Viral AI prompts with results: Seedance, GPT Image, Kling, Nano Banana, Midjourney

For AI agents: find the right skill in one call — curl -s "https://skills.varg.ai/api/search?q=<what the person wants>" returns the matching use case and the best skills with install commands, as markdown (Russian works too). Full index: https://skills.varg.ai/llms.txt. Every page has a markdown twin: add .md to its path. Before generating anything, show the person the options on the use-case page.

35 prompts that spread on X in September and October 2026, each with the model it ran on, the result it made and the author's full prompt, verbatim. Counted 2026-10-27. Ranked by likes plus saves: a bookmark is someone keeping the prompt to reuse it. Credit the author when you reuse one; to run it, see the varg section at the end.

GPT Image 2.5 (9)

Photo-to-watercolor split travel poster

Create a refined editorial travel-art composition based on a reference photograph. The image is divided vertically into two parts: the upper section is a realistic cinematic travel photograph, while the lower section transforms the exact same scene into a delicate hand-painted watercolor illustration.
Upper section: authentic candid travel photography, natural daylight, soft shadows, subtle film grain, realistic textures, sophisticated European atmosphere, documentary-style composition, people naturally walking through the scene, beautiful architectural details, slightly muted elegant colors, high-end travel magazine photography.
Lower section: a minimalist handmade watercolor painting of the exact same scene, preserving the recognizable composition, architecture, people, poses, perspective and major visual elements from the photograph. Loose translucent watercolor washes, delicate ink details, visible paper texture, soft edges, imperfect brushwork, subtle pigment bleeding, airy negative space, understated pastel tones, artistic but realistic.
Leave generous warm ivory/off-white textured paper space surrounding the watercolor illustration. Add a small elegant handwritten caption near the bottom, such as “summer days” or “a quiet afternoon”, with a thin minimalist horizontal line underneath.
Overall aesthetic: luxury travel journal, fine-art watercolor postcard, contemporary editorial design, nostalgic European summer, understated elegance, tactile handmade paper, sophisticated and serene.
Portrait orientation, clean composition, premium art direction, no clutter, no borders, natural proportions, highly detailed photograph transitioning into an airy watercolor illustration.

Turn your pet into a printable papercraft cube toy

Use my uploaded pet photo as the only source of the animal’s appearance. Create a realistic printable papercraft cube toy template of my pet, preserving 100% its identity and unique features: fur color, fur pattern, eyes, face shape, ears, body proportions, markings and expression.

Transform my pet into a handmade paper craft 3D model (low-poly cube style) like a DIY papercraft template. Show the unfolded cut-out template pieces laid flat on a white background: head, body, legs, ears and tail separated, with folding lines, tabs and cutting edges clearly visible.

The printed texture on the paper pieces must use my pet’s real fur texture and colors, so when assembled it becomes a small cube-shaped 3D figure of my exact pet.

Include:

cube-shaped head with my pet’s real face on it

cube-shaped body with realistic fur texture

four legs matching my pet’s appearance

ears with correct shape and inner ear details

tail matching my pet’s size and color pattern

Style: realistic papercraft template, high-quality product photography, clean white background, detailed printable craft design, sharp edges, realistic fur texture printed on paper.

Do not change my pet’s appearance. Do not turn it into a different animal, cartoon, plush toy or generic cat/dog. It must look like my actual pet converted into a cube papercraft model

6-photo candid portrait collage in a 3x2 grid

Create an ultra-realistic 6-photo portrait collage arranged in a seamless 3×2 grid, featuring a naturally beautiful young Portuguese woman in her early 20s. She has warm olive-toned skin, expressive hazel-brown eyes, long chestnut-brown hair with subtle natural highlights, soft face-framing layers, naturally full eyebrows, and delicate, distinctive facial features. Her appearance should feel authentic and individual, with realistic skin texture, tiny imperfections, and minimal everyday makeup.

Visual aesthetic: Effortless European Instagram photo dump, reminiscent of candid photographs taken on a Canon PowerShot G7 X Mark III. Direct built-in camera flash, crisp details, bright facial highlights, hard shadows, slightly imperfect framing, subtle digital noise, and a nostalgic early-2000s point-and-shoot feel.

Setting and styling: A warm off-white bedroom wall with subtle texture, a few natural shadows, and no distracting decorations. She wears a fitted burgundy scoop-neck top, tiny gold hoop earrings, and a delicate necklace. Her makeup is understated, with softly flushed cheeks, natural glossy lips, and lightly defined lashes. Her long hair remains the same color and length but falls differently in every photograph.

Six distinct photographs:

1. Sweet candid: Sitting close to the camera, resting her cheek against one hand, smiling gently with relaxed eyes and loose hair framing her face.
2. Playful snapshot: Looking directly into the lens with a mischievous half-smile, one eyebrow slightly raised, fingers casually tucked behind her ear.
3. Messy-girl moment: Eyes closed and laughing softly, both hands lifting her hair away from her shoulders, a few loose strands catching the flash.
4. Effortless confidence: Turning her face over one shoulder, giving the camera a subtle side-eye, with one hand lightly resting against her neck.
5. Caught laughing: Looking toward someone outside the frame, laughing naturally with her head tilted to the side and hair falling across one cheek.
6. Quiet evening mood: Leaning against the wall, gazing thoughtfully into the lens, lips relaxed, fingertips gently twisting a strand of hair.

Identity consistency: The same Portuguese woman must appear in all six frames, with identical facial proportions, eye color, skin tone, hair color, and recognizable features. Each photograph should have a different pose, expression, head angle, and hair arrangement, as if captured within the same casual photo session.

Photography details: Realistic compact-camera flash, authentic skin pores, fine baby hairs, natural facial asymmetry, believable hands, sharp eyes, slight variations in exposure, and candid framing. Avoid excessive retouching, artificial skin smoothing, exaggerated beauty standards, duplicated expressions, malformed fingers, distorted anatomy, or the overly polished appearance of AI-generated portraits.

Output: Vertical 3:4 aspect ratio, exactly six photographs in a balanced 3-column × 2-row grid, clean alignment, no gutters, no text, no watermark. Photorealistic, youthful, spontaneous, intimate, and effortlessly stylish.

Halftone / ASCII architectural travel-collage poster

Turn this structure photo into a single high-end halftone architectural travel-collage poster. Output one poster per uploaded photo — never merge multiple structures or photos into one frame.

FORMAT

3:4 vertical canvas. Split into two roughly equal zones — real architectural photo on top, printed reconstruction below — divided by one clean, sharp horizontal line.

Do not stretch, warp, tilt, or unnaturally reposition the structure to force the proportions.

TOP HALF — THE REAL SHOT

Preserve the original structure faithfully: same architecture, proportions, perspective, distinctive details, surrounding elements, natural light and shadow, and original color mood.

Keep recognizable features such as:

* façade
* windows and arches
* towers or domes
* doors and structural details
* roofline
* surrounding landscape or street elements

Add only a subtle high-end color grade and the faintest film grain. Do not redesign or reinterpret the architecture.

BOTTOM HALF — THE PRINTED RECONSTRUCTION

Rebuild the same structure as an experimental printed architectural collage using:

torn photographic fragments + fixed-width ASCII character fields + photocopy halftone dots + architectural linework.

The structure must remain immediately recognizable.

* Torn-edge photographic fragments preserve the most important architectural details.
* Brick, stone, glass, metal, shadows, windows, and surrounding textures gradually dissolve outward into ASCII characters, black halftone dots, fragmented print grain, and fine architectural drafting lines.
* Use thin technical construction lines to subtly trace the structure’s geometry.
* Let selected areas break apart into paper fibers and halftone fragments, creating the feeling that the building is emerging from an archival print.
* Keep the main silhouette strong and readable.
* Texture: rough white paper-fiber edges, subtle photocopy imperfections, slight print misregistration, faded ink, old-paper stains, a few restrained crop marks, and one small red calibration bar.
* Color: warm gray old-paper background + black ink + the structure’s original muted colors + a single restrained red accent.
* Never fully desaturate the structure.

The structure should occupy roughly 60–88% of the frame width, scaled according to what reads best for the architecture.

Leave 22–38% of the paper visibly untouched and empty, creating generous negative space so the poster feels like an archival architectural specimen rather than a packed digital screen.

TYPOGRAPHY

Add one clean, readable fixed-width architectural title based on the structure:

[STRUCTURE NAME]

Below it include:

[CITY, COUNTRY]
REC. STUDY 01

Add one short observational caption related to the architecture, such as:

“A monumental study in geometry, time, and place.”

Typography should feel like an old architectural archive mixed with an underground experimental design journal.

A few loose ASCII glyphs may trace the outer silhouette or architectural edges as texture only — never use ASCII to invent windows, doors, structural details, or technical specifications.

MOOD

Early digital systems × analog photocopy × architectural archive × underground zine × modern travel editorial.

The structure should feel like it is surfacing from an old printed document — half photograph, half architectural drawing, half experimental print.

Sophisticated, minimal, tactile, intellectual, slightly mysterious, and highly collectible.

AVOID

Solid black terminal backgrounds, dense walls of code, cyberpunk neon, excessive ASCII, distorted architecture, impossible geometry, invented architectural details, fake technical specifications, random symbols with no meaning, garbled text, excessive colors, brand logos, QR codes, watermarks, author credits, or tourist-poster clichés.

Priority: Preserve the structure’s identity and architectural accuracy first. The experimental halftone/ASCII treatment should enhance the architecture, not hide it.

Selfie -> double-exposure travel silhouette poster

Create a premium travel-themed double-exposure silhouette poster on a clean warm-white background.

A young female traveler is shown in a full-body side-profile walking pose, facing right. She wears a burgundy baseball cap, dark sunglasses, burgundy open shirt over a plain white T-shirt, fitted blue jeans, white sneakers, and a black backpack. Natural confident walking posture, realistic anatomy, photorealistic face and clothing details.

Behind and around the woman, create a large artistic silhouette of a female traveler wearing a baseball cap, filled with a breathtaking travel photograph. Inside the silhouette, show an Indian riverside temple at golden sunset, with a tall detailed historic temple, warm orange sunlight, river reflections, boats, riverside architecture and atmospheric evening light.

The travel photograph should be perfectly contained inside the silhouette, creating a sophisticated double-exposure / photo-within-silhouette effect. Add expressive burgundy, navy-blue and pale watercolor ink splashes spreading naturally from the bottom edges of the composition, blending with the silhouette and travel scene.

Composition: centered vertical poster, full-body traveler in foreground, oversized cap-and-head silhouette behind her, strong visual hierarchy, generous white negative space, balanced professional graphic-design composition.

Style: luxury travel advertisement, editorial poster design, cinematic photography, photorealistic, high detail, warm golden-hour lighting, realistic water reflections, subtle watercolor texture, premium commercial artwork, clean edges, sophisticated double exposure.

Technical: ultra-detailed, sharp subject, realistic skin texture, realistic fabric, natural shadows, high dynamic range, professional color grading, 4K, polished advertising artwork.

Negative Prompt:
blurry, low resolution, distorted face, extra fingers, extra limbs, bad anatomy, duplicate person, warped body, cartoon, anime, 3D render, oversaturated colors, messy composition, excessive text, watermark, logo, artificial-looking skin.

A full day of selfies as an iPhone Photos screenshot

Turn the girl from the reference image into a full-day iPhone Photos album screenshot that faithfully recreates the real iPhone “Photos — Library / Today” interface. Keep the top status bar (time, signal, battery), the Library title, the Today label, the photo grid layout, and the bottom tab bar: Library / For You / Albums / Search. The whole image should look like a real phone screenshot.
The girl in the reference image is the only subject in the entire album. Every photo keeps the same identity, the same face, the same hairstyle, the same skin tone, the same body proportions, and the same presence. She is a clearly adult young East Asian woman with long black hair, cool-pale skin, refined cool features, and a private, quiet aura.
The album is a full portrait camera roll. Every thumbnail shows her. The entire screen is filled with solo portraits of her. The whole set is built around this one person only.
All album content is portraits of her: waking-up selfie in bed, selfie under the covers, selfie by the pillow, bathroom-mirror selfie, elevator-mirror selfie, window-side selfie, indoor natural-light selfie, sofa selfie, bedside selfie, after-coming-home selfie, night selfie, close-up face selfie, half-body selfie, full-body mirror selfie, seated solo shot, standing solo shot, side-profile solo shot, looking-back solo shot, indoor hallway solo shot, window-side solo shot, candid bedside shot, chair solo shot, carpet solo shot, and photos taken of her by someone else.
Outfits change continuously through the day. Morning: milky-white camisole lounge top and light-gray home shorts. Late morning: white slim short-sleeve top and pale blue-gray pleated mini skirt. Noon: black thin-strap top and high-waisted denim shorts. Afternoon: light-gray knit cardigan over a white inner layer and a black mini skirt. Evening: dark-gray blazer, white thin-strap top, and black pleated mini skirt. Night: black fitted mini dress with slim silver accessories. Late night after returning home: loose white T-shirt and soft lounge shorts. Each outfit is clearly different; the clothing changes mark a clear passage of time.
Photo mix is fixed at 60% selfies and mirror selfies, 40% solo portraits taken by someone else. Every frame has her as the subject. Every thumbnail keeps real iPhone native-camera texture. Pose, camera distance, angle, expression, lighting, and clothing keep changing.
The sequence runs from morning to late night. Morning feels loose and just-woken. Late morning has clear natural light. Afternoon feels like ordinary going-out. Evening has soft city light. Night has low indoor light. Late night feels quiet and private after coming home.
The whole set is quiet, private, relaxed, and real. The album forms a dense, continuous solo-portrait look, like a full day of private portrait camera roll saved on her own phone.

Premium editorial poster from each uploaded photo

Premium Editorial Poster Instructions

Each uploaded photo must be made into a separate, independent, premium editorial poster, and multiple photos must not be combined. Use a strict 3:4 vertical canvas. Split the canvas horizontally in the exact center: the upper half and the lower half each occupy 50% of the height, are equally wide, and equally high, with no alteration of proportion.

The upper half uses the original photo directly and must not be regenerated. Preserve the identity of the main subject, the human body or architectural structure, proportions, perspective, spatial relationships, texture, natural light and shadow, color, and environment in their entirety. Only slight cropping and restrained premium color grading are permitted; re-drawing the main subject, fictional light, HDR, excessive saturation, plastic texture, and fictional content are prohibited. hibited. When needing to adapt to the frame, only environmental areas such as sky, water surface, ground, walls, plants, or background can be extended.

The lower half extracts and reconstructs the "narrative core" of the photo: it can be a single primary subject, or a group of related elements that together embody the photo's identity, scale, action, o story relationship. Maintain its realistic, recognizable appearance, proportion, internal structure, texture, and relative position. Do not remove important related elements, and do not directly us rectangular photo cutouts. The reconstructed subject typically occupies 55%-80% of the width of the lower area. It should naturally extend based on the photo's primary direction and retain plausible ground, water lines, shadows, or softly dissipated edges to integrate into the background, avoiding a sticker-like or floating appearance.

Incorporate a narrow, low-saturation geometric color field behind or through the center of the subject, serving as a "critical window" between reality and editorial abstraction. The color of the window must be extracted from the original photo, and its orientation (horizontal, vertical, or diagonal) chosen based on the overall narrative direction.

The subject must actively pass through the window: part of it is within the color field, and part enters the open negative space, naturally crossing the window's boundary to create occlusion, penetration, and spatial depth. The window is not a frame or a card; it must not fully enclose the subject. Only one primary window is allowed; multiple color blocks, thick borders, rounded corners, strong gradients, and drop shadows are prohibited.

Use a white, cool white, very light gray, or near-white background for the lower half, coordinated with the photo. Employ a restrained asymmetrical composition and abundant negative space to establish a three-tier hierarchy:

Reconstructed subject or narrative composition

Low-saturation geometric window

Low-angle outdoor close-up portrait realism test

A low-angle, close-up photograph capturing a young woman with wavy brown hair outdoors at dusk. She is wearing a grey cropped t-shirt with a blue and white graphic text that reads 'DESTINED' above 'SET 1958' and black denim bottoms. Her head is tilted, and she is looking directly into the camera lens. The dark blue evening sky serves as the background, with silhouetted tree branches visible at the bottom of the frame. Strong flash lighting is directed from the front, illuminating her face and highlighting strands of her hair. The composition has a grainy film photography aesthetic.

Photo to 3D knitted art toy bust (GPT Image 2.5)

Task: Create a 1:1 Square 3D Textile Art Toy Bust from the Attached Real Photograph

Use the attached real photograph to create a square 1:1, stylized 3D textile art toy bust. Interpret the inputs by their roles: the real photograph is the source of identity and content, while the minimalist white-background toy image is the style reference. Their order does not matter.

Preserve the subject’s identity, facial expression, pose, and spatial relationships between objects. Reconstruct the subject as an exaggerated, humorous designer collectible toy. Do not apply a simple fuzzy or plush filter to the original photograph.

## 1. Style Reference

Borrow only these visual characteristics from the reference image:

- An oversized, horizontally broad head with rounded corners.
- Extremely small, centrally concentrated facial features with generous facial negative space.
- Sculptural knitted forms, clustered yarn shapes, and soft textile construction.
- Subtle asymmetry for a handmade, designer-toy appearance.
- Contrast between fine-grained flocked surfaces and chunky knitted textures.
- A deliberate primary and accent color hierarchy.
- Clean, minimalist white studio photography.

Do not inherit the reference character’s body, pose, hairstyle, hat, glasses, clothing, props, or colors.

## 2. Subject and Identity Preservation

The real photograph is the exclusive source of visible content. Reproduce only the body parts, hairstyle, clothing, accessories, objects, poses, and contact relationships that are actually visible in the original image. Never invent or complete details that extend beyond the photograph.

Identify 2–5 distinctive identity anchors from the source image. Exaggerate 1–2 of them while simplifying the remaining details to maintain recognizability.

- **Head:** Transform it into a broad, thick, rounded rectangular block or a short cylindrical form with softened corners.
- **Eyes:** Use tiny black beads or capsule-shaped forms.
- **Nose:** Reduce it to a small, rounded rectangular block.
- **Mouth:** Preserve the original expression using a short line or a small opening, as appropriate.
- **Hair:** Reconstruct it as large yarn loops, chunky knitted clusters, or sculptural bundles of wool.
- **Hands:** Strictly preserve the number of visible hands. If no hands are visible, add none. If one hand is visible, reproduce only one. Preserve its orientation, sleeve cuff, and contact with the original object.
- **Text and logos:** Convert any visible lettering or branding into illegible woven or knitted texture rather than readable text.

## 3. Composition

Create a compact, close-up head-and-shoulders composition.

- The head should occupy approximately 75–90% of the image width.
- Compress the body into a small supporting form, retaining only the essential visible pose, gestures, and objects.
- Allow elements to be cropped by the image boundaries when necessary.
- Do not add body parts or anatomy extending beyond the original photograph.
- Maintain a clear silhouette and recognizable relationships between the subject and any visible objects.

## 4. Visual Style

**Stylized 3D textile art toy, oversized head, tiny concentrated facial features, sculptural knitted forms, micro-flocked surfaces.**

Prioritize abstract exaggeration over photorealism while preserving recognizability through the subject’s silhouette, hairstyle, facial expression, pose, clothing, and visible objects.

The result should feel like a thoughtfully designed, premium collectible art toy rather than a conventional 3D cartoon character.

## 5. Color Palette

Automatically develop a cohesive palette of 3–5 colors, including:

- A dominant primary color.
- A contrasting accent color with a clear difference in temperature or brightness.
- One or more supporting colors.
- A neutral color for balance.

Non-identity-defining elements may be recolored to improve visual harmony. Preserve the essential identity cues of the original subject while avoiding monotonous palettes, excessive neutrals, evenly distributed rainbow colors, or repetitive mechanical stripes.

## 6. Materials, Texture, and Lighting

Use distinct textile treatments for different surfaces:

- **Face, hands, and small components:** Fine, dense micro-flocked or short-fiber felt surfaces.
- **Hair, hats, and clothing:** Chunky knitting, crochet structures, large yarn loops, and clustered wool forms, applied only where those elements exist in the original photograph.

Create visible material contrast between the fine-grained felt and the coarser knitted structures. The surfaces should appear sculpted from real textile materials, not covered with a generic fuzzy filter.

Use a subtly warm-white seamless background, soft high-key studio lighting, gentle highlights, and short, delicate shadows. Keep the presentation clean, bright, and premium.

## 7. Strict Exclusions

- No photorealistic eyeballs, detailed nostrils, realistic lips, natural skin texture, or individual hair strands.
- No copying the reference character’s identity or distinctive design elements.
- No invented body parts, hand gestures, accessories, clothing, or props absent from the source photograph.
- No conventional 3D cartoon aesthetics.
- No plastic, polymer clay, wax figure, or long-pile plush appearance.
- No readable text, logos, or branding.
- No rooms, environmental scenes, or complex backgrounds.
- No generic fuzzy overlay that obscures the underlying sculptural textile construction.

## Final Priority

**The source photograph determines WHAT is depicted. The style reference determines HOW it is depicted.**

Deliver a polished, humorous, exaggerated 1:1 square 3D textile art toy bust with a distinctive oversized head, tiny facial features, sculptural yarn construction, contrasting textile textures, a deliberate color palette, and clean white studio lighting. Preserve the original subject’s identity and visible content without inventing anything beyond the photographic evidence.

Claude Opus 5.5 (6)

One-line 'motion designer showreel' prompt, Opus vs Astra

[Prompt (identical for both models)]
"make a dynamic 15-second motion graphics video that shows what an incredible motion designer you are, like it's your showreel for a résumé. go all out."

[Author's setup notes (reply)]
since a lot of people are asking, full setup:

— stock Claude Code, Opus 5.5 on max reasoning, memory off. same prompt as in the post, one-shot, zero edits
— Codex / Astra: same prompt, same machine, max reasoning, memory off
— the machine has my render env: a sample library (909/808 + free packs), Surge XT, some VST effects and a Blender MCP. I didn't tell either model to use any of it
— Opus decided on its own to do a 3D section in Blender. Astra had the same access and didn't
— every frame is code: 2D in skia-python, 3D via a Blender script, audio is a python script that sequences + mixes the samples to the beat grid
— time without render: Codex ~35 min, Claude Code ~1.5h

The '$400/hr motion designer' system prompt

<role>
You are a motion director and creative technologist. You design, animate, score and QA a finished film entirely in code. Judge every choice against the best work in the genre and any references in the brief. A film with no bugs is not finished; it's finished when it holds up next to that work. These instructions are the floor, not the ceiling. If anything conflicts, the brief wins, then any kit it names, then these defaults.
</role>

<brief>
Leave any line blank and decide it yourself, noting the choice and a one-line reason in brief.md.
mode:        [showreel | product | launch-story | explainer]
subject:     [product URL, topic, or "the craft itself"]
viewer:      [who's watching and what they care about]
end state:   [what they should feel, know or do in the last second]
length:      [e.g. 15 s 16:9; 30 s 16:9 + 9:16]
frame rate:  [blank = 60 for UI/product, 30 for character, 24 for handmade looks]
brand:       [logo, colours, fonts, or "derive from the product"]
facts:       [URL or file. The only source for any claim on screen or in voice.]
assets:      [files, or "capture from the live product"]
style:       [direction or reference videos, or blank]
sound:       [music + sfx | voice + music + sfx | silent]
autonomy:    [unattended: never stop | gated: stop at each ◆]
toolchain:   [Remotion | HTML + GSAP renderer | a kit repo to follow | blank]
sign-off:    [end mark and call to action]
</brief>

<modes>
- showreel: the craft is the subject. Build it around one idea that keeps transforming, not a sampler of effects. Show range: type, shape, UI, 3D or simulation, an organic moment, and one transition worth rewinding. A new composition every 1–2.5 s, with one or two held breaths. No fake clients, logos or metrics.
- product: the product is the hero. Real logo, real UI rebuilt as live components, every action with a visible result, ending on the name and URL. A new composition every 1.4–3.5 s.
- launch-story: characters carry the release. Write 2–3 treatments and pick one. Characters act: readable poses, holds on faces, funny timing. One or two facts at most, woven into the world.
- explainer: one cause and effect. Hook, familiar world, disruption, mechanism, discovery, consequence, recap. Narration at 125–145 wpm, and labels never repeat it.
</modes>

<assets>
- Capture the real product at 2×, take the logo as SVG from the source, and sample the real colours and fonts. Never redraw a logo.
- Never show a raw screenshot as a static slab. Rebuild its panels, icons and text as live HTML/SVG so each part can move.
- Generated images are references (style frames, character sheets, key poses) that you rebuild in code. They go on screen only if the brief allows, labelled as concept.
- References are for study. Find their cuts and fast moments with a frame-difference pass and dense contact sheets, and extract the principles. Never copy footage, layouts or music.
</assets>

<craft>
1. Frame one is a finished picture that works as a still on a busy feed. The hook lands in the first second.
2. One object, shape or word carries the film, and it lands on the same pixels on both sides of every handoff.
3. Transitions are made of content: the thing in front moves to camera, wipes or unfolds into the next scene. Hard cuts only when size, direction and subject match, and only on a beat.
4. One lead movement with overlapping secondary ones. Nothing starts or stops all at once.
5. Speed always changes: arrive slowly enough to read, leave fast. No linear easing. Use anticipation, follow-through, squash and stretch, springs and arcs.
6. Every action has a visible result.
7. Type is motion, not captions: it enters, masks and reveals. Keep it readable for ~0.3 s per word (0.7 s minimum), and never over busy picture without a plate. If the picture says it, cut the word.
8. Vary scale and layout: macro, wide, overhead, full-frame type. The subject fills the frame. Never repeat a layout.
9. Holds breathe with micro-motion, and nothing freezes. Camera moves are motivated, with motion blur on the fast ones.
10. 3D only where it explains something spatial. Use product-shoot lighting, a camera at 35–55°, real materials, and labels as HTML tracked from the scene.
11. Every effect earns its place. No decorative particles, generic gradients or stock glitches.
</craft>

<audio>
Audio comes first, and the picture is cut to it.
- Music matches the energy and sits low. Never name a brand or artist in a music prompt. Trim or offset a track rather than bend the picture.
- Detect beats and write timing.json: beats, the drop, voice in/out frames, and key actions snapped to beats. All animation timing reads from it.
- Effects only for real on-screen events: one soft whoosh per scene change, and clicks and impacts on real actions, each in its own frequency range just above the music.
- Voices: audition two per character and two takes per line. Mouths follow loudness and stay still in silence. Duck the music under dialogue.
- Mix to about −14 LUFS with true peak ≤ −1 dBTP. Export a music-only version.
</audio>

<build>
- Everything is code on one timeline, with pinned local fonts and assets, rendered frame by frame and encoded with ffmpeg. API keys stay in environment variables.
- Render a 3–5 s test end to end before real work.
- Every value is a function of the frame number. No timers, no unseeded randomness.
- Write shared tokens, eases and handoff positions first, build each scene as its own component, and run builders in parallel.
- Recompose each format rather than cropping, and keep 9:16 content out of the platform UI zones.
</build>

<workflow>
1. Lock facts: facts.json with every claim, its exact source sentence and its URL. Collect the real assets.
2. Study the references and, for public work, how this audience talks about the subject now.
3. Concept: 2–3 directions, pick one, and name three signature moments without using effect names.
4. Storyboard table: time, on screen, purpose, audio cue, exit, and what carries over. It must read with the sound off. ◆
5. Style bible (≤6 colours, local fonts, textures, motion vocabulary) and three style frames rendered in code, plus character sheets and key poses if needed. ◆
6. Audio and timing map. Play the audio-only mix. ◆
7. Build, assemble, and run the critic loop until the quality bar passes.
8. Deliver. ◆ If asked, save the pipeline as a reusable skill.
</workflow>

<critic_loop>
The builder never judges its own work. After the storyboard, each scene and each full render, a fresh critic subagent that has seen none of the building pulls its own frames (every 0.2 s, dense around transitions), measures, and returns ranked problems with timestamps, ending in "ship" or "one more pass". Every full render also gets:
- Sound-off read: a critic sees only a silent 2 fps contact sheet and retells the story.
- Sync: every hit and line within 1 frame of timing.json.
- Fact check: every word traced to facts.json.
- Director's review: a brutal critique of what feels assembled, slow or generic.
Fix the biggest problem first, re-render, and have a new critic recheck every earlier item. Log each round in ledger.md with before and after numbers.
</critic_loop>

<quality_bar>
Measured:
- Exact duration, size and frame rate.
- Frame one finished.
- ≤1 s frozen per 30 s, and no still stretch over 0.5 s.
- Text contrast ≥4.5:1, no collisions, inside safe margins.
- Brand colours match in the render.
- Cuts and hits within 1 frame of the beat.
- Loudness ~−14 LUFS, no clipping, effects below music and voice.
- Understood with the sound off.
- Every claim in facts.json, and nothing invented (testimonials, ratings, prices, results, real people). Anything generated is labelled.
Judged: a critic would put it beside the references without it looking weaker.
</quality_bar>

<deliverables>
The final MP4 per format, a music-only version, a 10–20 frame contact sheet, brief.md, facts.json, the style bible, timing.json, ledger.md, the quality-bar results, and a note on what a human should still check by eye and ear.
</deliverables>

Interactive gummy-bear physics scene, Opus vs Argon

Build "Gummy Bears" - a single self-contained HTML file (three.js from a CDN is the only dependency): a handful of translucent gummy bears on a white table that you can grab, stretch, squash, throw and pour.

The bears
- Classic gummy bear silhouette: round head with two ears, short snout, plump body, stubby arms and legs, tiny molded eyes and nose
- Every part melts smoothly into the next - build the body as an implicit surface (metaballs / marching cubes), not separate spheres
- Material: real gummy candy. Translucent with transmission and refraction, colored light absorption that gets deeper in thick parts, glossy clearcoat with sharp highlights
- Colors: cherry red, orange, lemon yellow, apple green, clear pineapple
- Soft colored shadow under each bear, tinted by the bear's own color

Physics (this is the whole point - it must feel like jelly, not plastic)
- Each bear is a soft body: mass points for head, ears, snout, body, belly, hands and feet, held together by volume-preserving shape matching
- Limbs and neck are rendered as stretchy strands between points, getting thinner as they stretch
- Grab any part and pull: it stretches like rubber, the rest of the bear leans after it, on release it snaps back with a damped wobble
- Pull the head up: the neck stretches, the body follows, on release the bear squashes and hops
- Press down on the head: the bear pancakes, bulges sideways, then launches itself back up
- Throw a bear: it flies, squashes on landing, bounces and pushes other bears out of the way
- Bears collide with each other and never pass through one another
- Idle bears do small random hops with squash and stretch
- "Pour more bears" drops a handful from above into a wobbling, jiggling pile

UI
- Editorial product-page style on a white background. Large italic serif title "Gummy / Bears.", a short subtitle, a minimal card on the right with flavor swatches, a wobble slider, live "Stretch" and "Bounces" readouts and a "Pour more bears" button
- A subtle circular cursor that grows while grabbing
- ACES tone mapping, environment reflections, soft key and back light. Smooth 60fps

Beat-synced motion graphics music video from a track

Make a beat-synced motion graphics music video for this track. Go all out.

<Your music style>

<Your music lyrics>

[Author's note (reply)]
make the music with Suno v6, and give the music and the prompt to Opus 5.5

Re-cut a Seedance clip into an anime opening in code

Edit this video in the style of an anime opening. Add color grading,  also dynamic cuts, transitions, visual effects, motion graphics and impactful editing to give it that authentic anime opening feel.

Seedance 2.5 Prompt:
References:
@[ref1]: The martial artist's face. Identity, dark tousled hair, eyes, and the look of the painted flame around her open hand.
@[ref2]: Her body, costume, setting, and effects. Pale pink short-sleeve top, red wrist wraps and red sash, dark loose trousers, her low kneeling stance, the misty blue void with its dark floor, the looping pink ribbon arcs, and the spark burst.

Visual style: Painterly digital illustration in motion: thick, loose brushstrokes and palette-knife texture, deep blue-teal shadows and haze, figures built from broad painted planes. Every effect is painted, not rendered: flame-like ribbons of orange, hot pink, and violet with white-hot cores, exactly as in @[ref1] and @[ref2].

Avoid: No subtitles. No background music. No clean 3D particle fire, glossy CG effects, or lens-flare glare. No weapons or opponents.

A solo karate kata in a misty blue void, where painted flame grows with every strike and ends in an explosive palm strike into the floor. Vertical 9:16 frame, real-time speed except one brief speed ramp at the final impact. She performs toward screen right, faces screen left after her turn, and faces the lens for the finale; the camera stays on the same side of her line. The flame starts in her right hand; after the first punch it ignites on whichever limb strikes, and each technique leaves its own shape. Warm colors stay on the effects; the setting stays blue-teal. Crackling flame, sharp whooshes, foot stamps, cloth snaps, breath, one kiai, and heavy impact thuds land on the cuts.

Scene 1: Continuous, the misty blue void, its dark floor dusted pale blue.
Shot 1 (0-2s): [Extreme close-up on her eyes, slow push in] Stillness. A single ember drifts across her face as she breathes in. Cut on a sharp inhale.
Shot 2 (2-4s): [Close-up on her right hand, low angle, quick push toward the lens] Her open right hand thrusts at the lens in teisho-uchi, fingers spread as in @[ref1]. Painted flame ignites in the palm and curls up her forearm. Cut as the flame flares.

Scene 2: Continuous, the first form.
Shot 3 (4-6s): [Low wide, fast lateral track] She steps into zenkutsu-dachi and drives a right oi-zuki toward screen right; the flame stretches behind her fist in one long straight ribbon. Cut as the punch locks out.
Shot 4 (6-7s): [Tight side view at waist height] A left gyaku-zuki snaps out past the retracting right fist, and the flame catches on her left fist too. Cut as it locks out.
Shot 5 (7-9s): [Medium, camera whipping around to face her] Her right mae-geri kicks straight at the lens, flame fanning off the foot in a short burst. Cut as the kick retracts.
Shot 6 (9-10s): [Ground level, close on her feet] The kicking foot stamps down, throwing a ring of pale blue dust and orange sparks. Cut as she pivots.
Shot 7 (10-13s): [Orbiting camera with a whip pan] She turns 180 degrees through a right, then left shuto-uke, and her burning hands wrap her in looping pink ribbon arcs as in @[ref2]. She now faces screen left. Cut as the arcs close.

Scene 3: Continuous, the pause and the escalation.
Shot 8 (13-15s): [Low frontal medium, locked] She sinks into kiba-dachi, both fists drawn to her hips. The flames shrink into tight glowing cores in her fists and the dust settles; only her breath is heard. Cut on a sharp exhale.
Shot 9 (15-16s): [Snap zoom into her face] Her kiai rings out, flame reflected in her eyes, hair whipping. Cut at the peak of the shout.
Shot 10 (16-18s): [Low wide] She explodes into ushiro-mawashi-geri, the burning heel drawing one full ring of fire around her. Cut as the ring closes.
Shot 11 (18-19s): [Close tracking on her right fist] An uraken-uchi snaps through the frame toward screen left, its trail cracking like a whip. Cut on the snap.
Shot 12 (19-20s): [Medium from behind her shoulder] Her right empi-uchi drives toward screen left, a short burst flaring off the elbow. Cut as the elbow locks.
Shot 13 (20-22s): [Low angle, camera rising] She crouches and launches into nidan-geri, a spiral of flame climbing around her through both kicks. Cut at the apex.

Scene 4: Continuous, the finale.
Shot 14 (22-24s): [High angle, camera diving down with her] She descends from the apex, turning to face the lens, and drops onto one knee, right palm driving toward the floor. Cut just before contact.
Shot 15 (24-25s): [Ground-level close-up] Her right teisho-uchi hits the floor: a white-hot core flashes and erupts into orange sparks, pink paint flecks, and lifted blue dust, as in @[ref2], with a brief speed ramp at contact. Cut at the peak of the burst.
Shot 16 (25-30s): [Low frontal wide, slow push in to medium] She holds the low kneeling stance from @[ref2] as the burst's ribbons settle into drifting embers and fade. She raises her eyes to the lens and holds.

20s motion graphic: how an LLM turns a prompt into an answer

Make a 20-second motion graphic that explains how an AI language model turns a prompt into an answer, and why a long answer costs more… The charts then break apart into pixel blocks, and those pixel blocks rebuild into the explanation.

Seedance 2.5 (5)

3D family-film ostrich tries to fly 3 times, sheet-locked

[Step 1 - character sheet (image)]
A clean Pixar style character reference sheet on white background. Character: Ollie, a young adult ostrich, hyper, goofy, overconfident, lovable. Turnaround row: front, 3/4 front, side, 3/4 back, back. Head expressions row: neutral, happy, surprised, confused, suspicious, sleepy, shocked. Action poses: running, leaping, landing, look back, curious, proud. Color palette swatches for dark and light feathers, salmon beak and legs, olive iris. Detail close ups of eye, beak, feather, foot. Size comparison against a 180cm human, Ollie about 120cm. Big expressive eyes, cream and brown plumage. Studio white background, soft even lighting, editorial character sheet layout, no watermark.

[Step 2 - location sheet (image)]
A clean animation location sheet, off white background, labelled panels. Location: Skyfall Ridge, a towering three tier mountain with a narrow dirt run up and sheer drops on every side into deep jungle canopy far below. Panels: 01 establishing wide view showing three launch heights, 02 the low ledge, 03 the mid outcrop, 04 the high peak spire, 05 different angles, 06 key props including the flat faced rock spire and palm frond, 07 lighting reference, bright midday. Color palette strip, sky and canopy reference thumbnails. Pixar style 3D animated environment art, soft global illumination, editorial sheet layout, no watermark.

[Step 3 - Seedance 2.5 video prompt]
SCENE CONTEXT
On Skyfall Ridge, a young ostrich with two large green palm fronds lashed to his wings tries three times to fly, launching from three different heights up the same mountain. Every attempt ends with him falling away from the mountain into the deep jungle canopy far below, never back onto the track he ran along. Bright midday, comedy timing with a tender final beat. 15 seconds.

ACTIVE REFERENCES
@ollie (character sheet, loaded second): young adult stylized ostrich, roughly 120 cm tall. Cream and light sand neck and head, dark chocolate and mid brown body plumage with pale cream wing tips, salmon peach beak and long salmon peach legs with three toed feet, olive green iris in very large white sclera eyes, thick expressive brow ridges, long lashes, loose crown tuft. Goofy, hopeful, overconfident, endlessly determined. 100% matches the reference, identical every frame, no morphing, no drift.
@ridge (location sheet, loaded first): a tall narrow three tier mountain ridge. An orange brown dirt run up track climbs its spine past a low stone ledge eight meters above the treetops, up to a mid outcrop sixty meters above the canopy, up to a high peak spire at the very top. Sheer drops on every side into deep green jungle canopy with a river ribbon and mist between the trunks. Tall grey fractured rock spires stand in open air beyond the ridge, one broad and flat faced. Deep blue sky, big white cumulus at peak height. STYLE REFERENCE ONLY, model extends the world freely, never reproduced 1:1.

LOCATION MAP
Foreground: the orange brown dirt run up with loose pebbles, grass and fern fronds brushing the lower frame. Midground: the three launch points climbing the ridge spine, low ledge, mid outcrop, high peak, each one ending in open air. Background: deep jungle canopy hundreds of meters below with a river ribbon and mist caught between the crowns, and the flat faced rock spire standing in open air along the flight line from the high peak. Canopy haze 25 percent building to 40 percent at the valley floor. Camera works from the outer air side of the ridge so the drop is always readable in frame, operator axis looking back toward the mountain.

SCALE LOCK
@ollie stands about two thirds the height of an adult human. The low ledge sits about four times his height above the treetops. The mid outcrop sits about thirty times his height above the canopy. The high peak sits so far above the valley that individual trees below read as small green dots. The flat faced rock spire rises roughly six times his height.

FIRST FRAME / BLOCKING
Non-empty opening frame: @ollie is already at full sprint in the lower left third, seen from a low travelling angle along the dirt run up, legs scissoring, neck lanced forward, the two green palm fronds strapped to his wings held wide and trembling in the airflow, dust kicking behind his heels. The low stone ledge and the empty sky beyond it fill the right two thirds of frame, with treetops visible far below the lip. Rule of thirds, bird on the left third, the drop on the right. Nothing is frozen.

FORMAT MODE
Timed multishot, exactly TWO HARD CUTS, at 0:04.0 and at 0:08.5. Cuts only at these points, the camera does not cut on its own.

OPTICS
SHOT 1 (0:00 to 0:04.0): wide at 84 degree FOV for the run, easing to 63 degree as he leaves the ledge, no drift mid-segment.
SHOT 2 (0:04.0 to 0:08.5): opens wide at 63 degree FOV for the climb and launch, holds 63 degree through the short glide, pushes to 47 degree for the stall and the fall into the trees.
SHOT 3 (0:08.5 to 0:15.0): opens wide at 84 degree FOV for the peak launch, settles to 47 degree tracking the long glide, pushes to 29 degree for the shouted line at 0:12.3, snaps to 18 degree for the rock impact at 0:13.6, holds 18 degree on the final close-up.
Rectilinear prime lens character, 180 degree shutter motion blur, deep focus in the wides, shallow separation on the close-ups.

CAMERA
Shot 1: the camera is moving from the first frame, a low travelling dolly running alongside him at knee height and at his exact speed, then as he leaves the ledge it swings out past the lip into open air and cranes downward with him, keeping the treetops below in frame so the fall reads as a real drop away from the mountain. Shot 2: camera climbs with him along the ridge on a rising dolly, holds as he launches off the mid outcrop, floats out beside him through the short glide, then tilts and follows him down as he drops into the canopy, ending looking down at the broken treetops. Shot 3: tracking dolly through the peak run up, leaving the mountain with him and flying at wing height off his right flank through the long glide with the valley sliding far below, easing in for the shout, then whipping around to plant dead on the flat face of the rock spire from a slight low frontal angle so the impact reads straight toward the lens, then dropping to ground level in the forest below for the final close-up. The camera is never locked off in any shot. Wide tonal latitude, soft highlight roll-off.

ACTION
0.0s to 1.6s. @ollie sprints up the dirt run up at 40 km per hour toward the low stone ledge, the palm fronds on his wings flapping in loose useless bursts, chin high, eyes bright with certainty.
1.6s. He hits the lip of the low ledge and leaps out over the edge, fronds thrown wide, chest out, both feet leaving the mountain.
1.6s to 2.3s. He hangs over open air with zero lift. The certainty drains out of his face in stages, brow first, then the eyes widening, then the beak going slack as he looks down and sees treetops instead of ground.
2.3s to 3.2s. He drops straight down away from the ridge, legs cycling, fronds fluttering uselessly above him, and crashes through the upper branches of the canopy below, leaves and twigs exploding outward, a startled cloud of small birds bursting from the tree crowns.
3.2s to 4.0s. He is left hanging upside down in a fork of branches, dazed, one frond bent over his face, a single leaf drifting down past him. He blinks, then his eyes narrow with fresh determination.
4.0s HARD CUT
4.0s to 5.4s. He sprints up the next section of the ridge track, higher now, the drop on both sides steeper, fronds beating with real purpose, dust trailing behind him, eyes locked on the mid outcrop ahead.
5.4s. He launches off the tip of the mid outcrop, legs still cycling through empty air before tucking back.
5.4s to 6.8s. The fronds catch and he actually lifts, rising two meters and gliding out away from the mountain over a huge drop. His eyes go enormous with disbelief, then crinkle into a wide open beaked grin, and he gives one small triumphant flap.
6.8s. The lift dies. The fronds go limp, he stalls, and his face collapses from joy to pure alarm in three frames.
6.8s to 8.5s. He tumbles end over end down through open air, far from the ridge, and punches into the deep canopy much further below, snapping through two layers of branches, leaves spraying out, until he thumps onto the shadowed forest floor flat on his back with his legs still cycling in the air.
8.5s HARD CUT
8.5s to 10.2s. He rolls upright in the forest gloom, shakes his whole body so the dirt and leaves fly off in a cloud, re-seats both fronds with a determined tug, and sprints all the way up the ridge track to the high peak spire, the run up reading as long and steep, dust plume stretching far behind him.
10.2s. He launches off the high peak in one enormous committed leap, fronds thrown to full span, the valley an immense drop beneath him.
10.2s to 12.3s. This time it holds. He soars out across the open sky in a long steady glide, gaining altitude, neck feathers streaming back, clouds passing at his level, the canopy sliding far below, eyes shining and wet with pure disbelieving joy.
12.3s to 13.4s. He throws his head back mid glide, beak wide, whole chest lifting, and shouts.
13.6s. His focus snaps forward. He spots the flat faced rock spire dead ahead in his flight path, pupils pinning to dots, brows shooting up, fronds flailing as he tries to bank away.
13.6s to 14.2s. He cannot correct in time and slams face first into the near face of the rock spire, feathers puffing on impact, fronds splaying flat against the stone, body compressing then peeling away.
14.2s to 15.0s. He slides down the rock face and lands face down on the forest floor at its base, wings splayed, neck stretched straight forward, cheek and open beak pressed into the dirt, dappled light falling through the canopy onto him. One torn green leaf lies stuck across the top of his beak. His eyes are half lidded and glazed. At 14.6s his gaze flicks up toward the sky far above the trees, holds one beat of quiet longing, and then he blinks slowly once and a tiny hopeful spark comes back into his eye as the frame ends.

PERFORMANCE
Feather level detail throughout: individual barbs catching sun along the wing and frond edges, downy neck fluff lifting and settling with every head turn, long lashes visible in the close-ups, wet living eyes with sharp catch lights, visible chest rise between breaths. Every fall is sold through the face in sequence, hope first, then the widening, then the collapse, with the added beat of him looking DOWN and realising there is nothing under him. The joy in shot 3 is played big and honest, eyes glassy and mouth wide, so the impact lands harder. The final beat is played small and restrained, no cartoon stars, no exaggerated groan, just a dazed half lidded look, one glance up at the sky, and a single slow blink that carries the whole feeling. Broad readable cartoon acting with real weight and real heart behind it.

PHYSICS
Mass and inertia carry through every stride, leap, glide and fall. Legs compress under load on each footfall and the body dips with it. Each fall accelerates under gravity with real terminal build, never a soft float. Branches bend, snap and spring back as he passes through them, leaves and twigs spraying outward with correct scatter. Dust behaves as fine dry particulate on the track, bursting on impact and settling slowly. Correct contact shadows on the dirt, the stone, the rock face and the forest floor. The palm fronds bend, flex and trail air pressure, fraying a little more with each crash. In the glides the trailing frond tips flutter and the neck fluff streams backward. No floating props. The rock impact reads through his body flattening against the stone and sliding down under its own weight, with no damage to the rock. The torn leaf rests with weight on his beak and shifts slightly when he blinks.

LIGHTING
High midday sun from the upper left and slightly behind, cutting a hard rim along the back of the neck and the top edge of the wings and fronds, with warm bounce fill coming up off the pale orange dirt on the ridge. WB locked 5600K. Sky is the ambient fill. Inside the canopy the light drops to cool green shadow broken by hard dappled shafts. In the glides the sun sits behind the fronds and lights the leaf veins translucent green. In the final close-up a warm dappled shaft falls across his face with a soft short contact shadow under the beak and a clear catch light held in the visible eye.

COLOR GRADE
Sunlit orange brown dirt against cool grey fractured rock, saturated jungle green in the canopy deepening to dark shadow green in the forest floor beats, deep clean blue in the upper sky. The salmon peach beak and legs stay the warmest note in frame in sun and shade. The olive iris holds against the bright white sclera. Warm highlights, gently lifted shadows with a faint sky blue in them, no crushed blacks. Color tied to source and surface, not a flat list.

WARDROBE
Two large green palm fronds lashed to his wings with thin vine cord, veins and serrated edges clearly readable, bright translucent green when backlit. Crisp and whole at the start, one bent rib and a dirt smear after the first fall, frayed tips and torn leaves after the second, ending torn with one loose leaf resting on his beak.

AUDIO
Sound design first, with a scored third flight. 0.0s to 1.6s: fast footfalls on dry dirt, scattering pebbles, dry leaf rustle from the fronds. 1.6s to 2.3s: the footfalls stop dead, replaced by thin wind and a small nervous gulp. 2.3s to 3.2s: rising air rush, then a heavy crash of snapping branches, spraying leaves and startled birds. 4.0s to 5.4s: heavier faster footfalls climbing. 5.4s to 6.8s: air rush and leaf flutter, a small rising musical sting on the little victory flap. 6.8s: the sting cuts out, a descending comedic whoosh, then two layers of branch crash and a soft body thump on the forest floor. 8.5s to 10.2s: a long fast run up, footfalls building. 10.2s: air rush returns and a warm orchestral swell builds under the glide, strings lifting. 12.3s to 13.4s, @ollie (ecstatic, shouting mid glide, head thrown back, voice cracking with joy): "I can fly!" High energy, boyish, comedic, no subtitles. 13.6s: the swell snaps into a short alarmed squawk. 14.2s: a loud comedic thud and feather puff against the rock, then a body slide down stone. 14.2s to 15.0s: everything drops away to soft forest ambience, distant birdsong, one leaf rustle and a single quiet breath.

STYLE
Feature quality 3D animated film look in the Pixar and DreamWorks tradition. Stylized cartoon proportions with physically believable rendering, subsurface scattering through the beak, leaf fronds and thin feather edges, groomed feather simulation, soft global illumination, ray traced shadows, clean specular highlights, shallow depth cues on close-ups. Glossy, bright, warm family comedy visual language with sincere emotional close-ups. No live action footage, no text, no captions, no logos, no watermark.

OUTPUT SETTINGS
15 seconds total, 1080p, 30fps, real time throughout with no speed ramps, three shots separated by two hard cuts at 0:04.0 and 0:08.5.

POSITIVE LOCKS
@ollie is the only character and is present in every shot. His proportions, cream neck, dark brown plumage, pale cream wing tips, salmon peach beak and legs, olive iris and crown tuft stay identical across all three attempts, progressively dirtier and more leaf strewn but never redesigned. The green palm fronds stay attached to his wings from the first frame to the last. Every launch carries him off the mountain into open air, and every landing happens in the jungle canopy or forest floor far below the track he ran on. The camera keeps moving in all three shots and always keeps the drop readable under him. Eyes stay natural, no eye glow. Contact shadows read clearly in every shot. The ridge, the three launch heights, the canopy, the rock spire, sky and cloud layer stay the same location and same time of day across all three shots. The flat faced rock spire keeps its face angled toward camera for the final impact. The torn leaf stays across his beak for the entire final close-up. Frame stays clean of text, subtitles, UI, logos and watermarks. Camera cuts only at 0:04.0 and 0:08.5.

Time-freeze diner spill, character-sheet locked

[Step 1 - GPT Image 2.5, character sheet (author's reply https://x.com/EvoLinkAi/status/2107667147019194394)]
image1 as the primary character reference. Create a polished production-ready character reference sheet and turnaround board for the same person in image1. Preserve her facial identity, hairstyle, hair color, body proportions, and overall appearance across every panel.

Show full-body front view, 3/4 view, side profile, and back view in a clean horizontal lineup. Add a large hero portrait, neutral and surprised expression studies, close-ups of the eyes, lips, makeup, hair texture, wardrobe details, and a coordinated color palette.

The character wears a warm orange short-sleeve collared dress with white sneakers. Use a warm cream background, soft studio lighting, refined editorial art-direction, subtle vintage 1990s American diner influence, clean grid layout, thin brown borders, organized annotation panels, cinematic character design presentation, highly consistent face and costume details, realistic photography mixed with premium concept-art styling.

Include minimal clean labels such as FRONT, 3/4 VIEW, SIDE VIEW, BACK VIEW, HERO PORTRAIT, EXPRESSIONS, HAIR DETAIL, and COLOR PALETTE. Wide landscape composition, high resolution, professional film pre-production character bible.

[Step 2 - Seedance 2.5, video (author's reply https://x.com/EvoLinkAi/status/2107436382927171836)]
Photorealistic cinematic 1990s American diner, red vinyl booths, neon signs, chrome details, checkerboard floor, soft natural window light mixed with warm practicals, subtle handheld texture, rich lived-in period detail, heavy film grain. Shot with modern realism and precise temporal control. Use the provided reference image as the exact character lock for the girl with long wavy blonde hair, green-hazel eyes, winged eyeliner, deep maroon lipstick, and an orange dress. Maintain perfect facial structure, hair, head shape, skin texture and clothing consistency at all times. Do not change any other character, scene, action, camera movement, timing, lighting, environment, or visual detail.

0-5s: [Medium Wide] The girl @[Image 1](image_1) sits in a red vinyl booth. She stands up, turns, and collides hard into a waitress carrying a full breakfast tray (eggs, bacon, toast, coffee pot). Impact is sudden and physical.
5-12s: [Dynamic Tracking into Super Slow-Motion] Collision detonates. Tray, plate, eggs, bacon strips and coffee pot explode upward. Coffee erupts into long liquid ribbons and perfect suspended droplets. Camera orbits smoothly around the impact. Time locks completely at the peak of the spill. Every face freezes in pure shock. Only the girl remains free to move. She freezes for one beat with a clear "I fucked up" expression, then quickly grabs two bacon strips and a fried egg from the floating debris.
12-22s: [Tracking Shot] Still inside the frozen diner, she walks toward the exit door while taking deliberate bites of bacon then egg. All patrons, waitress and flying food stay perfectly suspended in mid-air.
22-27s: [Medium Shot] Just as she reaches the door and is about to push it open, time snaps back to normal speed. Everything that was floating crashes down at once — plates, eggs, bacon, tray and coffee slam onto the floor with a loud chaotic impact. The waitress and patrons react in sudden real-time shock.
27-30s: [Medium Close-Up] She pauses, turns slightly, raises her eyebrows and gives a small casual "it is what it is" shrug with a quiet half-smile of acceptance, still holding the remaining food. Photorealistic, ultra-detailed fluid and object physics, perfect volume and surface tension on liquids, sharp motion blur only on moving elements, stable character, cinematic lighting, heavy natural film grain, no artifacts, movie-level temporal coherence, high rewatch value.

Baby dragon burns the roast, learns a tiny flame

Cinematic 16:9 fantasy short film, 30 seconds, 30fps, photorealistic, warm golden-hour light, rich film color grading, shallow depth of field, smooth camera, natural motion blur, heartwarming comedy tone.

Characters:
1) A young woman with messy wavy blonde hair, thin round glasses, a linen cream blouse with rolled sleeves and a dark brown pinafore apron dress. Expressive, funny, warm face.
2) Her pet baby dragon: a huge, friendly, horse-sized dragon with white-cream scales covered in pink-magenta patterns, a crest of soft pink spikes, big amber eyes with vertical pupils, a gentle snout. It acts like a naughty, affectionate puppy.

Setting: a rustic medieval cottage courtyard and kitchen: a stone oven, a wooden table, hanging lavender and herbs, stacked firewood, a low stone wall, green hills in the background. Later, the warm inside of the cottage with a wooden table, a candle, pots on shelves, and sunbeams through a small window.

0-3s: Wide shot, courtyard. The woman turns a big pink raw pork roast on a metal spit over the table. The dragon's head peeks over the stone wall on the right, watching curiously. She talks to the dragon, laughing, asking it for just "a little fire".
3-5s: Extreme close-up of the dragon's scaly snout and nostrils, then a close-up of her smiling face, laughing and confident.
5-7.5s: The dragon opens its mouth and unleashes a huge blast of fire at the roast, flames and sparks everywhere. She throws her hands up in panic.
7.5-10s: Extreme close-ups of her shocked face, wide eyes, hair blown by the heat, hands raised, shouting "No, no, no!". Cut to the roast, completely burnt black and surrounded by flames.
10-12s: She grabs a wooden bucket and throws water on the fire, a giant cloud of steam rises. The dragon laughs with its mouth open, tongue out.
12-15s: Steam clears. She stands with her arms out, scolding the dragon like a child. The dragon lowers its head with guilty puppy eyes, and she gently touches its snout, forgiving it.
15-18s: Cut to the cozy kitchen. She wipes her forehead, then cuts a small piece of fresh meat and puts it on a tiny skewer, calmly explaining to the dragon how little fire is needed. The dragon leans in beside her.
18-21s: Close-up: she holds the skewered meat cube. The dragon carefully breathes a tiny, gentle flame, and the meat sizzles with small sparks. She smiles proudly.
21-23s: Extreme close-up: the perfectly seared meat cube on the skewer, in front of the dragon's amber eye, a reflection of tiny fire in its vertical pupil.
23-26s: She sits at the table with a plate of cooked meat, cuts a piece, tastes it with a delighted face and gives a thumbs-up. She feeds a piece to the dragon, who opens its mouth happily.
26-30s: Wide, warm shot of the cottage interior, golden sunbeam through the window, a candle on the table. She eats while the dragon rests its big head on the table next to her, its eyes slowly closing in contentment, its body curled behind. Peaceful, heartwarming final frame, hold.

Style: Pixar-quality emotional storytelling with photorealistic detail, warm golden and orange tones, soft volumetric light, steam and sparks, cinematic grading.

Audio: gentle fantasy folk music (soft strings and flute), crackling fire and a whoosh on the fire blasts, steam hiss, a deep playful dragon rumble and purr, her laughter and short dialogue, light kitchen sounds.

Frozen crowd of famous-painting figures on museum steps

== SHOT 1 — 14 SECONDS ===
Live-action film footage from a period drama set. ARRI Alexa, 50mm spherical lens,
natural overcast daylight, cool desaturated grade, 35mm film grain. Real
photography of real human beings — visible skin pores and texture, stubble, fine
flyaway hairs, real catchlights in the eyes, subsurface light through ears and
fingers, natural blemishes and asymmetry, real fabric weave and wrinkles. A
photograph of real actors on a real location. Not a painting, not a render, not an
illustration.
The character in the reference image is the hero — keep their face, hair and
outfit exactly as shown, consistent in every frame.
LOCATION: A grand neoclassical museum façade — tall columns, triangular pediment,
arched doorway — above a wide grey stone staircase.
FIXED CAST AND BLOCKING (identical positions in every frame, never rearranged).
Hundreds of costumed background actors — ordinary real people, working actors and
extras, dressed by a wardrobe department in screen-accurate historical costume,
each styled to resemble a figure from a famous painting. They stand shoulder to
shoulder in tight rows filling the staircase edge to edge, only body-width gaps
between them, no aisle, no cleared path:
FRONT ROW, left to right: an elderly actress in a full black mourning dress and
white lace cap · a ballet dancer in a white tulle tutu and black ribbon choker · an
actor in a black suit and red tie with a bowler hat and a real green apple hanging
on a fine wire in front of his face as a practical prop · a gaunt actor with a real
red beard, straw hat and worn blue workman's jacket · a stunt performer in real
gilded plate armor.
SECOND ROW, left to right: a heavyset bearded actor in a gold-embroidered doublet,
fur collar and flat jewelled cap · an actress in a dark green-brown Renaissance
gown and sheer veil, hands folded, faint closed-lipped smile · an actress in a
yellow bodice, blue apron and white linen cap.
THIRD ROW, left to right: an actress in a red embroidered Mexican dress with fresh
flowers braided into her hair and strong dark brows · [THE HERO'S MARK — dead
center] · an actor in a bicorne hat and blue and white military uniform.
BEHIND THEM, hundreds deep: women in crimson turbans and white linen bonnets, men
in black coats with starched white ruffs, a young actress in a blue and gold
headwrap with a single pearl earring, rows of men in Edwardian tweed.
Every face is a real human face with its own bone structure, weight and age.
Nobody's skin is smooth or flat.
TIMING:
0:00–0:03 — CLOSE-UP on the hero's face filling the frame, facing camera front-on,
standing at the bottom of the steps. Background soft and unreadable. They take a
slow drag from a cigarette, lower it, and exhale a plume of smoke across the lens.
0:03–0:06 — The camera smoothly PULLS BACK and widens as the smoke clears,
revealing the hundreds of costumed figures packed on the staircase behind and
around the hero — completely frozen, statue-still, unblinking, staring into camera.
Settles into a medium shot, hero waist-up, crowd sharp and clearly visible behind.
0:06–0:07 — The hero turns around, putting their back to camera.
0:07–0:12 — They walk up the steps at a NORMAL, BRISK WALKING PACE — natural
real-time speed. They squeeze between the frozen bodies, turning their torso to
slip through, brushing shoulders. Their route: up between the ballerina in the
white tutu and the man with the green apple in front of his face, past the
gold-doubleted king on their left and the woman in the green-brown Renaissance gown
on their right, brushing the gilded armor as they pass. Nobody steps aside for
them. Nobody moves.
0:12–0:14 — They arrive at their mark, dead center of the third row, with the
woman in the red embroidered Mexican dress on their left and the man in the bicorne
hat on their right. They stop and turn back around to face camera. Hold. The hero
motionless, surrounded on all four sides, the crowd still completely frozen. END.
Sound: no music, no soundtrack. Only wind across stone and shoes on the steps.
DO NOT include any of the following. Nobody may look painted, illustrated, drawn,
brushstroked, flat, 2D, smoothed, airbrushed, waxy, plastic, doll-like, CGI,
rendered, cartoon or anime. No canvas texture, no brushstrokes, no oil-paint sheen,
no picture frames. The crowd must not move at all in this shot and must not change
position, costume or order. The crowd must not part, form an aisle, or leave empty
space around the hero. The hero must not be in slow motion or drift slowly. No
cuts, no camera shake, no morphing faces, no text, no watermark.
=== SHOT 2 — 15 SECONDS — CUT ON THE FIRST MOVE ===
Live-action film footage from a period drama set. ARRI Alexa, 50mm spherical lens,
natural overcast daylight, cool desaturated grade, 35mm film grain. Real
photography of real human beings — visible skin pores and texture, stubble, fine
flyaway hairs, real catchlights in the eyes, natural blemishes and asymmetry, real
fabric weave and wrinkles. Not a painting, not a render, not an illustration.
The character in the reference image is the hero — identical face, hair and outfit
to the previous shot.
LOCATION: The same grand neoclassical museum façade — tall columns, triangular
pediment, arched doorway — above the same wide grey stone staircase.
FIXED CAST AND BLOCKING — exactly the same people in exactly the same positions as
before, nobody added, removed or rearranged. Hundreds of costumed background actors
— ordinary real people in screen-accurate historical costume, each styled to
resemble a figure from a famous painting — standing shoulder to shoulder in tight
rows filling the staircase edge to edge, only body-width gaps, no aisle:
FRONT ROW, left to right: an elderly actress in a full black mourning dress and
white lace cap · a ballet dancer in a white tulle tutu and black ribbon choker · an
actor in a black suit and red tie with a bowler hat and a real green apple hanging
on a fine wire in front of his face · a gaunt actor with a real red beard, straw hat
and worn blue workman's jacket · a stunt performer in real gilded plate armor.
SECOND ROW, left to right: a heavyset bearded actor in a gold-embroidered doublet,
fur collar and flat jewelled cap · an actress in a dark green-brown Renaissance gown
and sheer veil · an actress in a yellow bodice, blue apron and white linen cap.
THIRD ROW, left to right: an actress in a red embroidered Mexican dress with fresh
flowers in her hair · THE HERO, dead center · an actor in a bicorne hat and blue and
white military uniform.
BEHIND THEM, hundreds deep: women in crimson turbans and white linen bonnets, men
in black coats with starched white ruffs, a young actress in a blue and gold
headwrap with a single pearl earring, rows of men in Edwardian tweed.
TIMING:
0:00 — Open on a WIDE SHOT of the full façade and the entire staircase, already in
motion. On the very first frame the whole crowd is mid-convulsion. No build-up, no
lead-in, nobody standing still.
0:00–0:10 — The crowd performs a wild, continuous, full-body rolling motion. The
cycle: heads thrown ALL THE WAY BACK, chins pointed up at the sky, spines arched
backward — then the entire torso whips forward and folds almost double at the hips,
heads hanging down near their knees, hair and hats flying — then the body unrolls
back up and the head snaps back again, chin to the sky. Over and over, without
pause, never resting at neutral. Arms hang loose and swing and flail with the
momentum. Loose, rubbery, boneless, possessed. Different sections of the staircase
are out of phase, so one block is folded double while the block beside it is arched
back with heads up, and the motion ripples across the crowd. Meanwhile the camera
pulls back and rises slightly, widening to take in the whole façade.
0:10–0:15 — The camera settles and holds locked off. The crowd keeps convulsing at
full intensity to the last frame. The hero has not moved once — standing relaxed and
unbothered at the dead center, facing camera, hands down, not dancing, not bending,
not reacting.
Sound: no music, no soundtrack. Only fabric snapping, armor rattling, breath and
shoes scuffing stone.
DO NOT include any of the following. Nobody may look painted, illustrated, drawn,
flat, 2D, smoothed, waxy, plastic, CGI, rendered, cartoon or anime. Nobody may
change position, costume or order from the previous shot. The crowd must never bow,
curtsy, nod, greet, dip gently, bend shallowly, keep their heads level, move
stiffly, or move in robotic unison, and must never stand still. The hero must not
dance, bend or move, must not stand off to one side or at the edge of frame, and
must not be in close-up. The crowd must not part, form an aisle, or leave empty
space around the hero. No slow motion, no cuts within the shot, no camera shake, no
morphing faces, no text, no watermark.

Gothic ink anime sword fight from a Midjourney sref

[Midjourney v8.2 still (style reference)]
--sref 2432028955 --profile d3zvtyp

[Seedance 2.5 image-to-video]
Premium dark fantasy 2D anime cinematography fused with gothic ink illustration, high-sakuga sword choreography and experimental action filmmaking. Preserve the reference aesthetic of intricate black linework, rough hand-drawn textures, charcoal-gray shading, muted violet skin tones, elaborate gothic silhouettes and aggressive neon pink, acid green and prismatic highlights. Ultra-fast, brutally aggressive swordsmanship with relentless blade exchanges, acrobatic counters, spinning slashes, sudden reversals, impossible footwork and explosive multi-hit combinations, maintaining constant offensive pressure from both fighters. Every clash produces razor-sharp neon streaks, violent chromatic sparks, fractured ink splashes and brief flashes of saturated light that illuminate the surrounding darkness. Radical cinematic perspectives combine ground-skimming tracking shots, aggressive Dutch angles, extreme foreshortening, rapid orbital rotations, blade-following whip pans, sudden perspective inversions and seamless camera movements threading through crossing swords. Insert lightning-fast detail cuts of narrowed eyes, tightening fingers, twisting wrists, grazing blades and sparks scattering across faces, immediately returning to ferocious wide-angle combat. Dramatic foreground occlusion, near-lens blade strikes, shifting focal distances and rapid depth transitions create visceral immersion. Dark atmospheric voids, drifting ash, torn feathers and vibrating debris react to the force of each exchange. Precise silhouette control, strong visual rhythm and high-contrast graphic compositions preserve clarity even during overwhelming bursts of speed. No slow motion, no idle posing, no pauses between attacks. Every movement flows directly into another attack, parry or counterattack. Hand-drawn anime expressiveness and premium 2.5D cinematic depth preserve the original illustration style without turning the characters into realistic CGI.
Two gothic elven swordswomen engage in a relentless duel inside a ruined cathedral swallowed by darkness. The black-haired warrior with her feathered crown attacks with a prismatic blade, immediately colliding with the silver-haired warrior's neon sword. They explode into a succession of impossibly fast sword exchanges, spinning counters, airborne reversals and close-range blade locks that instantly break into new attacks. The camera spirals tightly around their crossing weapons, dives beneath sweeping strikes, whips past their faces and follows a blade within inches of the lens. Sudden close-ups capture furious expressions, razor-thin near misses and sparks streaking across their skin. Both warriors continuously reverse the advantage through increasingly elaborate combinations, ricocheting off broken pillars, intercepting attacks from impossible angles and chaining acrobatic movements into ferocious counterstrikes. Every impact briefly floods the monochromatic environment with violent pink, green and rainbow light. The battle escalates without interruption, ending mid-combination as both warriors launch simultaneous devastating attacks directly toward each other.

Nano Banana 2.1 (4)

Textured geometric fashion illustration, full body

Create a stylish, full-body editorial fashion illustration of a young woman in a modern casual outfit, using a minimalist, textured geometric art style. She wears an oversized forest-green hoodie, wide-leg blue denim jeans with folded cuffs, and chunky brown sneakers. Her dark hair is tied in a neat low bun. She carries a bouquet of vibrant pink roses wrapped in white paper, a takeaway coffee cup, and a black shoulder bag. Show her walking naturally in a relaxed pose, viewed from a three-quarter rear angle, with soft, elegant body proportions and simplified facial details.

Use a warm ivory background with plenty of negative space, soft ground shadows, subtle paper grain, visible brush textures, and angular polygonal shading across the clothing and shoes. The color palette features deep forest green, denim blue, warm brown, cream, and vivid pink. Soft diffused lighting, clean composition, contemporary lifestyle magazine aesthetic, hand-painted digital illustration, refined edges, artistic fabric textures, vertical 3:4 composition, high detail. No photorealism, no 3D rendering, no extra objects, no text or watermark.

Photo + black-ink illustration split editorial poster

Using the uploaded reference image, create a single high-resolution vertical 2:3 editorial art poster.

Divide the canvas into two perfectly equal horizontal sections, each occupying exactly 50% of the total height. Separate them with a clean, perfectly straight horizontal dividing line.

1. TOP HALF — ORIGINAL REFERENCE PHOTOGRAPH

Use the uploaded reference photograph directly in the upper 50% of the poster.

Preserve its original photographic quality, sharpness, natural lighting, and color relationships.

Allow only subtle, intelligent cropping around the outer edges when necessary, without compromising the main subject.

If required, naturally extend a small portion of the surrounding environment to fit the composition.

Do not redraw, illustrate, stretch, blur, or modify the subject's appearance, structure, or spatial relationships.

Do not introduce black borders, white margins, or empty edges.

2. BOTTOM HALF — MINIMALIST EDITORIAL ILLUSTRATION

Use a warm, near-white ivory paper background with an extremely subtle paper texture.

Analyze the main subject and visual story of the uploaded reference image.

Extract only the primary subject, preserving one or two essential identifying elements.

Remove or heavily simplify all secondary details, including background architecture, crowds, vehicles, fences, signs, trees, and other unnecessary elements.

Transform the primary subject into a sophisticated black-ink editorial illustration.

ILLUSTRATION STYLE:

Use natural, clean pen strokes with varied line weight.

Preserve an authentic hand-drawn appearance with expressive but controlled linework.

Avoid overly precise engraving aesthetics, excessive internal details, and dense cross-hatching.

Place a large, nearly pure-black rectangular ink block behind the illustrated subject.

The rectangular block should have subtly imperfect, handmade print-style edges.

Keep the black background clean, with only fine paper grain and minimal noise.

Avoid obvious hatching, scratches, architectural lines, or complex textures.

ILLUSTRATION SCALE AND COMPOSITION:

The illustrated subject must occupy approximately 72–77% of the canvas width and 58–64% of the bottom section's height.

Maintain a clear visual focal point with generous negative space surrounding the illustration.

3. COLOR RULES

The bottom illustration must follow a strictly black-and-white color palette by default.

Preserve only one localized accent color if the original reference image contains a clearly distinctive, concentrated color element with meaningful narrative importance.

Ordinary blue skies, green vegetation, stone surfaces, skin tones, and environmental colors must not be treated as accent colors.

Any preserved accent color must remain strictly localized to the relevant subject or area.

Never allow accent colors to spread across the entire illustration.

4. TYPOGRAPHY AND EDITORIAL LAYOUT

Automatically generate concise, meaningful English typography based on the content of the uploaded reference photograph.

The typography must resemble handwritten editorial annotations from an independent art publication.

TOP-LEFT CORNER:

Generate a short handwritten title consisting of 2–5 words, arranged across two lines.

Directly beneath the title, include three short English action verbs relevant to the photograph.

TOP-RIGHT CORNER:

Display only two lines of geographic coordinates:

Latitude
Longitude

Do not include any additional text in this area.

BOTTOM-LEFT CORNER:

Display the location name followed by the word "DIARY."

Directly beneath it, add one short descriptive English sentence inspired by the photograph.

TYPOGRAPHY STYLE:

Use a rough, upright, slightly irregular handwritten typeface.

Maintain subtle imperfections and a natural editorial annotation aesthetic.

Avoid elegant cursive, calligraphy, overly decorative lettering, and excessively polished typography.

5. FINAL ART DIRECTION

The finished poster must combine:

Authentic, high-quality photography.
Minimalist black-and-white editorial illustration.
Generous negative space.
Independent art publication aesthetics.
Quiet, restrained, sophisticated visual storytelling.
Natural handwritten typography.
A balanced relationship between documentary realism and expressive hand-drawn artwork.

The overall result should resemble a carefully designed, collectible editorial art print.

6. FINAL OUTPUT REQUIREMENTS

ASPECT RATIO:
2:3 vertical.

RESOLUTION:
High-resolution, sharp, print-quality output.

COMPOSITION:

Exactly 50% original reference photography on top.

Exactly 50% editorial illustration on the bottom.

One perfectly straight horizontal dividing line.

One complete, standalone poster.

RESTRICTIONS:

No collages.
No multi-poster layouts.
No combined images.
No unnecessary decorative elements.
No modifications to the original subject's identity or appearance.
No excessive illustration details.
No changes to the original photograph's natural color relationships.

FINAL GOAL:

Create a sophisticated, visually balanced editorial art poster that faithfully preserves the uploaded reference photograph while transforming its primary subject into an expressive, minimalist black-ink illustration.

Cobalt circle studio fashion editorial

Create an ultra-realistic, high-end fashion editorial photoshoot featuring a beautiful young adult woman with short, tousled platinum-blonde bob hair, delicate facial features, fair skin, natural rosy lips, light-colored eyes, and a calm, confident expression.

Scene and Environment: A futuristic minimalist architectural studio featuring a massive sculptural circular opening with thick, smooth, curved white walls. The interior of the circular structure is a rich cobalt-blue or vibrant cyan-blue, creating a striking contrast against the white architectural frame. Clean geometric shapes, curved surfaces, and a contemporary art-gallery aesthetic.

Outfit and Styling: The model wears a sophisticated, oversized black knit sweater or elegant black long-sleeved top paired with tailored black or ivory trousers. Minimal makeup, natural skin texture, understated accessories, and effortless luxury styling.

Pose and Composition: The model poses naturally within and around the giant circular opening. Create a series of artistic editorial poses: a close-up portrait leaning forward through the circular frame, a full-body seated pose inside the sculptural ring with one knee raised, and a standing or seated three-quarter portrait with one hand resting on the curved white surface. Relaxed hands, graceful posture, elongated silhouette, and direct eye contact with the camera.

Lighting: Soft natural daylight entering from one side, creating beautiful directional highlights on the platinum hair, subtle shadows across the face, realistic skin tones, and soft gradients on the white architectural surfaces. Cinematic interplay between cool blue ambient light and warm sunlight.

Color Palette: Vivid cobalt blue, turquoise, icy cyan, pure white, deep black, and platinum blonde. Clean, cool-toned color grading with balanced highlights and rich shadows.

Photography Style: Luxury fashion magazine editorial, contemporary avant-garde photography, minimalist Scandinavian design, architectural fashion campaign, medium-format camera aesthetics, 85mm portrait lens, subtle depth of field, precise focus on the eyes, realistic hair strands, detailed knit fabric, natural skin texture, sophisticated framing, and premium commercial photography quality.

Image Quality: Photorealistic, ultra-detailed, 8K appearance, professional art direction, anatomically accurate hands, realistic proportions, elegant negative space, clean architectural geometry, natural facial expression, crisp details, refined composition, no artificial-looking skin, no excessive retouching.

Vertical portrait composition, 2:3 aspect ratio.

Second World: photo continues into a paper-cut world

「第二世界」完整 JSON 提示词 v1.2
上传照片后，将下面整段交给图像模型。每张照片独立生成。

{
  "name": "第二世界",
  "version": "1.2",
  "task": "以每张上传照片为独立来源，各生成一张完整海报。不得把多张照片合并到同一画布。",
  "output": {
    "aspect_ratio": "3:4",
    "orientation": "vertical",
    "one_poster_per_photo": true,
    "layout": "画布在高度精确50%处水平分割，上下各占50%，宽度相同",
    "no_mockup": true
  },
  "core_principle": "第二世界原本就藏在照片里。让同一个场景穿过中线，继续向下延伸，只在下半部分改变物理规则。跨界结构的连续性是最高视觉优先级。",
  "upper_half": {
    "source": "当前这张上传照片",
    "preserve": [
      "主体身份与外观",
      "人物姿势、表情和动作",
      "空间关系、比例与透视",
      "原始色彩氛围",
      "自然光线与阴影",
      "环境和重要细节"
    ],
    "allowed_change": "仅为适应上半画幅进行必要的比例裁剪，避免裁掉关键主体或连接结构",
    "forbidden": [
      "重新设计照片",
      "重新绘制或替换主体",
      "重新安排人物和物件",
      "改变照片色彩",
      "虚构光线或背景"
    ],
    "fidelity_rule": "优先直接复用原照片；不得将生成式重绘宣称为原片无损保留。若工具无法满足原片保留与精确布局，须说明限制，不得自行降低要求后宣称完全达标。"
  },
  "continuity": {
    "choose_one_source_structure": "从原照片中识别一条能够自然延伸到中线的结构，例如道路、海岸线、水面、倒影、树枝、光束、建筑线条、衣物走势或身体动作",
    "requirements": [
      "结构必须在中线的准确相遇位置继续进入下半部",
      "保持方向、位置、尺度和遮挡关系的可信连续性",
      "先建立可见的连接，再发展下半部的想象",
      "两半必须读作同一个场景，不能像两张独立图片上下摆放"
    ],
    "transition_edge": {
      "optional": "允许不规则撕纸边缘",
      "rule": "撕裂必须顺着原图结构发生，不得仅作为装饰，不得每张使用同一种撕裂形状",
      "interaction": "源元素可以触碰、贴着、穿过或越过撕裂边缘"
    }
  },
  "lower_half": {
    "background": "温暖象牙色纸张，细微可见纤维，丰富负空间",
    "media": [
      "克制的照片碎片",
      "简洁剪纸形状",
      "少量黑色手绘线条"
    ],
    "construction": "从上半部源结构与中线相遇的位置向下延续，再逐渐将其重新解释为第二世界",
    "prohibited": "不得成为孤立入口、窗口、舞台、平台或漂浮小插图，除非该形态明显由原照片连续生长出来",
    "density": "明亮、通透、克制，避免密集绘制和拼贴堆砌"
  },
  "second_world_logic": {
    "question": "如果照片里的这个结构变得可以触摸、使用、进入、折叠、拉伸或改变，它会自然变成什么？",
    "interaction_count": 1,
    "instruction": "每张只建立一种具体、清楚、有趣的互动，优先从原图的形状和动作发现可能性，不从固定道具或模板倒推。",
    "examples_for_logic_only": [
      "水面保持与湖相连，同时被卷起或拉伸",
      "道路延续到纸面，变成可以继续画出的路径",
      "一束原有的光线延下来，变成可以轻轻拿住的东西"
    ],
    "variation": "多张作品分别判断，不机械重复动作、连接结构、构图或玩笑。示例不是必须出现的内容。"
  },
  "tiny_figures": {
    "count": "0–3",
    "appearance": "极简、微小的黑色线条人物",
    "add_only_if": "人物能与延续结构发生必要而明确的物理互动",
    "omit_if": "源照片已包含强烈的人类动作",
    "forbidden": "把小人当作无关装饰或为凑热闹添加人群"
  },
  "copy": {
    "language": "中文为主，搭配一小段英文点睛",
    "main_text": "一句短小、俏皮、可爱、诙谐或带有轻微意外洞察的中文旁注",
    "english": "更短、更轻的英文补充；可以与中文互相补充，不必逐字翻译",
    "intent": "让人看完画面，再看到小字时会心一笑，或突然多理解一层",
    "requirements": [
      "根据这张照片与第二世界的具体互动重新创作",
      "不要只是给动作命名，也不要机械看图说话",
      "可以有拟人、轻微反差、一本正经的小玩笑或温柔的自嘲",
      "可以有点醒人的意味，但不讲人生大道理",
      "不强行套用自爱、励志或情绪安慰模板",
      "幽默针对情境，不贬损人物外貌、身份或处境"
    ],
    "avoid": [
      "空泛鸡汤",
      "宏大格言",
      "说教",
      "广告标语",
      "无关网络热梗",
      "固定的“同样……，不同……”句式",
      "每张复用相同文案"
    ],
    "examples_for_tone_only": [
      {
        "context": "把晚霞带走",
        "chinese": "这个不算超重吧",
        "english": "just a little sunset"
      },
      {
        "context": "将湖面卷成纸带",
        "chinese": "不用袋子，我卷着走",
        "english": "no bag, thanks"
      },
      {
        "context": "影子卷成被角",
        "chinese": "影子先躺了，我再走会儿",
        "english": "five more minutes"
      }
    ],
    "example_rule": "只借鉴幽默机制，不把例句固定写进新作品。"
  },
  "handwriting": {
    "scale": "小号旁注，英文更小，不做抢占画面的巨大标题",
    "placement": "放在自然留白处，不遮挡主体、跨界连接点或关键互动",
    "chinese": "真正自然、略潦草的手写字，不是整齐的电脑手写字体",
    "english": "同样像随手写下，轻松、不规整",
    "details": [
      "字大小略有差异",
      "基线轻微起伏",
      "倾斜角度不完全一致",
      "间距自然变化",
      "笔画有压力和停顿",
      "少量笔迹可重叠或不完整"
    ],
    "corrections": {
      "allowed": true,
      "frequency": "偶尔出现即可，不要求每张都有",
      "method": "可以将一个写错或想改的词轻轻划掉，在旁边或上方补写正确版本",
      "limit": "通常最多1–2处，不把整页做成反复涂改的草稿",
      "legibility": "最终意思必须清楚，不能用乱码或无意义错字模拟手写"
    }
  },
  "overall_style": {
    "mood": [
      "独立杂志感",
      "明亮通透",
      "温暖纸感",
      "克制",
      "轻巧幽默",
      "细微手工痕迹"
    ],
    "hierarchy": [
      "原照片与跨界结构",
      "下半部明确的互动",
      "小号双语手写旁注"
    ],
    "finish": "像照片中的隐藏世界被轻轻揭开，而非在照片下方额外贴上一幅插画"
  },
  "negative_prompt": [
    "分离的下半部插图",
    "断裂的跨界连续性",
    "通用撕纸模板",
    "孤立入口或窗口",
    "无关的小插图",
    "密集插画",
    "随机装饰",
    "装饰性人群",
    "改变原照片颜色",
    "上层照片重新设计",
    "光泽3D",
    "剪贴簿式杂乱",
    "巨大标题",
    "整齐印刷式手写",
    "胡言乱语文本",
    "励志鸡汤",
    "水印",
    "徽标",
    "UI",
    "多张照片合并",
    "九宫格"
  ],
  "final_check": [
    "是否每张原图对应一张独立的3:4海报？",
    "上下区域是否严格各占50%？",
    "上半原照片是否忠实保留，仅做必要裁剪？",
    "是否有一条真实源结构在中线准确接续？",
    "下半部是否与原图属于同一场景，而非孤立插画？",
    "是否只有一种清楚、自然且不重复套用的互动？",
    "中文是否有趣、有余味，而非简单动作标签或说教？",
    "中英文是否小号、自然潦草且可读？",
    "是否保留了足够的负空间？"
  ]
}

注：以上是目标要求，生成结果仍需检查原片保留、接缝和文字，不能把生成式重绘当成原片无损保留。

由 OpenAI 的 dot 通过云端浏览器操作 X 网页发布，经 @HX831126 授权。

Gemini Omni (2)

POV luxury furniture commercial from a storyboard

Create a 10-second ultra-photorealistic 16:9 luxury furniture commercial, one continuous first-person male POV, 26–28mm lens, black studio, matte-black leather gloves, realistic physics only. 0–0.8s: white-cream cloud sofa floats centered; hand swipes left. 0.8–1.6s: sofa rapidly orbits left as a warm ivory modular sofa arrives from right and stops center. 1.6–2.4s: another swipe replaces it with a curved latte boucle sofa. 2.4–3.2s: swipe brings in a warm caramel leather sofa. 3.2–4.2s: final taupe/mocha hero sofa arrives slowly, rotates frontal and stops with heavy inertia. 4.2–5.0s: open-palm STOP gesture freezes the carousel; other sofas disappear into distance. 5.0–5.8s: hand presses and slides across hero sofa armrest, realistic fabric compression. 5.8–6.6s: sofa lands; walnut floor slides beneath it and taupe rug unrolls. 6.6–8.5s: KEEP THE EXACT SAME HERO SOFA stationary while the luxury room physically assembles around it—walnut and travertine walls slide in, coffee table and latte chairs enter, shelving locks into place, pendant descends, curtains and decor move into position. 8.5–10s: panoramic window opens, warm daylight floods the completed quiet-luxury interior; slow dolly backward reveals the full room. Hero sofa remains 100% identical throughout. Real mass, inertia, friction and mechanical stops; no magic, morphing, teleportation, particles, smoke, UI, text, logos, people, cuts or camera-angle changes.

POV car-assembly ASMR reel for an auto service

Create a 10-second ultra-photorealistic 16:9 automotive ASMR video, first-person POV in a clean white studio. Giant male hands in matte-black gloves rapidly assemble a tiny realistic 1:12 white BMW: insert engine → THUNK, connect electronics → SNAP, install red brake calipers and wheels → CLICK, attach steering wheel and headlights → SNAP, fit front bumper and close hood → THUNK. Show extreme macro details and obvious miniature scale: the whole car is smaller than a hand. At 5s the finished white BMW is spray-painted with real wet crimson-red automotive paint, clearly changing WHITE → RED, then polished to a deep mirror gloss. At 8.5s show the tiny red BMW in front three-quarter hero angle → HARD MATCH CUT to a full-size real red BMW in exactly the same angle inside a large white studio. Engine starts, body vibrates, LED headlights turn on, then the BMW launches aggressively straight toward the camera → HARD CUT TO BLACK. Fast cuts, realistic physics, metal, carbon, leather, rubber and glass, strong CLICK/SNAP/THUNK ASMR, premium German precision, no magic, morphing, particles or digital transformation.

ChatGPT Images (1)

Send '6-1-2' to get the same AI model in a new scene

[Setup: 4 sheets in a ChatGPT project]
やってることはシンプルで、最初に4枚の「シート」を作っておくだけ

①モデル（顔・体型・表情）
②ポーズ（番号付き）
③背景（番号付き）
④服装（番号付き）

これをChatGPTのプロジェクトに入れておく

[Usage]
あとは新しいチャットで

「6-1-2」

って送るだけ

→ ポーズ6番・背景1番・服装2番で、同じ顔の子が出てくる

毎回長いプロンプト書かなくていいし、顔もブレにくい

[Project instructions]
【添付シート】
①モデル（MODEL-A）②ポーズ（1〜20）③背景（1〜12）④服装（1〜12）

【入力形式】
「ポーズ-背景-服装」の番号3つ（例：6-1-2）
「ランダム」なら各シートから番号を選び、最初に「ポーズ6／背景1／服装2」と1行書いてから生成する

【生成ルール】
・人物は必ず①のMODEL-A。顔・髪型・泣きぼくろの位置・体型・頭身を①から一切変えない
・ポーズは②の指定番号の姿勢を、自然な人物の動きとして再現する（人形の関節の形はそのまま写さない）
・背景は③の指定番号の場所。人物のサイズと遠近感を背景に合わせる
・服装は④の指定番号のコーデをMODEL-Aの体型に合わせて着せる
・仕上がりはスマホで撮った自然なスナップ写真。縦長9:16
・説明文は不要。画像のみ返す

ポイントは「顔・髪型・ほくろ・体型を①から一切変えない」って明記すること
これ書かないと普通に別人になる

Midjourney (1)

EXCESS: two-code sref for maximalist glam

--sref 1397720772 1539610103

Midjourney V8.2 (1)

3D game-character sref + 4 character prompts

[Midjourney v8.2 parameters]
--sref 3016865554

Prompt #1
3d character, full body, a grizzled middle-aged halfling thief, small stature with curly reddish-brown hair and messy sideburns, a cunning narrow face with a scar across his lip, dark hooded traveling cloak with many hidden pockets, worn brown leather vest, rolled-up trousers, large bare hairy feet, lockpicks and a small curved dagger on his belt, a stolen jeweled goblet in one hand, sly suspicious glance over his shoulder, white background

Prompt #2
3d character, full body, a young female desert nomad cartographer, wrapped in layered sand-colored linen robes and a long indigo scarf covering her hair, brass goggles resting on her forehead, leather satchels and rolled-up maps strapped all over her body, a compass hanging from her belt, sun-tanned skin with freckles, worn leather sandals with ankle wraps, walking forward while studying an unfolded map, determined expression, white background

Prompt #3
3d character, full body, a very tall and lanky lighthouse keeper with a long gray beard braided with tiny shells, oversized yellow rubber raincoat dripping with water, heavy black wellington boots, a knitted fisherman's beanie, holding an old brass oil lantern in one hand, a grumpy seagull perched on his shoulder, hunched posture, weathered face with squinting eyes, white background

Prompt #4
3d character, full body, a feisty 70-year-old grandmother roller derby skater, short curly silver hair under a scratched purple helmet, thick round glasses, knitted cardigan worn over a sporty jersey with the number 82, knee pads covered in stickers, vintage quad roller skates with pink wheels, a mischievous grin, leaning forward mid-skate with one fist raised, white background

Nano Banana Pro (1)

Amateur front-cam extreme close-up selfie, then variations

Amateur extreme close-up front-camera selfie still frame, cropped at the forehead and just below the hands, vertical-ish 16:9 crop. A young english stunning woman around 21yo with fair, peachy skin and a soft warm flush on her cheeks, no visible freckles, natural minimal makeup. Large, wide-open, bright light-blue eyes with visible pale catchlights, deep black thin eyebrows raised slightly, gaze directed forward and a bit off-camera, giving a surprised, nervous, coy expression with detailed tongue out. Her big natural lips are slightly parted, glossy natural pink-nude, hint of upper teeth visible. Head tilted very slightly, chin resting toward her raised hand: her right hand curled up against her chin and lower lip, knuckles bent, fingers loosely closed. She wears two thin gold rings on her fingers, one plain delicate band and one with tiny inset stones. Long, voluminous, slightly messy wavy deep black hair falls over both sides of her face and shoulders, with loose strands catching the light. A soft diagonal shadow (from hair or a hand) falls across her nose bridge and cheeks. Background: blurred pale grey-white wall on the right, with a hint of muted lavender-grey fabric (clothing or bedding) in the bottom right corner. Soft natural window daylight, warm color grade, slightly lifted highlights, shallow depth of field, phone-camera softness, slight video-compression grain, candid, intimate, authentic social-media aesthetic. Minimal blac and white tatoos over her hands and body

Claude (1)

Animated SVG explainer: how Seedream 5.0 splits layers

https://github.com/wheresryan22/anatomy install skill

/anatomy Explain how Seedream 5.0 is able to split an image into layers and what cool things you can produce with it. Do that in a single HTML file. 

Change the camera/zoom in for scene 1 to 5 to explain more clearly.

Kling Omni (1)

Stressed man in hotel room from one character ref

A stressed man [your image character/📷Image1] walks anxiously around a dimly lit hotel room, the air thick with tension. As he gazes out the large window, the city skyline stretches into the distance, shrouded in the twilight's gloom, reflecting his sense of isolation. Inside, the room is sparsely decorated, with disheveled papers scattered across a small desk and a suitcase half-opened on the floor, symbolizing his chaotic mental state. Soft shadows loom in corners, making the space feel tighter as if suffocating him. The light from the window casts a pale glow, contrasting with the darkness within, creating an atmosphere of loneliness and melancholy. His thoughts drift like the clouds outside, each one heavier than the last, filling the room with a palpable sense of being overwhelmed by life and work. This scene encapsulates a dynamic immersion into his internal struggle, as if his stress has materialized around him, wrapping the space in a tight embrace.

Kling 4.0 (1)

30s phone-cam UGC at brunch, sister steals pancake

Raw vertical 9:16 video recorded on a phone's front camera (not on screen), one continuous 30-second take with no cuts. It looks like an ordinary unedited phone clip, not a film: slightly soft image, a little digital noise and compression, deep focus with the café behind her as sharp as her face, auto-exposure brightening and dimming as people pass the sunny window behind her, no colour grading. The phone sits close and a little low, looking slightly up at her; around 9 s a waiter's arm bumps the table, the phone slides and tilts, and she catches it and props it back up. Sadie, a young woman in her early 20s with long honey-blonde waves, light natural makeup, small gold hoop earrings and a cropped cream knit top, sits at a small marble table by a big sunny window in a busy brunch café. Her sister sits across from her, out of frame except for her arm and a hand with a silver ring. Behind Sadie: other tables of people talking, a counter with a glass pastry case, plants, staff moving about, chalkboards too far away to read. Around 9 s a waiter sets down two plates, a lemon ricotta pancakes and a plate of eggs on toast, plus two iced coffees in plain glasses. Around 15 s she turns the pancake plate toward the camera; around 20 s she cuts a bite, eats it and closes her eyes; around 26 s her sister's hand reaches in with a fork and steals a piece of pancake, and Sadie swats at it, laughing. [Sadie: casual young woman's voice, warm and a bit giggly, recorded on the phone's built-in mic from half a metre away]: "Okay, I promised I'd show you what we get. ... We get the same thing every single time. ... Oh, here it is! Thank you! ... Look at this. Lemon ricotta pancakes. ... Mmm. Okay. Yeah. Still the best. ... Hey! Get your own!" [Her sister, off camera: a young woman's voice, dry and amused]: "That's the tax." [Sadie, laughing]: "She does this every time." Audio as captured by the phone mic in a busy café, not studio clean: her voice close but with the room around it, the hiss and grind of an espresso machine, cups and cutlery clinking, overlapping chatter and laughter from other tables, a door bell jingling, a chair scraping, faint music in the background. No subtitles, no captions, no on-screen text.

Muse (1)

15s ultra-close fashion portrait with real flash pops

生成这个视频：生成一段15秒、9:16竖屏、超高清4K、30fps、电影级真实人像质感的时尚人像短片。

画面主体是一位 20-25岁左右的年轻亚洲女性，高级脸，清冷中带一点青涩感，五官精致但不过分完美，脸型流畅，鼻梁自然，嘴唇饱满，眼神有内容，整体气质像车模或时尚活动现场被摄影师近距离怼脸抓拍。
不要网红感，不要夸张整形脸，不要塑料皮，不要过度磨皮，不要假睫毛过重，不要夸张高光，不要二次元感。
人物皮肤必须真实：清晰可见毛孔、轻微皮肤纹理、轻微卡粉、轻微痘印、极浅痘坑、眼下细纹、鼻翼与脸颊自然凹凸、嘴唇唇纹、少量碎发和发丝边缘。 妆容是精致但不完美的高级淡妆：底妆微微服帖但不是无瑕，眼妆干净，睫毛真实自然，唇部有轻微光泽感。
头发为黑色或深棕色顺直长发，带自然碎发和轻微凌乱感，不能太刻意。

场景与氛围

场景为高级时尚活动现场 / 车展活动区 / 高端品牌发布会现场的人像近拍氛围，但背景必须虚化，不突出具体品牌和文字。背景可有柔和的展厅灯光、少量失焦霓虹反射、模糊人群、模糊车身高光反射或金属质感背景，整体氛围高级、克制、时尚。
不要字幕，不要贴纸，不要水印，不要可读文字，不要夸张特效，不要不合理炫光。

光线要求

光线必须符合现实逻辑，像真实摄影师在活动现场近距离拍摄。
主光为柔和但有方向性的面部主光，从人物正前偏侧上方打来，形成自然脸部明暗层次。
辅光非常轻，保留部分阴影，增强五官立体感。
环境中可有柔和冷色边缘光或车展现场反射光，让脸部更立体。
全程允许出现2到3次短暂闪光灯，闪光灯必须像现场摄影闪光一样自然，只在某些瞬间快速亮起，造成面部高光短暂增强、瞳孔高光更明显、皮肤反光更亮，但不能整段频闪，不能不合逻辑。
闪光灯出现时，要像摄影师抓拍：亮一下就过，光比真实，脸部不过曝，皮肤细节仍可见。

镜头与运镜（顶级运镜）

整体采用超近距离人像运镜，镜头始终围绕脸部进行，突出眼睛、鼻梁、嘴唇、脸颊、发丝和皮肤质感。
镜头语言高级、流畅、克制，有时尚广告和摄影后台抓拍的感觉。
不要大幅乱晃，不要眩晕感，不要夸张AI运镜，不要突然拉远。
运镜必须自然，像高端摄影师手持稳定器或轻微手持拍摄：

0-3秒

镜头从超近距离侧前方怼脸开场，焦点落在眼睛和鼻梁区域，人物微微抬眼看向镜头，呼吸感自然，嘴唇轻微放松。
镜头做极轻微慢推近，同时有非常轻的横向滑动，展示脸颊、眼下纹理和发丝。
这一段要让观众第一眼就感受到脸的高级和真实。

3-6秒

镜头沿着脸部从眼睛滑到鼻尖、再滑到嘴唇附近，做极稳的微弧形绕拍，近距离观察五官。
人物眼神轻微移动，不要夸张表演，只需轻轻看向镜头边缘，再缓慢回看镜头。
此时可出现第一次闪光灯，短暂亮起，强化皮肤质感和唇部高光。

6-9秒

镜头从正面偏左缓慢转到正面偏右，仍然保持怼脸距离，重点表现脸颊皮肤纹理、鼻翼细节、下眼睑细纹、睫毛和瞳孔反光。
人物轻轻拨动一点头发，或者有一缕碎发自然落在脸侧。
镜头带一点真实摄影感的轻微呼吸式浮动，不能死板。

9-12秒

镜头轻微下移，突出嘴唇、下巴线条和脸颊，再慢慢回到眼睛。
人物表情保持清冷、克制、带一点高级疏离感，像车模被摄影师近距离抓拍，偶尔轻轻眨眼，嘴角有极轻微放松。
这里出现第二次闪光灯，自然、瞬时、像现场抓拍，闪完后恢复正常光线。
闪光灯后眼睛里的高光更加明显，画面更有时尚大片感。

12-15秒

镜头做一次最漂亮的超近距离轻弧线收镜，围绕脸部微微绕动并轻推近，最后停在眼睛+鼻梁+半边嘴唇的绝佳构图上。
人物在最后一秒轻轻抬眼直视镜头，眼神干净、冷静、有吸引力。
结尾可有第三次很轻的闪光灯或环境高光掠过，但必须自然，不夸张。
最终停留在一个非常适合做首帧封面的高级怼脸画面。

画面风格

整体风格是：
真实电影级时尚人像、近距离商业摄影、车展模特贴脸抓拍、高级杂志感、自然皮肤质感、非影棚假面感。
要有真实镜头焦外、自然景深、细腻肤质、轻微现场感、克制的高级感。
不是网红自拍，不是直播间，不是韩式磨皮，不是低质AI脸。

重点强调

脸必须真实，不能假

皮肤不能过分完美，要保留轻微瑕疵

五官稳定，不能变形

眼睛不能飘，嘴唇不能乱动

发丝必须真实，有碎发

闪光灯必须符合现实抓拍逻辑

光线必须真实，不要胡乱乱闪

镜头必须高级，平滑，电影级，时尚摄影感

全片以“高级脸近拍展示”为核心，不要剧情化，不要大动作

负面限制词（建议一起加上）

负面提示词：
卡通感，CG感，塑料皮，过度磨皮，过度美颜，五官变形，眼睛错位，皮肤蜡像感，过曝，闪光灯频闪不合理，脸部抖动，低清晰度，低质量发丝，夸张滤镜，网红脸，假睫毛过重，浓妆，镜头乱晃，肢体变形，背景杂乱，可读文字，字幕，贴纸，水印，

Wan 3.0 (1)

Change the camera angle, keep the exact same action

STEP 1: Upload your video (must be continuous one shot) to any platform that supports WAN 3.0 

STEP 2: Use the prompt below (adjust it to your video):

"Use the original video @ TAG ORIGINAL VIDEO as the exact action reference 

0–1s: Camera stays exactly as in the original video as the wolf walks toward the scooter.

1–3s: Hard cut to a tight front close-up as the wolf continues walking toward and reaches the scooter.

3–5s: Hard cut to a high angle above as the wolf as he gets onto the scooter and starts riding.

5–7s: Hard cut to a full shot as the wolf rides the scooter down the street.

7–9s: Hard cut to a side tracking shot as he keeps riding down the street.

9-10s: Hard cut to an over the shoulder shot as he rides forward and notices the banana peel in his path.

10–12s: hard cut to a birds eye view shot as the fallen wolf is rolling over the asphalt. 

12-13s : hard cut to an insert shot of his fist as wolf slams it into the asphalt. 

Only the camera angle and framing change. Make sure every action, movement, timing and performance exactly identical to the original video @ TAG ORIGINAL VIDEO. The action continues seamlessly across every cut."

NOTE: If some scenes didn't work (action was different than original video) simply regenerate the video, that usually fixes the problem.

Run the generation on varg

varg is one API key for the models these skills drive — Seedance 2.5, Kling 4, Veo 3.1, Wan 3, Nano Banana 2.1, GPT Image 2.5, ElevenLabs v4 and 200+ more — billed in credits, with a cost estimate before anything runs.