hodeehum

Exploring tech, AI, and agentic workflow design.

[Tugba @TGBA2023]

Dragon Day Fun ✨
Shell Crown Serenade & Seadragon 🐉
A sweet underwater violinist with a crown of seashells, playing a tiny ocean song for her curious seadragon friend 😍 and with every note, the coral reef glows a little brighter, like the sea is smiling back at them 🥹
Prompts are below. Try them and share your version♥️
@imagine

[Aleena Amir @aleenaamiir]

A fusion of fluid artistry and minimalist design.

PROMPT ↓
“Modern flat marbling-texture illustration of [SUBJECT], swirling organic pattern fill, muted jewel-tone palette, white background.”

[Tim Brooks @_tim_brooks]

Muse Image supports a lot of fun use cases, like interleaving text and many reference images in prompts. Try it today on https://t.co/XkqcJbxxnS!

[AI at Meta @AIatMeta]

Introducing Muse Image and Muse Video, the first media generation models developed by Meta Superintelligence Labs.

Muse Image is our most advanced image generation model yet. It follows instructions faithfully, edits with precision, composes from multiple references, and draws on Instagram for social context. It also brings agentic tool use capabilities to image generation and integrates with Muse Spark.

You can try Muse Image in the Meta AI app and web, as well as in Instagram Stories and WhatsApp – starting in limited countries with more locations on the way.

Today we’re also previewing Muse Video, which is built upon the same pretraining base as Muse Image to deliver exceptional visual fidelity with native audio support.

Learn more about both models: https://t.co/QtKDPDZP5v

[mariana Z @mariana057]

Parenting is buying 6 yogurts and watching them all get eaten In one day.

Then buying 12 yogurts and watching them expire because, “Nobody likes yogurt.”

[Primee32 @Primee32]

THE CHARACTER SHEET METHOD THAT KEEPS A BRAND ATHLETE CONSISTENT ACROSS AN ENTIRE RACE SEQUENCE IN SEEDANCE 2.0

One runner. One uniform. A full Tokyo Marathon sequence from start line to finish line victory.

No identity drift, no uniform changing color mid-race, no face losing detail between close-ups.

Most creators generate one hero shot and hope the character holds across every new scene.

The real consistency comes from building a full character sheet first — then feeding that sheet into Seedance 2.0, not a single portrait.

Here's the workflow:

1. Write the character bible — name, build, uniform, bib number, brand colors. locked before frame one
2. Build the full turnaround — front, 3/4 front, side, 3/4 back, back — all in one sheet
3. Add the expression sheet — focused, exertion, post-finish, victory, pre-race
4. Add action poses — full sprint, steady run, warm up, finish line
5. Lock costume detail + hex color codes — fabric, shoes, accessories, exact brand palette
6. Feed the finished sheet into Seedance 2.0 as the hard character reference
7. Stitch scenes in narrative order — start to finish

Why this works:

- Athlete gets designed once — Seedance can't reinvent what it's already been given
- Expression sheet locks emotional range before animation starts
- Hex color codes give exact uniform anchors instead of guessing from one lit photo
- Action poses encode movement logic so Seedance animates the right body mechanics per scene

Use cases:
⁃ Sports brand character campaigns
⁃ AI athlete or mascot content across multiple scenes
⁃ Animated shorts with a single hero and a clear narrative arc
⁃ Any project where one character has to hold from opening shot to final frame

Not every expression survived the close-up shots perfectly — victory face needed two regenerations.

But building the full sheet before touching Seedance turned a brand concept into a complete race story in one session.

[PixieVerse @itsPixieVerse]

GPT Image 2 Character Sheet Prompt:

Create a premium, modern high-fashion CHARACTER CONCEPT ART SHEET in a 16:9 widescreen layout on a pure white background, following this exact structured composition and visual aesthetic:

[STYLE]: high-contrast digital anime illustration with clean, slightly textured graphite outlines, sharp cel-shaded shadows, and soft hand-painted watercolor textures. The style should feel like premium, highly fashionable dark-fantasy concept art, featuring a bold, expressive color-clash harmony of deep midnight-blue, glowing spectral cyan, and charcoal gray.

[SUBJECT_DESCRIPTION]: A majestic, towering grave-titan demi-god king. His upper torso is muscular and weather-beaten, while his lower body dissolves into a swirling vortex of spectral neon-cyan flames and floating, cracked tombstones. Multiple spectral, translucent arms sprout from his back, grasping ancient, glowing stone relics. He wears a rotting, oversized royal shroud of deep midnight-blue, held together by tarnished gold chains. His face is a hollow, skull-like visage with piercing, glowing blue eyes, crowned by a floating halo of jagged obsidian shards. Ghostly mist and glowing embers of cyan fire drift from his robes, conveying a sense of tragic, ancient, and terrifying majesty.

Layout Composition (Organize strictly into these sections):

1. LEFT PANEL: METADATA & TURNAROUND

- Large, bold vertical typography of the character's Name in a clean, heavy sans-serif condensed font.

- Metadata Block: "ROLE", "CORE MOOD", and "VISUAL SIGNATURE" detailed in a clean sans-serif font.

- MINI TURNAROUND: 3 small, clean vertical figures showing "neutral", "back view", and "profile" stances.

- SILHOUETTE STUDY: 3 small, solid-black silhouette matchings of the turnaround stances underneath.

- EXPRESSION STUDY (Bottom Left): Exactly 4 small, close-up headshots showing subtle shifts in expression (e.g., Neutral, Pensive, Confident, Smile).

2. CENTRAL PANEL: THE CENTERPIECE

- One giant, dominant full-body figure of the character standing in a cool, relaxed signature pose (e.g., hands in pockets, looking down/aside). This figure must visually dominate the center of the sheet with maximum detail.

3. RIGHT PANEL: MULTI-POSE & THEMATIC STUDIES

- POSE STUDY: Exactly 4 to 5 small-to-medium figures in dynamic, casual stances (such as "seated", "crouching", "leaning", "low angle", or "top view"), each with a small handwritten-style descriptive label underneath.

- OPTIONAL SPECIALTY STUDY: A vertical strip of 4-6 artistic, painterly portraits with expressive brushstroke or paint-splatter borders showing morphing, stylized, or glitch-like variations of the character's face.

4. BOTTOM RIGHT PANEL: DETAIL STUDY

- Exactly 5 small square close-up crops highlighting the character's unique features, accessories, and textures (e.g., fabric weave, jewelry, specific utility pockets, unique footwear soles).

- Each crop must have a clean, neat handwriting-style label underneath describing the detail (e.g., braids medium length, layered fit, chunky sole).

Keep the entire presentation visually consistent, clean, and highly professional, mimicking a real studio-grade character design bible.

[Massimo @Rainmaker1973]

The carrom championship series

[📹 carromskills]

[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.

[OscarAI @Artedeingenio]

After watching this video, you'll understand why AI has changed military action anime forever.

I created it using a military anime style I developed in Midjourney (available to my subscribers), then brought it to life with Seedance 2.0 in @TopviewAIhq, the platform I recommend for getting the most out of this video model (more info in my bio).

Get ready for an adrenaline rush unlike anything you've seen before.

[Lyman Stone 石來民 🦬🦬🦬 @lymanstoneky]

“Grok inhabits a moral space shared with virtually no human societies” is my favorite part of this

[Massimo @Rainmaker1973]

Size comparison: Polar Bear vs. American Black Bear

[Massimo @Rainmaker1973]

Every state in the US has recorded temperatures of at least 100 degrees Fahrenheit including Alaska.

The highest temperature ever documented in the country reached 134 degrees Fahrenheit or 56.7 degrees Celsius at Death Valley California on July 10 1913. This measurement remains the highest reliably recorded air temperature on Earth.

Numerous states have experienced extremes well above 115 degrees Fahrenheit. In 1936 during the Dust Bowl era North Dakota recorded 121 degrees Fahrenheit even though the location sits only 150 miles from the Canadian border. Oklahoma registered 120 degrees Fahrenheit on multiple occasions that same summer.

The 2021 Pacific Northwest heat wave produced new milestones. Washington achieved an all time high of 120 degrees Fahrenheit while Oregon matched its statewide record of 119 degrees Fahrenheit.

Not all peak temperatures originate in the Southwest. Minnesota has reached 115 degrees Fahrenheit Missouri has hit 118 degrees Fahrenheit and Rhode Island has climbed to 104 degrees Fahrenheit.

Alaska recorded its highest temperature of exactly 100 degrees Fahrenheit in Fort Yukon on June 27 1915. Hawaii also attained 100 degrees Fahrenheit in Pahala on April 27 1931.

The majority of statewide heat records occurred during July or August. However some emerged during atypical early season heat waves demonstrating that extreme temperatures can appear outside the core summer months.

With more than 260 million Americans currently facing dangerous heat conditions meteorologists caution that long standing state records may face challenges in the future as extreme heat events increase in frequency.

[CryptoDiggerGod.eth @CDGalpha]

gemini 3.5 flash is free now 😳

google quietly made their newest flash model free-tier eligible. no credit card, 1M context window, and most people are still posting about the old 2.5 models.

what you get for $0:
- gemini 3.5 flash (google's newest, free tier)
- gemini 3.1 flash-lite for high-volume cheap calls
- full 1M token context on every free model
- 1,500 requests/day, native multimodal (text, image, audio, video)
- openai-compatible, one base url change

full setup (2 min):

step 1: grab a free api key
> go to https://t.co/bvoogVXOtj
> sign in with any google account
> click "get api key", no billing setup

step 2: point any client at gemini
> works in cursor, cline, claude code, anything openai-compatible
> or just prototype in the studio ui and hit "get code"

step 3: pick your model
> gemini-3.5-flash for the strong one
> gemini-3.1-flash-lite for cheap high-volume steps

important:
- pro models left the free tier april 1, flash and flash-lite only now
- free-tier prompts can be used to train google's models, don't send sensitive data
- limits are per project, extra keys don't add quota
- 1,500 req/day resets daily, plenty for building and testing

frontier-class model + 1M context + $0
while most guides are still telling you to use last-gen models

save this, the free tier shifts almost monthly

[Skye @skye0xz]

Baby hack every parent should know! 👶✨

[تُهَامِيّ | Touhami 𓂀 @IS7MI]

Prompt 🖼⤵️

Use the uploaded photo as the ONLY facial identity reference. Preserve the person's recognizable facial structure, skin tone, age, hairstyle, and unique facial features. Strict face-lock. Gender is determined automatically from the reference image.

Create a premium ultra-realistic 3×3 collage (nine unique panels). Each panel features the same person in a completely different outfit, pose, facial expression, hairstyle variation (while remaining recognizable), accessories, and composition. Every scene must feel unique and visually distinct.

Blend each photorealistic portrait seamlessly into imaginative black-and-white hand-drawn doodle environments made of clean ink line art. The person, clothing, tools, and key objects are rendered in rich vibrant colors with realistic textures, while the surrounding doodle world remains monochrome.

Generate nine different creative themes with varied compositions, such as adventure, travel, sports, science, fantasy, art, music, cooking, nature, architecture, exploration, hobbies, professions, and magical worlds. Randomize the activities, props, environments, camera angles, and perspectives so every generation produces fresh concepts.

Each panel should feature:
• Different clothing style and color palette
• Different pose and body language
• Different facial expression
• Different camera angle (close-up, eye level, low angle, high angle, overhead, side view, dynamic perspective)
• Different lighting mood
• Different colorful props integrated naturally with the doodle illustrations

Keep the person highly photorealistic with natural skin texture, realistic lighting, soft shadows, cinematic depth, and ultra-sharp facial details. The doodle backgrounds should be playful, clean, imaginative, and professionally illustrated without overpowering the subject.

Premium editorial quality, hyper-realistic, seamless color-to-doodle integration, vibrant colors, creative storytelling, soft natural lighting, ultra-detailed, 8K, masterpiece.

[Massimo @Rainmaker1973]

When you start a chess game, you have 20 possible moves available. After the first full move (White then Black), there are already over 400 possible positions. By the third move, that number jumps to around 8,900, and after the fourth it reaches nearly 200,000.

By the time you get to move #40, the total number of possible games explodes to roughly 10⁴⁰, a number comparable to the total number of atoms in the observable universe.

[Massimo @Rainmaker1973]

This creature lives inside you and controls literally every thing you do.

[Atenov int. @Atenov_D]

Anthropic CEO, Dario Amodei:

"Cancer is cured, the economy grows 10%… and 20% of people don't have jobs".

> That's the CEO of a $965B company, now worth more than OpenAI, describing the world he's building. Three things he keeps saying that most people still refuse to believe:

One: half of all entry-level white-collar jobs could be gone in one to five years, with unemployment spiking to 10–20%.

Two: it already started. Anthropic's own tools wiped $285 billion off software stocks in a single day, the worst day for the sector since the 2020 COVID crash.

Three: the real accelerant isn't jobs, it's AI that builds AI. And he puts the odds it ends badly at roughly 25%.

Save it. Reread in 2028 and see who was right.

[witcheer @witcheer]

I need as many updates as possible on this little Hermes robot