Illustration Strategy — Detail

The Step 2 collaboration in SKILL.md. Produces the Illustration Style Anchor block that gets written into the outline header.

Not every talk needs generated illustrations — demo-heavy, data-heavy, or screenshot-driven talks may not. When the author wants AI-generated illustrations, this protocol walks through the visual identity collaboratively.

Sub-step 1: Propose Style Ideas with Sample Prompts

Present 3–4 style options informed by three sources:

1. The talk's own concepts, metaphors, and narrative — the style should reinforce the thesis, not be decorative wallpaper.
2. The vault's visual history — read speaker-profile.json → visual_style_history for the structured data: default_illustration_style, style_departures[] (what styles the speaker has used and what triggered them), mode_visual_profiles[] (which modes tend toward which aesthetics), and confirmed_visual_intents[] (hard rules about visual design). Also read rhetoric-style-summary.md Section 13 (cross-talk visual patterns), slide-design-spec.md, and design_rules. Know what the speaker's default looks like so you can propose informed departures.
3. Historical precedent for this mode/context — read visual_style_history → mode_visual_profiles for the matching mode ID. If the vault shows the speaker uses a particular aesthetic for this talk type, surface that as a data point (e.g., "your vault shows you use terminal aesthetic for agent talks"). If this talk's mode/context has no visual precedent in style_departures, say so.

Each option includes: a name, why it fits this talk's concepts, how it relates to the speaker's visual history (continuation vs. departure), and a sample prompt excerpt showing a specific slide from THIS talk rendered in the style.

ILLUSTRATION STYLE OPTIONS for "{talk title}"
=========================================================

A. [STYLE NAME]
   CONCEPT FIT: [Why this style reinforces the talk's thesis,
   metaphors, and narrative arc — not just what it looks like]

   VAULT CONTEXT: [How this relates to the speaker's visual
   history — continuation of default, intentional departure,
   or precedent from similar talk types]

   Sample prompt (Slide N — [slide title]):
   "[Complete prompt showing this specific slide rendered
   in the proposed style]"

B. [STYLE NAME]
   CONCEPT FIT: [...]
   VAULT CONTEXT: [...]
   Sample prompt (Slide N — [slide title]):
   "[...]"

C. [STYLE NAME]
   ...

RECOMMENDATION: [Which option and why — grounded in concept
fit and vault context, not just aesthetic preference]
=========================================================

The key: each style option explains WHY it fits this specific talk's concepts, not just what it looks like. The author picks one (or mixes elements), then iterates on the anchor paragraph together.

Sub-step 2: Define Format Vocabulary & Aspect Ratios

Once the style is chosen, define the slide format types for this talk:

SLIDE FORMAT VOCABULARY
========================
FULL     — full-bleed illustration, 1-2 sentences overlaid
           → Landscape 16:9 (1920×1080)
IMG+TXT  — illustration ~60% of slide, text beside/below
           → Portrait 2:3 (1024×1536)
EXCEPTION — real photo, data table, bio, or primary source
           → No generated illustration; uses [IMAGE NN] placeholder
========================

Format names and ratios are talk-specific — the author may use different names or add formats (e.g., DIAGRAM for technical slides, QUOTE for attributed quotations).

Sub-step 3: Choose Image Generation Model

Different image-generation models produce meaningfully different aesthetics even from the same prompt — Gemini 3 Pro tends toward photorealism, Imagen toward painterly, the Gemini 2.x flash variants toward illustrated. The choice matters and is best made visually, not from prose descriptions.

The decision protocol is side-by-side generation, then speaker picks:

1. Pick the comparison slide. Choose a representative FULL-format slide from the outline — central to the deck's concept, not the title slide, not an edge case. The slide must already have a complete [STYLE ANCHOR]. <scene description> Image prompt drafted (using the anchor paragraph from Sub-step 1). Skip slides without prompts; the comparison needs concrete input to be meaningful.
2. Run the comparison. From the talk's working directory:

   python3 skills/illustrations/scripts/generate-illustrations.py \
     presentation-outline.md --compare <slide-num>

   The script renders the same prompt across the curated COMPARE_MODELS list (currently three Gemini-family models — see the script's constant block; updated as new image-capable models become available). Output lands in illustrations/model-comparison/slide-<NN>-<model>.<ext>.
3. Present the candidates. Show the speaker the side-by-side outputs. Lead with a brief read of each — what each model emphasized about the prompt — but the visual decision is the speaker's, not the agent's.
4. Bake the choice. The selected model goes into the outline header's **Model:** \<model-name>`` line. Every subsequent generation in Step 3 (deck illustrations) and Step 4 (builds) uses this model.

If the speaker dislikes all three candidates, iterate by either changing the comparison slide (a different slide may render better across the same model set) or revising the anchor paragraph (Sub-step 1) and re-running the comparison. Don't widen the model list ad-hoc — the curated set is what the script supports; new entries belong in the script's constant.

Model-specific prompt conventions (e.g., negative prompts, aspect-ratio tokens, reserved keywords) should be baked into the anchor paragraph immediately after the model is chosen, before Step 3 generation runs.

Sub-step 4: Visual Continuity Devices

Define recurring elements that tie the deck together as a coherent visual artifact:

• Sequential numbering (e.g., "FIG. N" numbering) — ties the deck together as one coherent document. The generation model may render numbers imperfectly; that's acceptable.
• Recurring characters/motifs in consistent style (same uniforms, same species, same rendering approach across all appearances).
• Checklist progression — a shared base image that gets edited to add checkmarks, fill-ins, or stamps across the talk. Use image editing (not regeneration) to preserve visual consistency. Track which slide is the base image.
• Progressive visual elements with explicit base-image tracking: document which slide is the "source" image for each progression, so edits chain correctly.
• Annotation style (callout labels, footnotes, stamps) — keep labels funny/deadpan if that's the tone; the gags in labels ARE the point.

Gate

Author approves the style anchor paragraphs, format vocabulary, and model choice. These become the Illustration Style Anchor section in the outline header. Once written, Step 3 (generation) can run.