commercial_ui_ad · v1
yuan.studio brand launch · static-composite layered motion
Brief
Replicate the DStudioproject "static-composite → layered-motion" pipeline through our ApiMart relay (vs. their Higgsfield platform), and use the result as yuan.studio's first brand asset. The "product" is the studio itself: a fictional landing-page mockup + an animated 5s version that holds UI text pixel-stable while the cinematic vista breathes and 3 mini-thumbnails micro-loop.
Gate 1 · static composite
Gate 2 · 5s i2v probe
Process · 4 gates
6 字段创意输入: walking-B Studio framing · brand=yuan.studio · headline=PERSONAL IP. AI-DIRECTED. · vibe=evidence-based creative AI studio · style=CAV · aspect=16:9
GPT Image 2 (ApiMart relay) · 1672×941 PNG · 全 4 项 pass: UI legibility · brand filter · 4-region layout · CAV style anchor
Seedance 2.0 i2v + native audio · 1920×1080 @ 5.04s · 全 4 项 pass: 0px UI 漂移 · cloud/wave/ticker 3 motion 自然 · brand 保持 · ambient audio
同 prompt · --duration 10 · 待跑 · 用作未来 brand asset 升级版
Verdict
First end-to-end project-internal sample. Pipeline previously held only an external observation (DStudioproject @ X · Higgsfield). Capability now graduates from 0 项目内 · 1 外部 to observed_once in our ledger. To upgrade to validated_2plus, repeat with a second independent brand under same gates.
Next
- → Production 10s 跑实(关 G2)
- → 第 2 个 brand 试做(升 validated_2plus)
- → 多 aspect 1:1 + 9:16(关 G4)
Featured IPs
canonical_works in _atlas.yaml to surface it here.
Built using
_atlas.yaml as the migration completes.
Disclosure
AI-generated content. Image: GPT Image 2 (text-to-image, no real-person reference). Video: Seedance 2.0 (image-to-video from gate1_v1.png reference). Audio: Seedance 2.0 native ambient generation. No real subjects, no real brands, no real client involved. yuan.studio is a self-promo asset.