Andrei Zdetoveţchi

Since it's Father's Day, this could work well. It's about a father trying to reunite with his daughter :)

About a week ago, I tested the workflow from @aimikoda and made my own version of Jeddica. Now, I added a few more cuts, slapped a song of mine from Suno and decided to publish it. Made with Seedance 2.0 Fast on @Hailuo_AI and @dreamina_ai

@aimikoda but I think I'll stick with Codex, for a while, since I've discovered that it can create the images (charsheet and storyboard) directly there :)

@azdeto I'll share it soon. Just upload the file to Claude and ask it to add it as a skill. It should adjust and integrate it for you. You can also ask it to make it Claude-compatible or fix whatever error you're getting.

Bad Taste GPT Image 2 + Seedance 2.0 workflow and prompts in the replies.

@aimikoda lol, that was easier than expected. First I tried to add it in the app and gave some error about xml codes. Then I uploaded in the chat and ask it to install. But it didn't put among the existing skills so hallucinated. Last I put in the Code tab and it worked :)

@Alin_Reaper05 las' ca ma lamuresc eu pana la urma. Voiam doar sa am o idee cum functioneaza treaba ca sa nu cheltui aiurea pe alea de le-am primit la inregistrare :))

@azdeto Habar nu am, sunt putin confuz acuma 😂 cred ca 800 credite normale, asa ar fii logic...dar, probabil 8k credite de hub da 😁

Hello there, everyone! Minimax Hub is here, and I made a very short documentary about alien life with it in just a few minutes... I know the voice is a bit off, but this is just the beginning. 😉

@Alin_Reaper05 Da, vad ca rata de conversie e de 1:10. Deci e normal. Deci tie ti-a consumat 8k de hub credits?

@azdeto Da, nu stiu daca sunt cumva diferite, ca am vazut ca am luat de pe Hailuo app si am transferat aici 3k credite si mi-a dat 30k 😁

@Alin_Reaper05 pfff... mda. eu abia am 3000. tre sa ma gandesc la altceva 🤣

@azdeto Aoleu, acuma vad, mi-a luat Seedance 2.0 8100 credite...hmm, cred ca e mai okay fara hub 😅

@mitte_ai @DaraChaww actually, is $3.26 - I made the math on the anual plan on the Creative. So for monthly subs, the cost is even higher.

@mitte_ai @DaraChaww actually I see different numbers... 2041 credits. And if I change the aspect ratio to 16:9, I get 3629. That's about $2.268.

Seedance 2 is now 25% cheaper on Mitte. Best price anywhere. Plus tooling built for filmmakers: face references and voice cloning, both supported natively. Create your next film on Mitte:

@aimikoda Care to elaborate more about image replication system for NBP? Also, how do you add your storyboard skill to Claude, I tried to add it in their Windows app and got some error...?

Actually, I used an image replication system I created for Nano Banana Pro to generate a prompt that recreates a visual I found on Pinterest with different characters, and I used that prompt in GPT Image 2. Prompt: CHARACTER = skinny teenage scavenger with cyan twin-tail hair CREATURE = colossal stitched kaiju plush guardian PALETTE = electric cyan, neon pink, acid yellow, deep violet BACKGROUND = flat turquoise sky DO: Place CHARACTER standing in the foreground with a tired, annoyed expression. Give CHARACTER a unique silhouette separate from CREATURE. Add worn streetwear, oversized geometric shirt, bandages, scratches, patched sleeves, and cyberpunk accessories. Position CREATURE looming behind CHARACTER, towering over her like a living mountain. Design CREATURE as a gigantic patchwork beast, not humanoid. Give CREATURE a bulky asymmetrical body, mismatched horns, stitched seams, fabric patches, exposed stuffing, and oversized claws. Create huge glowing yellow eyes with different shapes and proportions. Open CREATURE's massive mouth behind CHARACTER as a dramatic backdrop. Fill the mouth with broken stone-like teeth and deep shadow. Show signs of age through tears, repairs, stitches, and patched fabric. Make CREATURE feel protective rather than aggressive. Use sketchy linework, rough fabric folds, and graphic shapes throughout. Push the color contrast with bold cyan, pink, and yellow accents. STYLE: punk anime illustration, street-art poster, graphic novel cover, hand-drawn ink lines, flat cel shading, highly stylized CAMERA: low-angle medium-full shot, centered composition, slight wide lens MOOD: lonely, rebellious, strange friendship, post-apocalyptic whimsy RULES: CHARACTER and CREATURE must have completely different silhouettes. CHARACTER must feel human and fragile. CREATURE must feel ancient, gigantic, and plush-like. CREATURE occupies most of the frame. Colors remain highly saturated and graphic. No realistic rendering. NO: matching facial features, matching proportions, humanoid monster, photorealism, cinematic lighting, gradients, text, logos, watermarks, extra characters ---- Then I created a storyboard from that image (Shared the storyboard skill with my subscribers. ). GPT Image 2 Prompt for Storyboard: Create a 16:9 image. [PROJECT CARD] Create a compact designed masthead, not a table. TITLE: BAD TASTE META LINE: absurd dread -> slapstick horror-comedy / locked-off single take PRIORITY: identical fixed framing across all panels; the gag reads through change inside one unmoving frame MICRO BRIEF: a girl stands alone in a locked frame; a giant plush monster rises behind her, swallows her whole, fails to digest her, and spits her back out drenched in drool. [CONTINUITY HEADER] SEQUENCE ID: GULP-01 REFERENCE PRIORITY: identity reference controls character identity; this storyboard controls staging, motion, geography, continuity. [SCENE PACKET] PREMISE: a locked camera watches a girl get swallowed whole by a giant patchwork plush monster and spat back out when his stomach rebels. LOCATION: flat open cracked stone ground, empty flat backdrop, no walls, no props; usable surfaces are ground plane only; the monster occupies the full background when present. START -> END: C1 stands alone at frame center, bored posture, monster absent -> C2 slumps queasy in background, C1 sprawled in foreground center soaked in thick drool strands, frame otherwise unchanged. ACTION CHAIN: empty calm -> C2 rises from behind ground line filling background -> jaws lunge down and engulf C1 in one gulp -> mouth closed, throat and belly bulge as C2 works to swallow -> C2's face sours, cheeks balloon, body heaves with nausea -> C2 retches and ejects C1 forward; C1 lands dripping, drool strands connecting her to the open mouth. PROP / EFFECT STATE: drool/saliva is the key effect: absent P01-P02, strands inside jaws P03, dribble from sealed lips P04, leaking heavily P05, massive spray plus strands and puddle around C1 in P06; C1's clothing dry P01-P03, soaked and matted P06. MUST READ: the camera never moves; only the monster and the girl change inside one identical fixed frame. [CHARACTER SANITIZATION] C1: young woman, slim small silhouette, very long twin-tail hair, oversized boxy graphic tee over shorts, mismatched tall socks, chunky platform sneakers, adhesive bandages on face and legs, slouched deadpan posture, light loose movement. C2: colossal round patchwork plush monster, body wider than frame, mismatched stitched fabric patches, two stubby striped horns, one flat button left eye, one wide slit-pupil right eye, enormous mouth of large blunt jagged teeth, stubby clawed paws, heavy lumbering movement. Remove contradictory traits, invisible psychology, excessive costume detail, and backstory that cannot appear in a panel. [IDENTITY CONSISTENCY] Identity reference controls face, body, wardrobe, and proportions for both characters; keep C1 and C2 IDs, silhouettes, wardrobe, key props, and screen positions consistent across all panels; C1 stays frame center, C2 fills the background; do not redesign or merge characters. [STORYBOARD PURITY] Panel images are visual-only low-detail monochrome light-gray rough sketches. Put panel numbers, beat names, and lens tags in the header strip outside each panel image. No color, labels, arrows, captions, subtitles, logos, watermarks, timing marks, diagrams, UI, ghost poses, duplicate bodies, or technical overlays inside panels. [MASTER SHOT RULE] Panel 01 is the master: full geography of the locked frame, C1 small at center on the ground line, empty background, generous headroom where C2 will later rise. Every later panel keeps this exact geography. [EMOTIONAL ARC] bored stillness -> looming dread -> violent engulfment -> uneasy strained swallowing -> rising disgust and nausea -> messy comic rejection; read through C1's slack posture vanishing into scale contrast, C2's eye and cheek changes, and the growing drool state in an unchanging frame. [STYLE LOCKS] STYLE LOCK: final video style is vibrant patchwork anime illustration: bold dark linework, saturated cyan-magenta-purple-yellow palette, stitched-fabric texture on the monster, grungy distressed detail, flat solid cyan backdrop, flat graphic lighting with minimal shading. EFFECT LOCK: drool reads as thick glossy viscous strands and ropes with soft highlights, stretchy and heavy, never mist or spray particles; consistent thickness and stringy behavior across panels. ENVIRONMENT LOCK: flat cracked stone ground plane and plain flat backdrop only; no set redesign, no added props, no horizon change; same ground cracks in every panel. [SPATIAL CONTINUITY LOCK] P01 through P06 share one identical locked-off camera setup: same low wide frontal angle, same lens, same framing, same ground line, same crack pattern, zero camera movement, zero cuts. No panel is a new establishing shot. Locked anchors: C1's center floor mark, ground cracks, flat backdrop, frame edges. Only allowed changes: C2 entering/filling the background, character poses, mouth open/closed state, drool state, C1 visible/inside/ejected, damage and wetness on C1. [DIRECTOR STRIP] Bottom animatic track board aligned to panel columns. Tracks: BEAT LINE, CAMERA PATH, ACTION PATH, RHYTHM TRACK, ESCALATION MAP, STATE TRACK, STYLE TRACK. Use shot chips, thin lines, rhythm blocks, small intensity bars, one-to-three-word labels. No seconds or timestamps. RHYTHM TRACK format: `RHY P##: [hold|slow reveal|build|burst|impact|pause|recover|final hit] / [short block|medium block|long block] / [clean beat|match beat|smash beat|held beat|whip beat]`. ESCALATION MAP format: `ESC P##: [L1 calm|L2 tension|L3 rise|L4 surge|L5 peak] / [flat|rise|spike|drop|release|unresolved]`. PANEL HEADERS: P01 / 24mm wide / Empty master -> P02 / 24mm wide / Monster rises -> P03 / 24mm wide / One-gulp swallow -> P04 / 24mm wide / Straining to digest -> P05 / 24mm wide / Nausea turn -> P06 / 24mm wide / Spit-out finale CAMERA + LENS PLAN: P01 locked-off low wide, same-lens -> P02 same locked frame, hold -> P03 same locked frame, hold -> P04 same locked frame, hold -> P05 same locked frame, hold -> P06 same locked frame, final hold ACTION PATH: P01 C1 stands slack at center, alone -> P02 C2 looms up from behind ground line, filling background above C1, jaws parting -> P03 C2's open jaws crash down over C1, swallowing her in one gulp, drool strands snapping -> P04 mouth sealed, C2 upright, throat and belly bulging as he gulps and grinds -> P05 C2's cheeks balloon, eyes wince and water, body buckles forward in nausea, drool leaking through teeth -> P06 C2 retches and ejects C1 forward; C1 lands sprawled at center foreground coated in drool, thick strands stretching back to C2's hanging-open mouth RHYTHM TRACK: P01 RHY: hold / medium block / held beat -> P02 RHY: slow reveal / medium block / held beat -> P03 RHY: burst / short block / smash beat -> P04 RHY: pause / long block / held beat -> P05 RHY: build / medium block / match beat -> P06 RHY: final hit / short block / smash beat ESCALATION MAP: P01 ESC: L1 calm / flat -> P02 ESC: L2 tension / rise -> P03 ESC: L4 surge / spike -> P04 ESC: L3 rise / flat -> P05 ESC: L4 surge / rise -> P06 ESC: L5 peak / release STATE TRACK: P01 C1 dry, frame empty behind -> P02 C2 enters background, C1 unaware -> P03 C1 inside jaws, drool strands active -> P04 C1 hidden, bulge state, lip dribble -> P05 bulge high, heavy leak, queasy face -> P06 C1 ejected soaked, drool puddle, C2 slumped sick STYLE TRACK: P01 flat cyan calm -> P02 looming patchwork mass -> P03 jaw spike -> P04 strained bulge -> P05 queasy green tint cue -> P06 glossy drool finale [SEQUENCE] Grid: 6 panels in a 3x2 grid; one locked-off continuous single take sampled at six phases, identical framing in every panel, no cuts. --- After that, I used both the original image and the storyboard image to generate the video with the following prompt: Seedance 2.0 Prompt: Use @[storyboard ref] as the authoritative director-approved storyboard blueprint for the sequence. Treat every storyboard panel as a consecutive shot within a single cinematic sequence. Follow panel order exactly and do not invent alternative coverage. Do not render the storyboard sheet itself. Preserve camera placement, framing, lens intent, shot scale, character staging, screen direction, environmental geography, prop placement, action choreography, continuity and emotional escalation shown by the storyboard. The storyboard is the primary source of truth for visual storytelling. Recreate the filmed sequence implied by the panels rather than the physical storyboard artwork. The entire video is one continuous locked-off shot with no visible cuts; the camera never moves, pans, zooms, or shakes; each panel is a sampled phase of the same unmoving frame. Use one virtual lens / same-lens locked camera. Use @[ref image] as the authoritative character reference for C1, the cyan twin-tail girl, and C2, the giant patchwork plush monster. ENVIRONMENT: flat cracked stone ground, plain solid cyan backdrop, no props; C1's floor mark stays at frame center, same ground cracks throughout; C2 fills the background when present. EMOTIONAL GUIDANCE: Valence: bored neutral to threatening to disgusted comic relief. Arousal: low and still, spiking at the swallow, simmering through digestion, surging at the nausea, releasing into deadpan calm; shown through posture, C2's eye and cheek changes, bulge motion, and drool state while the frame stays still. VISUAL STYLE: vibrant patchwork anime illustration matching @[ref image]: bold dark linework, saturated cyan-magenta-purple-yellow palette, stitched-fabric texture on the monster, grungy distressed detail, flat graphic lighting, minimal shading; drool is thick glossy viscous strands with soft highlights. AUDIO: No background music or score. Only diegetic ambience, foley, impacts, texture, and silence: faint wind, fabric thumps, wet gulps, gurgles, a huge retch, splattering drool. PANEL BEATS: P01: Locked low wide frame; C1 stands alone at center, slouched and idle, weight shifting slightly; empty backdrop; soft wind. P02: Same frame; C2 rises silently from behind the ground line, his patchwork mass filling the background above C1; jaws begin to part; C1 does not react; low fabric creak. P03: C2's open jaws crash down over C1 and he swallows her whole in one fast gulp; drool strands snap; mouth slams shut; heavy wet impact. P04: C2 sits upright, mouth sealed, throat and belly bulging and shifting as he gulps and grinds; muffled gurgles; drool dribbles from his lips. P05: His cheeks balloon, eyes wince and water, body buckles forward heaving with nausea; drool leaks through his teeth; rising groan. P06: C2 retches and ejects C1 forward; she lands sprawled at center foreground soaked in drool, thick strands stretching back to his hanging-open mouth as he slumps queasy behind her; splatter, then still silence. -- Used my storyboard skill on Claude Fable 5.

@Alin_Reaper05 evident ca sunt😂 Cam cate credite ti-a consumat tie filmuletul asta, sa vad daca merita sau mai bine incerc altceva?

@azdeto Interesant, tehnologia o avansat foarte repede, abia astept sa vad noua versiune 😉 esti roman, nu? 😁

@aimikoda Thanks, it was just a test for the workflow. Now I need to come up with more original ideas, that's the challenge! :))

@azdeto Haha, that turned out great. I made three different versions and I still have one more idea in mind. I absolutely loved your Jeddica too. Let's see what you come up with next.

GPT Image 2 + Seedance 2.0 - Prompt Share White Saber × Droideka You can find the prompts below.

@Framer_X Animations I can do and I did a few. But the difference between a thing that moves and a good animation is the story and how it's told. So, this is what I want to learn, how to write a script and direct it. Then the animation will come easier :)

With only 48 hours left, I’m giving away 3 lifetime memberships to the Cartoom Hero course + community ($199 value) 🎁 To enter: • Like this post • Quote repost it • Comment why you want to learn AI animation The 3 best answers win. Good luck. Doors close in 48 hours. 🚨

@RobotCleopatra Cool, but... you gave it a photo with a blurred background and asked it to blur the background? Also, the pictures you gave it are already good enough, I would've used some real life cases that really needs editing :)

Current portrait retouching workflow: Firefly AI Assistant for batch processing across Photoshop and Lightroom. Retouched an entire headshot session from a single prompt. No app switching. No manual steps. The key is treating AI like a creative director, not a magic button: Describe your standard clearly Let the agent orchestrate the technical work Refine with follow-up prompts Stay in control of the result The time savings come from precision. One clear prompt. One consistent standard. The agent handles the repetitive execution. That's where AI assistance starts feeling like a real workflow upgrade. This post is sponsored by @Adobe through the Adobe Firefly Ambassadors program. #AdobeFireflyAmbassadors #Ad #FireflyAIAssistant

@maxescu @dreamina_ai Cool! :) But I'm confused, I knew they wouldn't allow references to real people. So how is it, how did you put inside the generated videos?

3PM WEDDING My first full short with @dreamina_ai Seedance 2.0. SD 2.0 is live on the web and in the app. Rolling out in select regions. I got early access to Dreamina Seedance 2.0 through their CPP. Watch and enjoy: #DreaminaSeedance2 #DreaminaAI #DreaminaCPP #Seedance2