▶ demo at sourceSeedance (ByteDance Seed)
ByteDance Seed's official home for the Seedance video models. Start here for official release notes and demos before you pick a hosted endpoint.
▶ demo at sourceSeedance 1.0 Pro on fal
fal's hosted image-to-video endpoint for Seedance 1.0 Pro, ByteDance's higher-quality tier. You can call it by API without running the model yourself.
Feature your AI video tool
One labelled sponsor slot in the grid. Placeholder: no sponsor has been sold.
▶ demo at sourceKling AI
Kuaishou's own Kling studio, where you make text-, image- and reference-driven video and images in one web app. It's the place to try new Kling versions first.
▶ YouTube demoVeo (Google DeepMind)
Google DeepMind's page for the Veo video model family, with capability notes and sample clips straight from the lab.
▶ demo at sourceVeo 3 on fal
Veo 3 served through fal's API, with text-to-video and optional generated audio. Price depends on whether audio is on.
▶ demo at sourceVeo 3 Fast on fal
A faster, cheaper Veo 3 variant on fal. It's a sensible default for drafts before you spend on the full model.
▶ YouTube demoVeo on Gemini API docs
Google's developer docs for generating video through the Gemini API, including when to pick Veo 3.1 versus faster editing-oriented models.
▶ demo on XGoogle Flow
Google's filmmaking app built on Veo and its image models. You assemble shots and scenes in one place instead of generating clips one by one.
▶ example at sourceRunway
Runway's main site: a web studio and API for generating and editing video with its Gen-series models.
▶ example at sourceRunway pricing
Runway's plan page (Free, Standard, Pro, Max), credit-based. Check it before estimating per-clip costs, since credits per second vary by model.
▶ demo on XRunway API docs
Developer docs for Runway's API: auth, endpoints and model options for building video generation into your own product.
▶ example at sourceRunway Changelog
Runway's running log of releases. It's the quickest way to see which models, clip lengths and features are new.
▶ demo at sourceHailuo AI
MiniMax's consumer app for its Hailuo video models, with text- and image-to-video plus social-oriented tools.
▶ example at sourceMiniMax video API docs
MiniMax's platform overview, where to find API access for its video, speech and language models.
▶ demo at sourceLuma Dream Machine
Luma's Dream Machine app for its Ray video models. It runs on credit-based monthly plans.
▶ demo at sourceLuma Ray2 on fal
Luma Ray 2 through fal's API. It's known for natural motion and coherent scenes, and fal's page quotes the cost per 5-second clip.
▶ demo on XLuma API docs
Luma's API reference, for calling Ray video generation directly rather than through a reseller.
▶ demo at sourcePika
Pika's creator-focused video app, with short effects-driven clips and multi-shot generation.
▶ demo at sourcePika 2.2 on fal
Pika 2.2 text-to-video on fal, at up to 1080p. fal lists a per-video price range.
▶ demo at sourcefal video models
fal's catalog of hosted video models (Kling, Veo, Seedance, Wan, Hailuo and more) behind a single API. It's useful for switching models without new integrations.
▶ demo at sourceReplicate video collection
Replicate's curated collection of text-to-video models you can run by API, an alternative host to fal.
▶ demo at sourceComfyUI
ComfyUI is the open, node-based app for building image and video pipelines on your own GPU or in the cloud. Most open-model video workflows assume it.
▶ demo at sourceComfyUI docs: video
Official ComfyUI tutorial for Wan 2.1: text-to-video with the 1.3B model and image-to-video with the 14B model, including where each model file goes.
▶ demo on XComfyUI Changelog
ComfyUI's release log. Check it to see when new video models get native nodes.
▶ example at sourceLTX Video on fal
Lightricks' open LTX Video model on fal. It's one of the cheapest hosted options, at a flat per-video price.
▶ example at sourceHunyuanVideo on fal
Tencent's open HunyuanVideo on fal, at a flat per-video price. fal notes it's free for commercial use.
▶ demo at sourceHiggsfield
Higgsfield is a creator platform that wraps several video and image models with camera-move presets, editing, upscaling and an agent for automating runs.
▶ demo on XHedra
Hedra's platform and models, best known for expressive character and talking-avatar video.
▶ demo at sourceMoonvalley Marey
Moonvalley, the lab behind the Marey video model, which it positions for professional film work.
▶ demo at sourceVidu
Shengshu's Vidu video generator, with text-, image- and reference-to-video aimed at social content and ads.
▶ YouTube demoPixVerse
PixVerse's video generation models and app, now also pushing into real-time and interactive video.
▶ demo at sourceKrea Video
Krea's creative suite, which gives you access to several third-party video models (including Seedance) alongside its own image tools.
▶ demo at sourceTopaz Video AI
Topaz's video enhancement models for upscaling, sharpening, stabilizing and frame interpolation. It's a common last step before delivering AI clips.
▶ demo at sourceElevenLabs
ElevenLabs' voice generation platform, the usual pick for narration and character voices in AI video pipelines.
▶ example at sourceVeo prompt guide (Google Cloud)
Google Cloud's official prompting guide for Veo, covering how to describe subject, action, camera, style and audio to get more controllable clips.
▶ demo at sourceVeo 3.1 on fal
Veo 3.1 on fal for text-to-video. fal lists separate per-second prices with and without audio, plus a higher 4K tier.
▶ demo at sourceVeo 3.1 Fast on fal
The fast Veo 3.1 tier on fal, at roughly half the price of full Veo 3.1 per second. It's a good fit for iterating before a final render.
▶ demo at sourceKling 2.5 Turbo Pro on fal
Kling 2.5 Turbo Pro text-to-video on fal, Kling's speed-optimised premium tier. The price wasn't captured from the page.
▶ demo at sourceHailuo 2.3 Pro on fal
MiniMax Hailuo 2.3 Pro image-to-video on fal at 1080p, billed at a flat price per generation.
▶ demo at sourceWan 2.5 Preview on fal
A preview of Wan 2.5 text-to-video on fal, with per-second prices that rise from 480p to 1080p. The page also accepts an audio file input.
▶ demo at sourceLTX-2 on fal
Lightricks' LTX-2 Pro on fal: text-to-video with audio from 1080p up to 2160p, billed per second by resolution.
▶ demo at sourceWan 2.2 on GitHub
Alibaba's official open-weights repo for Wan 2.2, the successor to 2.1, with code and model downloads for local runs.
▶ demo at sourceLTX-2 on GitHub
Lightricks' official Python inference package and LoRA trainer for LTX-2, its open audio-video model.
▶ demo on X2011 smartphone street vlog look (Seedance 2.5)
A prompt that recreates early-2010s phone footage on purpose: front-camera selfie framing, 480p noise, compression blocks, rolling shutter and white-balance shifts.
▶ demo on XAlley fight action thriller (Kling 3.0)
A night-alley action prompt where one woman faces down a crew, with detailed blocking, lighting and camera notes.
▶ demo on XSunrise breakfast vlog in home-video style (Seedance 2.5)
A detailed prompt for a raw, handheld early-2000s home-video look in a sunny kitchen, with strict rules to keep the reference character's face and outfit unchanged.
▶ demo on XFeeding stray cats, camcorder vlog (Seedance 2.5)
A prompt for a gentle, ASMR-style outdoor vlog shot like handheld mini-DV footage, with the sounds to emphasise listed out.
▶ demo on XSlow-motion lipstick beauty commercial (Seedance 2.5)
A two-reference product prompt (one image for the lipstick, one for the model and car) for a slow-motion beauty ad with precise product tracking.
▶ demo on XEarly-2000s Korean MiniDV home video look (Seedance 2.5)
A detailed prompt for an original character filmed like 2002–2005 MiniDV home footage. It's useful for nailing a period camcorder look.
▶ demo on XMiniature aerospace diorama commercial, JSON prompt (Veo 3.1)
A JSON-structured Veo 3.1 prompt for a 10-second hyper-real miniature aerospace diorama ad. The fields split title, duration, format, style and prompt.
▶ demo on XMiniDV home video, same character (Seedance 2.5)
A 15-second early-2000s MiniDV prompt that repeats exact character details to keep her consistent across shots.
▶ demo on XMultishot fashion UGC party vlog (Kling 4.0 Flash)
A structured 20-second vertical prompt for a creator vlogging a party outfit across several shots. It shows how to lay out a multishot UGC brief.
▶ demo on XWordless post-apocalyptic love story (Seedance 2.5)
A 24-second, 24fps prompt that tells a love story between two survivors purely through visuals, with no dialogue.
▶ demo on XTimed multi-move fight scene with a locked character (Kling 4.0)
A long, second-by-second action prompt that pins an original character to a reference image and chains strikes without resets, written for Kling 4.0.
▶ demo on XDojo martial arts training scene (Seedance 2.0)
A realistic training-scene prompt with two martial artists sparring in a wooden Japanese dojo, with notes on lighting and movement.
▶ demo on XSchool hallway K-drama comedy beat (Seedance 2.5)
A 30-second comedy prompt set in a school hallway that uses an attached image to lock both characters' looks.
▶ example at sourceManimGX – Manim's API on a Rust/wgpu engine, for 3D videos in Python
A re-implementation of the Manim animation API on a Rust/wgpu engine, for code-driven 2D/3D explainer videos in Python. It also runs in the browser.
▶ example at sourceSunny London street anime clip (Seedance 2.5)
A 20-second anime prompt with labelled character blocks (look, outfit, behaviour) on a bright London street.
▶ example at sourceGothic ink anime sword fight (Seedance 2.5)
A style-heavy prompt that fuses 2D dark-fantasy anime with gothic ink illustration for high-sakuga sword choreography, preserving a reference aesthetic.
▶ demo on XMiniature banana gelato food ad, JSON prompt (Veo 3.1)
A JSON-format Veo 3.1 prompt for a 10-second artisanal gelato ad in a miniature diorama style. The structure is reusable across products.
▶ demo on XMiniature lychee sorbet food ad, JSON prompt (Veo 3.1)
A JSON-structured Veo 3.1 food-commercial prompt that builds a hyper-real miniature diorama around a sorbet.
▶ demo on XTimes Square 1985 VHS tourist footage (Seedance 2.5)
A historical-recreation prompt for gritty 1985 Times Square shot like consumer VHS camcorder footage.
▶ demo on XMiniature luxury tech diorama, JSON prompt (Veo 3.1)
Another JSON-format Veo 3.1 prompt from the same creator, this time a desert-themed miniature tech product ad.
▶ demo on XOne-take kitchen UGC: cat cake vs real cat (Seedance 2.5)
A front-camera, single-take UGC prompt where cutting a cat-shaped cake meets a real cat's reaction.
▶ demo on XRainy anime heartbreak short (MiniMax H3)
A scene-timed vertical anime prompt for a short emotional story, starting with a rain scene at 0–5s.
▶ example at sourceDark-fantasy samurai fight with locked characters (Seedance 2.0)
A fast-cut samurai action prompt that defines its characters once and keeps them identical, with an explicit no-gore note.
▶ demo on X25-shot dark-comedy 'video flow' (Seedance 2.5)
An ambitious 15-second, 25-shot prompt written as a shot-by-shot flow with varied focal lengths and rapid editorial cuts.
▶ example at source12-shot anime rom-com classroom scene (Seedance 2.5)
A vertical anime prompt that specifies exactly 12 shots and 11 cuts for a small classroom comedy beat. It's a good template for cut control.
▶ demo on XGoPro shot from a helicopter over a lava field (Seedance 2.5)
A comedic vertical action-cam prompt filmed from a hovering helicopter doorway above a lava field.
▶ demo on XTop-down pastel domino spiral (LTX 2.5 Pro)
A satisfying top-down push-in prompt with hundreds of pastel dominoes toppling in slow motion.
▶ example at sourceopen-higgsfield
An open-source studio with one prompt bar, per-model settings and a single gallery of finished runs.
▶ demo on XUGC camera product walk-and-talk (Seedance 2.5)
A short prompt for a natural UGC video where a creator shows off a fictional camera brand while exploring a city.
▶ demo at sourceawesome-seedance
A Seedance 2.x prompt library that traces cases to their original posts and publishes cross-model retest verdicts.
▶ demo on XStoryboard-guided vertical sequence (Seedance 2.5)
A 12-second prompt that uses a character reference plus a master storyboard image as strict guides for the whole sequence.
▶ demo at sourceOpenMontage
An agentic video production system that gives AI coding assistants a set of pipelines, tools and production-knowledge files for making finished videos.
▶ example at sourcecinematic-video-prompt-skill
A cheat sheet and Claude skill covering camera angles, moves, lighting, composition and grading terms for video prompts.
▶ demo on XHandheld NYC city-walk vlog
A shaky, vlog-style follow shot of a stylishly dressed woman walking through New York, with exact wardrobe details for consistency.
▶ demo at sourcedrama-skills
Open-source skills for producing AI short dramas: script, character assets, storyboards and image/video prompts, with review steps.
▶ demo on XProfessional vertical product commercial (Seedance 2.5)
A long, shot-specified commercial prompt written to look like real professional camera footage.
▶ demo on XGolden-hour travel vlog detour (Seedance 2.5)
A photoreal travel-vlog prompt where an ordinary sunset walk turns into a small adventure on a hidden trail.
▶ example at sourceComfyUI-MiniMaxH3-TimelineDirector
A ComfyUI plugin and all-in-one workflow for MiniMax H3. It covers text-to-video, reference images, audio-driven generation, editing and motion transfer, and builds long videos from continuous segments in one run.
▶ demo on XRainy neon pop music video (Seedance 2.0)
A structured 25-second music-video prompt with sections for style, story and shots, built around rainy neon streets.
▶ example at sourceSeedance 1.0 Lite on fal
The cheaper, faster Lite tier of Seedance 1.0 on fal, for text-to-video. The page lists 5 or 10 second clips at up to 720p.
▶ demo at sourceKling 2.1 Master on fal
The premium Kling 2.1 Master endpoint on fal, for text-to-video. fal pitches it for the smoothest motion and closest prompt adherence of the 2.1 tiers.
▶ demo at sourceKling 2.1 Standard on fal
A lower-cost Kling 2.1 image-to-video endpoint on fal. fal's page gives a clear per-clip and per-second price, which makes it easy to budget for batches.
▶ demo at sourceMiniMax Hailuo 02 on fal
Hailuo 02 image-to-video through fal, billed per second at 512p or 768p.
▶ demo at sourceWan 2.1 on fal
Wan 2.1 image-to-video hosted on fal, with roughly 5–6 second clips at 480p or 720p and a flat price per video.
▶ demo at sourceWan (Alibaba) GitHub
Alibaba's official open-weights repo for Wan 2.1, with code, model downloads and hardware notes for running it locally.
▶ example at sourceComfyUI-MiniMaxH3-Easy
A single compact ComfyUI workflow for MiniMax H3 that covers text-to-video, image-to-video, first/last-frame and reference video, with one shared input panel.
▶ example at sourceComfyUI-Spectrum-MiniMax-H3
ComfyUI nodes that apply Spectrum, a training-free sampling speed-up from a 2026 paper, to the native MiniMax H3 audio-video model. They handle H3's combined audio/video latents.
▶ demo on XPayday mini-vlog with timed scenes (Seedance 2.5)
A UGC mini-vlog prompt broken into timed scenes (0–5s and so on) across one character's day. It's a useful template for scene timing.
▶ demo at sourceComfyUI_MiniMax_H3_Extender
A ComfyUI node that chains several MiniMax H3 clips into one long sequence, carrying motion, visuals and audio across the joins. Recent versions add an optional latent refine/upscale step.
▶ example at sourceComfyui-MMH3-UltimateUpscale
A single ComfyUI node that upscales finished MiniMax H3 videos on low-VRAM GPUs by processing time chunks and spatial tiles. The audio track is preserved.
▶ YouTube demodramaclaw
A script-to-finished-film engine for dramas, ads and product videos, built as one end-to-end pipeline.
▶ demo on XUGC talking-head beverage ad via Kling MCP
A product-ad prompt asking Kling for an influencer-style talking-head video that introduces a mango passionfruit drink.
▶ example at sourceawesome-MiniMax-H3-cases
A large, traceable gallery of MiniMax/Hailuo H3 cases with their public prompts and checked tutorials, in two languages.
▶ example at sourceMaestro
A fully local video, image and music studio on the WanGP pipeline. Its director mode plans a whole music video or short from one prompt.
▶ demo on XNeon-city music video with three singers (Seedance 2.5)
A long music-video prompt with three singers performing at night in a neon city, with lip-sync and high-budget MV direction.
▶ example at sourceMiniMax H3 Day-0 Support in ComfyUI: Open Weights, Native Audio, and 2K Video
Comfy's post on day-one ComfyUI support for MiniMax H3: open weights, native audio and 2K video.
▶ example at sourceTwo-reference character shot, first take (Seedance 2.5)
A prompt that maps @image1 (full body) and @image2 (close-up) to one character for 30 seconds at 720p.
▶ example at sourceComfyUI-H3-Motion-Context-MultiRef
MiniMax H3 nodes and workflows for extending clips, one-shot music videos, video-to-video motion transfer and custom keyframes.
▶ example at sourceGoohai-MiniMax-H3_Integration
ComfyUI nodes for MiniMax H3's audio-video conditioning and reference-driven generation.
▶ example at sourceseedance2-skill
An agent skill focused only on writing strong Seedance 2.0 prompts.
▶ YouTube demoCalliope
A local story-to-video app. It turns an idea into a storyline and script, splits scenes into shots, generates each shot through your own ComfyUI, then stitches the clips into a film with ffmpeg.
▶ example at sourceai-shortVideo-pipeline
Backend plumbing for an AI short-video service: multi-model failover, metering and quality gating.
▶ demo on XCreature breaks out of a cinema screen, one take (Seedance 2.5)
A strict single-take prompt that forbids any cut: a creature appears to break out of a theatre screen, filmed on a phone.
▶ example at sourceagnes-video-generator
A self-hosted web app that turns text into multi-scene videos with narration, subtitles and a presenter, powered by Agnes AI.
▶ demo at sourceawesome-seedance
A large community collection of Seedance 2.0 prompts and resources across film, anime, UGC and ad styles.
▶ example at sourceMix-Studio
A polished front-end that runs tuned ComfyUI image and video workflows from desktop or phone without touching the node graph.
▶ demo on XPack goat rescue, one handheld take (Seedance 2.5)
A 20-second vertical, single-take, phone-camera prompt that tells a small rescue story (a goat leads a hiker to a trapped man) in documentary style.
▶ example at sourceawesome-seedance-2-prompts
A curated library of Seedance 2.0 prompts by style, plus notes on API use and keeping characters consistent.
▶ demo at sourceHunyuanVideo-1.5
HunyuanVideo 1.5, Tencent's lighter open video model aimed at running on more accessible hardware.
▶ example at sourcehiggsfield-seedance2-jineng
Fifteen Claude prompt skills for Seedance 2.0 on Higgsfield, each built around a hook-first structure for genres like CGI, anime and ads.
▶ example at sourcewind-comic
A multi-agent pipeline that turns a one-line idea into a short drama: script, storyboards and character-consistent video, with swappable providers.
▶ example at sourcelanshu-awesome-ai-video-kit
A Chinese-language kit from enterprise video projects: hundreds of prompts across 15 models, Claude skills and methodology notes.
▶ example at sourceai-shortfilm-prompts
A Claude Code skill that turns a rough idea into a structured, model-ready video prompt, with 21 genre templates.
▶ example at sourceNomi
An open-source video workbench that lets coding agents direct any model or local ComfyUI over MCP, from storyboard to an editable first cut.
▶ demo on XLuxury perfume bottle product shot (Seedance 2.0)
A compact product-ad prompt for a perfume bottle on glossy black with amber light and golden smoke.
▶ example at sourcevisual-skills
Agent skills that pair film-craft ideas (blocking, montage) with exact prompt syntax for Seedance, Kling, Veo and others.
▶ example at sourceflowboard
A local infinite-canvas tool where you wire model, product and scene nodes into product videos.
▶ example at sourcehiggsfield-ai-prompt-skill
A Claude skill bundle (32 sub-skills) for writing cinematic Higgsfield prompts, centred on Seedance 2.x features like omni-reference and extend.
▶ example at sourcecomfy-skills
Comfy's official plugin for coding agents. It adds slash commands and an MCP connection so an agent can generate images, video, audio and 3D, or run ComfyUI workflows on Comfy Cloud.
▶ example at sourcecomfyui-mcp
An MCP server plus sidebar agent that lets an AI assistant build, run and edit ComfyUI graphs in plain language, for images, video and audio.
▶ example at sourceawesome-ad-video-prompts
A curated set of original, high-craft prompts for product and ad videos, written for Seedance 2.0, Veo 3, Kling and Runway.
▶ example at sourceai-media-generator
A Claude Code skill that drafts director-style prompts for image, video and music and can run them in a browser.
▶ example at sourcevibe-creating-skill
A portable, bilingual agent skill that rewrites ideas into model-ready text-to-video prompts.
▶ example at sourceDirectorSKILL
A Claude Code skill for shot lists and keyframe/video prompts, with ten selectable director-style overlays.
▶ example at sourcevelorn
An AI-native video editor organised around a real timeline, with generative steps and local agent control.
▶ example at sourceComfyUI-Easy-Media
ComfyUI nodes that simplify loading media and assembling video pipelines, with friendlier parameters than the defaults.
▶ demo on XLTX-Video
Lightricks' official LTX-Video repo: open weights and inference code for one of the fastest open video models.
▶ demo at sourceComfyUI-WanVideoWrapper
Kijai's widely used ComfyUI wrapper for Wan video models. New Wan features and optimisations often land here before native nodes.
▶ example at sourceBlowtorch on salmon macro physics test (Veo 3.1)
A locked-off macro prompt that tests whether a blowtorch flame hits and spreads across food realistically, with lighting and film stock spelled out.
▶ example at sourceComfyUI-Bernini
A standalone ComfyUI plugin for the Wan 2.2 Bernini model, for generating and editing video.
▶ example at sourceawesome-ltx2
A maintained index of LTX-2 models, text encoders, LoRAs and ComfyUI workflows in one place.
▶ example at sourceVibe-Workflow
An open-source, node-based builder for chaining image and video generation steps, pitched as a free alternative to Weavy and Krea Nodes.
▶ example at sourceLightX2V
A lightweight inference framework for text-to-video, image-to-video and image generation that supports several open models. There's also a hosted demo studio.
▶ YouTube democomfyui-deno-custom-nodes
A grab-bag of ComfyUI nodes: RTX Video Super Resolution, LTX 2.3 helpers, Wan 2.2 reference editing, and video compare and preview.
▶ example at sourceComfyUI-LTXVideo
Lightricks' official ComfyUI support for LTX-Video, the reference way to run LTX models in a graph.
▶ demo at sourcegoku
Code for the Goku video generation foundation models, a CVPR 2025 highlight paper.
▶ example at sourceFastVideo
An open framework for faster video generation, covering both post-training and real-time inference. Docs and a cookbook are linked from the repo.
▶ demo at sourceComfyUI-HunyuanVideoWrapper
Kijai's ComfyUI wrapper for Tencent's HunyuanVideo, an early way to run it locally with memory optimisations.
▶ demo at sourceID-LoRA-LTX2.3-ComfyUI
A ComfyUI node for LTX 2.3 that makes video carrying one person's identity, from a reference image and voice. Use only with consent of the person shown.
▶ example at sourceegaki
A command-line tool for generating images and video from the terminal across Google, OpenAI, fal, Replicate and Seedance. Agents can use it too.
▶ example at sourcecomfyui-vrgamedevgirl
A large ComfyUI node pack for AI video production. Its main tool, the AI Video Builder, plans, prompts, generates, reviews and assembles a video scene by scene, with LTX 2.3 and MiniMax H3 support.
▶ example at sourceComfyUI-PainterNodes
A ComfyUI toolkit spanning video generation, image editing and audio-driven lip-sync across Flux, MiniMax, LTX and Wan 2.2.
▶ demo at sourceseedance-2-generator
A starter SaaS codebase (Next.js, Stripe, NextAuth, Prisma) for selling Seedance 2.x video generation.
▶ demo on XRainy-night urban thriller walk (Seedance 2.5)
A 30-second suspense prompt that follows a woman from a subway exit through rain-soaked streets, beat by beat.
▶ example at sourceLance – image/video generation and understanding in one model
Lance, from ByteDance, is a single model for both generating and understanding images and video, with code, paper and weights linked.
▶ example at sourcevibeframe
A CLI and MCP server that lets a coding agent turn a brief into a finished MP4 using your own Seedance, Runway, Kling or Gemini keys. Every paid step has a dry-run price and a hard spending cap.
▶ demo at sourceawesome-video-prompts
A prompt collection for Veo 3/3.1, Kling and Hailuo, organised by scene type.
▶ YouTube demoA Pipeline for Making 10-minute AI Movies with Claude Code and Seedance
A Claude Code pipeline that makes 10-minute films with Claude as director, combining Seedance for video, Nano Banana for images and ElevenLabs for voice.
▶ example at sourceText-to-video model from scratch (2 brothers, 2 years, 2B params)
Two brothers trained a 2B-parameter text-to-video model from scratch. Their write-up shows both good and bad samples.
▶ YouTube demoAI-NovelFlow
A Chinese open-source platform that turns novels into videos by chaining open image and image-to-video models and community ComfyUI workflows.
▶ demo on XCrane descent: soldiers emerge from smoke (Hailuo 2.3)
A slow-motion crane-down prompt with soldiers marching out of smoke as a building collapses behind them.
▶ example at sourcestory-shot-agent
An agent that breaks a script into shots and writes consistent text-to-video prompts for each, keeping characters stable across segments.
▶ demo on XCyberpunk android at a vending machine (Luma Ray 3)
A moody rain prompt with holographic glitches, a metallic faceplate and a pulsing neon skyline.
▶ demo on XWar mechs under twin suns (Hailuo 2.3)
A tracking prompt that follows one mech's heavy stride across a desert battlefield.
▶ demo on XSunset reunion love story (Kling 3.0)
A scene-by-scene prompt for a 15-second emotional reunion at sunset on a quiet golden-lit street.
▶ demo on XMage crossing a floating sky bridge (Hailuo 2.3)
An anime tracking prompt across a rune-lit stone bridge between sky islands.
▶ demo on XText VFX: word forms in volcanic ash (Runway Gen-4.5)
A dive-through text-effect prompt where swirling ash and embers form the word "WRATH" before the eruption swallows it. Swap the word for your own title card.
▶ demo on XMountaineer in a blizzard, slow push-in (Luma Ray 3)
A wide aerial-to-push-in prompt where snow whips across the lens as a climber collapses on a ridge.
▶ demo at sourceComfyUI-Wan-SVI2Pro-FLF
Nodes for Wan 2.2 SVI 2 Pro that add first/last-frame keyframe control and seamless stitching across clips.
▶ example at sourceopenjourney
A Midjourney-style web UI clone that generates images with Imagen 4 and video with Veo through Google's Gemini SDK.
▶ example at sourceComfyUI-PainterLTXV2
Sampler and latent-prep nodes for LTXV that separate audio and video sampling.
▶ demo on XEagle escaping an erupting volcano (Luma Ray 3)
A rapid ascending follow-shot prompt that tracks an eagle banking through ash and lava fountains.
▶ demo on XText VFX: word forms from glacier shards (Runway Gen-4.5)
A hyperspeed FPV prompt through a collapsing glacier where floating ice shards line up to spell "FROST" and then shatter.
▶ demo at sourceComfyUI-Wan-VACE-Prep
Helper nodes that make common edits (transitions, extensions, outpainting) easier to set up for Wan VACE.
▶ demo on XZero-gravity orbital greenhouse (Luma Ray 3)
A floating-camera prompt through glowing alien plants in an orbital lab as a researcher works.
▶ demo on XDesert highway motorcyclist at golden hour (Luma Ray 3)
A live-action wide-shot prompt with dust trails, long shadows and post-apocalyptic road signs.
▶ example at sourceComfyUI-PainterLongVideo
A ComfyUI node for long-form Wan 2.2 videos that keeps motion and scenes consistent and corrects unwanted slow motion.
▶ example at sourceComfyUI-VideoHelperSuite
Nodes for loading, combining and exporting video frames in ComfyUI. Nearly every ComfyUI video workflow depends on it.
▶ demo on XMech battle overhead drone shot (Luma Ray 3)
A one-sentence action prompt: two armoured mechs clash in a burning city, shot from an overhead drone.
▶ demo on XKnight with molten armor, three-shot sequence (Kling)
A multi-shot prompt written as a shot list: a wide battlefield, a close-up as lava-like cracks spread across the armor, then a wide walk through rising flames.
▶ demo at sourceComfyUI-HunyuanLoom
Nodes for editing existing videos with HunyuanVideo (video-to-video) in ComfyUI.
▶ example at sourceComfyUI_HunyuanVideoFoley
Generates matching sound effects for a video from the clip plus a text prompt, using HunyuanVideo-Foley in ComfyUI.
▶ demo at sourceComfyUI-LTXTricks
Extra ComfyUI nodes that add finer control over LTX Video generation.
▶ demo at sourceComfyUI-Wan-VACE-Video-Joiner
Workflows that use Wan VACE to build smooth transitions between any two clips, AI-generated or real footage.
▶ demo on XAnime scooter chase through Tokyo (Luma Ray 3)
A low, fast tracking prompt that weaves through alleys and food stalls in an anime-style chase.
▶ demo on XLow-angle desert motorcycle at dawn (Hailuo 2.3)
A low tracking prompt with dust trails, heat shimmer and long dawn shadows.
▶ demo on XDrone orbit of a drifting sports car (Hailuo 2.3)
A sweeping circling-drone prompt around a car drifting through mountain curves at sunset.
▶ demo on XFantasy armies and glowing arrows, overhead (Hailuo 2.3)
An overhead war-fantasy prompt with arrow volleys, siege magic and flying beasts.
▶ example at sourcevapourkit
A front-end for VapourSynth focused on video upscaling.
▶ YouTube demovideo2x
A long-running open framework for AI video upscaling and frame interpolation, with several backends in one tool.
▶ demo on XText VFX: word from cracked desert earth (Runway Gen-4.5)
A ground-skimming prompt where dust from opening fissures locks into the word "DRY" and then collapses.
▶ demo on XSkateboard rail grind POV (Hailuo 2.3)
A first-person skatepark prompt with camera shake on landings and sparks off the rail.
▶ demo on XOrbit shot sunset duel with kanji shockwaves (Hailuo 2.3)
A compact orbit-camera prompt for a stylised duel where each clash sends kanji rippling into the sky.
▶ example at sourceComfyUI-Frame-Interpolation
ComfyUI nodes for frame interpolation (RIFE, FILM and others) to raise the frame rate of generated clips.
▶ demo at sourceLTX2-Infinity
A ComfyUI workflow that chains LTX-2 generations into effectively unlimited-length video, based on the Stable-Video-Infinity idea.
▶ demo on XText VFX: word forms in valley fog (Runway Gen-4.5)
A slow, quiet title-reveal prompt where morning mist reshapes into "BEGIN" and dissolves.
▶ demo on XLava-lit underworld throne room (Luma Ray 3)
An overhead gliding prompt over a demon king's court with molten rivers and obsidian columns.
▶ demo on XShrine maiden on a storm cliff, ink-brush style (Hailuo 2.3)
A low-angle prompt with exaggerated robe motion and lightning drawn in thick ink-brush strokes.
▶ demo on XText VFX: word ignites in a fire vortex (Runway Gen-4.5)
A punch-through-flames prompt where fire spirals into a vortex that ignites "INFERNO" and then explodes.
▶ demo on XUpside-down forest walk (Luma Ray 3)
A surreal follow-cam prompt through a forest with inverted trees and leaves that change colour with each step.
▶ demo on XJapanese rooftop sunset timelapse (Luma Ray 3)
A locked-off wide timelapse prompt over an old Japanese neighbourhood as golden hour turns to neon night.
▶ demo on XFPV flight through Persepolis (Veo 3.1)
A fast FPV prompt sweeping through Achaemenid palaces, columns and reliefs, with motion blur.
▶ example at sourcefree-ai-video-upscaler
Source for a free browser-based AI video upscaler.
▶ demo on XBioluminescent exoplanet ocean (Kling 2.6)
A short sci-fi prompt with glowing sea life, twin moons in time-lapse and an underwater camera glide.
▶ demo on XDesert-noir drone shot of a classic car (Hailuo 2.3)
An overhead follow prompt with a blood-red sunset and a lone motel on the horizon.
▶ demo on XMonk at sunrise on a misty cliff (Luma Ray 3)
A calm push-in from behind a meditating monk as the sun rises over mist-filled valleys.
▶ demo on XFPV flight through the ancient Lydian Kingdom (Veo 3.1)
A fast FPV prompt weaving through temples, markets and coin workshops in ancient Sardis. It's part of a series of historical FPV flights.
▶ example at sourcemuapi-comfyui
ComfyUI nodes for 600+ hosted models, including Seedance, Kling and Veo 3, via the muapi.ai service.
▶ demo on XText VFX: word in a volcanic lightning storm (Runway Gen-4.5)
A prompt that pierces ash clouds above an eruption, where sparks and static briefly form "FURY".
▶ demo on XPastel rooftop sunset, watercolour anime (Luma Ray 3)
A gentle push-in with lens flares, drifting petals and watercolour tones for a dreamy anime rooftop scene.
▶ demo on XTilt-shift ships over a futuristic city (Runway Gen-4.5)
A tilt-shift prompt where a narrow focus band on passing ships makes the city below look like a miniature.
▶ example at sourceExtreme close-up miniature sushi character (Kling)
A detailed macro prompt: handheld extreme close-up with a slow push-in on a tiny anthropomorphic sushi character. It's a good model for breathing-camera micro shots.
▶ example at sourceComfyUI-fal-API
ComfyUI nodes that call fal's 1,400+ hosted models, including Kling, Veo and Seedance, with caching and cost controls.
▶ example at sourcefal-mcp-server
An MCP server that lets Claude generate images, video, music and audio through fal.
▶ demo on XNeon light burst over mountains (Runway Gen-4.5)
A short abstract prompt: a small neon circle in darkness explodes into waves that silhouette mountains.
▶ demo on XRotating beverage can with floating fruit (Kling 2.6)
An image-to-video product prompt: a can rotates 360° while the camera orbits and fruit pieces float through frame.
▶ YouTube demoTheAnimeScripter
An open anime-video toolkit for AI upscaling, RIFE interpolation and restoration in one pass, as CLI, standalone app or After Effects plugin.
▶ demo on XNeon maglev tunnel sprint (Luma Ray 3)
A breakneck side-tracking prompt with a pilot chased by drones through a streaking neon tunnel.
▶ example at sourceComfyUI-HunyuanVideoMultiLora
ComfyUI nodes for stacking several LoRAs on HunyuanVideo at once.
▶ demo on XNight train station with parallax (Luma Ray 3)
An anime medium shot with background parallax, wet platform reflections and glowing vending machines.
▶ example at sourceOpen-AI-UGC
A self-hostable Next.js app for making UGC-style video ads with AI actors, using models like Veo 3.1 and Seedance 2. Disclose AI actors in ads.
▶ demo on XPOV horse ride through the ancient Kuru Kingdom (Veo 3.1)
A rider's-eye POV prompt galloping through an ancient Indian kingdom, with the horse's head and reins in frame.
▶ example at sourceimg-prompt
Pick tags in your own language and it assembles the English prompt that image and video models expect.
▶ demo at sourceComfyUI-TeaCacheHunyuanVideo
Applies the TeaCache caching trick to HunyuanVideo in ComfyUI to speed up generation.
▶ demo on XDragon in a storm: same-prompt test across four models
An epic fantasy prompt with a dragon in lightning-lit clouds and a sweeping camera, used as a stress test for physics, volumetric light and camera movement.
▶ example at sourceWarehouse fight storyboard: Seedance 2.0 vs Kling V3 Pro
A long storyboard-style prompt for a crime-drama fight in a dim warehouse, written to test action choreography.
▶ example at sourceDJZ-Workflows
A large library of ready-made ComfyUI workflows across image and video models, including Hunyuan and CogVideo.
▶ example at sourcewavespeed-comfyui
WaveSpeed's ComfyUI nodes for calling hundreds of hosted image and video models (Seedance, Wan and others) inside a local graph.
▶ demo on XSubway platform shot with a locked character reference (Kling)
A reference-locked prompt that keeps one character's face, outfit and proportions identical in a realistic subway platform scene. The creator built the concept in ChatGPT first.
▶ demo at sourceComfyUI-HunyuanVideo-Nyan
Nodes that let you scale how strongly the CLIP and LLM text encoders steer HunyuanVideo, for tighter prompt control.
▶ example at sourcesynergize-motion-appearance
Official code for a CVPR 2025 paper on talking-head video generation with multi-scale codebooks. Research code; use only with consent of the person shown.
▶ demo on XAnime power-up leap with lightning (Luma Ray 3)
A dynamic-angle prompt with speed lines and exaggerated perspective, ending in a lightning-filled leap.
▶ demo on XMoonlit rooftop chase in slow motion (Luma Ray 3)
An anime slow-motion side shot of a robed girl fleeing shadow spirits, her staff leaving a light trail.
▶ example at sourceDescending drone shot over a flooded city (Veo)
An aerial prompt slowly descending through fog over a flooded historic European city with rescue boats, in an ash-grey IMAX look.
▶ demo on XMagical-girl transformation spin (Luma Ray 3)
An orbiting-camera prompt for an anime transformation with crystals, ribbons of light and energy wings.
▶ demo on XRainstorm golf shot physics test
A brutal physics prompt (a golfer driving off a cliff tee in a violent rainstorm) designed to test ball tracking and weather.
▶ example at sourceSlow-motion walking shot with Kling 3.0 Motion Control
A one-line cinematic prompt for a slow-motion walk with dramatic lighting, run through Kling 3.0 Motion Control on a phone.
▶ demo on XPOV bike ride into ancient Kosala (Veo 3.1)
A first-person bicycle ride toward a fortified ancient Indian city with period architecture.
▶ example at sourceNeon city character turn (Kling 3.0 Motion Control)
A short prompt for a character turning toward camera in a neon-lit futuristic city, built for Kling 3.0 Motion Control.
▶ example at sourceRunway Gen-4.5: A new frontier for video generation
Runway's announcement page for Gen-4.5, its next-generation video model.
▶ example at sourceRain-freeze samurai dolly-in (Seedance 2.0)
A short, comma-list prompt for a samurai drawing his sword in heavy rain while the droplets freeze mid-air, with a slow dolly-in and desaturated grade.
▶ demo on XChameleon colour shift, slow zoom (Runway Gen-4.5)
An image-to-video nature prompt with a slow zoom as a chameleon's skin shifts through vivid colours.
▶ demo on XGirl on a bike with her dog (Kling 2.6)
A simple suburban prompt (child on a bicycle, colourful outfit, a Jack Russell running alongside) that's handy for testing character detail.
▶ demo on XLow-angle cyclist tracking shot on a mountain road (Veo 3.1)
A tracking-shot prompt where the camera follows a cyclist's curve down a foggy, wet mountain road with spray and sun flashes.
▶ demo on XSteadicam night-market walk (LTX-2, local)
A 15-second Steadicam prompt moving through a Southeast Asian night market, run locally on LTX-2.
▶ demo on XMisty valley sunrise drone shot (LTX-2, local)
An aerial drone prompt over a misty mountain valley at sunrise with a forward glide.
▶ demo at sourceLTXV 13B Distilled – Generate 5s Videos in Under 10s
Lightricks' distilled LTXV 13B, a faster version of its open model.
▶ demo at sourceHunyuanVideo
Tencent's original open HunyuanVideo repo: weights, code and a technical report for a large open video model.
▶ example at sourceLTX-Video: Realtime Video Latent Diffusion
The arXiv paper behind LTX-Video, explaining how it generates video faster than real time.
▶ YouTube demoVeo 3 and Imagen 4, and a new tool for filmmaking called Flow
Google's announcement of Veo 3 with native audio, Imagen 4, and the Flow filmmaking app.
▶ demo at sourceCogVideo
The official repo for CogVideo and CogVideoX, the open text- and image-to-video models from Zhipu/THUDM.
▶ demo at sourceVGen
Alibaba's VGen research codebase covering several diffusion video methods.
▶ example at sourceComfyUI-AnimateDiff-Evolved
The improved AnimateDiff implementation for ComfyUI with advanced sampling. It's the backbone of most AnimateDiff-era workflows.
▶ demo at sourceComfyUI-WanVideoStartEndFrames
Nodes to generate a Wan 2.1 clip between a chosen first and last frame, built on a modified Kijai wrapper.
▶ YouTube demoDAIN
Depth-Aware Video Frame Interpolation (CVPR 2019), the classic research code.
▶ example at sourceComfyUI-Cloud
Runs local ComfyUI workflows on cloud GPUs when your machine lacks the VRAM.
▶ example at sourceComfyUI-ADMotionDirector
Nodes for training AnimateDiff motion LoRAs inside ComfyUI.
▶ example at sourcecomfyui-dream-video-batches
Nodes for batch video generation with SVD and AnimateDiff.
▶ demo at sourceI made an Image2Video version of HunyuanVideo and tonight I open-sourced it
A community-built image-to-video variant of HunyuanVideo, released as open source on GitHub.
▶ example at sourceWanX open weight sota 14B video model release
The Hugging Face release of the open 14B Wan (WanX) video model.