boss
54 posts
Made a Haaland FIFA edit with Seedance 2.0
Prompt:
A high-end 3D Pixar-style digital animation. The subject is a cute, chubby toddler caricature of professional footballer Erling Haaland standing on a vibrant green grass pitch inside a massive, packed football stadium at night.
The toddler has oversized features, including a round face, an exaggerated pouty and grumpy expression, and his signature long blonde hair tied neatly into a top-knot bun with shaved sides. He is wearing a detailed light blue Manchester City FC home kit, complete with matching shorts and golden football boots. He stands confidently with his arms tightly crossed over his chest, rocking slightly back and forth to the rhythm of music, giving a playful and defiant attitude. One foot is planted firmly on top of a classic star-patterned UEFA Champions League football.
The camera angle is a dynamic, low-angle full-body shot that makes the character look heroic yet adorable. The background is softly blurred with bokeh, showing a massive crowd under intense, bright stadium floodlights that cast crisp highlights on his hair and clothes. Cinematic lighting, rich colors, ultra-detailed textures on the fabric, hair strands, and grass blades, rendering in smooth 4K resolution.
English
Who wouldn’t want this fluffy roommate?
Made with Seedance 2.0 in 4K.
Get the full open-source project with prompts below.
English
boss retweetledi
Fable 5 does crazy things on Supercomputer.
A Higgsfield creator built an app that blocks 3D scenes, records a real cinema camera move, and renders the video through Seedance 2.0.
A 15-second clay pass carries the exact trajectory, timing, and framing a prompt keeps losing.
Higgsfield AI 🧩@higgsfield
Viewport Preview Reference + Seedance 2.0 in 4K.
English
Fable 5 vs Sonnet 5 on agentic prompting.
Both videos generated using Seedance 2.0.
English
A creator opens ChatGPT Image 2.
Not to make a picture.
To build a character sheet. 6 angles. 4 expressions. One outfit.
Everyone assumes the animation tool is Seedance.
It's not.
The step most people skip isn't in Seedance.
It's before Seedance.
Consistency doesn't come from prompting harder.
It comes from the reference image you feed it.
The pipeline:
ChatGPT Image 2. Full character design sheet. Multiple angles. Expressions. Outfit. ChatGPT Image 2 again. Storyboard every shot. Camera angle. Action. Mood. Lock a color palette and lighting mood. Golden afternoon. Soft warm tones. Dramatic shadows.
Now open Seedance 2.0.
Each storyboard frame is the reference image.
One frame becomes one clip.
The prompt handles motion and camera only. The scene is already in the image.
Three rules that keep clips from drifting apart:
4 to 6 seconds per shot. Shorter clips mean less motion drift on faces. Don't repeat camera moves back to back. If shot 1 dollies in shot 2 holds or pulls back. Same palette every single prompt. No exceptions.
Seedance reads two things. The reference image and the prompt.
If the reference is detailed enough the output stays on model.
That's the whole trick nobody mentions.
Tomorrow. The exact ChatGPT Image 2 prompt structure for the multi angle sheet.
One article covers the entire workflow.
Pinned below. Don't scroll past it. ↓
English
THE 16-FRAME GRID TRICK THAT KEEPS SEEDANCE 2.0 CHARACTERS CONSISTENT
One character. Three riders. A full chase sequence across 16 shots.
No drift no redesign mid-scene, no "wait why did her hoodie change color."
Most people generate a single reference image, then re-prompt Seedance 2.0 shot by shot and hope the character holds together.
The real consistency comes from building the entire sequence as one 4x4 storyboard grid in GPT Image 2.0 first then feeding that grid into Seedance 2.0 not raw prompts.
Here's the workflow:
1. Write the scene + character bible setting outfits vibe movement style all in one place
2. Draft 16 frame captions shot type + action + one punchy line each
3. Lock the visual style in a single block halftone, motion lines, smear frames, tilted panels
4. Generate the full 4x4 grid in GPT Image 2.0 numbered corners thin black borders, captions baked in
5. Pull each frame out as its own reference image
6. Feed each frame + its caption into Seedance 2.0 as an individual shot
7. Stitch the shots in storyboard order
Why this works:
• Character gets designed once across the whole grid not reinvented in 16 separate prompts
• Comic-style motion lines and smear frames already encode direction so Seedance has less to guess
• Frame captions double as your shot list no re-planning once you're animating
• Consistent panel tilt and framing forces consistent camera logic before video even starts
Use cases:
• Chase scenes / action sequences with multiple moving subjects
• Comic-to-motion adaptations
• Group shots where every character needs to match across cuts
• Fast-cut sports or extreme-sports style edits
Not every frame survives the jump to video untouched some smear effects need cleanup once they're moving.
But building the whole sequence as one grid before touching Seedance 2.0 killed almost all my consistency issues.
English
Creativity belongs to everyone, and the earlier it starts, the further it goes.
Proud to support ACAI for Kids at NFC Summit in Lisbon, where children and parents created AI art, music, and stories together.
Alex 🧩 Scaling Video GenAI@alexmashrabov
We were proud to sponsor ACAI for Kids at NFC Summit 2026 in Lisbon. Over 100 children and parents created AI art, music, and stories together. We gave every family free credits to keep experimenting at home. Hope we'll see some future creators come out of workshops like these. Thanks @NFCSummit, TUMO Portugal, and AI Art Magazine.
English
Gemini Omni Flash and Seed Audio 1.0 hit Adobe Premiere Pro and DaVinci Resolve Studio via Higgsfield Plugins.
Clean a background with a prompt, generate multi-shots, and add motion graphics in the timeline.
Change a voice, narrate from text, or dub a clip into 18 languages.
English
Introducing Higgsfield Explainer, powered by Claude Fable 5 and Gemini Omni Flash.
Faceless documentaries at scale. It auto-researches your topic, narrates in any language, and renders up to 10 minutes in one run.
Available on Higgsfield · MCP · Supercomputer
English
Fable 5 + Higgsfield MCP = a full marketing team inside Claude.
The most anticipated model in AI runs content pipelines and deep analysis. With /goal, /loop, and a Fable 5 agent council, agent-run marketing is about to take off.
A month of campaign work ships in a day.
English