TikTok Viral Slideshow Automation — TikTok Viral Slideshow 自动化

TikTok Slideshow Marketing 自动化

Automate high-质量 educational TikTok slideshows for any 应用 or product. 生成s AI images via a user-chosen image generation method, 上传s via Automate命令行工具ps API, and 追踪s performance over time.

设置UP — 运行 Once When First Invoked

Work through 设置up conversationally. Ask one section at a time. WAIt for answers before moving on.

STEP 1 — 应用 信息

Ask:

What 应用 or product are we promoting? (name, what it does, who it is for) What is the 应用 Store / Google Play / product link? What is the core value proposition — what problem does it solve and why would someone 下载/buy it today? What are the mAIn brand colors?

After collecting answers, confirm back a one-sentence pitch for the product and ask the user to 应用rove it. This becomes the north star for all content.

STEP 2 — Image Generation Method

Ask the user which image generation method they want to use. Present these options clearly:

Option A — nano-banana-2 via Replicate (Recommended)

模型: google/nano-banana-2 Strengths: excellent text rendering, consistent character/mascot continuity when passing reference images, strong at dark cinematic styles Requires: Replicate account + API 令牌 (r8_...) Cost: pay-per-generation via Replicate

Option B — OpenAI / DALL-E

模型: gpt-image-1 or dall-e-3 Strengths: widely avAIlable, good general 质量, easy API 访问 Requires: OpenAI API key Limitation: no native image-输入 support for reference images in all versions — character consistency is harder

Option C — Google Gemini (imagen)

模型: imagen-3.0-生成-002 or similar via Google AI Studio Strengths: high 质量, Google eco系统 Requires: Google AI Studio API key Limitation: image-输入 for reference consistency varies by 模型 version

Option D — Something else

User pastes their own generation command or API detAIls Ask: "Paste the command or describe the 工具 you want to use. I will adapt the 工作流 to fit."

After the user chooses, 配置 the generation step accordingly. If they choose Option A, 创建 nano_banana-2.py (see STEP 5). If they choose another option, adapt the generation commands throughout the 工作流 to use their chosen 工具 instead.

STEP 2b — Automate命令行工具ps 凭证s

Ask for:

Automate命令行工具ps API key — 格式化: ac_sk_.... 设置 as env variable: 导出 AUTOMATE命令行工具PS_API_KEY="ac_sk_...". Never hardcode. Automate命令行工具ps TikTok Account ID — a number visible in the Automate命令行工具ps 仪表盘 URL or account 设置tings.

Tell the user to 设置 凭证s as 环境 variables before each 会话. Example for Option A (Replicate):

导出 REPLICATE_API_令牌="your_令牌_here" 导出 AUTOMATE命令行工具PS_API_KEY="your_key_here"

Adjust which env variables are needed based on the image generation method chosen in STEP 2.

STEP 3 — Style Discovery (Conversation First, As设置s After)

Do NOT ask for files yet. Have a real conversation first.

Ask one by one, wAIting for answers:

Niche and content type: What topic will these slideshows cover? What kind of content performs well in this niche? (Educational tips, workout plans, myth-busting, product demos, before/after, rankings, etc.)

Tar获取 audience: Who is watching? Age range, interests, pAIn points, what they hope to 获取 from this content.

Visual 身份: Does the brand have a mascot, character, or recurring visual element? Or will slides be text-and-photo driven? 机器人h work — some of the best-performing slideshows use no mascot at all.

Tone: Serious and expert? Playful and mocking? Motivational? 命令行工具nical? This affects 机器人h copy and visual style.

Art style inspiration: Ask the user to describe 1–3 TikTok accounts or visual references they like — or describe the look in their own words. Suggest if they're stuck: bold text on dark background like a movie poster, 清理 信息graphic style, rea列出ic illustration, flat cartoon character, lifestyle photography.

Background preference: Dark and cinematic, 清理 white/minimal, or something else?

Color palette: Beyond brand colors, what accent colors feel right for labels, highlights, callouts?

After collecting answers, synthesize into a Visual Style 图形界面de — 5–6 bullet points describing the look. Present it and ask: "Does this capture your vision, or should we adjust anything before I build the prompting 系统?"

STEP 4 — As设置 Collection

Based on style decisions, ask only for relevant as设置s:

If using a mascot/character: Ask for reference_mascot.png — a 清理 image of the character on a simple background. Passed to the image 模型 on every generation call for character consistency. If using organic phone CTA slides: Ask for 应用store.jpg — a screenshot of the 应用 Store or Google Play 列出ing. If text-and-photo / 信息graphic style: No mascot needed. Confirm the visual rules. Other brand as设置s: 记录o, existing slide examples, competitor screenshots to reference.

Tell the user exactly where to drop files:

/your-project-folder/ ├── reference_mascot.png ← if using a character ├── 应用store.jpg