Use this reference whenever a deck may contain images. The goal is to prevent template-like empty boxes while still supporting real user images, later replacement, and optional AI-generated visuals.
Always classify every image into one role before planning the deck:
Design Reference Image
- used to extract Design DNA
- contributes mood, palette, composition, density, texture, image treatment, and visual psychology
- never inserted into the deck by default
- identifiable subjects are not reusable visual content and must not be redrawn, traced, stylized, approximated, abstracted into mascots, or rebuilt as CSS/SVG/HTML/AI visuals
Content Image / Slide Asset
- used as actual slide content
- must be explicitly approved by the user as slide material
- participates in Blueprint and Page Specs from the beginning
Do not infer that a reference image should appear in the deck. If unclear, ask one short clarification.
When an image is a Design Reference Image, extract only style traits:
- palette and contrast behavior
- typography mood and title treatment
- composition rhythm and crop behavior
- line quality, borders, shadows, texture, grain, material, and motion feeling
- information density and visual hierarchy
Do not extract the image's subject as reusable slide content. Forbidden subject matter includes people, animals, characters, mascots, specific objects, products, buildings, vehicles, toys, and recognizable subject parts such as faces, eyes, ears, tails, fins, horns, wings, paws, posture, clothing, fur, scales, and silhouettes.
This firewall blocks both direct asset use and redraws. A cat, fish, dinosaur, person, mascot, or product reference cannot become a CSS/SVG illustration, AI-generated approximation, icon, diagram character, decorative motif, or "inspired by" mascot. If the user wants the subject itself, it must be approved as a Content Image / Slide Asset or requested as a new generated image in the Image Intent step.
Ask image-intent questions only after:
- Design DNA exists
- the Design DNA panel has been accepted or tuned
- the active Design DNA candidate is accepted, or a saved profile/adapter has been selected
- PPT requirements are being collected
Do not ask image-intent questions during Design Intake unless the user explicitly says an image should appear in the slides.
During PPT Requirement Discovery, ask:
How should this deck use content images?
A. No content images; use Design DNA visuals, typography, diagrams, and whitespace
B. I will provide real images; plan slides around them
C. Reserve replaceable image slots for later
D. Generate AI images for concept/atmosphere
E. Mixed
If the user chooses B or E, ask what type of images they will provide:
A. Product / website / software screenshots
B. People / team / event photos
C. Data charts / paper figures / technical diagrams
D. Brand assets / logos / visual identity material
E. Mixed
If the user chooses C, ask:
How should missing images appear in the draft?
A. Show designed placeholders that match the Design DNA
B. Hide empty boxes and show CSS fallback visuals
C. Keep slots only in image-manifest metadata
Default to B if the user does not specify.
For each approved content image, create an asset record:
{
"asset_id": "img_03_dashboard",
"file": "assets/images/03-dashboard.png",
"type": "product_screenshot",
"semantic_role": "evidence",
"best_use": ["feature_slide", "proof_slide"],
"avoid_use": ["full_bleed_background"],
"recommended_ratio": "16:10",
"fit": "contain",
"crop_policy": "preserve_ui_text",
"caption_required": true,
"safe_text_overlay": false
}Asset type guidance:
| Type | Default treatment | Avoid |
|---|---|---|
product_screenshot |
framed contain, browser/device chrome when useful | full-bleed crop that cuts UI text |
data_chart |
contain, large enough labels, caption outside | cover crop, decorative background use |
people_photo |
hero subject or editorial crop | text over face, crop through eyes/hands |
brand_logo |
small identity mark, cover/closing, preview styling | stretching, busy backgrounds |
atmosphere_photo |
background image with scrim or crop reveal | pretending it is evidence |
object_product |
object focus, shadow/material integration | repeated side placement without purpose |
Choose a slot type based on page purpose, not on a fixed template.
| Slot type | Use for | Rules |
|---|---|---|
hero_bleed |
cover, section opener, atmospheric transition | mandatory scrim/plate for text |
object_focus |
product, person, 3D object, artifact | reserve exclusion zone; no text behind subject |
screenshot_frame |
website, app, dashboard, code, UI | contain by default; preserve important text |
evidence_card |
case proof, research figure, source image | caption outside; high contrast frame |
image_grid |
portfolio, multiple cases, before/after | consistent ratios and spacing |
editorial_cutout |
magazine-style people/object crop | intentional crop; protected title area |
background_texture |
low-salience atmosphere | opacity/scrim; never reduce text contrast |
side_visual |
deliberate split page | allowed only when the page purpose calls for it; not the default |
inline_thumbnail |
list/process/case card | small, aligned, not decorative clutter |
diagram_image |
technical figure/system diagram | caption and label legibility first |
Never repeat side_visual on many slides just because images might exist.
If no approved content image exists, the slide should still feel finished.
Use:
- typographic visual
- CSS/SVG diagram
- data visualization built in HTML/CSS/SVG
- abstract material object derived from Design DNA
- controlled texture/pattern
- intentional whitespace
- optional AI-generated image only after user approval
Do not render a large empty rectangle, plus sign, "drop image here", "image placeholder", or a meaningless decorative side panel.
Do not use missing-image strategies to redraw reference subjects. Abstract material objects may include paper strips, tape, labels, slices, grid blocks, hand-drawn lines, halftone fields, crop marks, and geometric modules. They must not include animal/person/character/object silhouettes or subject parts from the reference.
Use replaceable slots only when the user wants to add images later.
Preferred default:
CSS fallback is visible.
Image slot metadata is saved.
No ugly empty frame appears in the final deck.
If the user wants visible editable slots, the placeholder must follow the active Design DNA:
| Style direction | Placeholder treatment |
|---|---|
| minimal/premium | hairline border, pale grid, small slot id, no loud label |
| cyber/tech | dark panel, corner brackets, subtle glow, restrained scanline |
| editorial/magazine | crop marks, figure number, caption baseline, paper tone |
| academic/report | Figure N frame, caption area, clean border |
| playful/3D | soft inset panel, rounded material surface, small index marker |
Visible placeholder text must be minimal: slot id, ratio, or short semantic label. Do not place large instructional text in the slide.
When slots are replaceable or images are external assets, create an image-manifest.json or equivalent internal spec:
{
"image_slots": [
{
"slot_id": "p03-main-visual",
"page": 3,
"purpose": "show product interface or system screenshot",
"recommended_file": "assets/images/03-main-visual.png",
"slot_type": "screenshot_frame",
"ratio": "16:10",
"fit": "contain",
"safe_area": "center-right",
"caption_required": true,
"fallback_if_missing": "css_wireframe_dashboard",
"visible_placeholder": false
}
]
}For folder-based HTML decks, use stable names:
assets/images/01-cover-visual.jpg
assets/images/03-dashboard.png
assets/images/05-team-photo.jpg
assets/images/07-proof-chart.png
Use {page}-{semantic-name}.{ext}. If users replace the image with the same file name, the HTML should not need path edits.
When a slide has image or future image behavior, include:
{
"image_strategy": {
"intent": "none | approved_content_image | replaceable_slot | generated_image | fallback_visual",
"slot_type": "screenshot_frame",
"asset_id": "img_03_dashboard",
"slot_id": "p03-main-visual",
"semantic_role": "evidence",
"ratio": "16:10",
"fit": "contain",
"placement": "center-right",
"safe_area": { "x": 980, "y": 170, "w": 780, "h": 620 },
"caption": "Model result dashboard",
"text_overlay_allowed": false,
"fallback_if_missing": "css_wireframe_dashboard",
"visible_placeholder": false
}
}For intent: none, state the non-image visual strategy instead of leaving this blank.
When the active Design Profile came from reference images with identifiable subjects, also include:
{
"visual_subject_policy": {
"uses_reference_subject": false,
"subject_replication_allowed": false,
"forbidden_reference_subjects": ["cat", "fish", "dinosaur", "person", "mascot", "product object"],
"visual_surrogate_strategy": "abstract_material_shape | typography | diagram | texture | whitespace",
"forbid_subject_silhouette": true,
"forbid_subject_parts": true
}
}Rules:
- Text must not sit directly on busy images, screenshots, gradients, glow, or patterned areas.
- Text over images requires a scrim, solid plate, translucent plate, outline plate, or a separate protected text area.
- Evidence images and screenshots should usually have captions outside the image, not overlaid on top.
covercrop is allowed for atmosphere;containis default for screenshots, charts, UI, formulas, and evidence.- If image and text are both important and crowded, split into more slides rather than shrinking text.
AI images are allowed for:
- concept covers
- atmosphere
- metaphors
- abstract technology visuals
- product vision imagery
- section visuals
AI images are not allowed to pretend to be:
- real screenshots
- factual evidence
- financial charts
- academic results
- real people or events
Ask for approval before generating AI images unless the user has already requested generated visuals.