Learning Objectives
By the end of this module, you will be able to:
- Route any still image task to the optimal tool using the Image Decision Tree
- Route any video task to the optimal tool using the Video Decision Tree
- Route any audio task to the optimal tool using the Audio Decision Tree
- Apply task-specific quick references for the 12 most common creative scenarios
- Make routing decisions that account for quality, speed, and cost trade-offs
The Master Decision Tree
This is the single most important reference in the entire course. Bookmark it. Print it. Tape it to your monitor. Start here for ANY AI creative task.
WHAT ARE YOU MAKING?
│
├─── STILL IMAGE ──────────────→ [Image Decision Tree]
│
├─── VIDEO CLIP ───────────────→ [Video Decision Tree]
│
├─── UGC / AVATAR VIDEO ──────→ [UGC Decision Tree]
│
├─── VOICE / DIALOGUE ────────→ ElevenLabs (always)
│
├─── MUSIC ────────────────────→ [Music Decision Tree]
│
└─── MULTI-SHOT NARRATIVE ────→ Google Flow (planning)
+ Multi-model routing (execution)
+ NLE (finishing)
→ See Course 1 & Course 7
The Image Decision Tree
GENERATING A STILL IMAGE
│
├── Does it need readable text in the image?
│ │ (logos, labels, signage, headlines, product names)
│ │
│ ├── YES ──→ Nano Banana Pro
│ │ (~94% text accuracy — only reliable option)
│ │
│ └── NO ──→ Continue below ↓
│
├── What's the primary goal?
│ │
│ ├── PHOTOREALISM (looks like a real photograph)
│ │ │
│ │ ├── Need to edit it conversationally after generation?
│ │ │ ├── YES ──→ Nano Banana Pro (multi-turn editing)
│ │ │ └── NO ──→ Continue ↓
│ │ │
│ │ ├── Need maximum technical control / LoRA / ControlNet?
│ │ │ ├── YES ──→ FLUX (via ComfyUI)
│ │ │ └── NO ──→ Continue ↓
│ │ │
│ │ ├── Need to combine 3+ reference images?
│ │ │ ├── YES ──→ Nano Banana Pro (up to 14 references)
│ │ │ └── NO ──→ Either Nano Banana Pro or FLUX
│ │ │
│ │ └── Quick concept, no fuss?
│ │ └── GPT Image (ChatGPT)
│ │
│ ├── ARTISTIC / STYLIZED (mood, editorial, concept art)
│ │ │
│ │ ├── Need a specific aesthetic that you can control precisely?
│ │ │ └── YES ──→ Midjourney V7
│ │ │ (--sref for style, --p for personalization,
│ │ │ --exp for exploration)
│ │ │
│ │ └── Need style transfer from a reference image?
│ │ ├── For still image ──→ Midjourney V7 --sref
│ │ └── For video later ──→ Runway Gen-4.5 References
│ │
│ ├── PRODUCT PHOTOGRAPHY
│ │ │
│ │ ├── Single SKU, highest quality?
│ │ │ └── Nano Banana Pro (4K, text-safe, multi-turn refinement)
│ │ │
│ │ ├── Catalog scale (50+ SKUs)?
│ │ │ └── Nano Banana 2 (Flash) via API batch
│ │ │ or Claid.ai / Pebblely for automated backgrounds
│ │ │
│ │ └── Need one-click lifestyle backgrounds?
│ │ └── Pebblely or Flair.ai
│ │
│ └── CHARACTER REFERENCE (for video production)
│ │
│ ├── Need a consistent character across 5-10 images?
│ │ └── Midjourney --oref (--ow 200) or Nano Banana Pro multi-ref
│ │
│ ├── Need a locked character across 50+ images?
│ │ └── FLUX LoRA training or Higgsfield SOUL ID
│ │
│ └── Need to transfer a character to video?
│ └── Build a 6-image reference package (Course 5)
│ → Use as keyframe inputs for video generation
The Video Decision Tree
GENERATING A VIDEO CLIP
│
├── Does it need synchronized dialogue (lip sync)?
│ │
│ ├── YES, with AI-generated voice
│ │ └── Veo 3.1 (best native dialogue + lip sync)
│ │
│ ├── YES, with a specific pre-recorded voice
│ │ └── Generate voice in ElevenLabs first
│ │ → Use as audio reference in Veo 3.1 / Google Flow
│ │ OR sync in post-production
│ │
│ └── NO dialogue ──→ Continue below ↓
│
├── Does it need ambient audio / sound effects?
│ │
│ ├── YES, generated with the video
│ │ └── Veo 3.1 (best native environmental audio)
│ │ or Kling 2.6 (good native audio, faster)
│ │
│ └── NO, audio will be added in post
│ └── Route by visual needs below ↓
│
├── What kind of camera work does it need?
│ │
│ ├── PRECISE camera path (specific orbit, exact dolly, defined crane)
│ │ └── Kling 2.6 (1,296 camera presets, intensity sliders)
│ │
│ ├── NATURAL / CINEMATIC camera (described in text, model interprets)
│ │ └── Veo 3.1 (best at interpreting natural language camera direction)
│ │
│ └── CREATIVE / EXPERIMENTAL camera (unusual, stylized, dream-like)
│ └── Runway Gen-4.5 (most artistic motion interpretation)
│
├── What's the subject?
│ │
│ ├── PRODUCT (orbit, demo, hero shot)
│ │ └── Kling 2.6 (precise control, product stays stationary)
│ │
│ ├── CHARACTER (person moving, emoting, speaking)
│ │ └── Veo 3.1 (best facial rendering, best character-driven output)
│ │
│ ├── ENVIRONMENT (landscape, establishing shot, atmosphere)
│ │ └── Veo 3.1 (best atmospheric quality)
│ │
│ └── ABSTRACT / EFFECTS (style transfer, surreal, experimental)
│ └── Runway Gen-4.5 (strongest creative effects)
│
├── Does it need realistic physics?
│ │ (water pouring, fabric draping, objects falling, collisions)
│ │
│ ├── YES ──→ Sora 2 or Veo 3.1 (best physics simulation)
│ └── NO ──→ Route by other criteria above
│
└── What's the speed priority?
│
├── FASTEST POSSIBLE (testing, iteration, rough cuts)
│ └── Kling 2.6 (fastest generation time)
│
└── HIGHEST QUALITY (hero content, final delivery)
└── Veo 3.1 (highest cinematic quality ceiling)
The UGC / Avatar Decision Tree
GENERATING UGC-STYLE VIDEO
│
├── How many variants do you need?
│ │
│ ├── 20+ variants fast (ad testing at scale)
│ │ │
│ │ ├── Have a product URL to work from?
│ │ │ └── Creatify AI (URL-to-ad automation)
│ │ │
│ │ └── Need maximum authentic UGC feel?
│ │ └── Arcads (ad-optimized avatars)
│ │
│ ├── 5-15 variants (standard campaign)
│ │ └── HeyGen (best avatar quality + voice cloning)
│ │
│ └── 1-3 variants (highest possible quality)
│ └── Full AI production pipeline
│ (Nano Banana Pro character + Veo 3.1 animation)
│ → See Course 3, Module 2
│
├── Need multi-language versions?
│ └── YES ──→ HeyGen (built-in translation + lip sync)
│
└── Need specific demographic not in avatar libraries?
└── Full production: generate custom character in Nano Banana Pro
→ Animate in Veo 3.1 with directed performance
The Music Decision Tree
GENERATING MUSIC
│
├── Does it need vocals / lyrics?
│ ├── YES ──→ Suno (best vocal quality + song structure)
│ └── NO ──→ Continue below ↓
│
├── Is it background score / underscore?
│ └── YES ──→ Udio (best instrumental, better genre precision)
│
├── Is it ambient / atmospheric?
│ └── YES ──→ Udio (excellent ambient generation)
│
└── Is it a jingle / brand audio?
└── Suno (with specific BPM, duration, and style direction)
Task-Specific Quick References
"I need product photos"
Single product, hero quality → Nano Banana Pro (4K, multi-turn)
Catalog batch (50+ SKUs) → Nano Banana 2 Flash API
Product on white background → Nano Banana Pro or Claid.ai
Product in lifestyle context → Nano Banana Pro (environment-first)
Product on model → Nano Banana Pro + character reference
Quick social media product post → Pebblely (one-click backgrounds)
"I need a brand video"
30-second hero / campaign → Course 1 pipeline (NBP → Veo/Kling → NLE)
Character-driven narrative → Course 7 pipeline (Flow + multi-model)
Product demo video → Kling 2.6 (orbits) + Veo 3.1 (lifestyle)
Website hero video (looping) → Veo 3.1 or Kling, export as loop
Investor / pitch video → Course 4 approach (brand-defining = Tier 1)
"I need social media content"
Instagram feed imagery → Midjourney (artistic) or NBP (product)
Instagram Reels / TikTok → Kling 2.6 (fast, 9:16 native)
YouTube pre-roll (15-30s) → Full pipeline (Course 1)
LinkedIn brand video → Veo 3.1 (cinematic, professional)
Pinterest pins → Midjourney (aspirational, 2:3 vertical)
"I need ad creative"
UGC testimonial ads (volume) → Creatify or Arcads
UGC testimonial (premium quality) → Full production (Course 3, Module 2)
Product video ads → Kling 2.6 (orbits) + text overlays
Brand awareness ads → Veo 3.1 (cinematic quality)
Multi-variant hook testing → Course 3, Module 4 system
"I need consistent characters"
Quick series (5-10 images) → Midjourney --oref (--ow 200)
Campaign work (10-20 images) → NBP multi-reference + Identity Header
Long-running IP (50+ images) → FLUX LoRA or SOUL ID
Transferring character to video → Keyframe bridge method (Course 5)
"I need audio"
Dialogue / voiceover → ElevenLabs (always)
Brand voice / voice clone → ElevenLabs Voice Design or Clone
Background music (instrumental) → Udio
Full song with vocals → Suno
Sound effects → Native AI (Veo/Kling) or Freesound.org
Practical Exercise
Exercise: Route 5 Real Projects
Think of 5 creative projects you have upcoming (or invent plausible ones). For each:
- Describe the project in one sentence
- Walk through the appropriate decision tree
- Name the tool(s) you'd use
- Justify in one sentence why that tool is optimal for this specific task
- Identify a backup tool if your first choice isn't available
This exercise builds the routing muscle. After doing it 5 times consciously, it becomes intuitive.
Key Takeaways
- The Master Decision Tree routes any creative task to the right tool by asking 3-5 questions about the task's specific requirements.
- Start with the output type (image, video, UGC, audio, narrative), then narrow by requirements (text, camera, style, physics, speed).
- No tool appears in every branch. Each tool wins specific scenarios — the skill is matching the scenario to the strength.
- The task-specific quick references give instant answers for the 12 most common creative production scenarios.
- When in doubt between two tools: choose the one that handles your MOST IMPORTANT requirement best, and accept compromise on secondary requirements.
References & Resources
- Apostle.io Course 1: Multi-Model Routing Module
- Apostle.io Course 5: Tool-Specific Consistency Techniques
- Pinterest board — AI Tool Comparison: https://pinterest.com/search/pins/?q=ai%20creative%20tools%20comparison