut deviations up to about 1 s in the listed measurements (versus up to 2 s on H3), finer liquids and detail, its own sound effects in sync — took 7 min against 9 s on H3 Max 480P. Not yet measured on every format; write the same shot-by-shot prompt in plain text and refer to the references as @Image1, @Audio1. Compare the routes before choosing a generation. ## Classify | The job | Go to | |---|---| | Make the product reference | `ad-product-sheet` | | A creator talks about the product to camera | `ad-ugc` | | A cinematic product video, no people | `ad-spot` | | Banners, posts, stories with text | `ad-static` | ## Global rules 1. The product reference is one picture with every side; the prompt says "exactly this can, its label and text, every time it is on screen; only one can". 2. Text on screen only where asked; the end card text comes from a keyframe, not from the video prompt. 3. Never say "phone" in a UGC prompt — the camera "is never seen". 4. The source used cloud tools (fal, ElevenLabs) for heavy compute and local ffmpeg to mix and trim. Use your own supported environment; no Mac or author-specific runner is required. No demo product, character, voice files, music or local runner is delivered or required. Prepare your own references and permissioned audio.