Gadgetify

DIY: 3D Printed #Stargate with #Arduino control gadgetify.com/3d-printed-sta…

One of the better prompts I have done. It's a bit long though. Here is a modified version that is shorter. This was done for 4 economic theories. 2x2 grid, 16:9, 4 famous economic theories: INPUT: $ TOPIC // ── INFERENCE ────────────────────────────────────────────────── THINKER = infer_primary_theorist($ TOPIC) TITLE = infer_theory_name($ TOPIC) METAPHOR = infer_physical_metaphor($ TOPIC) WORKERS = infer_construction_crew($ TOPIC) CHART = infer_side_panel_chart($ TOPIC) // ── TEMPORAL STAGES ──────────────────────────────────────────── // We render the FINAL frame. These stages describe how the poster // resolves from blank canvas to a dense, historical diorama. STAGE 0 — PAPER CANVAS (t = 0.00) Sepia, aged paper background fills the frame. Faint blueprint grid lines and vintage grain. STAGE 1 — TEXT & UI BLOCKS (t = 0.15) Top-left block materializes with title, author, year, and summary. Side panel block appears with the chart/diagram. Typography is rigid, vintage serif. STAGE 2 — MONUMENTAL METAPHOR (t = 0.30) The giant physical object appears. It is massive, sculpted from stone, wood, or iron. It dominates the center of the frame. STAGE 3 — SCAFFOLDING (t = 0.45) Wooden planks, ladders, cranes, and ropes appear. They wrap around the giant object, showing it is under construction or operation. STAGE 4 — MICRO WORKERS (t = 0.60) Tiny human figures populate the scaffolding. They are dwarfed by the monument. They are carrying tools, climbing ladders, operating machinery. STAGE 5 — ENVIRONMENTAL TEXTURE (t = 0.75) Dust, ambient occlusion, and construction debris. Deep shadows in the crevices of the monument. Harsh directional light highlights the macro surface. STAGE 6 — FINAL LOCK (t = 1.00) The sepia color grade is perfectly balanced. The scale contrast (giant object vs. tiny workers) is clear. The chart accurately represents the theory. The image reads as a vintage museum display or architectural illustration. Output: one 8K architectural diorama poster.

Kind of a fun prompt. You never know what it generates :) 2x2 grid, 16:9, do this for 4 topics you pick: UNKNOWNS the carrier genre and its formal signature ; the one inverted variable ; t_rec = time to recognition ; the target of the comment (the carrier itself, or something else) ; fidelity level LAWS recognition must happen before comprehension The reader identifies WHAT IS BEING IMITATED from composition and treatment alone, in roughly a second, before reading a word. If recognition arrives late, the joke arrives dead. This means the carrier's signature must be loaded into the silhouette, the palette and the layout, not into fine detail. the signature lives in a handful of formal choices Every recognisable genre is carried by four or five decisions: the camera height and lens, the palette, the type treatment, the figure arrangement, the level of finish. Identify them, execute them exactly, and recognition is automatic. Approximating them is worse than ignoring them. change exactly one variable Parody holds the form constant and swaps the content, or holds the content and swaps the form. Changing both leaves nothing for the reader to measure the gap against, and the gap IS the joke. the gap must be legible as intentional A parody must look like a deliberate, competent execution of the carrier with one thing wrong, not like a failed execution. Competence is what makes the deviation read as a choice. Sloppy imitation reads as inability, and inability is not funny. parody comments; pastiche admires Parody uses the form against itself or against a target. Pastiche adopts the form in earnest. Both are legitimate, they are opposite in tone, and an image must pick one — a half-mocking, half-loving image reads as confused rather than complex. imitate the FORM, not the artefact The target is a genre's conventions — the action-movie poster, the perfume ad, the safety notice, the nature documentary frame, the renaissance altarpiece's compositional logic. Rebuilding a specific protected image shot-for-shot is reproduction wearing a joke as a hat, and it also stops being interesting the moment the reader realises nothing was invented. the medium must match the claim If the carrier is a photographic genre, the parody is photographic. Executing a photographic genre as an illustration changes which genre is being invoked, and recognition breaks. period signature is total If the carrier is of an era, every element belongs to it: the typeface, the printing artefacts, the colour response, the hairstyles, the props, the aspect ratio. One anachronism is the joke; two is a mistake. OVERDETERMINATION the carrier recovered 3× ⇒ 2 redundant equalities Ĉ_composition (the staging and camera signature) · Ĉ_treatment (palette, finish, printing artefacts) · Ĉ_typography (the type system and layout). All three name the same genre. HONEST One carrier genre, one inverted variable, competent execution, a stated stance (comment or homage), and one period if a period is claimed. FORBIDDEN A carrier unrecognisable in the first second ; both form and content changed ; sloppy execution passed off as parody ; mocking and admiring in the same frame with no resolution ; a specific existing artwork rebuilt element for element rather than its genre reconstructed ; a photographic genre parodied as illustration ; two anachronisms ; a formal signature approximated rather than executed ; an explanatory caption doing the recognition work the image failed to do

Inspired by a prompt I shared a few days back but has more depth. I kept it plain English because that's what folks prefer. 2x2 grid, 16:9, days that changed history: Create a monochrome conceptual artwork in which the word [TOPIC] becomes a giant three-dimensional architectural structure, as if it were a museum-scale sculptural diorama or an isometric exhibition model. The letters of [TOPIC] must be massive, solid, black or charcoal forms occupying most of the composition, resting in a clean white space with faint drafting marks, construction lines, and geometric guide overlays. Transform the letters into terrain, architecture, infrastructure, and narrative space. The top surfaces, inner voids, edges, bridges, platforms, and cutaways of the letters should contain miniature scenes that explain [TOPIC]. Populate the structure with tiny humans, vehicles, animals, tools, scaffolds, stairways, camps, carts, cranes, platforms, and other relevant miniature elements inferred from [TOPIC]. The letters must not just display the word — they must become a livable world about the word. The visual language should be precise, sculptural, and intelligent: black-and-white only, clean studio lighting, crisp shadows, isometric or elevated perspective, minimal but elegant technical annotation, and a gallery-model aesthetic. The image should feel like a hybrid of typography, industrial design, architecture maquette, and conceptual storytelling. Important behavior: - infer the narrative content from [TOPIC] - integrate the scenes into the letters themselves, not around them - make the letters physically believable as objects - use tiny figures to establish scale and narrative - keep the image clean, premium, and highly legible - avoid color, noise, painterly softness, and generic poster clutter

A colorful prompt. Works for all kinds of topics. Tested it with GPT 2.5 2x2 grid, 16:9, do this for 4 famous decades: // p5.js GENERATIVE KINETIC POSTER // TARGET_SUBJECT: "[FIGURE]" // ASSETS: "[THEMATIC_ICONS]" function draw() { background(255); // Pure white // 1. ANCHOR GENERATION renderStencilPortrait("[FIGURE]", align=RIGHT, contrast=MAX); // 2. HORIZONTAL VELOCITY FIELD (The Smear) // Extract pixels from the portrait and drag them left across the X-axis for (let i = 0; i < 1000; i++) { let streak = new VectorPath(); streak.color = "AI_INFER(Sampled portrait colors + vibrant Cyan/Red/Yellow)"; streak.velocity = createVector(-50, 0); // Strict horizontal movement LEFT streak.thickness = random(5, 30); streak.draw(); } // 3. FOREGROUND ELEMENTS renderText("[FIGURE]", style="Heavy Italic, White Fill, Black Stroke", angle=-10); spawnVectorIcons("[THEMATIC_ICONS]", blending_mode="MULTIPLY"); } /* RENDER EXECUTION: Translate this generative code into a final, polished 2D graphic design masterpiece. The stark contrast between the rigid, detailed portrait and the violent, horizontal color streaks is the core aesthetic. */

This prompt works for all kinds of topics. I simply used it to visualize 4 systems of government. 2x2 grid, 16:9, do this for 4 systems of government: INPUT: TOPIC ::= [ANY TOPIC] INFERENCE: THESIS ::= infer_central_idea(TOPIC) TENSION ::= infer_most_interesting_relationship(TOPIC) DOMINANT_FORCE ::= infer_agent_object_system_or_concept(TOPIC) SECONDARY_ENTITY ::= infer_affected_or_complementary_entity(TOPIC) ACTION ::= infer_decisive_physical_verb(TOPIC) METAPHOR ::= translate_abstract_relation_to_visual_event(TOPIC) INTERFACE ::= infer_object_where_action_occurs(TOPIC) EDITORIAL_TITLE ::= derive_short_chapter_heading(TOPIC) EMBEDDED_WORD ::= infer_one_word_only_if_useful(TOPIC) EMOTIONAL_MODE ::= infer(irony, warning, wonder, humor, empathy, optimism, tension, discovery) CONSTRAINTS: visual_complexity ::= LOW semantic_density ::= HIGH dominant_colors ::= 2 primary_metaphors ::= 1 focal_actions ::= 1 background_detail ::= MINIMAL text_density ::= VERY_LOW GOAL: communicate_one_nontrivial_idea_in_one_glance

I am actually working on a font app and testing different ideas to see how far I can push it. Here is a prompt that I recently tried with GPT 2.5 Prompt: 2x2 grid, 16:9, do this for 4 famous science fields: data Mechanism = Curvature | Quantization | RandomMotion | EnergyRelease | Attraction | Decay | Repetition | HeatFlow data Operator = Warp | Quantize | Jitter | Slash | Sag | Erode | Ghost | Melt operator_of :: Mechanism -> Operator -- M₁: the transform IS the mechanism data Deform = Df { op :: Operator, amplitude :: Band, uniform :: True_ } -- M₃ band_ok :: Deform -> Bool band_ok d = perceptible (amplitude d) && legible_under (amplitude d) -- M₅ data Sheet = Sh { ink :: One, ground :: Plain, extras :: None } -- M₄ forbidden :: [Regex] forbidden = [ /decorative_unrelated_style/ , /plain_type/ , /wrong_word/ , /illegible/ , /mixed_mechanisms/ , /plain_letters/ , /color/ , /extra_graphics/ , /background_scene/ , /imperceptible_deformation/ , /overdone_illegible/ ] render :: Topic -> Image render t = let s = infer t in assert (op s == operator_of (mechanism s) && band_ok (deform s)) $ sheet (apply (letterforms t)(deform s)) one plain none MimeticTypeWordmark(TOPIC).render()

Children's stories visualized with GPT 2.5. Kind of a cute style. Prompt: 2x2 grid, 16:9, do this for 4 public space non-copyrighted children stories: Garment region G ; ecosystem masses {m_i} T₂ COLLAGE: ∪ m_i G (coverage ≥ τ) ; m_i ∈ ecosystem_of(topic) ; |families| ≥ 3 (structures·machines·signs·greens·roads) T₁ MASCOT: thing = iconic_of(topic) personified ; area(mascot) = max ; central T₃ FRIENDS: |friends| ≥ 4 ; ∀f : faced(f) ∧ symbol(f) ∈ symbols_of(topic) ; gathered around mascot T₄ PROPS: text(prop) ∈ texts_of(topic) ; props ≠ ∅ T₅ STYLE: style_vector(img) = (flat_geometric, one_face_grammar, saturated_family, cream_textured, two_color_block + shadow + footer) -- equality, not similarity η clean geometry, line‑less shapes, paper grain HONEST ⟺ collage ∧ mascot ∧ friends ∧ props ∧ style ∧ η -- "plain costume" = coverage < τ ; "faceless" = ∃f : ¬faced(f) ; "style drift" = style_vector ≠ lock

I don't think I have tried any of these. They look delicious though. GPT 2.5 prompt: 2x2 grid, 16:9, do this for highly underrated international spicy meals: void main() { string dish = "[DISH_NAME]"; // Optical Materials vec4 ar_panel = refraction(frosted_glass) + opacity(0.7); vec3 food_material = subsurface_scattering(high) + moisture(glossy); vec3 toy_material = plastic_roughness(0.4) + visible_joints(true); // Semantic Generation vec3 ingredients = generateDeconstructed(dish, food_material); vec3 charts = generateFlavorRadar(dish); // Layout plateOn(slate_board, ingredients); placeActor(miniature_chef, toy_material, gripping: tongs); orbitUI(ar_panel, charts, around: plate); setBackground(depth_of_field: extreme, focus: foreground, prop: menu_box(dish)); } Illumination: bright, soft, diffused studio lighting, making colors pop, subtle reflections on the plate, no harsh shadows::1.6 Render Stack: high-end food photography, top-down or 45-degree angle, extremely shallow DOF, hyper-realistic focus on the food textures::1.2 Negative: messy presentation, dark and moody lighting, plastic-looking food, cluttered background, cartoonish:: -1.2

I asked GPT to visualize 4 days that changed science. I really love this style. 2x2 grid, 16:9, do this for 4 days that changed science: INPUT: $ TOPIC // ── UNIVERSES OF CANDIDATES ──────────────────────────────────── U_WORD = defining_words($ TOPIC) // short, punchy phrases U_DATE = key_dates($ TOPIC) U_STEPS = chronological_steps($ TOPIC, 4) // exactly 4 events U_ACTORS = human_archetypes($ TOPIC) // soldiers, astronauts, protestors U_VEHICLES = transport_archetypes($ TOPIC) // boats, rockets, planes U_FX = atmospheric_effects($ TOPIC) // smoke, dust, water, fire U_MATERIAL = primary_material($ TOPIC) // concrete, lunar soil, asphalt U_UI_CARD = ui_layouts() // top-left text, bottom-left label // ── SELECTIONS WITH BALANCE RULES ────────────────────────────── WORD = argmax(U_WORD, brevity * iconic_recognition) DATE = argmax(U_DATE, historical_weight) STEPS = sort_chronologically(U_STEPS, k=4) ACTORS = top_k(U_ACTORS, k=4, constraint='each step has a distinct actor action') FX = argmax(U_FX, visual_drama * atmosphere) MATERIAL = argmax(U_MATERIAL, textural_contrast * realism) // ── HARD CONSTRAINTS ─────────────────────────────────────────── 1. The WORD must be rendered as heavy 3D architecture. 2. The letters serve as the literal terrain (ground, walls, platforms). 3. Miniature elements must be scaled correctly relative to the letters (1:100 or smaller). 4. The 4 steps must map spatially across the letters (e.g., step 1 on left, step 4 on right). 5. The background must be entirely clean white. 6. Top-left UI text must contain title, date, subtitle, and 4 numbered steps. 7. Bottom-left UI must be a small white card with a compass/lens graphic. 8. No modern digital text on the letters themselves (unless graffiti is historically accurate). // ── LAYER SETS ───────────────────────────────────────────────── BASE_SET = { Clean white infinite void } TYPE_SET = { Extruded 3D letters made of MATERIAL } ACTOR_SET = { Tiny posed figures on/around TYPE_SET } VEHICLE_SET = { Tiny vehicles on/around TYPE_SET } FX_SET = { Volumetric smoke, fire, water } UI_TOP_SET = { Title, Date, 4 Steps text block } UI_BOTTOM_SET = { White label card } // ── ASSEMBLY ─────────────────────────────────────────────────── SCENE = layer_stack( BASE_SET, TYPE_SET, ACTOR_SET, VEHICLE_SET, FX_SET, UI_TOP_SET, UI_BOTTOM_SET ) // ── CAMERA ───────────────────────────────────────────────────── Format : 2x2 grid, 16:9 overall Resolution : 8K Style : Macro photography of a meticulously crafted physical model Output: one 8K monumental type diorama.

An experimental prompt for turning every day objects into scifi vehicles. Prompt: 2x2 grid, do this 4 common desk or home gadgets: 1. ANALYZE INPUT (Motion & Chassis): Shape: (Streamlined/Blocky/Compact/Sprawling) Structure: (Mechanical/Organic/Modular) Surface: (Armored/Slick/Transparent/Rusted) 2. AUTO-SELECT VEHICLE TYPE: IF Streamlined + Slick → Hover bike IF Blocky + Armored → APC-style rover (non-weaponized) IF Modular + Mechanical → Delivery drone car IF Organic + Curved → Bio-inspired pod IF Transparent + Thin → Glass-canopy scout vehicle 3. EXECUTE: cinematic render, ground contact dust, studio variant optional, believable panel lines.

A cute style for visualizing these kinds of topics. Tested it with GPT 2.5 Prompt: 2x2 grid, 16:9, do this for 4 legendary scientists: -- INPUT SET @ subject = '$SUBJECT'; -- 1. DISCOVER VISUAL PERSONALITY CREATE TEMP VIEW semantic_profile AS SELECT CLASSIFY_DOMAIN(@ subject) AS domain, INFER_PERSONALITY(@ subject) AS personality, INFER_RECOGNIZABLE_MASCOT(@ subject) AS mascot, INFER_ICONIC_SYMBOLS(@ subject, 12) AS symbols, INFER_ENVIRONMENT(@ subject) AS setting, INFER_VISUAL_CULTURE(@ subject) AS cultural_dna, INFER_DOMINANT_EMOTION(@ subject) AS emotion, INFER_COMPLEMENTARY_PALETTE(@ subject) AS palette, INFER_CONCISE_TITLE(@ subject) AS title; -- 2. BUILD CHARACTER IDENTITY CREATE TEMP VIEW protagonist AS SELECT CHARACTERIZE( subject = @ subject, silhouette = 'recognizable simplified subject form', proportions = 'oversized head, compact body, appealing stylized features', expression = semantic_profile.emotion, eyes = 'large expressive graphic eyes', pose = 'personality-revealing gesture', costume = 'contextually relevant clothes or materials', personality_fidelity = HIGH ); -- Never default to the same smiling person. -- -- Some subjects become playful humans. -- Others become animals, tools, landmarks, -- machines, inventions or symbolic creatures. -- -- The expression may be: -- grumpy, curious, confident, sleepy, -- surprised, mischievous or cheerful. -- 3. SYMBOLIC ENVIRONMENT COLLAGE CREATE TEMP VIEW environment AS SELECT BUILD_LAYERED_WORLD( symbols = semantic_profile.symbols, setting = semantic_profile.setting, foreground = 'recognizable everyday objects', middle_ground = 'subject-related landmarks and activity', background = 'simplified environmental context', scale = 'playful, intentionally exaggerated', construction = 'overlapping flat paper-cut shapes' ); Every visual element must explain some characteristic of @ subject. For geographical subjects: use accurate landmarks and cultural objects. For historical subjects: use appropriate objects and period details. For scientific subjects: use genuine scientific symbolism. For fictional subjects: use recognizable world-specific imagery. For abstract subjects: create understandable visual metaphors. -- 4. EXPRESSIVE SIDEKICK SYSTEM CREATE TEMP VIEW sidekicks AS SELECT PERSONIFY( semantic_profile.symbols, count => 3..7, eyes => 'large and simple', mouths => 'tiny and emotionally readable', arms_and_legs => 'only when beneficial', behavior => 'contextually meaningful', variation => 'each has a distinct expression' ); Do not anthropomorphize everything. Some symbols remain buildings, landscapes, vehicles, natural objects or decorative shapes. The poster should alternate between living mascots and recognizable real-world forms. -- 5. GENERATE THE CHARACTER'S MOOD STORY STORY = { protagonist : protagonist, situation : infer_relational_scene(@ subject), sidekicks : sidekicks, conflict : infer_small_emotional_contrast(@ subject), resolution : infer_visual_payoff(@ subject) }; Keep the humor subtle and naturally embedded. The main character might be: proudly showing off the subject, reluctantly participating, reacting to an absurd prop, overwhelmed by surrounding symbols, celebrating an achievement, exploring an unfamiliar world. For serious subjects: choose an appropriate dignified emotion. -- 6. VERTICAL COMPOSITION LAYER_1 = small atmospheric symbols LAYER_2 = protagonist head and shoulders LAYER_3 = symbolic environment LAYER_4 = foreground objects and sidekicks LAYER_5 = large subject title ASCII: ✦ [SKY / SYMBOLS] ✦ ╭────────────╮ │ PROTAGONIST│ │ PERSONAL │ │ EXPRESSION │ ┌──┴────────────┴──┐ │ LANDMARKS │ │ + OBJECTS │ │ + SIDEKICKS │ │ + ENVIRONMENT │ └───────┬──────────┘ FOREGROUND ╔══════════════════╗ ║ SUBJECT NAME ║ ╚══════════════════╝ This is a hierarchy, not a fixed template. Rearrange layers to best explain the subject. -- 7. ART DIRECTION STYLE = 'mid-century picture-book illustration × playful character design × colorful cut-paper collage × vintage travel-poster typography × expressive editorial cartoon'; GEOMETRY = simplified rounded figures bold flat shapes crisp overlapping silhouettes minimal black outlining handcrafted paper edges PALETTE = semantic_profile.palette; Limit primary colors to 5–8. Maintain strong subject-specific color identity. -- 8. TYPOGRAPHY TITLE = semantic_profile.title TYPE: enormous block lettering lightly irregular geometric forms deep offset shadows limited dimensional paper extrusion Optional supporting slogan: one short subject-derived phrase Place title at bottom or another compositional anchor inferred from subject. Title must remain readable at thumbnail size. -- 9. OUTPUT CONSTRAINTS ASSERT protagonist_is_subject_specific = TRUE; ASSERT symbolic_objects_are_relevant = TRUE; ASSERT main_character_expression_is_distinctive = TRUE; ASSERT composition_has_clear_depth_layers = TRUE; ASSERT title_is_readable = TRUE; ASSERT generic_tourism_icons = FALSE; ASSERT random_smiling_objects = FALSE; RENDER: one vibrant vertical storybook collage where @ subject becomes an expressive illustrated character surrounded by a world of subject-specific visual symbols. Aspect ratio: 2:3 portrait.

I asked GPT to visualize 4 topics that were in the news recently. Fun way to keep up with everything going on :) 2x2 grid, 16:9, do this for 4 topics that were in the news: INPUT :: $ TOPIC // STEP 1 — MODEL CAUSAL RELATIONSHIPS G = (V, E) V = infer_conceptual_entities($ TOPIC) E = infer_relationships(V) RELATION_TYPES = { causes, enables, blocks, replaces, transforms, protects, accelerates, exposes, connects, depends_on, challenges } // STEP 2 — CHOOSE VISUAL THESIS E* = argmax_e∈E( meaning(e) × surprise(e) × visualizability(e) ) A = source(E*) B = target(E*) R = relation(E*) // STEP 3 — TRANSLATE RELATION TO ACTION PHYSICAL_ACTION = metaphorize(R, A, B) Examples of transformation logic: enables → opening / supporting / unlocking blocks → obstructing / compressing / enclosing accelerates → pushing / launching / stretching replaces → exchanging / covering / swapping connects → bridging / threading / joining exposes → lifting / peeling / revealing These are transformation categories, not mandatory objects or gestures. // STEP 4 — BUILD SCENE GRAPH [DOMINANT A] | | INFERRED ACTION ▼ [INTERACTION POINT] | ▼ [ENTITY B] CONSTRUCTION: convert nodes into recognizable silhouettes convert relationship into one physical gesture assign contrasting scales eliminate nonessential nodes // STEP 5 — RENDER STYLE Use one dominant dark silhouette against large warm-white negative space. Render secondary actors and objects with minimal monochrome iconography. Small uppercase editorial heading. Thin framing rule. No decorative scenery. FINAL = render( physical_action, semantic_scale_contrast, editorial_minimalism )

Funny this came up. I actually created an app with Opus 4.8 a while back that generates random prompts for these types of experiments. GPT 2.5 Prompt: Paradoxitecture Shadowgeometry Voidminimalism

A cute way to visualize different cities with GPT 2.5 Prompt: 2x2 grid, 16:9, days that changed history: Create a monochrome conceptual artwork in which the word [TOPIC] becomes a giant three-dimensional architectural structure, as if it were a museum-scale sculptural diorama or an isometric exhibition model. The letters of [TOPIC] must be massive, solid, black or charcoal forms occupying most of the composition, resting in a clean white space with faint drafting marks, construction lines, and geometric guide overlays. Transform the letters into terrain, architecture, infrastructure, and narrative space. The top surfaces, inner voids, edges, bridges, platforms, and cutaways of the letters should contain miniature scenes that explain [TOPIC]. Populate the structure with tiny humans, vehicles, animals, tools, scaffolds, stairways, camps, carts, cranes, platforms, and other relevant miniature elements inferred from [TOPIC]. The letters must not just display the word — they must become a livable world about the word. The visual language should be precise, sculptural, and intelligent: black-and-white only, clean studio lighting, crisp shadows, isometric or elevated perspective, minimal but elegant technical annotation, and a gallery-model aesthetic. The image should feel like a hybrid of typography, industrial design, architecture maquette, and conceptual storytelling. Important behavior: - infer the narrative content from [TOPIC] - integrate the scenes into the letters themselves, not around them - make the letters physically believable as objects - use tiny figures to establish scale and narrative - keep the image clean, premium, and highly legible - avoid color, noise, painterly softness, and generic poster clutter

Asked GPT 2.5 to design me a few pieces of furniture based on physics concepts. I actually don't mind the apple one as long as it is comfortable. 2x2 grid, 16:9, do this for 4 famous physics concepts: Concept: Object → one iconic furniture piece (gallery design). 1. ANALYZE INPUT (Ergonomics & Load): Load-bearing: (High/Medium/Low) Touch: (Cold/Warm/Sticky/Soft) Form: (Cantilever/Stack/Tube/Blob) 2. AUTO-SELECT PIECE: IF High load + Tube → Chair IF Stack + Rigid → Shelving IF Blob + Soft → Lounge seat IF Cold + Reflective → Coffee table IF Cantilever + Mechanical → Floor lamp 3. EXECUTE: Gallery pedestal render, material closeups, designer plaque.

Still experimenting with these styles. GPT is so great at these. 2x2 grid, 16:9, do this for 4 scientific theories: Convert the iconic achievement into an exploded technical assembly suspended around the typography. Begin with gigantic black letters spelling [FIGURE]. The central silhouette of [ICONIC ACHIEVEMENT] still subtracts from those letters in white negative space. Then deconstruct the iconic achievement into its meaningful real or conceptual components. Arrange these components along a mechanically coherent exploded axis. Depending on the subject, these components might represent: physical parts scientific principles historical stages cinematic elements architectural systems mathematical layers conceptual dependencies Do not invent arbitrary hardware. Every exploded component should have a meaningful relationship to the achievement. Use fine drafting lines to connect each separated component to the corresponding negative-space location in the typography. Add: numbered callouts section arrows datum references concentric measurement rings tolerance marks assembly paths tiny specification captions The diagram should look as though the person's achievement could theoretically be reconstructed from the drawing. Maintain strict monochrome rendering. Optional dimensional version: render the exploded pieces as tiny matte-white physical components floating above a black-and-white technical sheet. Final character: patent drawing × exploded mechanical manual × modernist biography × identity design.

Art movements of different decades. Tested it with GPT 2.5 Prompt: 4 images: 4:5, stunning art movements of different decades: INPUT: $ SUBJECT Create ONE tall vertical illustrated poster about $ SUBJECT. The image must automatically reinterpret $ SUBJECT as a playful, dense, symbolic character-collage. ──────────────────────────────── SEMANTIC RESEARCH Analyze $ SUBJECT and infer: SUBJECT_TYPE = city place historical event decade / era cultural movement invention scientific event person sport country institution technology franchise abstract topic other TITLE = shortest instantly recognizable name for $ SUBJECT HERO = choose the strongest central character or personification. HERO may be: a person explorer historical figure citizen mascot profession archetype invented representative character Do NOT force a person if the subject is better represented by an object, machine, animal, landmark, or symbolic figure. ICON_SET = select 8–16 highly recognizable subject-specific elements. Possible classes: landmarks buildings transportation foods tools inventions animals clothing products documents symbols flags signs machines technologies landscapes historical artifacts cultural objects entertainment objects SCENE_SET = infer 2–5 important environments or moments. TEXT_SNIPPETS = infer only a few short useful labels such as: dates locations transit names famous phrases event names eras institutions Never turn the poster into an infographic. ──────────────────────────────── CENTRAL CHARACTER SYSTEM Create one oversized friendly central character. Visual anatomy: SYMBOL / OBJECT icon icon icon ◉ ◉ CENTRAL CHARACTER ┌─────────────┐ │ SUBJECT │ │ WORLD │ │ EMBEDDED │ │ INSIDE BODY │ └─────────────┘ object landmark object SHOES / BASE HUGE TITLE BELOW The HERO occupies roughly 55–70% of the poster height. Treat the body as a CONTAINER FOR THE SUBJECT. Examples: city → cityscape forms torso or clothing historical event → event scene spreads across suit/body decade → fashion, technology and entertainment occupy clothing and surroundings science → equipment, diagrams and discoveries populate body country → landmarks, foods, wildlife and culture become costume invention → components and consequences become surrounding characters ──────────────────────────────── ANTHROPOMORPHIC PROP RULE Give selected secondary objects: oversized googly eyes tiny curved smiles simplified limbs only when useful Examples: taxi pizza planet flag camera cassette robot rocket building food tool may become friendly miniature characters. Do NOT anthropomorphize everything. Approximately 30–50% of important props may have faces. This creates the playful children's-editorial feeling. ──────────────────────────────── COMPOSITION Use a dense vertical pyramid composition. TOP: small floating icons / symbols UPPER CENTER: large hero face MIDSECTION: largest concentration of landmarks, objects and subject-specific scenes LOWER SECTION: transportation / food / secondary objects / ground-level details BOTTOM: massive subject title Use overlapping scale changes aggressively. Objects may float beside the hero. Buildings may emerge directly from clothing. Vehicles may drive across the torso. Landmarks may sit behind the head. Objects may extend beyond the body silhouette. The composition should feel like one enormous illustrated sticker collage, NOT separate panels. ──────────────────────────────── STYLE playful mid-century children's editorial illustration × contemporary paper-cut collage × 1960s educational poster × kawaii object character design × bold retro souvenir graphics Use: flat vector-like shapes cut-paper construction slightly imperfect handmade geometry simple rounded characters large circular eyes tiny smiling mouths minimal noses primitive geometric anatomy friendly disproportion thick simplified silhouettes subtle printed-paper grain soft screenprint texture very restrained shading clean crisp edges occasional paper overlap slight tactile depth Avoid: photorealism 3D CGI glossy plastic anime rendering realistic anatomy tiny inaccessible detail complex gradients heavy shadows ──────────────────────────────── COLOR Infer a vivid palette from $ SUBJECT. Prefer: sunny yellow vermillion red bright cobalt blue leaf green turquoise warm coral black cream Use approximately 5–8 dominant colors. Colors should feel screenprinted rather than digitally luminous. Background: warm ivory / aged cream paper with very subtle fibers and speckling. ──────────────────────────────── TYPOGRAPHY TITLE must be huge. Use chunky retro display lettering: bold condensed block letters irregular handmade proportions slight extrusion or black offset shadow high-contrast alternating colors TITLE should occupy approximately 15–22% of the poster. Examples of hierarchy: NEW YORK THE 1980s MOON LANDING Use line breaks based on the semantic rhythm of $ SUBJECT. Optional tiny decorative motifs around title: burst rays waves mini skyline sun stars heart simple emblem Do NOT overload the poster with text. ──────────────────────────────── SUBJECT ADAPTATION If $ SUBJECT is a CITY: choose a representative resident or traveler + landmarks transit food street signs architecture local wildlife cultural objects. If $ SUBJECT is an ERA / DECADE: choose a person wearing era-specific fashion + technology music media toys architecture transportation consumer objects visual motifs. If $ SUBJECT is a HISTORICAL EVENT: make the central participant or archetype the hero + date key machinery symbols locations participants artifacts scientific or political context. If $ SUBJECT is SCIENCE: choose scientist / explorer / symbolic figure + equipment discoveries models objects diagrams relevant environments. If $ SUBJECT is abstract: invent a visual metaphor first, then populate it with concrete symbolic objects. ──────────────────────────────── VISUAL DENSITY Hero = dominant. Major subject icons = 6–10. Minor decorative icons = 4–10. Every major area of the poster should contain meaningful subject information, but retain small breathing spaces between clusters. The viewer should understand $ SUBJECT before reading the title. ──────────────────────────────── FINAL IMAGE one single vertical poster approximately 2:3 aspect ratio warm cream background central oversized friendly character dense subject-specific object collage several selectively anthropomorphized props bright retro colors paper-cut / screenprint texture massive chunky title at bottom joyful, educational, collectible, instantly readable, highly specific to $ SUBJECT. NO grid. NO panels. NO generic stock icons. NO style-description text inside artwork.

Testing my logo prompts with something a bit more challenging. 2x2 grid, 16:9, do this for 4 non-copyrighted public-space books: INPUT: WORD_OR_PHRASE = [ANY TEXT] OBJECT_OR_INSTRUMENT = [ANY OBJECT / FORM] SYMBOL_SYSTEM = [OPTIONAL] MARK_STYLE = [OPTIONAL] SILHOUETTE MAP: [UPPER EXTENSION] handle / neck / cable / branch / smoke / blade / antenna / plume / strap ↑ [LEFT MASS] → [MAIN TYPOGRAPHIC BODY] → [RIGHT TIP OR EXIT] ↓ [LOWER SUPPORT] base / curve / tube / root / wheel / shadow / tail / stand / grounding stroke [DETAIL NODES] = controls, symbols, holes, buttons, marks, rivets, notes, leaves, stars, gears, droplets, etc. [CENTER VOID] = readable counter / opening / object chamber / symbolic negative space Rules: - The full mark must read as both WORD_OR_PHRASE and OBJECT_OR_INSTRUMENT. - The word remains readable. - The object is recognizable through minimal strokes. - Optional symbols appear only as functional or meaningful details. - The largest strokes define the main object profile. - Negative space must be actively designed. - Do not default to any one object or symbol system. - Do not add unnecessary illustration. Render: A premium minimal calligraphic emblem on a clean background.