02 comfyui (EN+ZH) - unify the asset numbers end to end: 108 cover pairs = 216 banner+OG assets. The intro, the frontmatter metric, the description and the `ws.close()` comment all still carried the old "89", which contradicted the 216/25.9-min figure in the result section (consultant finding). - state the timing scope explicitly: end-to-end per slug, diffusion plus the Pillow clearance audit and the two-size WebP encode. - disclose the limit of the mtime evidence: mtime keeps only the last write, so gate-rejected re-queues from an earlier pass do not lengthen the measured window. 01 payments (EN+ZH) - the idempotency comment now names the 72h bank reconciliation window, which is the boundary that makes EXPIRED a legitimate fulfilment entry point (line count unchanged at 27).
This commit is contained in:
hoelee committed
1 parent
bfcf61c7ea
commit
403d4c0d9f
4 files changed
+16
-16
No files matched your search
@@ -106,7 +106,7 @@ public function callback(): ResponseInterface
|
||||
return $this->response->setStatusCode(404)->setBody('UNKNOWN_BILL');
|
||||
}
|
||||
// Exactly two states may enter fulfilment: a fresh PENDING order, or an
|
||||
// EXPIRED bill the bank settled late. Everything else (PROCESSING already
|
||||
// EXPIRED bill the bank settled late, inside the 72h reconciliation window.
|
||||
// holds a queue slot, FULFILLED is done, RETRY_QUEUE is retrying) is a
|
||||
// replay and must be absorbed here — checking only for 'FULFILLED' leaves
|
||||
// the whole PROCESSING window open to a second queue push.
|
||||
|
||||
@@ -2,7 +2,7 @@
|
||||
ogImage: "/og/batch-generating-80-technical-covers-with-comfyui.jpg"
|
||||
banner: "/banners/batch-generating-80-technical-covers-with-comfyui.jpg"
|
||||
title: "Batch Generating 80+ Technical Covers with ComfyUI: Constraining Style, Space, and Typography"
|
||||
description: "How I automated the generation of 89 high-contrast technical article covers using a headless ComfyUI API pipeline: enforcing negative space, dark palettes, and zero-hallucination text boundaries."
|
||||
description: "How I automated the generation of 108 cover pairs (216 banner and Open Graph assets) using a headless ComfyUI API pipeline: enforcing negative space, dark palettes, and zero-hallucination text boundaries."
|
||||
pubDate: 2026-10-27
|
||||
category: ai
|
||||
blueprint: ai-agent-automation
|
||||
@@ -14,8 +14,8 @@ summary:
|
||||
audience: "AI Engineers, Content Automation Leads & Creative Technologists"
|
||||
painPoint: "Generating article covers via diffusion models produces random visual styles, garbled hallucinatory text, and zero compositional space for typography headlines."
|
||||
impact:
|
||||
- metric: "89 Covers Automated"
|
||||
detail: "Built a headless ComfyUI workflow executing via WebSocket API on a local 12GB VRAM node, generating 89 consistent technical covers in 48 minutes."
|
||||
- metric: "108 Cover Pairs Automated"
|
||||
detail: "Built a headless ComfyUI workflow executing via WebSocket API on a local 12GB VRAM node: a single batch run wrote 216 cover assets (108 banner + OG pairs) inside a 25.9-minute window, timestamped by the filesystem."
|
||||
- metric: "100% Text Separation"
|
||||
detail: "Banned text generation in diffusion prompts; forced visual models to generate clean background negative space and composited typography programmatically."
|
||||
- metric: "Dark System Cohesion"
|
||||
@@ -34,10 +34,10 @@ When building a high-credibility technical knowledge base or blog archive, visua
|
||||
is non-negotiable. If one article features a glossy 3D cartoon, the next an anime illustration,
|
||||
and a third a photorealistic macro lens shot, the site feels like an uncurated content farm.
|
||||
|
||||
Recently, the technical learning portal I manage (`hoelee-learning`) needed 89 individual cover assets
|
||||
to replace generic SVG placeholders.
|
||||
Recently, the technical learning portal I manage (`hoelee-learning`) needed 108 cover pairs — 216 individual banner and
|
||||
Open Graph assets — to replace generic SVG placeholders.
|
||||
|
||||
Attempting to generate 89 covers manually via WebUI interfaces or Discord bots introduces
|
||||
Attempting to generate over a hundred covers manually via WebUI interfaces or Discord bots introduces
|
||||
three structural failure modes:
|
||||
1. **Style Drift**: Different seeds and random prompts cause lighting, camera angles, and art directions to wander uncontrollably.
|
||||
2. **Diffusion Text Hallucination**: Diffusion models notoriously struggle with text. They produce garbled alien glyphs, pseudo-English letters, and smeared gibberish across the focal plane.
|
||||
@@ -158,7 +158,7 @@ def generate_cover(slug, topic_metaphor):
|
||||
return prompt_id
|
||||
raise
|
||||
finally:
|
||||
ws.close() # 89 covers x 1 socket: never closed, it leaks handles silently
|
||||
ws.close() # 108 covers x 1 socket: never closed, it leaks handles silently
|
||||
|
||||
def prompt_in_history(prompt_id):
|
||||
with urllib.request.urlopen(f"http://{COMFY_HOST}/history/{prompt_id}", timeout=10) as r:
|
||||
@@ -233,7 +233,7 @@ def audit_top_negative_space(img_path: str) -> tuple[bool, str]:
|
||||
## The result
|
||||
|
||||
By turning visual asset creation into a deterministic, API-driven engineering pipeline:
|
||||
* **216 cover assets in a 25.9-minute window**: a single batch run wrote 108 banner + OG pairs (timestamped by the filesystem, so the figure is auditable rather than estimated) on one on-premise RTX 3060 node.
|
||||
* **216 cover assets in a 25.9-minute window**: a single batch run wrote 108 banner + OG pairs on one on-premise RTX 3060 node. The figure is end-to-end per slug — diffusion inference plus the Pillow clearance audit and the two-size WebP encode — and is taken from filesystem timestamps rather than estimated, so it is auditable. One honest limit of that evidence: mtime records only the last write, so covers rejected and re-queued by the clearance gate in an earlier pass would not lengthen the measured window.
|
||||
* **100% brand consistency**: Every single article shares the identical dark-mode color temperature, technical mood, and compositional negative space.
|
||||
* **Zero manual prompting**: Once the topic metadata was defined, the entire batch executed unattended, and the resulting WebP assets were verified and committed straight to Git.
|
||||
* **The pipeline is its own proof**: every cover image on this blog — including the one above this article — came out of this exact loop, not out of a design tool.
|
||||
|
||||
@@ -97,7 +97,7 @@ public function callback(): ResponseInterface
|
||||
$this->db->transRollback();
|
||||
return $this->response->setStatusCode(404)->setBody('UNKNOWN_BILL');
|
||||
}
|
||||
// 只有两种状态可以进入履约：全新的 PENDING，或银行延迟结算的 EXPIRED。
|
||||
// 只有两种状态可以进入履约：全新的 PENDING，或 72 小时对账窗口内银行延迟结算的 EXPIRED。
|
||||
// 其余状态（PROCESSING 已占一个队列位、FULFILLED 已完成、RETRY_QUEUE 在重试）
|
||||
// 都属于重放，必须在此吸收——只判 'FULFILLED' 会留下整个 PROCESSING 窗口，
|
||||
// 让第二次推送穿透幂等检查。
|
||||
|
||||
@@ -2,7 +2,7 @@
|
||||
ogImage: "/og/zh-batch-generating-80-technical-covers-with-comfyui.jpg"
|
||||
banner: "/banners/batch-generating-80-technical-covers-with-comfyui.jpg"
|
||||
title: "用 ComfyUI 批量闭环 80+ 张技术图文封面：设计规范约束、留白控制与批处理工作流"
|
||||
description: "如何通过 Headless ComfyUI API 流水线自动化生成 89 张高对比度技术博客封面：深度实战留白预留、暗色调约束与零文本幻觉的批处理闭环方案。"
|
||||
description: "如何通过 Headless ComfyUI API 流水线自动化生成 108 组（216 张 Banner 与 Open Graph）高对比度技术博客封面：深度实战留白预留、暗色调约束与零文本幻觉的批处理闭环方案。"
|
||||
pubDate: 2026-10-27
|
||||
category: ai
|
||||
blueprint: ai-agent-automation
|
||||
@@ -14,8 +14,8 @@ summary:
|
||||
audience: "AI 工程师、内容自动化负责人与创意技术架构师"
|
||||
painPoint: "直接用扩散模型生成文章封面会导致风格混乱飘忽、画面充斥外星乱码伪文字，且视觉中心过于拥挤导致无法排版叠加标题文本。"
|
||||
impact:
|
||||
- metric: "89 张封面全量批处理"
|
||||
detail: "基于本地 12GB VRAM 生产节点构建无头 ComfyUI 异步 WebSocket 批处理管线，单批次在 25.9 分钟窗口内写出 216 个封面资产（108 个 slug × banner + OG 卡片）。"
|
||||
- metric: "108 组封面全量批处理"
|
||||
detail: "基于本地 12GB VRAM 生产节点构建无头 ComfyUI 异步 WebSocket 批处理管线，单批次在 25.9 分钟窗口内写出 216 个封面资产（108 个 slug × banner + OG 卡片），耗时取自文件系统时间戳可核验。"
|
||||
- metric: "图像与文本排版解耦"
|
||||
detail: "提示词中严厉禁止模型生成任何文字，专注于渲染上部 45% 的干净负空间，后续标题完全交由代码程序化叠加。"
|
||||
- metric: "品牌视觉系统一致性"
|
||||
@@ -32,7 +32,7 @@ dynamicCta: true
|
||||
|
||||
在维护高专业度的技术知识库或技术博客矩阵时，视觉设计的一致性是确立专业信任感的重要基石。如果第一篇文章配的是糖果色 3D 卡通插图，第二篇变成了日系二次元动漫，第三篇又跳成了写实的微距摄影，整个站点就会立刻呈现出缺乏把控的粗糙拼凑感。
|
||||
|
||||
近期，我所维护的技术学习门户（`hoelee-learning`）需要为全量存量文章补充 89 张独一无二的高清图文封面，以彻底替换早期的通用占位图。
|
||||
近期，我所维护的技术学习门户（`hoelee-learning`）需要为全量存量文章补充 108 组（共 216 张 Banner 与 Open Graph）高清图文封面，以彻底替换早期的通用占位图。
|
||||
|
||||
如果依赖人工在 WebUI 界面或群聊机器人中逐张抽卡生成，会立刻陷入三大结构性困境：
|
||||
1. **风格漂移（Style Drift）**：不同的随机种子与自由发散的提示词，会导致光影色温、镜头视角与艺术风格四处乱飘。
|
||||
@@ -149,7 +149,7 @@ def generate_cover(slug, topic_metaphor):
|
||||
return prompt_id
|
||||
raise
|
||||
finally:
|
||||
ws.close() # 89 张图 × 1 个套接字，不关闭就是静默句柄泄漏
|
||||
ws.close() # 108 张图 × 1 个套接字，不关闭就是静默句柄泄漏
|
||||
|
||||
def prompt_in_history(prompt_id):
|
||||
with urllib.request.urlopen(f"http://{COMFY_HOST}/history/{prompt_id}", timeout=10) as r:
|
||||
@@ -216,7 +216,7 @@ def audit_top_negative_space(img_path: str) -> tuple[bool, str]:
|
||||
## 交付结果
|
||||
|
||||
通过将视觉物料生产工程化为确定性 API 流水线：
|
||||
* **25.9 分钟窗口内写出 216 个封面资产**：单批次产出 108 组 banner + OG 卡片（时间戳来自文件系统，可核验而非估算），全部在单块 12GB 显卡节点上完成。
|
||||
* **25.9 分钟窗口内写出 216 个封面资产**：单批次产出 108 组 banner + OG 卡片，全部在单块 12GB 显卡节点上完成。该口径是端到端每 slug 耗时——扩散推理加上 Pillow 留白审计与双尺寸 WebP 转码——取自文件系统时间戳而非估算，因此可核验。这条证据有一处必须承认的局限：mtime 只记录最后一次写入，若某些封面在更早一轮被留白门禁判为不合格并重跑，那部分耗时不会体现在这个窗口里。
|
||||
* **100% 保持品牌设计调性一致**：所有配图在暗色调对比、光影重心以及排版空间上呈现出高度收敛的专业秩序感。
|
||||
* **实现无人工值守**：一旦文章的元数据与核心隐喻就绪，流水线即可全自动批处理、校验合规性并直接回流 Git。
|
||||
* **这条管线本身就是证据**：本站每一张封面图——包括本文题图——都出自这条闭环，而不是出自设计软件。
|
||||
|
||||
Reference in new issue
Block a user