All skills
hoodini avatar

/image-master

@83e2083

Master prompt-engineer for photoreal, artifact-free AI still images on ANY tool (Reve, Midjourney, Flux, GPT-image, Imagen, Nano Banana, Stable Diffusion). Builds prompts that hit National-Geographic-grade realism — true skin/fur texture (no plastic), correct anatomy/hands/faces, physically coherent light/shadows/reflections, clean legible text — while engineering deliberate visual IMPACT (color contrast, composition, awe/adrenaline, a striking point of view). Runs a gated process: lock the point-of-view and build the 8-block Capture Stack BEFORE generating, then run a forensic pre-submit inspection against the known artifact list. Use whenever the goal is a single still image that must look REAL and hold up under close inspection — contest entries, hero shots, product/character/wildlife/architecture/concept art, or any time AI images come out plastic, distorted, or fake. Composes with nano-banana-2 (execution) and director (motion). Hebrew triggers: תמונה, תמונות, פוטוריאליזם, ריאליזם, לייצר תמונה, פרומפט לתמונה, בלי עיוותים, עור פלסטיק, ידיים מעוותות, פרצוף מעוות, חדות, נשיונל ג'יאוגרפיק.

Use this Skill: https://skilld.dev/gh/hoodini/ai-agents-skills/image-master

This session only. Nothing lands on disk.

references10-directing-emotion.md

≈1.8k tokens on demand. Your agent reads this file only when SKILL.md points to it.

10 — Directing Emotion: the Heart-Wrenching Frame

Technical realism gets an image PAST inspection. Emotional direction is what makes it WIN. The most common failure of a clean prompt: it describes blocking ("a cheetah runs", "a lion sits") and leaves expression to the model — which defaults to NEUTRAL. The result is technically fine and emotionally dead. Fix: direct the decisive emotional moment, the gaze, and one signature micro-detail in every frame.

0. Anchor every frame to a SPECIFIC, REAL pain — research it first

Generic sadness is forgettable; a specific, true, lived pain is unforgettable. Before designing a series about any real human experience, RESEARCH the actual pains of the people who lived it (testimonies, reporting) and anchor each image to ONE concrete pain. The image then carries a truth the viewer recognizes in their own body — the highest form of "distinct visual narrative." Translating the pain through an allegory (here: animals) makes it bearable AND universal, but the pain underneath must be real and specific — never invented.

Worked anchors — the Israeli home-front pains this series is built on (each → ONE image, all sourced in the project research):

  • the parental dilemma of waking a just-asleep child for the siren, or gambling on letting them sleep;
  • the elderly/disabled who cannot reach shelter in the 15–90s window ("Go, don't wait for me");
  • 56% of homes have no safe room (mamad);
  • the public shelter crammed like "a tin of sardines," no toilet, strangers at 3am;
  • caught in the open, told to lie face-down and cover your head;
  • hit despite doing everything right (a struck safe room);
  • bone-deep exhaustion from nightly alarms; the parent who performs calm and never sleeps;
  • the siren as a Pavlovian trauma trigger (≈60% report phantom sirens);
  • pets paralyzed by terror they cannot decode.

1. Direct the decisive EMOTIONAL moment (not the action)

Never "X does Y." Always the peak instant of feeling: not "a cheetah runs" but "a cheetah at full desperate stretch twists her head back toward the incoming light, eyes wide." The verb carries the emotion. Freeze the moment a heart breaks, not a generic activity.

2. Animal affect cues — the believable vocabulary (real ethology, not cartoon)

  • Fear: ears pinned flat back · eyes wide with the whites/sclera showing · pupils dilated · head lowered, body crouched · flared nostrils · rapid shallow breathing, flank heaving · trembling muscles · tail tucked · frozen mid-flinch.
  • Protection (parent): body curled tight over the young · head lowered over them · muscles tensed · one paw drawing a cub in · eyes scanning, afraid-FOR-them, not aggressive.
  • Desperation / exhaustion: panting open mouth · foam or saliva flecking the muzzle · heaving flanks · strained trembling limbs · stumbling.
  • Grief / shock: frozen stillness · a wet glistening eye · a single tear cutting a track through dust or ash · the thousand-yard stare. Avoid bared-teeth aggression unless intended — it reads as anger, not fear, and open mouths risk teeth artifacts.

3. The signature micro-detail — the thing that breaks the heart

Every frame needs ONE, named explicitly (the model will not invent it): a tear welling or tracking through dust · the airburst/fire mirrored in the wet eye · breath fogging in cold air · a cub's tiny paw gripping a parent's mane · foam at the mouth · dust caked on a tear-streaked face · trembling limbs · a single catchlight that is actually the war.

4. The eye is the story (and a craft flex)

Lead the viewer to ONE eye — large, sharp, wet, catchlit. Put the reflection of the airburst/strobe inside the pupil: simultaneously the most devastating emotional beat AND a showcase of fine-detail + reflection physics the judges reward. Reuse this "war-in-the-eye" device across the set for cohesion and a striking point of view. (The reflection must match the real light direction — see 02.)

5. Emotion vs artifact — resolve the tension on purpose

Expressive faces raise facial-render risk; backs-turned is safe but emotion-poor. The resolution: keep the BODY/pose safe (turned, cropped, head-down for hands/limbs) while granting ONE readable, well-directed eye — face/head turned into frame, single catchlight, slight natural asymmetry (01). You can have both drama and safety; you just have to direct it. Don't trade away the eye for safety — trade away the hands.

6. The emotion gate (add to every generation)

Before generating any frame, answer in one line:

"What is the animal FEELING in this exact instant, which eye shows it, and what single detail makes me ache?" If the prompt doesn't answer all three, it will render neutral. Don't generate until it does.

7. Reality check

No prompt guarantees emotion — the model still has to deliver, and you will re-roll and pick the frame where the eye and the tear land. Generate a batch, then choose the one that aches. A clean-but-neutral frame is a reject, however technically perfect.

8. Legibility beats subtlety — make the key story element UNMISTAKABLE

A frame fails if a stranger can't tell what's happening in one second. If the core story element (here: the war) must be understood, render it BIG, clear and iconic — never "distant," "soft," or "subtle." Hard lessons from the first Reve batch:

  • A "distant airburst" renders as a meaningless bright star / lens-flare. Instead write: "the sky filled with the unmistakable Iron Dome interception — many curved interceptor-missile trails criss-crossing, orange airburst explosions, drifting smoke trails, over a city skyline." Iconic and instantly readable.
  • But make it read as an INTERCEPTION, not FIREWORKS: cascading sparks read as celebration. Specify "rising interceptor streaks meeting incoming rockets in sharp mid-air airburst flashes, plus incoming rocket trails and smoke — a missile interception over a city, not fireworks, no cascading sparks."
  • An extreme close-up has NO room for context — the war vanishes and a red strobe just reads as "a lamp." Pull back to a MEDIUM shot so the subject's emotion AND the war sky are both in frame. Reserve extreme close-ups only when the war is dramatically, clearly reflected in the eye.
  • Anchor the setting so it reads as the real place: an Israeli street, a concrete bomb shelter, a city skyline — never an ambiguous "cave/den."
  • Wild animals (lion, tiger) read as "Israel at war" ONLY with the unmistakable war sky + a city behind them. Domestic animals (dog, cat) in a real street/shelter read fastest — lean the set toward maximum legibility.
  • Text colour: don't default to flat white (reads generic). Specify it — masthead white, the recurring tagline in alarm RED (pops like a real magazine, ties to the Red-Alert siren). Test before generating: would a stranger glancing for ONE second say "that animal is in a war"? If not, the war isn't big enough.

Source: SKILL.md on GitHub

No alerts3mo3 checks · Risk SAFE
  • Gen Agent Trust Hub3mo

    The skill is a comprehensive instructional framework for AI image prompt engineering. It consists entirely of Markdown documentation and does not contain any executable code, scripts, or mechanisms for data exfiltration. All external references are informative and point to legitimate AI and photography resources.

  • Socket3mo

    No alerts

  • Snyk3mo

    Risk: LOW · No issues

Signed by skilld at 83e2083. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub 2 days ago.

Steadyupdated 4 months ago

README badge

README badge for hoodini/ai-agents-skills/image-master