video-generation

AI video generation via Replicate — 17 models, editing, and production workflows

Zur Installation springen

Repository: fabioc-aloha/ChessCoach
Letzte Quellaktivität: 5. Mai 2026 um 20:58
Erkannte Sprache von SKILL.md: Englisch
Sterne: 1
Forks: 0

Installationsoptionen

Standardmäßig ist der Prompt ausgewählt, der zuerst die Quelle prüft. Sie können zu einem direkten Befehl wechseln oder eine lokale Kopie herunterladen.

Quelldateien prüfen

Lesen Sie SKILL.md und alle von SkillsMP angezeigten Begleitdateien, bevor Sie sich für eine Installation entscheiden.

Zurück zur Suche Zurück zur Installation

type	skill
lifecycle	external-dependent
inheritance	inheritable
name	video-generation
description	AI video generation via Replicate — 17 models, editing, and production workflows
tier	standard
applyTo	*/video,/animate,/clip*
$schema	../SKILL-SCHEMA.json
currency	2026-04-22T00:00:00.000Z

Video Generation Skill

Domain: AI Video Generation
Version: 2.0.0
Last Updated: 2026-04-15
Author: Alex (Master Alex)
Source: Patterns from AlexVideos CLI toolkit
Related: image-handling (images), text-to-speech (audio)

Overview

Generate AI videos using 17 cloud models on Replicate. Supports text-to-video, image-to-video, video editing, and talking-head workflows.

Model Catalog (17 Models)

Key	Model	Replicate ID	Duration	Audio	Cost
`veo3fast`	Veo 3.1 Fast	`google/veo-3.1-fast`	4/6/8s	✅ Auto	$0.10–0.15/sec
`veo3`	Veo 3.1	`google/veo-3.1`	4/6/8s	✅ Auto	$0.20–0.40/sec
`grok`	Grok Video	`xai/grok-imagine-video`	1–15s	✅ Lip-sync	$0.05/sec
`gen45`	Gen-4.5 Runway	`runwayml/gen-4.5`	5–10s	❌	$0.12/sec
`kling`	Kling v3	`kwaivgi/kling-v3-video`	3–15s	Optional	$0.17–0.22/sec
`kling26`	Kling v2.6	`kwaivgi/kling-v2.6`	5–10s	✅ Auto	variable
`kling3omni`

Model	prompt	duration	image	aspect	resolution	negative	audio
`veo3fast`	✅	4/6/8	✅	✅	✅	✅	✅ auto
`veo3`	✅	4/6/8	✅	✅	✅	✅	✅ auto
`grok`	✅	1–15	✅	✅	✅	—	✅ auto
`gen45`	✅	5–10	✅	✅	—	—	—
`kling`	✅	3–15	✅	✅	✅ (mode)	✅	optional
`kling26`	✅	5–10	✅	✅	—	✅	✅ auto
`kling3omni`	✅	3–15	✅	✅	✅ (mode)	—	—
`sora`	✅	4–12	✅	✅	—	—	✅ auto
`sora2pro`	✅	4/8/12	✅	✅	✅	—	✅ auto
`seedance`	✅	2–12	✅	✅	✅	—	—
`seedpro`	✅	2–12	✅	✅	✅	—	—
`pixverse`	✅	5/10	✅	✅	✅ (quality)	✅	✅ auto
`hailuo`	✅	6/10	✅	—	✅	—	—
`hailuo23`	✅	6/10	✅	—	✅	—	—
`ray2`	✅	5/9	✅	✅	—	—	—
`rayflash`	✅	5/9	✅	✅	—	—	—
`wan`	✅	5–10	—	✅	—	✅	✅ auto

Need	Model	Why
Default / Quick preview	veo3fast	Best balance: speed, quality, cost
Talking head / lip-sync	grok, kling26	Grok: best lip-sync; Kling26: best motion
High quality cinematic	sora2pro, veo3	Premium quality, synced audio
Image animation	kling, gen45	Strong image-to-video
Budget production	rayflash, seedance	Cheapest per-second
Longest duration (15s)	grok, kling	Only models supporting 15s
Real-world physics	hailuo, hailuo23	Physics simulation, VFX

# Step 1 — Age-progress reference photo (if needed)
node scripts/generate-image.js "A 26-year-old professional in studio" --model nanapro --image ref.png

# Step 2 — Generate video with Kling v2.6 (best talking-head i2v)
node scripts/generate-video.js "Person looking directly into camera, speaking confidently" --model kling26 --image aged.jpg --duration 10

# Step 3 — Generate speech
node scripts/generate-voice.js "[script]" --model mm28turbo

# Step 4 — Merge video + speech
node scripts/generate-edit-video.js --model avmerge --video clip.mp4 --audio speech.mp3

Key	Model	Purpose	Cost
`modify`	Luma Modify	AI video style transfer	variable
`reframe`	Luma Reframe	AI crop to new aspect ratio	$0.06/sec
`trim`	Trim Video	Extract segment by time	<$0.001
`merge`	Video Merge	Concatenate videos	variable
`avmerge`	ffmpeg-static	Combine audio + video	free (local)
`extract`	Extract Audio	Strip audio from video	variable
`frames`	Frame Extractor	Export frames as images	<$0.001
`upscale`	Real-ESRGAN	AI upscale to 4K	~$0.46
`caption`	AutoCaption	Auto subtitles	~$0.07
`utils`	Video Utils	Convert, misc ops	<$0.002

import Replicate from "replicate";

const replicate = new Replicate();

const output = await replicate.run("google/veo-3.1-fast", {
  input: {
    prompt: "A person walking through a futuristic city at sunset, cinematic lighting",
    duration: 5, // seconds
  },
});

console.log("Video URL:", output);

import { readFileSync } from "fs";

const output = await replicate.run("kwaivgi/kling-v3-video", {
  input: {
    prompt: "Gentle wind blowing through hair, subtle movement",
    image: readFileSync("source-image.png"), // or URL
    duration: 5,
  },
});

const output = await replicate.run("xai/grok-imagine-video", {
  input: {
    prompt: "Person speaking to camera with natural expressions",
    audio: readFileSync("narration.mp3"), // Voice audio
    duration: 10,
  },
});

Element	Good Example	Avoid
Camera motion	"slow dolly forward", "tracking shot"	"camera moves"
Lighting	"golden hour", "dramatic side lighting"	"good lighting"
Style	"cinematic 4K", "documentary style"	"nice video"
Subject	"a person with short brown hair"	"someone"
Action	"walking slowly through", "gesturing while talking"	"doing something"

// 1. Generate video (no audio)
const video = await replicate.run("kwaivgi/kling-v3-video", {
  input: { prompt: "...", duration: 5 },
});

// 2. Generate narration via TTS
const audio = await replicate.run("minimax/speech-2.8-turbo", {
  input: { text: "Narration text", voice: "Wise_Woman" },
});

// 3. Combine with ffmpeg
// ffmpeg -i video.mp4 -i audio.mp3 -c:v copy -c:a aac output.mp4

import { writeFileSync } from "fs";

const response = await fetch(output);
const buffer = await response.arrayBuffer();
writeFileSync("output.mp4", Buffer.from(buffer));

# Check video properties with ffprobe
ffprobe -v quiet -print_format json -show_format -show_streams output.mp4

Error	Cause	Solution
Generation timeout	Complex prompt	Simplify, reduce duration
NSFW rejection	Content policy	Adjust prompt
Low quality output	Vague prompt	Add specific details
Audio desync	Wrong model	Use lip-sync model