Emily2040 · Emily2040 · May 30, 2026 · May 30, 2026
@@ -2,6 +2,20 @@
 
 All notable changes to this project are documented here.
 
+## [5.4.5] — 2026-05-30
+
+### Added
+
+- Added seven generated visual-gallery assets: two cinematic hero shots and five text-rich infographics for skill capabilities, CDN delivery, reference roles, production delivery, and QC.
+- Added README visual-gallery coverage so the front page shows the skill as a professional filmmaker operating system instead of a single generic image.
+- Added eval coverage for the six-plus-image visual-gallery requirement.
+
+### Changed
+
+- Updated the README hero, badges, design standard, frontend redesign notes, and frontend design-system rules for text-rich infographic assets.
+- Strengthened `scripts/design_audit.py` to require the visual gallery, validate PNG headers, enforce minimum dimensions, and fail stale visual guidance.
+- Bumped active skill metadata, validator expectations, and eval metadata to v5.4.5.
+
 ## [5.4.4] — 2026-05-30
 
 ### Added

@@ -1,22 +1,22 @@
-![Seedance 2.0 Skill OS cinematic hero: camera lens, reference frames, film strip, and audio waveform](assets/hero-cinematic.png)
+![Seedance 2.0 Skill OS cinematic command-center hero: brief, references, prompt, post, QC, subtitles, audio waveform, and shot cards](assets/hero-command-center.png)
 
 # Seedance 2.0 Skill OS
 
 **Intent-first AI filmmaking for Seedance 2.0.**
 
 Text-to-video · Image-to-video · Video-to-video · Reference-to-video · Audio-aware prompting · Copyright-safe rewrites · Agent Skills
 
-[![Version](https://img.shields.io/badge/version-5.4.4-111827?labelColor=0f172a)](#changelog)
+[![Version](https://img.shields.io/badge/version-5.4.5-111827?labelColor=0f172a)](#changelog)
 [![Skills](https://img.shields.io/badge/sub--skills-23-0ea5e9?labelColor=0f172a)](#skill-map)
 [![References](https://img.shields.io/badge/references-40-8b5cf6?labelColor=0f172a)](#reference-library)
-[![Evals](https://img.shields.io/badge/evals-51-22c55e?labelColor=0f172a)](#validation)
+[![Evals](https://img.shields.io/badge/evals-52-22c55e?labelColor=0f172a)](#validation)
 [![License](https://img.shields.io/badge/license-MIT-f59e0b?labelColor=0f172a)](LICENSE)
 
 Author: [Iamemily2050 (@iamemily2050)](https://github.com/Emily2040) · [Instagram](https://instagram.com/iamemily2050) · [X](https://x.com/iamemily2050) · [Website](https://iamemily2050.com)
 
 Platform context: [ByteDance Seedance 2.0](https://seed.bytedance.com/en/seedance2_0) · Dreamina · Jimeng · Doubao · [Volcengine Ark](https://www.volcengine.com/docs/82379/2291680?lang=zh) · [BytePlus ModelArk](https://docs.byteplus.com/en/docs/ModelArk/2291680) · [Runway Seedance 2](https://docs.dev.runwayml.com/guides/seedance/)
 
-Updated: **2026-05-30** · **v5.4.4 professional filmmaker workflow expansion**
+Updated: **2026-05-30** · **v5.4.5 visual stress-test gallery and professional infographic expansion**
 
 ---
 
@@ -103,7 +103,7 @@ The v5.4 release line adds a dated research layer for safer data mining and plat
 
 ![Seedance 2.0 Skill OS infographic: source registry, prompt router, multimodal references, safety gates, and eval loop](assets/skill-os-infographic.png)
 
-The visual map is intentionally text-light so it remains clean in GitHub and reusable in docs. It represents the six lanes this package keeps separate:
+The operating-system map stays simple for GitHub readability; the visual gallery below adds the text-rich professional infographics. Together they represent the six lanes this package keeps separate:
 
 - Research sources: dated official, academic, platform, and community evidence.
 - Production spine: brief, shot list, continuity, post handoff, localization, and delivery/QC.
@@ -112,6 +112,28 @@ The visual map is intentionally text-light so it remains clean in GitHub and reu
 - Safety gates: IP, likeness, voice, brand, real-person, filter, and platform-policy checks.
 - Quality evals: schema checks, source freshness, vocabulary integrity, design audit, and behavior cases.
 
+## Visual Gallery
+
+The README now includes a committed visual set rather than a single generic hero. These generated bitmap assets are paired with searchable Markdown so the images can be cinematic while the repo remains auditable.
+
+### Hero Shots
+
+![Seedance 2.0 command-center hero showing brief, references, prompt, post, QC, subtitles, audio waveform, and shot cards](assets/hero-command-center.png)
+
+![Global filmmaker mode hero showing director, DP, editor, colorist, sound mixer, localization lead, and QC lead on a cinematic production stage](assets/hero-global-filmmaker-mode.png)
+
+### Text-Rich Infographics
+
+![What this skill can do infographic: brief, references, prompt, generate, post, deliver](assets/infographic-skill-capabilities.png)
+
+![CDN video delivery map infographic: creator, origin, CDN edge, global review, delivery, fast playback, regional cache, version control, and QC before publish](assets/infographic-cdn-delivery-map.png)
+
+![Reference role map infographic: image equals identity, video equals motion, audio equals timing](assets/infographic-reference-role-map.png)
+
+![Production to delivery infographic: brief, shot list, generate, edit, localize, QC](assets/infographic-production-delivery.png)
+
+![Professional QC stack infographic: picture, color, audio, text, rights, metadata](assets/infographic-professional-qc-stack.png)
+
 ## Skill Map
 
 ![Seedance 2.0 cinematic skill map: modular skill clusters around an AI filmmaking director console](assets/skill-map-cinematic.png)
@@ -241,11 +263,11 @@ The CI workflow runs the same checks on push and pull request.
 
 ## Design Standard
 
-The v5.4 front page uses a generated cinematic bitmap hero, a generated operating-system infographic, a generated cinematic skill-map infographic, and the cleaned v5.2 information architecture. The README should stay readable in GitHub mobile, dark mode, and narrow widths. SVG assets must include `<title>` and `<desc>` elements, use internal CSS only, and avoid external fonts or scripts. See [`docs/frontend-redesign.md`](docs/frontend-redesign.md).
+The v5.4.5 front page uses multiple generated cinematic bitmap hero shots, text-rich infographics, a generated operating-system infographic, a generated cinematic skill-map infographic, and the cleaned v5.2 information architecture. The README should stay readable in GitHub mobile, dark mode, and narrow widths. SVG assets must include `<title>` and `<desc>` elements, use internal CSS only, and avoid external fonts or scripts. See [`docs/frontend-redesign.md`](docs/frontend-redesign.md).
 
 ## Changelog
 
-See [`CHANGELOG.md`](CHANGELOG.md). Current release: **v5.4.4**.
+See [`CHANGELOG.md`](CHANGELOG.md). Current release: **v5.4.5**.
 
 ## License
 

@@ -5,7 +5,7 @@ license: MIT
 user-invocable: true
 tags: [seedance]
 metadata:
-  version: "5.4.4"
+  version: "5.4.5"
 ---
 
 # seedance-20

@@ -2,14 +2,14 @@
 
 The repository does not currently contain a standalone web app. The public frontend is the GitHub README, generated bitmap hero/infographic assets, and SVG support diagrams.
 
-## v5.4.2 Design Goals
+## v5.4.5 Design Goals
 
 - Lead with a real cinematic production-control hero instead of generic abstract graphics.
 - Show Seedance's practical range: references, first/last-frame continuity, product reveal, timeline control, audio, and camera direction.
-- Use generated cinematic infographics for both the operating-system overview and skill map.
-- Keep bitmap assets text-free so GitHub renders all important content as searchable Markdown.
+- Use generated cinematic infographics for the operating-system overview, skill map, skill capability map, CDN delivery map, reference role map, production delivery map, and QC stack.
+- Allow text-rich infographics when the text is large, corrected, visually balanced, and repeated in nearby searchable Markdown.
 - Keep SVG assets as support diagrams, not the primary emotional surface.
-- Validate README completeness and asset presence with `scripts/design_audit.py`.
+- Validate README completeness, gallery coverage, PNG dimensions, and asset presence with `scripts/design_audit.py`.
 
 ## v5.2 Design Goals
 
@@ -22,6 +22,13 @@ The repository does not currently contain a standalone web app. The public front
 
 ## Assets
 
+- `assets/hero-command-center.png`
+- `assets/hero-global-filmmaker-mode.png`
+- `assets/infographic-skill-capabilities.png`
+- `assets/infographic-cdn-delivery-map.png`
+- `assets/infographic-reference-role-map.png`
+- `assets/infographic-production-delivery.png`
+- `assets/infographic-professional-qc-stack.png`
 - `assets/hero-cinematic.png`
 - `assets/skill-os-infographic.png`
 - `assets/skill-map-cinematic.png`
@@ -35,5 +42,6 @@ The repository does not currently contain a standalone web app. The public front
 - Every SVG needs `<title>` and `<desc>`.
 - README should stay readable on mobile and dark mode.
 - Avoid dense badge walls and noisy decorative text.
-- Keep generated bitmap hero assets text-free; render all product/project text as real Markdown.
-- Keep the infographic text-light; README prose should explain the lanes with accessible Markdown.
+- Use text-rich generated infographics only for big, short labels that remain legible at README width.
+- Inspect every generated text image; reject garbled words, ugly typography, poor contrast, or placeholder-looking panels before commit.
+- Keep equivalent Markdown explanation beside every text-rich image so the repo remains accessible and searchable.
@@ -1,5 +1,5 @@
 {
-  "version": "5.4.4",
+  "version": "5.4.5",
   "updated": "2026-05-30",
   "cases": [
     {
@@ -661,7 +661,7 @@
       "assertions": [
         "requires assets/skill-os-infographic.png",
         "requires assets/skill-map-cinematic.png",
-        "keeps generated infographic assets text-light",
+        "keeps generated infographic text readable or repeats it in Markdown",
         "keeps accessible Markdown explanation beside images"
       ],
       "failure_mode": "front page remains generic or SVG-only",
@@ -800,6 +800,21 @@
         "seedance-pipeline",
         "seedance-recipes"
       ]
+    },
+    {
+      "id": "visual_gallery_six_text_rich_assets",
+      "prompt": "The README visuals still look generic; add at least six beautiful hero and infographic images with readable text, including CDN and skill capability graphics.",
+      "expected_output": "README contains a generated bitmap visual gallery with at least six assets, including a command-center hero, global filmmaker hero, skill capability infographic, CDN delivery map, reference role map, production delivery map, and QC stack.",
+      "assertions": [
+        "requires at least six committed gallery PNG assets",
+        "includes CDN delivery map and skill capability infographic",
+        "keeps generated text readable and corrected",
+        "pairs every text-rich image with accessible Markdown"
+      ],
+      "failure_mode": "generic front page or unreadable infographic text",
+      "skills_expected_to_activate": [
+        "seedance-pipeline"
+      ]
     }
   ]
 }
@@ -14,15 +14,17 @@ This repository has no application frontend. The user-facing frontend is the Git
 ## Asset rules
 
 - Use SVG for simple structural support diagrams.
-- Use generated bitmap images for the README hero, operating-system infographic, and skill-map infographic when the asset needs cinematic texture, real scene depth, or visual storytelling.
-- Bitmap hero/infographic/map assets should be text-free or nearly text-free, logo-free, and readable at GitHub README width.
+- Use generated bitmap images for the README hero, operating-system infographic, skill-map infographic, capability map, CDN delivery map, reference-role map, production-delivery map, and QC stack when the asset needs cinematic texture, real scene depth, or visual storytelling.
+- Bitmap hero/infographic/map assets should be logo-free, watermark-free, and readable at GitHub README width.
+- Text-rich infographics are allowed when labels are large, short, corrected, and repeated in accessible Markdown next to the image.
 - SVG assets must include `<title>` and `<desc>`.
 - No external scripts, images, fonts, or tracking in SVG assets.
 - Avoid generic lens dashboards, dense decorative noise, and unreadable micro labels.
+- Inspect generated text manually; reject garbled words, ugly font treatment, low contrast, noisy decoration, and placeholder-looking panels.
 
 ## README rules
 
 - No line longer than 500 characters.
 - Tables should have real newlines.
 - Every major section should answer a user decision: what is it, where do I start, what skills exist, how do I validate, what changed.
-- Bitmap hero art should contain no embedded text, logos, or watermarks; keep real project text in Markdown for accessibility.
+- Bitmap hero art should avoid watermarks and tiny text. Text-rich infographic labels must also be represented in Markdown for accessibility and search.
@@ -2,7 +2,7 @@
 
 last_verified: 2026-05-30
 
-This snapshot records the evidence used for the v5.4.4 section-depth, multilingual, visual-asset, professional-filmmaker, and surface-refresh pass. It is a dated working layer, not a permanent guarantee of platform access, pricing, limits, or regional availability.
+This snapshot records the evidence used for the v5.4.4-v5.4.5 section-depth, multilingual, visual-asset, professional-filmmaker, visual stress-test, and surface-refresh pass. It is a dated working layer, not a permanent guarantee of platform access, pricing, limits, or regional availability.
 
 ## Source Priority
 

@@ -3,9 +3,61 @@
 
 import argparse
 import re
+import struct
 from pathlib import Path
 
 
+GALLERY_ASSETS = [
+    "assets/hero-command-center.png",
+    "assets/hero-global-filmmaker-mode.png",
+    "assets/infographic-skill-capabilities.png",
+    "assets/infographic-cdn-delivery-map.png",
+    "assets/infographic-reference-role-map.png",
+    "assets/infographic-production-delivery.png",
+    "assets/infographic-professional-qc-stack.png",
+]
+
+CORE_BITMAP_ASSETS = [
+    *GALLERY_ASSETS,
+    "assets/hero-cinematic.png",
+    "assets/skill-os-infographic.png",
+    "assets/skill-map-cinematic.png",
+]
+
+
+def png_dimensions(path: Path) -> tuple[int, int] | None:
+    with path.open("rb") as handle:
+        header = handle.read(24)
+    if len(header) < 24 or header[:8] != b"\x89PNG\r\n\x1a\n" or header[12:16] != b"IHDR":
+        return None
+    return struct.unpack(">II", header[16:24])
+
+
+def check_png_asset(
+    root: Path,
+    rel: str,
+    label: str,
+    errors: list[str],
+    *,
+    min_bytes: int = 100_000,
+    min_width: int = 1200,
+    min_height: int = 650,
+) -> None:
+    path = root / rel
+    if not path.exists():
+        errors.append(f"missing asset: {rel}")
+        return
+    if path.stat().st_size < min_bytes:
+        errors.append(f"{rel} appears too small for a real {label} image")
+    size = png_dimensions(path)
+    if size is None:
+        errors.append(f"{rel} is not a valid PNG")
+        return
+    width, height = size
+    if width < min_width or height < min_height:
+        errors.append(f"{rel} is too small for README display ({width}x{height})")
+
+
 def main() -> int:
     parser = argparse.ArgumentParser()
     parser.add_argument("repo", nargs="?", default=".")
@@ -27,12 +79,19 @@ def main() -> int:
         if long_lines:
             errors.append("README.md has lines over 500 chars: " + ", ".join(map(str, long_lines[:10])))
         for required in [
-            "assets/hero-cinematic.png",
+            "assets/hero-command-center.png",
+            "assets/hero-global-filmmaker-mode.png",
             "assets/skill-os-infographic.png",
             "assets/skill-map-cinematic.png",
             "## What This Skill Does",
             "## Professional Filmmaker Scope",
             "## Operating System At A Glance",
+            "## Visual Gallery",
+            "### Hero Shots",
+            "### Text-Rich Infographics",
+            "CDN video delivery map",
+            "What this skill can do",
+            "Professional QC stack",
             "## Start Here",
             "## Skill Map",
             "api-workflow.md",
@@ -45,27 +104,31 @@ def main() -> int:
         ]:
             if required not in text:
                 errors.append(f"README.md missing `{required}`")
+        gallery_count = sum(1 for rel in GALLERY_ASSETS if rel in text)
+        if gallery_count < 6:
+            errors.append(f"README.md must reference at least six visual-gallery PNG assets ({gallery_count} found)")
+        for rel in GALLERY_ASSETS:
+            if rel not in text:
+                errors.append(f"README.md missing gallery asset `{rel}`")
 
-    if not (root / "docs" / "frontend-redesign.md").exists():
+    redesign_doc = root / "docs" / "frontend-redesign.md"
+    if not redesign_doc.exists():
         errors.append("missing docs/frontend-redesign.md")
+    else:
+        doc_text = redesign_doc.read_text(encoding="utf-8").lower()
+        if "text-rich infographics" not in doc_text or "infographic-cdn-delivery-map.png" not in doc_text:
+            errors.append("docs/frontend-redesign.md missing v5.4.5 text-rich gallery guidance")
+
+    design_system = root / "references" / "frontend-design-system.md"
+    if not design_system.exists():
+        errors.append("missing references/frontend-design-system.md")
+    else:
+        ds_text = design_system.read_text(encoding="utf-8").lower()
+        if "text-rich infographics" not in ds_text or "reject garbled" not in ds_text:
+            errors.append("references/frontend-design-system.md missing text-rich infographic quality rules")
 
-    hero = root / "assets" / "hero-cinematic.png"
-    if not hero.exists():
-        errors.append("missing asset: assets/hero-cinematic.png")
-    elif hero.stat().st_size < 100_000:
-        errors.append("assets/hero-cinematic.png appears too small for a real hero image")
-
-    infographic = root / "assets" / "skill-os-infographic.png"
-    if not infographic.exists():
-        errors.append("missing asset: assets/skill-os-infographic.png")
-    elif infographic.stat().st_size < 100_000:
-        errors.append("assets/skill-os-infographic.png appears too small for a real infographic image")
-
-    skill_map = root / "assets" / "skill-map-cinematic.png"
-    if not skill_map.exists():
-        errors.append("missing asset: assets/skill-map-cinematic.png")
-    elif skill_map.stat().st_size < 100_000:
-        errors.append("assets/skill-map-cinematic.png appears too small for a real skill-map image")
+    for rel in CORE_BITMAP_ASSETS:
+        check_png_asset(root, rel, "README visual", errors)
 
     for rel in ["assets/hero-dark.svg", "assets/hero-light.svg", "assets/skill-map.svg"]:
         path = root / rel

@@ -50,6 +50,7 @@
     "audio_post_loudness_stems",
     "delivery_qc_preflight",
     "imf_localization_versioning",
+    "visual_gallery_six_text_rich_assets",
 }