Trending AI photo and video styles: the looks people ask for now
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.mdto its path. Before generating anything, show the person the options on the use-case page.
32 looks from AI photo and video posts on X and Pinterest boards, September–October 2026, ranked by how hot they are now. Each one has a clause to add to any prompt (with a [subject] slot), what breaks the look, the models that do it best and the posts that show it.
1. Code-made motion design
rising · video · also called Opus motion design, kinetic typography, product launch video, After Effects killer, HyperFrames / Remotion motion
Product-launch motion graphics rendered from HTML, GSAP or three.js instead of a video model: big kinetic type, UI cards sliding on a strict grid, isometric devices and Swiss-style layouts. Neutral or near-black backgrounds with one accent color, crisp vector edges and tight, deliberate easing; every frame is pixel-sharp because nothing is generated.
- Prompt clause:
a 15-second product-launch motion graphic for [subject]: bold kinetic typography, UI cards sliding on a precise grid, crisp vector shapes, one accent color on a neutral background, snappy ease-in-out timing, built in HTML/GSAP - Keep out: AI-video morphing, blurry or misspelled text, random stock footage, more than two typefaces, linear robotic easing, cluttered layouts
- Best on: Claude Opus 5.5 (code: HTML/GSAP/three.js), Claude Opus 5.5 + HyperFrames/Remotion, Higgsfield Katana + Claude Motion
- Trend: Biggest AI-visual cluster this month: motion-design probe since Sep 15 (>=300 likes) has 10 posts in Oct 13-27 (20.4k likes total) vs 4 in Sep 29-Oct 12 (8.5k); 12.7k (Sep 24), 6.9k (Oct 10), 6.5k (Oct 19), 4.4k (Oct 17) like posts.
- Posts: @twoclipping, 12,717 likes · @flxrnc, 6,880 likes · @0xwhrrari, 6,521 likes · @rexan_wong, 6,893 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5
2. Unretouched phone snapshot
rising · photo and video · also called iPhone photo, doesn't look AI, raw realism, candid phone pic, AI influencer realism
Looks like an unedited phone photo: mixed ambient light, slight noise and phone over-sharpening, natural skin with pores and blemishes, casual off-center framing and ordinary backgrounds. The point is that nobody can tell it is AI.
- Prompt clause:
[subject] in a casual unedited smartphone photo, natural mixed ambient light, slight noise and phone sharpening, real skin texture with pores, off-center candid framing, ordinary background, no retouching - Keep out: studio lighting, airbrushed skin, perfect symmetry, cinematic bokeh, heavy color grading, model poses
- Best on: Grok Imagine (varg
grok_imagine_1_5), Seedance 2.5 (stills from video) (vargseedance_2_5), GPT Image 2.5 vs Nano Banana Pro (varggpt_image_2_5) - Trend: Realism probe since Sep 25 (>=400 likes): A 7 posts / 6.1k L, B 7 / 5.2k L, C 6 / 4.2k L; 'pushed realism too far' GPT Image x Seedance 2.5 clip 2.6k likes / 332k views (Oct 16, x.com/D_studioproject/status/2111000390099738889).
- Posts: @DavidRunsAds, 644 likes · @jackzhang123vip, 1,227 likes · @flowersslop, 1,067 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
- Prompts in this style: Send '6-1-2' to get the same AI model in a new scene (ChatGPT Images); 6-photo candid portrait collage in a 3x2 grid (GPT Image 2.5); Low-angle outdoor close-up portrait realism test (GPT Image 2.5); A full day of selfies as an iPhone Photos screenshot (GPT Image 2.5); Amateur front-cam extreme close-up selfie, then variations (Nano Banana Pro) — full text on the prompts page
3. Hollywood blockbuster realism
at its peak · video · also called cinematic, Hollywood blockbuster, IMAX look, teal and orange, movie still
Photoreal live-action frames graded like a studio tentpole: teal shadows against warm skin and practical fire, anamorphic flares, shallow depth and heavy atmosphere (rain, smoke, sparks). Motion is kinetic but readable: dolly pushes, whip pans, slow-motion impacts and hard-cut montages that keep one hero subject consistent.
- Prompt clause:
[subject] in an ultra-photoreal Hollywood blockbuster shot, anamorphic 2.39:1 lens, teal-and-orange grade, volumetric smoke and practical sparks, low-angle dolly push, shallow depth of field, 24fps motion blur - Keep out: flat even lighting, plastic CG skin, video-game sheen, oversaturated HDR, morphing between shots, cartoon proportions
- Best on: Seedance 2.5 (varg
seedance_2_5), Kling 4.0 (vargkling_v4), Grok Imagine (varggrok_imagine_1_5) - Trend: The default look of the Seedance 2.5 wave; cinematic/anamorphic probe since Sep 1 (>=200 likes): A 5 / 1.1k L, B 9 / 3.1k L, C 6 / 2.6k L, while single organic posts still hit 3.1k (Oct 10) and 2.4k (Sep 15). The biggest one in the window (7.0k, Oct 21) was part of a paid Flova campaign.
- Posts: @auqibhabib, 2,423 likes · @Noor_ul_ain43, 1,026 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
- Prompts in this style: Time-freeze diner spill, character-sheet locked (Seedance 2.5) — full text on the prompts page
4. Modern TV anime
at its peak · video · also called anime style, anime one-shot, anime edit, AMV
Crisp digital-cel anime: clean line art, two-tone cel shadows, saturated palettes, soft compositing glow and camera moves through 3D-assisted backgrounds. Expressive faces and comedic reaction cuts; long single shots are new with 30-second native generation.
- Prompt clause:
[subject] as a modern TV anime scene, clean digital line art, two-tone cel shading, saturated colors, soft compositing glow, dynamic camera move, expressive anime faces - Keep out: photoreal skin, plastic 3D render, muddy colors, inconsistent line weight, western cartoon proportions
- Best on: MiniMax H3 + GPT Image 2.5 character sheet (varg
minimax_h3), Kling 4.0 (native 30s) (vargkling_v4), Seedance 2.5 + Opus 5.5 (vargseedance_2_5) - Trend: 12.6k likes (Oct 4) for a MiniMax H3 anime test, 1.3k / 523k views (Oct 20) for a Kling 4.0 one-shot, plus anime-opening and edit posts (801 Oct 13, 464 Oct 23); anime probe A 7 / 5.3k L vs B 4 / 18.2k L.
- Posts: @kevinleguillou, 12,599 likes · @ShamiWeb3, 1,268 likes · @aimikoda, 801 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
5. Cross-section cutaway
rising · photo · also called cutaway illustration, coupe transversale, cross-section book, exploded view
A castle, ship or machine sliced open to show every room and mechanism at once, drawn like a classic children's reference book: fine ink lines, watercolor washes, tiny busy figures, numbered labels and a parchment page. Code variants pull a 3D model apart layer by layer.
- Prompt clause:
a detailed cross-section cutaway of [subject] showing every room and mechanism, classic illustrated reference-book style, fine ink linework with watercolor washes, tiny people at work, numbered labels, parchment background - Keep out: photoreal render, closed exterior with no interior, flat vector icons, empty rooms, illegible clutter
- Best on: image model not named (likely GPT Image 2.5 / Nano Banana) (varg
gpt_image_2_5), Claude Opus 5.5 (interactive 3D cutaway) - Trend: 12.4k likes / 1.33M views / 8.5k bookmarks for AI cross-section plates (Oct 24) plus 1.6k for an Opus 5.5 cutaway jet engine (Oct 2); very high bookmark ratio signals reuse.
- Posts: @Frenchie_, 12,415 likes · @claudeai, 1,584 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
6. Retro 90s anime
at its peak · video · also called 90s anime, retro anime, cel anime, anime opening, 80s anime
Hand-painted cels over painted backgrounds with film grain, soft halation and slightly washed-out color: dusk purples, neon city nights, mecha cockpits, sweat drops and speed lines. Framed like a TV opening, with held poses and pans across stills.
- Prompt clause:
[subject] as a 1990s cel-animated anime scene, hand-painted background, thick ink outlines, limited cel shading, film grain and soft halation, washed-out dusk palette, 4:3 TV-opening framing - Keep out: modern digital glow and particles, 3D CG look, ultra-clean lines, HDR saturation, 60fps smoothness
- Best on: Seedance 2.5 + Opus 5.5 Blender MCP (shot blocking) (varg
seedance_2_5), Midjourney + Kling 4 + Seedance (vargseedance_2_5) - Trend: Anime probe since Sep 1 (>=200 likes): A 5 / 6.1k L, B 4 / 7.9k L, C 11 / 15.5k L; retro-specific posts 4.4k (Oct 3), 1.6k (Oct 5), 1.6k (Sep 26), 3.9k (Oct 24).
- Posts: @AIWarper, 4,375 likes · @botcreates, 3,902 likes · @emmanuel_2m, 1,647 likes · @domenic, 1,567 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
7. Jelly / gummy 3D
rising · video · also called jelly world, gummy bear test, squishy 3D, Gimme Gummy
Translucent candy-colored solids with subsurface glow, glossy highlights and wobble physics: things squash, jiggle and slice cleanly. Pastel seamless backdrops, macro framing and side-by-side model comparisons.
- Prompt clause:
[subject] made of translucent jelly, glossy gummy material with subsurface glow, soft wobble and squash physics, candy colors, macro studio shot on a pastel seamless background - Keep out: opaque matte plastic, rigid motion, realistic food texture, dark moody lighting, cluttered scene
- Best on: Claude Opus 5.5 (three.js code), Claude Opus 5.5 vs GPT-6 Astra
- Trend: Jelly/gummy probe since Sep 10 (>=150 likes): A 6 posts / 6.6k L, B 9 / 5.7k L, C 5 / 4.1k L; 4.0k likes / 630k views (Oct 13), 1.1k / 489k (Oct 14). Pinterest Predicts 2026 lists 'Gimme Gummy' as a 2026 trend.
- Posts: @semichenkko, 3,977 likes · @vib3coded, 2,095 likes · @vib3coded, 1,139 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
8. Selfie UGC / AI influencer
at its peak · video · also called UGC, AI influencer, front-camera vlog, talking to camera, AI character
Vertical front-camera footage of one consistent character talking or reacting to camera: arm's-length framing, phone-mic audio, natural room light, slight wide-angle distortion and micro-shake, repeated across many everyday locations to look like a real creator's feed.
- Prompt clause:
vertical 9:16 front-camera selfie video of [subject] talking to camera at arm's length, natural room light, slight wide-angle distortion, handheld micro-shake, casual everyday location, phone microphone sound - Keep out: cinematic camera moves, studio lighting, multiple camera angles, color grading, flawless skin
- Best on: Higgsfield AI Influencer, GPT Image 2.5 character + Seedance 2.5 (Higgsfield) (varg
seedance_2_5), Seedance 2.5 (vargseedance_2_5) - Trend: AI-influencer/UGC probe since Sep 25 (>=400 likes): A 3 posts / 2.4k L vs B 15 / 21.5k L vs C 2 / 1.1k L; peaked with the Higgsfield AI Influencer launch (7.0k likes / 3.4M views Oct 2; 5.9k / 5.1M Oct 5) and the 'Jean Phil' character's 100M+ views.
- Posts: @maubaron, 1,137 likes · @Just_sharon7, 566 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
- Prompts in this style: Send '6-1-2' to get the same AI model in a new scene (ChatGPT Images); Amateur front-cam extreme close-up selfie, then variations (Nano Banana Pro); 30s phone-cam UGC at brunch, sister steals pancake (Kling 4.0) — full text on the prompts page
9. Pixel art
rising · photo and video · also called 16-bit, 8-bit, sprite art, retro game art
Low-resolution images on a hard pixel grid with limited palettes, dithered gradients and no anti-aliasing; SNES-era scenes like moonlit castles and sunsets, plus tiny sprites with 2-6 frame loops.
- Prompt clause:
[subject] as 16-bit pixel art, strict pixel grid, limited 32-color palette, dithered gradients, no anti-aliasing, retro RPG scene, crisp nearest-neighbor scaling - Keep out: blurry upscaled pixels, anti-aliased edges, painterly brush strokes, smooth gradients, mixed pixel sizes
- Best on: Claude Opus 5.5 (code-drawn pixel art), Nano Banana (sprites; 'always pick a Nano Banana based model') (varg
nano_banana_2_1) - Trend: Pixel-art probe since Sep 1 (>=200 likes): A 9 posts / 6.1k L vs B 2 / 833 L vs C 9 / 7.6k L; 2.6k (Sep 22), 1.2k (Oct 26), 838 (Oct 24), 736 (Oct 23).
- Posts: @majidmanzarpour, 2,574 likes · @asynkimo, 1,192 likes · @peekcell, 999 likes · @zer0point_eth, 736 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
10. Video-game footage look
rising · video · also called GTA style, GTA 6 style, game HUD, gameplay footage, fighting game
A real-time game-engine render with HUD overlays (minimap, mission text, health bars, wanted stars, 'FINISH HIM' banners) over a third-person or side-on camera, sun-baked saturated cities and slightly too-clean geometry. Most popular as parodies that turn everyday chores into missions.
- Prompt clause:
[subject] as third-person open-world video game footage, real-time engine render, HUD with minimap, mission text and health bar, sun-baked saturated city, chase camera behind the character - Keep out: film grain, shallow cinematic depth of field, handheld shake, no UI, painterly textures
- Best on: Seedance 2.5 (varg
seedance_2_5), Seedance + Codex + Higgsfield (vargseedance_2_5) - Trend: GTA/game-graphics probe since Sep 15 (>=300 likes): A 7 posts / 7.5k L vs B 1 / 312 L vs C 12 / 9.5k L, pushed by the GTA 6 trailer on Oct 15 (1.5k likes that day, 1.2k / 738k views Oct 21).
- Posts: @Abmankendrick, 1,221 likes · @Stefan_3D_AI, 885 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
11. Isometric wireframe
at its peak · photo and video · also called isometric SVG, technical isometric, exploded isometric, isometric objects skill
Objects in true 30-degree isometric projection drawn with thin monoline strokes, mostly white and light grey with one accent color, often exploded into layers that assemble with smooth eased motion. Reads like premium hardware or SaaS product illustration.
- Prompt clause:
[subject] as an isometric wireframe illustration, true 30-degree isometric projection, thin monoline strokes, white and light-grey palette with one accent color, exploded layers, clean technical look - Keep out: perspective distortion, sketchy uneven lines, heavy shading, photoreal textures, cluttered labels, multiple accent colors
- Best on: Claude Opus 5.5 (SVG skill), Claude Opus 5.5 + anatomy skill
- Trend: Cluster of 2.4k (Oct 3), 2.3k (Oct 11), 1.1k (Oct 8) like posts; isometric/cutaway probe since Sep 1 (>=200 likes): A 5 posts / 9.5k L, B 11 / 11.4k L, C 5 / 1.3k L.
- Posts: @wheresryan22, 2,428 likes · @0xMovez, 2,334 likes · @itsnotryan, 1,118 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
12. Flat vector science explainer
rising · video · also called Kurzgesagt style, Kurzgesagt-like, flat vector animation, science explainer
Flat geometric vector shapes with soft inner gradients on deep navy or space-blue backgrounds, rounded friendly characters (birds, planets, cells), glowing accent rims and big numeric callouts. Motion is smooth parallax, scaling and orbiting rather than camera moves.
- Prompt clause:
[subject] explained in a flat vector science-explainer animation, geometric shapes with soft gradients, deep navy background with glowing accents, rounded friendly characters, big number callouts, smooth parallax motion - Keep out: photoreal footage, sketchy hand-drawn lines, dense text blocks, harsh black outlines, 3D realism, muddy palette
- Best on: Claude Opus 5.5 (code), Claude Opus 5.5 + HyperFrames
- Trend: New this week: 3.0k likes / 236k views (Oct 25) and 1.1k / 332k (Oct 26) for 'recreate a science-explainer channel with Claude' posts; no comparable posts before Oct 25.
- Posts: @finnardley, 3,022 likes · @WaterrrMiao, 1,116 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
13. 3D animated family film
at its peak · video · also called Pixar style, Disney 3D, 3D cartoon, 3D toddler video, nursery-rhyme 3D
Big-eyed stylized characters with soft subsurface skin, rounded proportions and plush fabrics, lit like a studio feature with a warm key, bounce fill and rim light. Bright clean palettes and gentle comic timing; the kids-channel variant adds bold educational text and nursery-rhyme pacing.
- Prompt clause:
[subject] as a stylized 3D animated family-film character, big expressive eyes, rounded proportions, soft subsurface skin, warm key light with rim light, bright clean palette, gentle comedic timing - Keep out: photoreal humans, uncanny skin pores, dark gritty grade, stiff mocap motion, anime line art
- Best on: Seedance 2.5 (varg
seedance_2_5), Nano Banana Pro + Seedance 2.5 + HyperFrames (vargseedance_2_5) - Trend: 3D-animation probe since Sep 1 (>=200 likes): A 1 / 270 L, B 13 / 4.7k L, C 6 / 2.7k L; peaked late Sep to early Oct with toddler/educational channels (1.5k Sep 29, 1.4k Oct 4, 1.4k / 544k views Sep 26) and is still posted daily at lower engagement.
- Posts: @Zarnab_with_Ai, 1,481 likes · @AngryTomtweets, 1,424 likes · @Zarnab_with_Ai, 1,379 likes · @ArchiveExplorer, 1,314 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
- Prompts in this style: 3D family-film ostrich tries to fly 3 times, sheet-locked (Seedance 2.5); Baby dragon burns the roast, learns a tiny flame (Seedance 2.5) — full text on the prompts page
14. 2000s digicam flash photo dump
steady · photo · also called digicam, Canon G7X look, CCD camera, point-and-shoot flash, Y2K flash photo, photo dump grid
Direct built-in flash from a compact digital camera: bright flat faces, hard shadows on the wall behind, slightly blown highlights, cool digital noise in the dark areas and tilted, imperfect framing. Usually delivered as a 3x2 or 3x3 grid of candid frames from one session.
- Prompt clause:
candid photo of [subject] taken with a 2000s compact digital camera, harsh direct built-in flash, hard shadow on the wall behind, slight overexposure, digital noise, tilted imperfect framing, presented as a 3x2 photo-dump grid - Keep out: softbox studio light, smooth retouched skin, portrait-lens bokeh, cinematic color grade, perfect symmetry
- Best on: GPT Image 2.5 (ChatGPT) (varg
gpt_image_2_5), GPT Image 2 (ChatGPT) (varggpt_image_2_5) - Trend: Steady weekly grid prompts: flash/digicam probe since Sep 1 (>=150 likes): A 6 posts / 2.2k L, B 3 / 780 L, C 11 / 3.9k L; 827 likes Oct 15, 649 Oct 21.
- Posts: @MissDelulu9, 827 likes · @john_my07, 649 likes · @MissDelulu9, 595 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
15. Vintage TV broadcast
rising · video · also called 1960s TV show, 80s commercial, technicolor sitcom, retro TV pilot, old TV ad
Footage that looks taped off an old broadcast: soft tube-camera image with bloom, saturated period palette (avocado, mustard, teal), multi-camera sitcom staging on bright sets, title cards and announcer framing. The 1980s variant adds VHS chroma noise and beige computer-era props.
- Prompt clause:
[subject] as a 1960s color TV show, multi-camera sitcom set with saturated period colors, soft tube-camera image, slight bloom and scanlines, 4:3 frame, vintage title card - Keep out: modern widescreen, digital sharpness, contemporary clothes or phones, cinematic shallow depth
- Best on: Seedance (varg
seedance_2_5), Seedance 2.5 (vargseedance_2_5) - Trend: Small but new: a retro-decade TV/commercial probe since Sep 15 (>=300 likes) finds 3 posts, 2 of them after Oct 13 (1.3k Oct 22, 1.5k Oct 15), helped by Halloween retro-horror crossovers.
- Posts: @SeharShinwari, 284 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
16. Luxury product commercial
steady · photo and video · also called product splash, perfume ad, beverage commercial, hero product shot
One product as the hero on a dark or saturated seamless: high-speed liquid splashes and droplets frozen mid-air, crisp rim light along glass edges, macro detail, and slow push-ins or orbiting moves.
- Prompt clause:
[subject] as the hero of a luxury commercial, high-speed liquid splash frozen mid-air, crisp rim light along glass edges, dark glossy seamless background, macro detail, slow orbiting camera - Keep out: cluttered background, flat lighting, warped or misspelled labels, extra products, handheld shake
- Best on: Seedance 2.5 (varg
seedance_2_5), Nano Banana 2 + Gemini Omni Flash 1.1 (vargomni_flash_1_1) - Trend: Product-commercial probe since Sep 25 (>=300 likes): A 5 posts / 2.8k L, B 12 / 5.0k L, C 3 / 1.4k L; reliable 300-700 like prompt shares.
- Posts: @Strength04_X, 663 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
- Prompts in this style: POV luxury furniture commercial from a storyboard (Gemini Omni) — full text on the prompts page
17. Stop-motion craft (clay, felt, knit)
steady · photo and video · also called claymation, stop motion, needle felt, knitted puppet, stitched stop-motion, 粘土风格
Handmade miniatures with fingerprints in clay, fuzzy felt fibers or knit stitches, on tabletop sets with a soft key light and shallow macro focus. Motion is stepped at about 12fps with slight 'boiling' between frames.
- Prompt clause:
[subject] as a handmade stop-motion puppet of plasticine, needle felt or knitted yarn, visible fingerprints and fibers, tabletop miniature set, soft key light, macro shallow depth of field, 12fps stepped stop-motion movement - Keep out: smooth CGI surfaces, fluid 60fps motion, photoreal skin, glossy plastic, sharp vector edges
- Best on: GPT Image 2.5 (frames) (varg
gpt_image_2_5), Kling 4.0 (claymation) (vargkling_v4) - Trend: Sep peak (7.0k likes / 1.0M views Sep 8) then steady posts; craft/stop-motion probe since Sep 1 (>=150 likes): A 6 / 1.9k L, B 3 / 688 L, C 11 / 9.5k L. Disney Animation's Oct 21 'Clay' news (1.8k likes, not AI) adds attention.
- Posts: @charlierguo, 7,000 likes · @cheerselflin, 380 likes · @oggii_0, 326 likes · @reach_vb, 338 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
18. Paper-cut / papercraft
steady · photo and video · also called paper cut-out, 剪纸风格, papercraft toy, layered paper, pop-up book
Flat shapes cut from matte colored paper, layered with small drop shadows to fake depth, visible fiber texture and slightly imperfect edges. Ranges from red Chinese paper-cut silhouettes to fold-up cube toys and cut-out animation with hinged limbs.
- Prompt clause:
[subject] made of layered cut paper, matte colored cardstock with visible fiber texture, soft drop shadows between layers, slightly imperfect hand-cut edges, flat frontal composition - Keep out: glossy 3D render, photographic textures, airbrushed gradients, realistic lighting falloff
- Best on: GPT Image 2.5 (varg
gpt_image_2_5), GPT Image 2 (ChatGPT) (varggpt_image_2_5) - Trend: Posts every couple of weeks without a breakout: 1.2k likes (Sep 19), 1.1k (Oct 21, 剪纸), 493 (Sep 3), 459 (Sep 5).
- Posts: @icreatelife, 1,178 likes · @daaihq, 1,110 likes · @azed_ai, 459 likes · @xiaoxiaodong, 493 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
- Prompts in this style: Turn your pet into a printable papercraft cube toy (GPT Image 2.5); Second World: photo continues into a paper-cut world (Nano Banana 2.1) — full text on the prompts page
19. Hand-painted anime film
steady · photo and video · also called Ghibli style, Ghibli-like, 2D animation, golden age 2D
Pastoral, watercolor-like painted backgrounds with towering cumulus skies, soft daylight and gentle colors; characters drawn with simple rounded lines and minimal shading. Calm everyday moments such as bike rides, seaside towns and meals, with lots of air in the frame.
- Prompt clause:
[subject] in a hand-painted 2D animated film still, watercolor-like painted background, big cumulus clouds, soft natural daylight, simple rounded character lines, gentle pastoral palette, calm wide composition - Keep out: 3D render, harsh neon, heavy black outlines, photoreal textures, busy action composition
- Best on: Kling, Seedance + Dreamina (varg
seedance_2_5) - Trend: Evergreen since the 2025 wave; 2D/hand-drawn probe since Sep 1 (>=300 likes): A 1 / 1.5k L (news), B 4 / 7.0k L, C 15 / 11.3k L; 1.5-1.6k like posts in September, 2.6M views for a 2D-animation reel (Sep 27).
- Posts: @q_a_i_m_01, 1,525 likes · @knightama, 1,460 likes · @coryaslan, 1,580 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
20. Painterly storybook
steady · photo · also called Midjourney painterly sref, --sref 2120427644::4 837554414::2, --sref 2116795492, gouache fantasy
Thick gouache-like brushwork with visible strokes, lush greens and soft blue shadows, small figures and animals inside big quiet landscapes, like a picture-book spread. Edges are textured and slightly crisp from blended style references.
- Prompt clause:
[subject] as a painterly storybook illustration, thick gouache brush strokes, lush greens and soft blue shadows, a small figure in a vast quiet landscape, textured paper, warm afternoon light - Keep out: photoreal detail, smooth airbrushed gradients, harsh black outlines, neon colors, 3D render look
- Best on: Midjourney v8.2 with --sref blends
- Trend: Midjourney sref probe since Sep 1 (>=150 likes): A 5 posts / 1.3k L, B 4 / 2.9k L, C 11 / 6.1k L; top painterly sref sets 1.8k (Oct 2) and 1.0k (Sep 20).
- Posts: @GlitterPixely, 1,823 likes · @GlitterPixely, 1,001 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
21. Risograph / grain print
steady · photo · also called risograph, riso, halftone, duotone poster, screen print
Two or three flat spot inks (fluoro pink, cobalt, lavender) with grainy ink texture, slight misregistration and halftone dots. Oversized simple shapes and bold figures on off-white paper, often with tiny sans-serif captions.
- Prompt clause:
[subject] as a two-color risograph print, fluorescent pink and cobalt blue spot inks, grainy ink texture, slight misregistration, halftone shading, oversized simple shapes, off-white paper - Keep out: smooth digital gradients, photoreal detail, more than three colors, glossy 3D, perfectly clean vector edges
- Best on: Midjourney (sref exploration), GPT Image 2.5 (halftone travel collage) (varg
gpt_image_2_5) - Trend: Steady mid-size posts: 1.7k (Oct 1), 780 (Sep 12), 497 (Sep 16); risograph/duotone probe since Sep 1 (>=100 likes): A 3 / 1.4k L, B 3 / 2.6k L, C 14 / 3.2k L.
- Posts: @VigoCreativeAI, 1,705 likes · @GabeTognon, 780 likes · @Naiknelofar788, 497 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
- Prompts in this style: Halftone / ASCII architectural travel-collage poster (GPT Image 2.5) — full text on the prompts page
22. Chibi sticker pack
steady · photo · also called LINE stamps, LINEスタンプ, kawaii stickers, chat stickers, emoji pack
A grid of one character in 8 to 16 exaggerated expressions: big head, tiny body, thick white die-cut border, flat pastel colors and a short reaction word in each sticker, on a clean white sheet.
- Prompt clause:
a sticker sheet of [subject] as a chibi character in 12 different expressions, big head and small body, thick white die-cut outline, flat pastel colors, short reaction text, clean white background, the same character in every sticker - Keep out: realistic proportions, painterly shading, busy background, inconsistent character design, misspelled text
- Best on: ChatGPT image (GPT Image 2/2.5) (varg
gpt_image_2_5) - Trend: Sep 1 LINE-stamp wave (1.3k likes / 341k views) then steady 400-500 like prompt shares; sticker probe since Sep 1 (>=200 likes): A 4 / 3.9k L, B 3 / 828 L, C 13 / 5.8k L.
- Posts: @halukik_0520, 1,281 likes · @azed_ai, 528 likes · @rascal8023, 497 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
23. Manga page
steady · photo · also called AI manga, 漫画, コマ割り, comic panels
Multi-panel page layouts with speech balloons and SFX lettering, black-and-white ink with screentone or limited color, dynamic panel shapes, speed lines and exaggerated faces. AI posts focus on keeping one character consistent across panels, often from a rough layout sketch.
- Prompt clause:
[subject] as a manga page, 5-panel layout with varied panel shapes, black-and-white ink with screentone shading, speed lines, speech balloons with Japanese lettering, the same character consistent across panels - Keep out: single illustration, photoreal shading, western comic coloring, inconsistent faces between panels, gibberish lettering
- Best on: GPT Image 2.5 (ChatGPT) with a panel-layout reference (varg
gpt_image_2_5), GPT Image 2.5 + Claude manga skill (varggpt_image_2_5) - Trend: Manga/comic probe since Sep 1 (>=200 likes): A 4 posts / 1.5k L, B 3 / 944 L, C 13 / 5.7k L; steady JP/ZH creator posts at 250-800 likes.
- Posts: @blogad55266101, 789 likes · @UNfukashigi, 254 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
24. FPV drone one-take
steady · video · also called FPV, drone fly-through, one continuous shot, POV ride
One uninterrupted first-person flight that dives, banks and threads gaps between buildings, under bridges and over cliffs, with ultra-wide 14-16mm distortion, edge motion blur and horizon roll. Usually golden hour or neon night, with speed ramps instead of cuts.
- Prompt clause:
one continuous FPV drone shot flying through [subject], fast dives and banking turns, threading narrow gaps, ultra-wide 14mm lens, motion blur at the frame edges, golden-hour light, no cuts - Keep out: static tripod framing, cuts or crossfades, warping buildings, weightless camera without inertia
- Best on: Seedance 2.5 (varg
seedance_2_5), Seedance 2.5 (FishCreative) (vargseedance_2_5) - Trend: Mid-size posts every week, no breakout: FPV/one-take probe since Sep 1 (>=200 likes): A 5 / 1.1k L, B 9 / 3.1k L, C 6 / 2.6k L.
- Posts: @Naiknelofar788, 438 likes · @abxxai, 394 likes · @q_a_i_m_01, 377 likes
- Pinterest board: pin 1 · pin 2 · pin 3
25. Surreal giant scale
steady · photo and video · also called giant hand, giant objects, impossible scale
Ordinary objects or body parts at landscape scale (a giant hand lifting a flooded car, a lime as big as a wave), rendered photoreal with consistent light and atmospheric haze to sell the size.
- Prompt clause:
a photoreal surreal scene where [subject] is enormous, towering over tiny people and vehicles, consistent natural light, atmospheric haze for scale, wide-angle low viewpoint - Keep out: inconsistent shadows, cartoon rendering, cramped framing, no scale references
- Best on: model not named, Seedance 2.5 / Kling (Flova) (varg
seedance_2_5) - Trend: Low, steady volume: giant/scale probe since Sep 1 (>=200 likes): A 2 posts / 448 L, B 4 / 963 L, C 2 / 897 L; best post 1.5k likes / 422k views (Sep 23).
- Posts: @ElaraVisionAI, 1,524 likes · @ShamiWeb3, 788 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
26. ASMR cutting
steady · video · also called glass fruit cutting, satisfying cutting, crystal ASMR
Macro close-ups of a knife slicing unexpected materials (glass fruit, crystal, gummy, wax) on a cutting board, with clean soft light, slow controlled motion and crisp foley.
- Prompt clause:
macro close-up of a sharp knife slowly slicing [subject] made of translucent glass on a wooden board, clean soft light, satisfying smooth cuts, crisp ASMR cutting sounds - Keep out: fast cuts, shaky camera, messy background, distorted hands, muffled audio
- Best on: model not named
- Trend: Still posted (477 likes / 61k views Oct 23, 436 / 51k Oct 25) but far below its 2025 peak; ASMR/cutting probe since Sep 1 (>=200 likes): A 3 posts / 1.1k L, B 7 / 8.4k L (mostly non-ASMR), C 10 / 7.9k L.
- Posts: @Andres_Ai5, 477 likes · @Andres_Ai5, 436 likes
- Pinterest board: pin 1 · pin 2
27. Photo-to-editorial art poster
fading · photo · also called split poster, photo + illustration poster, minimalist lifestyle art poster, independent art publication cover
The original photo kept intact (top half or inset) beside a minimalist redraw of the same scene: ink lines or three to four flat colors on rough off-white paper. Lots of negative space and small handwritten or editorial type such as a title, coordinates or 'diary'.
- Prompt clause:
an editorial art poster of [subject]: top half the original photo, bottom half a minimalist hand-drawn illustration of the same scene on rough off-white paper, max 4 flat colors, generous negative space, small handwritten title and coordinates - Keep out: busy collage, heavy typography, full-bleed illustration, glossy gradients, changing the subject's face, extra decorations
- Best on: GPT Image 2.5 (ChatGPT) (varg
gpt_image_2_5), Nano Banana 2.1 (vargnano_banana_2_1) - Trend: Peaked in September (1.7k Sep 19, 1.0k Sep 2, 994 Sep 9); poster probe since Sep 1 (>=200 likes): A 3 posts / 934 L, B 1 / 2.1k L, C 16 / 7.4k L.
- Posts: @Naiknelofar788, 1,698 likes · @MissDelulu9, 1,035 likes · @harboriis, 417 likes · @ChillaiKalan__, 414 likes · @MahnoorAi12, 465 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
- Prompts in this style: Photo-to-watercolor split travel poster (GPT Image 2.5); Halftone / ASCII architectural travel-collage poster (GPT Image 2.5); Selfie -> double-exposure travel silhouette poster (GPT Image 2.5); Premium editorial poster from each uploaded photo (GPT Image 2.5); Photo + black-ink illustration split editorial poster (Nano Banana 2.1) — full text on the prompts page
28. Watercolor travel journal
fading · photo · also called photo + watercolor vignette, travel sketchbook, urban sketch, travel field notes
Loose ink lines with transparent watercolor washes on cream paper, unfinished edges that fade to white, handwritten captions, dates and small stamps. Usually shown with the source photo above to sell the 'photo to sketchbook page' transformation.
- Prompt clause:
[subject] as a travel-journal page, loose ink linework with transparent watercolor washes, unfinished edges fading into cream paper, handwritten caption and date, small stamp, generous white space - Keep out: full-bleed digital painting, saturated flat colors, heavy outlines, photoreal rendering, cluttered layout
- Best on: GPT Image 2 (ChatGPT) (varg
gpt_image_2_5) - Trend: Concentrated in early September (994 likes Sep 9, 438 Sep 6, 417 Sep 7); the same probe finds few October posts (A 3 / 934 L vs C 16 / 7.4k L across the photo-to-art poster family).
- Posts: @Sairah_0, 994 likes · @Taaruk_, 438 likes · @MissDelulu9, 417 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
- Prompts in this style: Photo-to-watercolor split travel poster (GPT Image 2.5) — full text on the prompts page
29. Tiny people, giant world
fading · photo and video · also called miniature world, tilt-shift, tiny world, micro adventure
Small figures in an everyday environment shot like macro photography: coffee cups as skateparks, leaves as rooftops. Tilt-shift blur, crisp texture on the giant object and warm natural light.
- Prompt clause:
a tiny miniature [subject] in a giant everyday world, macro lens, tilt-shift shallow depth of field, oversized everyday objects as landscape, warm natural light, crisp material texture - Keep out: human-scale perspective, deep focus everywhere, flat lighting, cartoon rendering
- Best on: Seedance 2.5 (varg
seedance_2_5), Seedance 2.5 (ElevenCreative) (vargseedance_2_5) - Trend: Tiny/miniature probe since Sep 10 (>=150 likes): A 1 post / 173 L, B 2 / 658 L, C 17 / 8.2k L; the September 'your selfie as a miniature world' prompt wave has cooled.
- Posts: @itxabdullaa, 352 likes · @harboriis, 227 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
30. Early-2000s DV camcorder
fading · video · also called MiniDV, 2000s home video, camcorder footage, POV: you're in the 2000s, Y2K camcorder
4:3 standard-definition video with soft detail, interlaced edges, slightly crushed blacks and cool fluorescent or warm tungsten color, plus handheld wobble and on-camera mic sound. Ordinary moments (birthdays, convenience stores, schoolyards) staged as personal memories, often with a date stamp.
- Prompt clause:
[subject] filmed on an early-2000s MiniDV camcorder, 4:3 standard definition, handheld wobble, soft focus and interlacing, slight color bleed, on-camera light, orange date stamp in the corner, candid home-video moment - Keep out: 4K sharpness, cinematic depth of field, modern phone HDR, film-grade color, widescreen
- Best on: Seedance 2.5 (varg
seedance_2_5), Seedance 2.5 on Higgsfield (vargseedance_2_5) - Trend: Peaked late Sep to early Oct (2.9k likes Oct 2, 918 Sep 30); DV/camcorder probe since Sep 10 (>=100 likes): A 3 posts / 700 L, B 8 / 4.6k L, C 9 / 1.6k L.
- Posts: @saniaspeaks_, 918 likes · @AIwithkhan, 348 likes · @SimplyAnnisa, 273 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
31. CCTV / dashcam footage
fading · video · also called security cam footage, doorbell cam, bodycam, dashcam, surveillance POV
A fixed high-corner or windshield-level wide lens with low dynamic range, blown highlights, compression blocking and a burned-in timestamp or camera ID. Desaturated green or sodium color, barrel distortion and no operator: events just happen in frame, which sells absurd incidents as real.
- Prompt clause:
[subject] captured on a fixed CCTV security camera, high corner angle, wide fisheye distortion, low-resolution compressed video, timestamp and CAM 03 overlay, flat fluorescent light, slight frame drops - Keep out: cinematic grade, moving camera, shallow depth of field, clean 4K detail, music-video cuts
- Best on: GPT Image 2.5 first frame + Seedance 2.5 (varg
seedance_2_5), Seedance 2.5 (via Magnific) (vargseedance_2_5) - Trend: September wave (22.5k likes / 64.4M views Sep 17; 3.4k Sep 8); a probe since Sep 29 (>=100 likes) returns 7 posts, 5 of them news about a fake AI 'CCTV' clip in Delhi and only one creative post (a doorbell-cam Halloween clip, 204 likes).
- Posts: @Stephan_MAJ, 3,362 likes · @rourke_heath, 858 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
32. Liminal / Backrooms
fading · photo and video · also called backrooms, liminal space, poolrooms, dreamcore, analog horror
Empty, endlessly repeating interiors (yellow wallpaper and damp carpet under buzzing fluorescents, or white-tiled pools) with no people, flat even light and a faint wrongness. Video versions are VHS-degraded found footage with a timestamp and slow handheld walking.
- Prompt clause:
an empty liminal space of [subject], endless repeating rooms, buzzing fluorescent ceiling light, yellowed walls and damp carpet, no people, flat even lighting, VHS camcorder found-footage texture with timestamp - Keep out: people or crowds, dramatic colored light, clutter, clean cinematic grade, recognizable landmarks
- Best on: GPT-6 Astra (Blender scene + VHS effects), Seedance 2.0 + Nano Banana Pro + Seedream (varg
nano_banana_pro), Claude Opus 5.5 (code) - Trend: September spike around the Backrooms feature film (9.5k likes / 6.4M views Sep 4; 2.1k / 1.4M Sep 18); probe since Sep 1 (>=100 likes): A 7 posts / 1.7k L, B 2 / 1.0k L, C 11 / 16.7k L.
- Posts: @duncantrussell, 9,486 likes · @goodside, 331 likes · @kepochnik, 254 likes
- Pinterest board: pin 1 · pin 2 · pin 3 · pin 4 · pin 5 · pin 6
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.
- Claude Code:
claude plugin marketplace add vargHQ/skills && claude plugin install varg@varg(skill + hosted MCP) - Codex, Cursor, Gemini CLI, OpenClaw:
npx skills add vargHQ/skills --skill varg-ai -a <agent> -y - Claude.ai, ChatGPT, any MCP client: connect
https://mcp.varg.ai/mcp - API key: https://varg.ai/dashboard/api · Model specs: https://varg.ai/models/llms.txt