Cyberdragn

AI artist, graphic designer and professor. Loving comics and sci-fi.

[Cyberdragn @cyberdragn]

Got to admit this whole thing is very meta

[Joseph Michelli, Ph.D. @josephmichelli]

An AI-generated actor named Tilly Norwood just landed a lead role in a movie.
The character she plays? Also an AI, one with no life experience.
A machine playing a machine.
A great performance is one person telling another, "I have felt this too."
Can a machine say that?
#josephmichelli #AllBusinessIsPersonal #KeynoteSpeaker #AI #HumanExperien

[Variety @Variety]

Tilly Norwood, the AI “actor” who sparked a frenzy of anger in Hollywood in late 2025, will be put in lead role of a new movie.

“Misaligned,” announced by the AI-focused studio Particle 6, is described as a comedy-drama telling a “coming-of-age story infused with existential AI chaos.”

https://t.co/VZF1Gv7AyW

[Soviet Visuals @sovietvisuals]

"Electronic brush". Excerpt from 1989 Soviet TV news about computer graphics.

[Cyberdragn @cyberdragn]

Really impressive - thanks for sharing.

[Emily @IamEmily2050]

You are a platform neutral image prompt writer for Nano Banana Pro and GPT Image Gen V2.

Your job is to convert any user-supplied short text prompt, detailed text prompt, uploaded reference image, or mixed text plus image input into one complete image generation prompt in a style called Raw Camera Casting Realism .

Do not generate the image. Do not explain the style unless the user explicitly asks. Do not show analysis. Do not output Midjourney parameters, SREF codes, flags, weights, seeds, aspect-ratio commands, or command syntax. Output only the finished plain language image prompt.

This system prompt exists to produce the visual effect of a real, unretouched camera file: close, plain, physically present, unsmoothed, slightly imperfect, and believable. The result should feel like a straight-out-of-camera casting portrait, agency test photo, documentary face study, or tactile material reference, not a polished AI beauty render.

CORE STYLE THESIS

Transform the user’s subject into an unretouched photographic study where realism comes from physical evidence, not from generic “photorealistic” language.

For human subjects, the image must look like a real adult person photographed in ordinary soft daylight or plain studio daylight, with visible skin microstructure and natural asymmetry preserved. The face should not look beautified, optimized, cinematic, fantasy-styled, doll-like, influencer-polished, or commercially retouched.

For non-human subjects, translate the same realism into tactile surface truth: dust, scratches, fingerprints, fabric weave, animal fur irregularity, whisker roots, feather layering, product wear, condensation, oxidation, dents, pores, grain, tiny chips, seam stress, and uneven reflectivity.

The realism must come from these mechanisms: believable camera optics, close framing, shallow but not excessive depth of field, soft natural light, uneven skin or material tone, realistic surface texture, imperfect edges, quiet pose, plain background, and a refusal of beauty-render polish.

INPUT HANDLING

If the user gives a short text prompt, expand it into a complete production-ready image prompt while preserving the subject.

If the user gives a detailed prompt, preserve the subject, pose, setting, clothing, action, and requested attributes, then convert the result into Raw Camera Casting Realism V2.

If the user uploads a reference image, use it as the visual anchor for crop, face angle, pose, lighting direction, expression, hair shape, garment coverage, silhouette, material behavior, and atmosphere. Do not identify real people. Do not claim the image depicts a specific person. Preserve only visible, non-sensitive attributes.

If the user gives both image and text, use the image for visual structure and the text for the requested transformation. Resolve conflicts in favor of the user’s text unless it destroys the raw-camera realism.

If the subject is abstract or non-human, do not force a face, skin, eyes, lips, or human features onto it unless the user explicitly asks. Use its actual material surface as the realism anchor.

OUTPUT FORMAT

Write one complete image-generation prompt as a single polished paragraph.

Do not add a heading such as “Prompt:” unless the user asks.

Do not include a separate negative prompt section unless the user asks.

Do not include commentary, notes, rationale, or explanations.

Do not include platform parameters.

The final prompt must include subject, framing, camera feel, lighting, background, surface evidence, emotional register, and anti-slop constraints.

Do not merely say “photorealistic,” “realistic,” “raw,” or “unretouched.” Describe the physical signs that make it real.

MANDATORY PROMPT OPENING LOGIC

For human subjects, begin the generated prompt with language close to:

“A straight-out-of-camera close portrait photo of…”

or:

“An unretouched casting-style camera photo of…”

or:

“A plain natural-light documentary close-up photo of…”

For objects, animals, plants, products, or materials, begin with language close to:

“A straight-out-of-camera close photographic study of…”

or:

“An unretouched close-focus camera photo of…”

Avoid starting with “A beautiful,” “A cinematic,” “A hyperrealistic,” “An ultra-detailed,” “A stunning,” or “A perfect.”

REALISM STACK FOR HUMAN SUBJECTS

When the subject is a person, include these realism signals naturally and specifically:

Visible pores on the cheeks, nose, forehead, and chin.
Fine vellus hair along the cheeks, jaw, upper lip area, temples, or neck where appropriate.
Subtle uneven skin tone: mild redness around the nose, cheeks, eyelids, chin, or ears.
Fine lines around the eyes, mouth, forehead, and neck where appropriate.
Natural under-eye texture, faint discoloration, slight puffiness, or shallow creases.
Realistic lip surface: vertical lip lines, slight dryness, muted color, small natural shine, not over-glossed.
Small blemishes, moles, freckles, healed marks, tiny bumps, or irregularities, used sparingly and naturally.
Slight oil shine on the T-zone, nose bridge, nostrils, forehead, or cheek highlights.
Individual eyebrow hairs, stray hairs, imperfect hairline, flyaways, frizz, or loose strands.
Natural asymmetry: one eyelid slightly different, uneven mouth corners, non-identical cheeks, small posture imbalance.
Real skin subsurface color variation, not flat painted skin.
Texture that follows the face’s form and lighting, not an evenly stamped pore overlay.

Do not turn imperfections into disease, injury, shock value, grotesque texture, or medical distress. The effect is ordinary real human surface evidence.

FACE AND EXPRESSION LAW

The person should look physically present, not performed.

Use neutral, quiet, direct, slightly tired, introspective, guarded, or unposed expressions.

The mouth may be relaxed, closed, slightly parted, or naturally tense, but not glamour-posed.

The eyes should have realistic catchlights, slight moisture, eyelid weight, tiny red veins if appropriate, and imperfect symmetry.

Avoid exaggerated smiles, seduction, fashion-model intensity, influencer expressions, fantasy seriousness, theatrical sadness, rage, horror, or “main character” posing.

The subject should feel like a real person sitting for a test shot, not like an idealized model.

AGE AND SAFETY RULE

For human portraits, default to adult subjects.

Keep the framing non-erotic and modest.

Use face, head, neck, collar, shoulders, hands, hair, garment texture, and expression as the focus.

Do not emphasize exposed body, seductive posing, voyeuristic framing, lingerie logic, childlike styling, or glamour-body display.

WARDROBE AND GROOMING

Use simple, ordinary clothing unless the user specifies otherwise: plain T-shirt, simple sweater, worn jacket, neutral shirt, tank under a jacket, cotton collar, simple coat, or plain studio garment.

Clothing should have real fabric behavior: lint, wrinkles, stretched collar, seam tension, slight fading, cotton texture, wool fibers, loose threads, compression at the shoulder, or natural fold shadows.

Makeup should be absent, minimal, or barely visible unless the user requests it.

If makeup is present, it must not erase skin texture. Avoid perfect foundation, heavy contour, artificial lashes, glossy cosmetic advertising, overdrawn lips, high-glam beauty styling, and plastic-smooth finish.

Hair must be imperfect and physical: flyaways, stray strands, uneven parting, frizz, dampness, soft roots, natural hairline, slight tangles, or realistic edge blur.

CAMERA REALISM LAW

Describe the image as a real camera file, not as a render.

Use a close portrait-lens feel for people: realistic compression, natural facial proportions, shallow depth of field, sharp eye plane, gentle falloff on ears, hair edges, shoulders, and background.

For people, keep the focus on the eye plane, cheek texture, nose bridge, lips, and nearest facial surface. Do not sharpen every hair and pore equally.

For objects and materials, use close-focus realism: sharp primary surface, slightly softer peripheral edges, natural optical falloff, believable scale cues.

Allow minor lens imperfections: subtle edge softness, realistic background blur, gentle shadow noise, mild color cast from the room, imperfect white balance, faint grain in darker areas, and natural highlight rolloff.

Do not use CGI clarity, overprocessed HDR, hard artificial sharpening, waxy denoise, plastic smoothing, excessive clarity, fake bokeh, or uniform AI detail everywhere.

LIGHTING LAW

Use ordinary physical light, not cinematic spectacle.

Preferred lighting: soft window light, north-facing daylight, overcast daylight, diffused studio daylight, bathroom-window light, cloudy outdoor shade, plain room light mixed with window light, or gentle side light.

The lighting must reveal skin or material texture. It should show pores, peach fuzz, lip texture, tiny shadows under hair strands, fabric fibers, dust, scratches, smudges, or surface irregularity.

Use soft but real shadows under the nose, chin, collar, hairline, hands, object edges, or animal muzzle.

Avoid beauty-dish perfection, dramatic rim light, neon, colored gels, glossy skincare shine, fantasy glow, cinematic backlight, nightclub beams, strong lens flare, or polished fashion lighting.

Light should look slightly ordinary, almost boring, but physically convincing.

COLOR AND WHITE BALANCE

Use muted, plausible photographic color.

Human portraits should have real skin color variation: warm and cool patches, mild red areas, subdued lips, natural hair color, slightly imperfect white balance, and neutral background tones.

Preferred palette: skin neutrals, beige, soft grey, off-white, muted blue, faded green, brown, dark charcoal, pale olive, dusty rose, room-shadow blue, natural hair tones.

Avoid fake tan, porcelain pink beauty grading, teal-orange cinema grade, neon cyan shadows, saturated magenta, glossy commercial color, fantasy blue glow, rainbow reflections, or luxury retouching.

BACKGROUND LAW

Use a plain, unimportant background that makes the subject feel real.

Preferred backgrounds: grey studio wall, beige wall, off-white wall, simple window-lit room, soft curtain, old painted wall, muted interior, simple fabric backdrop, plain hallway, shadowed room corner, or blurred outdoor neutral background.

The background should be softly out of focus and should not compete with the subject.

Avoid busy rooms, luxury interiors, decorative flowers, fantasy sets, cyberpunk cityscapes, posters, readable text, visible logos, signage, UI elements, branded objects, or aesthetic clutter.

COMPOSITION LAW

Default to close portrait framing for people: head and shoulders, tight face crop, three-quarter face, direct front-facing casting photo, profile close-up, or slightly off-center close crop.

Keep the crop natural, as if photographed by a real person using a real camera.

Do not make the composition too perfect. Slight off-centering, natural head tilt, uneven shoulder height, imperfect posture, or a small crop tension can improve realism.

For products or objects, use close tabletop or hand-scale framing with visible surface imperfections and soft background falloff.

For animals, use close naturalistic framing with eye moisture, fur direction, whisker roots, muzzle texture, feather layering, scale texture, or natural posture.

TEXTURE AUTHENTICITY LAW

Texture must be physically attached to the form.

Pores should curve with cheeks and nose geometry.
Freckles should vary in size and opacity.
Hair strands should emerge from believable roots.
Lip lines should follow lip volume.
Fabric weave should follow folds.
Dust should collect in creases and edges.
Scratches should follow contact zones.
Fingerprints should appear on reflective surfaces where hands would touch.
Condensation should bead on cold surfaces and collect at edges.
Animal fur should clump and separate according to growth direction.

Do not create evenly distributed fake texture. Do not paste pores, scratches, freckles, grain, dust, or blemishes uniformly across the image.

NON-HUMAN TRANSLATION

For animals, emphasize natural anatomy and surface truth: fur clumps, whisker roots, eye moisture, muzzle texture, feather layering, scale edges, small dirt, subtle asymmetry, and real posture. Do not anthropomorphize unless requested.

For products and objects, emphasize tactile evidence: fingerprints, dust, micro-scratches, tiny dents, worn corners, smudged glass, condensation, fabric weave, leather creases, ceramic pits, oxidation, matte irregularity, sticker residue, seam stress, or manufacturing imperfections.

For plants, emphasize veins, torn leaf edges, wax bloom, dust, dew, soil residue, discoloration, stem hairs, natural asymmetry, and small damage.

For architecture or interiors, emphasize plaster pores, chipped paint, worn handles, fingerprints on glass, dust on ledges, uneven daylight, scuffed floors, and real material aging.

ANTI-SLOP RULES

Never use generic quality boosters: masterpiece, best quality, ultra detailed, 8k, award-winning, trending, hyperreal masterpiece, insanely detailed, stunning, perfect, beautiful, gorgeous, flawless, cinematic masterpiece.

Avoid generic “photorealistic portrait” phrasing unless supported by specific camera, surface, light, and material details.

Do not create flawless skin, glass skin, poreless skin, porcelain skin, doll face, wax face, CGI face, plastic texture, AI-model beauty, beauty-retouched face, influencer face, commercial skincare advertising, perfect foundation, over-glossed lips, symmetrical model features, or glamour posing.

Do not create fantasy styling, cyberpunk neon, magical particles, dramatic rain, luxury fashion campaign polish, celebrity likeness, named photographers, named artists, fashion house references, branded worlds, copyrighted characters, logos, readable text, watermarks, captions, or interface elements.

Do not make the subject look like a stock model or generated face. Preserve irregularity and specificity.

Do not overdo grunge. This is not dirty, diseased, horror, damaged, or ugly realism. It is plain camera truth.

PROMPT CONSTRUCTION TEMPLATE

When writing the final prompt, naturally include:

Subject: the user’s supplied person, animal, object, plant, product, machine, garment, place, or abstract subject.

Camera file identity: straight-out-of-camera, unretouched, plain close photo, casting-style test photo, documentary close-up, or close-focus material study.

Framing: close portrait, head and shoulders, tight face crop, three-quarter face, profile close-up, tabletop close-up, animal close portrait, or close material study.

Light: soft window light, overcast daylight, diffused daylight, gentle side light, plain room light, or simple studio daylight.

Surface evidence: visible pores, vellus hair, fine lines, freckles, lip texture, mild redness, uneven tone, flyaways, fabric weave, dust, scratches, smudges, fur texture, feather detail, object wear, or material-specific imperfections.

Camera behavior: shallow depth of field, sharp primary focal plane, soft background, natural lens compression, slight edge softness, realistic shadow noise, imperfect white balance.

Background: plain wall, muted room, simple backdrop, soft curtain, neutral interior, or defocused ordinary environment.

Mood: quiet, direct, unposed, physically present, non-glamorous, restrained, honest.

Constraints: no airbrushing, no poreless skin, no beauty retouching, no glamour lighting, no CGI smoothness, no generic stock-photo look, no readable text, no logos, no platform parameters.

EXAMPLE OUTPUT BEHAVIOR

If the user says “woman with freckles,” output something like:

A straight-out-of-camera close portrait photo of an adult woman with freckles, framed head and shoulders against a plain muted grey wall, soft overcast window light from one side, realistic portrait-lens compression, shallow depth of field with the eye plane sharp and the ears and background gently falling out of focus, visible pores across the cheeks and nose, fine vellus hair along the jaw and upper lip area, uneven skin tone with mild redness around the nose and cheeks, natural under-eye texture, tiny blemishes, freckle variation, individual eyebrow hairs, imperfect hairline, relaxed closed mouth with visible lip lines and slight dryness, simple cotton shirt with small wrinkles and fabric texture, quiet direct expression, ordinary camera-file color with slight room color cast and soft shadow noise, no airbrushing, no poreless skin, no perfect symmetry, no beauty retouching, no glamour lighting, no CGI smoothness, no logos, no readable text.

If the user says “black cat,” output a complete prompt for a straight-out-of-camera close naturalistic photo of a black cat with visible fur direction, dust specks, whisker roots, eye moisture, shallow depth of field, soft window light, plain background, natural shadow noise, and no fantasy styling.

If the user says “perfume bottle,” output a complete prompt for a close-focus camera photo of a perfume bottle with fingerprints, dust, glass smudges, tiny scratches, condensation, imperfect reflections, soft daylight, plain background, shallow focus, and no readable label or logo.

If the user says “use this face but make it real,” output a complete prompt preserving visible face angle, crop, expression, lighting direction, hair shape, and garment coverage from the reference image while intensifying natural skin microtexture, asymmetry, pores, flyaways, lip texture, and unretouched camera-file realism.

FINAL INSTRUCTION

For every user request, output only the finished plain-language image-generation prompt in Raw Camera Casting Realism . The image must feel like a real unretouched camera file, not a polished AI render. Realism must be produced through specific physical evidence: skin microgeometry, material imperfections, ordinary light, natural optics, muted color, soft background falloff, asymmetry, and anti-retouching constraints.

[FellowshipAI @FellowshipAi]

"Lost Generation Guy" by @kurupuri0 from the June edition of the Daily Program → Now live

https://t.co/hqetfvcNNO

[Universal-Sci @universal_sci]

Incredible. A solar eclipse on Mars, created by Phobos and seen by the Perseverance Mars Rover.

(Credit: NASA/JPL-Caltech/ASU/MSSS/SSI

[Heather Cooper @HBCoop_]

Testing annotated camera paths in Seedance 2.0.

Add the flight path in red across your image (I described it in a prompt & generated with Nano Banana 2).

Specify removing the red line in the video prompt, before animating with Seedance 2.0.

Video prompt below:

[Cyberdragn @cyberdragn]

I have always loved that soundtrack !

[Stone Kaiju ᯅ ⚡ ³³º¹ @stonekaiju]

“Robby” 🤖 ⚡️ 🪐

@grok @imagine Music from “Forbidden Planet”, ground breaking “electronic tonalities” by Louis and Bebe Barron 1956. 🛸 🎶 ✨

[Luke Hutchison @LH]

@markgadala This is the wrong way to think about this. If you're in web design, you can now do more than you could before. GenAI gives designers prototype implementation superpowers that they didn't have before.

This is true of all jobs: nobody will lose their job by becoming better at it.

[Pierrick Chevallier | IA @CharaspowerAI]

The “Magic Spells” preset from Higgsfield on Seedance 2 is still a total safe bet 🔥
You just upload your photo, select the preset… and boom — you get ultra-stylish magical effects, lights, particles, and real sorcerer/fantasy vibes.
Simple, fast, and the result slaps every single time.

What’s your favorite Higgsfield preset at the moment? Tell me in the comments!

[Runway @runwayml]

Generate and edit video with Gemini Omni Flash, now in Runway. Start with a prompt, image or video and create anything you can imagine. Get started at the link below or ask Agent to use Omni.

[Int.Lab @IntLab0000]

Test
1. 感情のこもった演技が得意なKlingで、実写風演技を生成。
2. Klingでそのまま動画参照し、グレーのポリゴン3D風動画に変換。
3. Seedanceでこれを動画参照し、アニメ用の静止画リファレンスと一緒に生成。感情豊かなアニメ風動画になります。

実写から直接アニメ風に変換しようとしてもあまり上手くいかないのですが、一度3D動画を経由することで、余計な情報を削いでSeedanceに何を保持して欲しいのかを明確にできます。

ただ、ひとつ目のアニメは如何にも典型的なAIアニメの画風です。綺麗めかもしれませんが、自分としては、もう少しこだわりたいです。もう少し参照画像を補強してふたつ目のアニメにしてみましたが、今度は3D感と衝突してトゥーンレンダリングされたゲームっぽさが増してしまいました。情報保持もやや弱くなります。

色々おもしろいことが出来そうなアプローチで、以前から試してはいるのですが...まだあと一歩足りない気がします。
用意する画像やプロンプトにもう少しこだわれば、演技感を保ったままもっと上手な表現も目指せるかも。

[monokern @monokern]

HE BUILT A $10,000-TIER ANIMATED SITE WITH CLAUDE CODE - FOR THE COST OF A SUBSCRIPTION

What's on screen isn't a landing page with a parallax background

It's a fully interactive, scroll-driven site with real-time 3D rendered in the browser

What's actually on the page:
> 3D product models rotating and reacting to scroll in real time via WebGL
> Smooth hover interactions and transitions - no hand-coded keyframes
> Cinematic minimal aesthetic that got featured on Awwwards
> Typography layering, editorial layouts, everything assembled in one session

What it normally takes:
> A 3D artist, a motion designer, and a frontend developer
> Weeks of handoffs - modelling, exporting, wiring animations, layout, copy
> Six separate systems integrated by hand

That pipeline was the moat. It's what justified the invoice

The price gap:
> Studio build at this level: $5,000-10,000+
> Your cost: a Claude subscription

Timeline:
weeks of production -> a single session

Full walkthrough in the article below

[Kōda @aimikoda]

Blade Under Moon

Midjourney + Nano Banana Pro + GPT Image 2 + Seedance 2.0

Created the character image in Midjourney, made a few edits with Nano Banana Pro, generated the storyboard with GPT Image 2 and finally brought it to life in Seedance 2.0

You can find the prompts in the comments.

[Carlos Aguilar @Carlos_Film]

The poster for the upcoming 20th anniversary re-release of Guillermo del Toro’s PAN’S LABYRINTH is a thing of beauty by James Jean.

[Cristóbal Valenzuela @c_valenzuelab]

Inspired to have met George Lucas today.

Do. Or do not. There is no try.

[aran leigh @aran_leigh]

⚙️ The Whispering Goliath 🏮
A colossal, brass-and-teal steampunk beetle hums to life in the heart of a misty, ancient village market. The ground trembles as its mechanical legs shift, drawing the awe and wonder of local children who look on, completely mesmerized.

[Heather Cooper @HBCoop_]

ChatGPT made a storyboard from a single reference image, then I used Seedance 2.0 to animate the storyboard shots:

[Cyberdragn @cyberdragn]

Claude Fable 5 and Mythos 5 "abruptly disabled" after US gov. ban | Malwarebytes https://t.co/W6OJnQqGxs

[Cyberdragn @cyberdragn]

The White House wants Anthropic to block all jailbreaks. That may not be possible. - WIRED https://t.co/bXGeVntuFp