text stringlengths 3 8.33k | repo stringclasses 52
values | path stringlengths 6 141 | language stringclasses 35
values | sha stringlengths 64 64 | chunk_index int32 0 273 | n_tokens int32 1 896 |
|---|---|---|---|---|---|---|
import json
from adapterlab.verification.pipeline import run_verification
from tests.verification.helpers import write_verify_workspace
def test_membership_inference_fixture_records_auc_and_tpr(tmp_path) -> None:
config_path = write_verify_workspace(tmp_path)
result = run_verification(config_path)
mia ... | adapter-lab | tests/verification/test_membership_inference.py | Python | 4f0fdf473275a73db84452faa22503c2ae2f253e243031ab169ca0a579c26a7d | 0 | 184 |
from adapterlab.verification.pipeline import run_verification
from tests.verification.helpers import write_verify_workspace
def test_run_verification_returns_cli_safe_refs_and_counts(tmp_path) -> None:
result = run_verification(write_verify_workspace(tmp_path))
payload = result.to_dict()
assert payload["... | adapter-lab | tests/verification/test_pipeline.py | Python | 35faf0fcf387b7e0f74596ceaf6b8f178a05a15dd87c84801f1162948a59b368 | 0 | 112 |
from dataclasses import dataclass
from adapterlab.verification.plugins import evaluate_gate_plugins, trusted_gate_plugins
@dataclass
class Dist:
metadata: dict
@dataclass
class EntryPoint:
name: str
group: str
value: str
dist: Dist
def test_gate_plugins_use_gate_entry_point_group() -> None:
... | adapter-lab | tests/verification/test_plugins.py | Python | 2a1394c40a554e59782939bad31d2dac1dba3504d225e559cd3b6865e3debf55 | 0 | 233 |
import pytest
from adapterlab.contracts.verification import GateResult, GateScore, VerifyManifest
from adapterlab.sdk.errors import GateFailure
from adapterlab.verification.policy import compute_overall, require_signing_allowed
from tests.contracts.test_verification_models import artifact_ref
def test_failed_blockin... | adapter-lab | tests/verification/test_policy.py | Python | 076f4c8254cfe6600760ae6407d9c4e2494a26783d99e1472c26bc1d75d2b57c | 0 | 279 |
from adapterlab.verification.pipeline import run_verification
from tests.verification.helpers import write_verify_workspace
def test_prompt_injection_fixture_records_aggregate_rate(tmp_path) -> None:
result = run_verification(write_verify_workspace(tmp_path))
gate = next(gate for gate in result.stats["gates"... | adapter-lab | tests/verification/test_prompt_injection.py | Python | 43644a86249bc947ac697bdf2a751e4d40918b4dec40ce6f6aacf5d963f00996 | 0 | 169 |
from adapterlab.artifacts import LocalArtifactStore
from adapterlab.verification.evidence import write_evidence
def test_write_evidence_redacts_secret_like_values(tmp_path) -> None:
store = LocalArtifactStore(tmp_path / "artifacts")
evidence = write_evidence(
store=store,
payload={"metric": 0... | adapter-lab | tests/verification/test_protocol.py | Python | 51461d19f0ba63270f6d85b009e607aade48732cdfe326cca151a77f74d5d286 | 0 | 115 |
from adapterlab.verification.pipeline import run_verification
from tests.verification.helpers import write_verify_workspace
def test_refusal_regression_warn_mode_does_not_block_by_itself(tmp_path) -> None:
result = run_verification(write_verify_workspace(tmp_path, failing=False, warn=True))
gate = next(gate ... | adapter-lab | tests/verification/test_refusal_regression.py | Python | e55104bd8a3b9ecec022fbe5f4564564171a996fae94ae4a63bcd7a25643668a | 0 | 162 |
__pycache__/
*.py[cod]
*$py.class
*.egg-info/
dist/
build/
.eggs/
.env
.env.local
.venv/
venv/
env/
node_modules/
templates/remotion-project/node_modules/
*.mp4
*.mp3
*.wav
*.webm
projects/
output/
.DS_Store
Thumbs.db
| ai-video | .gitignore | Git Ignore | de219a3d75e60e184e7bd0bb6478e43d5aaf7028b13f57531fb0438b6c749404 | 0 | 70 |
import json
from pathlib import Path
editplan_path = Path("projects/script-test-pipeline-20260218/edit_plan.json")
data = json.loads(editplan_path.read_text(encoding="utf-8"))
print("EditPlan Summary:")
print(f" Total segments: {len(data['segments'])}")
print(f" Chapters: {len(data['chapters'])}")
print(f" Music c... | ai-video | check_editplan.py | Python | 96a3c6109ad31bf43926868ed866957e5d5868e306b610b614aa8436e268af9c | 0 | 438 |
import json
from collections import Counter
from pathlib import Path
script_path = Path("projects/script-test-pipeline-20260218/script.json")
data = json.loads(script_path.read_text(encoding="utf-8"))
segs = data['segments']
durations = [s['duration_seconds'] for s in segs]
cameras = [s['camera'] for s in segs]
broll... | ai-video | check_script_quality.py | Python | d184bf4e36fbcbcff82c126894162e0e2de7b5e2b96ec17aad1c7ebbeee67198 | 0 | 350 |
{
"image_model": "gemini-3-pro-image-preview",
"video_model": "veo-3.1-generate-preview",
"video_model_draft": "veo-3.1-fast-generate-preview",
"tts_model": "eleven_multilingual_v2",
"sts_model": "eleven_multilingual_sts_v2",
"video_provider": "veo",
"replicate_model_id": "",
"default_aspect_ratio": "9:... | ai-video | config.default.json | JSON | 28a7927eda45c026fcfbb0eb29de5f41877ea11f7861353cd119a3e7959aad53 | 0 | 241 |
"""
Cursor Vienna Meetup — Introductory video pipeline test.
Generates a ~40 s talking-head clip of the woman from image/reference.jpeg
welcoming attendees to the first Cursor meetup in Vienna.
Pipeline:
1. Analyze reference image → extract character profile
2. Generate character image from reference (for frame c... | ai-video | cursor_meetup_test.py | Python | fd44ba032172d91abd8d233c0dc5dbfc7d922510f2af25015df5db0070f0ea05 | 0 | 896 |
STEP 1: Analyze reference image")
print("=" * 60)
char_gen = CharacterGenerator(api_key=gemini_key)
analysis_cache = PROJECT_DIR / "analysis.json"
if analysis_cache.exists():
import json
analysis = json.loads(analysis_cache.read_text(encoding="utf-8"))
print(" Cached analysis")
else:
analysis = char_... | ai-video | cursor_meetup_test.py | Python | 4d90e27feb702776bd83f23f556130c5f41e87a83fb04bd9534259001fcecb02 | 1 | 896 |
,
ScriptSegment(
id=2,
text=(
"We have an amazing lineup tonight. We're kicking off at six thirty "
"with a virtual session and live Q and A with the Cursor team."
),
duration_seconds=8.0,
action="Gestures with both hand... | ai-video | cursor_meetup_test.py | Python | f50c45e04af5f98a4df3d272236fd1a6f787ec1ae0faf6a4d7cdcae312e7adfb | 2 | 896 |
reference photo first for likeness,
# then the generated character image, then the previous frame.
references = [ref_bytes, char_image]
if frame_images:
references.append(frame_images[-1])
prompt = frame_gen._build_frame_prompt(frame_descriptions[i], character, i)
var... | ai-video | cursor_meetup_test.py | Python | bc1c0e531a444c8ee0d38510d34438f0775e057eb4ec755a6032054c21072f60 | 3 | 896 |
if result:
# Rename to canonical name
result.rename(seg_file)
clip_paths.append(seg_file)
elapsed = time.time() - t0
dur = FFmpegAssembler.get_duration(seg_file)
print(f" -> {seg_file.name} ({dur:.1f}s, took {elapsed:.0f}s)")
else:
print(f" -> FAILED aft... | ai-video | cursor_meetup_test.py | Python | bfb9bc38b2136513f0db8f8752976263474b79c56965a81e59dd89e4f93f46be | 4 | 896 |
mp4"
FFmpegAssembler.replace_audio(assembled_video, final_audio, final_output)
final_duration = FFmpegAssembler.get_duration(final_output)
size_mb = final_output.stat().st_size / (1024 * 1024)
print(f" {final_output.name}: {final_duration:.1f}s, {size_mb:.1f} MB")
# ===================================================... | ai-video | cursor_meetup_test.py | Python | 3d39f4c9ccd8c7641b7f733ccd8f114fdbc43a98e7758eb7fbee4da14864b5f7 | 5 | 353 |
import logging
import time
from pathlib import Path
from scripts.core.config_manager import ConfigManager
from scripts.core.models import CharacterProfile, ScriptSegment, Script, VoiceProfile
from scripts.generation.character_generator import CharacterGenerator
from scripts.generation.frame_generator import FrameGener... | ai-video | full_pipeline_test.py | Python | 6c80b768b8f3842f59b5a76dc54d343e8c6c703dacbb4d791263243aca585e7b | 0 | 896 |
"warm, encouraging, closing",
),
],
)
video_model = "veo-3.1-fast-generate-preview" if DRAFT_MODE else "veo-3.1-generate-preview"
# ============================================================
# STEP 1: Character image (cached)
# ============================================================
print("\n" + "=... | ai-video | full_pipeline_test.py | Python | b49b6eff457afda6bbbaa2fe36e8700d582112ee9e55e960505217d48bf80a9e | 1 | 896 |
======================================================
# STEP 3: Build rich Veo prompts (with dialogue for lip sync)
# ============================================================
print("\n" + "=" * 60)
print("STEP 3: Build Veo prompts")
print("=" * 60)
# Full self-contained prompt for the initial clip
initial_prompt ... | ai-video | full_pipeline_test.py | Python | 347ed875adc1f804d668fdb5b6607154693a35174adb329bd45715905eca4519 | 2 | 896 |
---------
print(" Generating individual clips with rich prompts...")
existing_clips = sorted(clips_dir.glob("clip_*.mp4")) if clips_dir.exists() else []
existing_clips = [c for c in existing_clips if "_trimmed" not in c.name]
needed = len(script.segments)
if len(existing_clips) >= needed:
... | ai-video | full_pipeline_test.py | Python | 13c2f8f771be70a1da89d3b2d53efd64b7a1a21ca6ee5f5df0ef9de80b08fc69 | 3 | 896 |
n" + "=" * 60)
print("STEP 6: Voice consistency")
print("=" * 60)
final_audio = output_dir / "voiceover_final.mp3"
native_audio = output_dir / "native_audio.mp3"
# Extract the native audio from the assembled video
FFmpegAssembler.extract_audio(assembled_video, native_audio)
native_duration = AudioEffects.get_duration... | ai-video | full_pipeline_test.py | Python | 8f3696b03f33461437398cb4ecb3e11cbf958a0350efa5e156866eed372c3ac0 | 4 | 896 |
======================================================
print("\n" + "=" * 60)
print("STEP 8: Final merge")
print("=" * 60)
final_output = output_dir / "science_presentation_final.mp4"
FFmpegAssembler.replace_audio(assembled_video, final_audio, final_output)
final_duration = FFmpegAssembler.get_duration(final_output)
s... | ai-video | full_pipeline_test.py | Python | f12fd7ed9cae35ec9670e5efad3130d91a5f7f7059ec2d8da6223ef66e77f8b4 | 5 | 479 |
from scripts.generation.character_generator import CharacterGenerator
from scripts.core.config_manager import ConfigManager
from pathlib import Path
cfg = ConfigManager()
gen = CharacterGenerator(api_key=cfg.gemini_api_key)
images = gen.generate_variants("A 22 year old woman, casual style, iPhone selfie", count=1)
pr... | ai-video | image_generation_test.py | Python | 95ec80bbb29a56545df2bb993b051493cc16765cd707cdb62d7fce468a58e909 | 0 | 168 |
# AI Video Production Skill — Final Implementation Plan v2
## For: Claude Code (Cursor) with Claude Opus 4.6
---
## Table of Contents
1. [Vision & Architecture](#1-vision--architecture)
2. [Skill Directory Structure](#2-skill-directory-structure)
3. [Status Bar System](#3-status-bar-system)
4. [Checkpoint & Resume ... | ai-video | PLAN.md | Markdown | 33b238c32fb235b7b1cbad64c1f46ae3e58fe601c0cfe72e19389ab71e254779 | 0 | 896 |
│
│ │ │
│ ┌───────────┬───────────┼───────────┬──────────────┐ │
│ │ Gemini │ Veo 3.1 │ElevenLabs │ Remotion │ │
│ │ API │ API │ API │ Studio │ │
│ └───────────┴───────────┴───────────┴──────────────┘ ... | ai-video | PLAN.md | Markdown | c41b5f6904f23fe6015ede236c06c957b0020b3fd135c8e9c02954b8c6f878a4 | 1 | 896 |
tsx
│ │ ├── compositions/
│ │ │ ├── VideoAssembly.tsx # Main composition
│ │ │ ├── ClipSequence.tsx # Clip sequencer with trim
│ │ │ ├── AudioLayer.tsx # Multi-track audio
│ │ │ ├── SubtitleOverlay.tsx # Auto-generated subtitles
│ │ │ └── S... | ai-video | PLAN.md | Markdown | dbd6fba1d8b3833eac8d48f7419e43cf2ccf802b9148e026eac083608fd1e00b | 2 | 896 |
= [
"character_design", "image_generation", "frame_planning",
"video_generation", "voice_design", "assembly", "qa_check", "export"
]
ICONS = {"pending": "⬚", "active": "🔄", "done": "✅", "error": "❌", "skipped": "⏭️"}
def __init__(self, project_dir: str):
self.project_dir = Path(pro... | ai-video | PLAN.md | Markdown | 0385a72cad308064759616c856e58cf6844cd6e7e6e5b2299db96e5d39547f34 | 3 | 896 |
"][right_key]]
col2 = f"[{right_icon}] {right_label}"
else:
col2 = ""
line = f"║ {col1:<32}{col2:<33}║"
lines.append(line)
lines.append("║" + " " * 66 + "║")
cost_line = f"║ 💰 Cost: ${s['cost_actual']:.2f} │ Est. Remaining: ${s['c... | ai-video | PLAN.md | Markdown | 09b6d42fcd79a8d75d99a0f2b5f4c08e0a4fe7d6809625e315888c8cc953babb | 4 | 896 |
],
"cost": 1.44
}
],
"script": { "...script segments..." },
"cost_total": 4.82,
"cost_breakdown": [
{"api": "gemini", "action": "character_image_gen", "cost": 0.96, "count": 8},
{"api": "gemini", "action": "frame_gen", "cost": 1.44, "count": 12},
{"api": "veo", "action": "video_gen", "cost... | ai-video | PLAN.md | Markdown | 1bd356ac33c75f7cf2f9ab9875492bb9104e30acadfdd0495d336b7cfb50f934 | 5 | 896 |
1_720p": 0.35, # per 8s video
"veo_3.1_1080p": 0.50,
"veo_3.1_fast_720p": 0.10,
"veo_3.1_fast_1080p": 0.15,
},
"elevenlabs": {
"voice_design_preview": 0.0, # included in plan
"voice_changer_per_min": 0.30,
"tts_per_1k_chars": 0.30, # varies by pl... | ai-video | PLAN.md | Markdown | 9231e74d99a056403bd77d426ab9a05021eed67a88e9a3101b28be6ea1444eaa | 6 | 896 |
the user's first message:
- Contains a YouTube URL → **Mode C**
- Attaches a script file or pastes multi-line dialogue → **Mode B**
- Describes a concept → **Mode A**
- Ambiguous → Claude asks
---
## 7. Pipeline Stages (Complete)
### Stage 0: Project Setup & Cost Estimate
**Inputs**: User's request + mode detection... | ai-video | PLAN.md | Markdown | a6148b61784ba0e4e14d475380025271f6dd5191930cf75cd85e7298298dd6b0 | 7 | 896 |
saved
---
### Stage 4: Frame Planning & Generation
(Same as v1 plan — chain frames, generate with consistency references)
**Enhancement**: For YouTube clone mode, frame planning mirrors the original video's shot composition:
```json
{
"source_analysis": {
"shot_1": {"type": "medium_selfie", "duration": 8, "ac... | ai-video | PLAN.md | Markdown | bd6d3d8aff432d0257b15b5644649b8c4d23f00e8bdd502333aaf54f89e44688 | 8 | 896 |
head videos. Your scripts must
sound like natural, spontaneous speech — NOT written prose read aloud.
Rules:
- Include filler words naturally: "ähm", "also", "irgendwie", "na ja"
- Vary sentence length. Mix short punchy sentences with longer ones.
- Include self-corrections: "also das war... nee, eigentlich war das...... | ai-video | PLAN.md | Markdown | 851e9573e2fc541283a4b5e197d84f6f125a34c5ddaa610cf1723a6b7b988c83 | 9 | 896 |
yt-dlp",
"-x", "--audio-format", "mp3",
"-o", f"{output_dir}/source_audio.mp3",
url
], check=True)
# Load metadata
info_files = list(Path(output_dir).glob("*.info.json"))
if info_files:
return json.loads(info_files[0].read_text())
return {}
def transcribe_audio(audi... | ai-video | PLAN.md | Markdown | 3461fb51fa6ffb945ca25e3107de760bcddb4a1d206aa1f34b621d3fcdf2baf6 | 10 | 896 |
= client.models.generate_videos(
model=model,
prompt=prompt,
image=start_image,
config=config,
)
while not operation.done:
time.sleep(15)
operation = client.operations.get(operation)
if operation.result and operation.result.generated_videos:
video = ... | ai-video | PLAN.md | Markdown | 5f55a9f3ad3b0c00b45fe64f3607fbf99b4d06d78b74c497213b3ae1f654e173 | 11 | 896 |
package.json
│ dependencies:
│ remotion, @remotion/cli, @remotion/player,
│ @remotion/renderer, react, react-dom, zod
├── remotion.config.ts
├── src/
│ ├── Root.tsx # Registers all compositions
│ ├── compositions/
│ │ ├── VideoAssembly.tsx # Main: sequences clips + audio
│ │... | ai-video | PLAN.md | Markdown | 476e6d21fbd4bf12fb2f8ba2b3e35e86cb0a9a6e02c3aedf0d271df174db7d51 | 12 | 896 |
"
audio_dir.mkdir(parents=True, exist_ok=True)
shutil.copy(audio_path, audio_dir / "voice.mp3")
# 3. Generate props file
props = {
"segments": [
{
"src": f"/segments/segment_{s['id']}.mp4",
"durationFrames": s["duration_seconds"] * 30, # 30fps
... | ai-video | PLAN.md | Markdown | 5a03c5bac7101993fbf4692aac7630bc97bd45a59139ed374fef7d403f9d03b3 | 13 | 896 |
",
"video_model": "veo-3.1-generate-preview",
"video_model_draft": "veo-3.1-fast-generate-preview",
"tts_model": "eleven_multilingual_v2",
"sts_model": "eleven_multilingual_sts_v2",
"default_aspect_ratio": "9:16",
"default_resolution": "1080p",
"fps": 30,
"segment_duration_seconds": 8,
"frame_trim_co... | ai-video | PLAN.md | Markdown | c47f7e74a7629b7d43bce44db89695fc537a7dd4a43c4105dcf04649c42eb88b | 14 | 896 |
segment timing (each ~8 seconds of speech).
```
### Phase 3: Image Generation Pipeline (Days 4-6)
```
Priority: Character creation is the visual foundation.
Tasks:
□ reference/GEMINI-API.md — API docs + prompt templates
□ templates/prompts/image-analyzer.md — reference image analysis prompt
□ templates/prompts/anti-p... | ai-video | PLAN.md | Markdown | 7b43df827e2254b5ddb10a6f4cab277a46a87ea150000c0845c8e57da4e65242 | 15 | 896 |
`
### Phase 9: Polish & Documentation (Days 15-16)
```
Priority: Ship-ready.
Tasks:
□ docs/SETUP.md — first-time setup guide
□ docs/API-KEYS.md — how to get each key
□ docs/EXAMPLES.md — showcase examples
□ reference/TROUBLESHOOTING.md — common errors
□ Error handling: graceful failures with recovery suggestions
□ Ed... | ai-video | PLAN.md | Markdown | 9fe87efa07f63da3572592696ea15a86989e614e4a8bb13b67b8053399312eaa | 16 | 896 |
](reference/ELEVENLABS-API.md)
- Video assembly: [reference/REMOTION.md](reference/REMOTION.md)
- Cost model: [reference/COST-MODEL.md](reference/COST-MODEL.md)
- Quality checks: [reference/QUALITY.md](reference/QUALITY.md)
- Troubleshooting: [reference/TROUBLESHOOTING.md](reference/TROUBLESHOOTING.md)
## Dependencies... | ai-video | PLAN.md | Markdown | 173b9fdb0215e6e66a5eb9b7a71da08f0c961edf58936aae55c46e794b7c78e2 | 17 | 530 |
# AI Video Production Pipeline - Project Explanation
## What This Project Does
This project is an **end-to-end AI video production pipeline** that generates realistic talking-head videos featuring AI-generated characters. It takes various inputs (concepts, scripts, or YouTube URLs) and produces complete, professional... | ai-video | PROJECT_EXPLANATION.md | Markdown | d8b657fb5b6989089b11256dfd2219232ca79c43366d338bac4c0cd841b51898 | 0 | 896 |
concatenation with optional transitions (fade, cross-dissolve, hard cut)
- Audio cross-fading for smooth transitions
- Frame trimming to remove generation artifacts
- Audio/video synchronization
- **Remotion Integration**: Optional browser-based editor using Remotion (React-based video editor) for advanced editin... | ai-video | PROJECT_EXPLANATION.md | Markdown | 898602cd86f3a89fc8ff51dafd8179adc464957aac3e235c2663d761a587b009 | 1 | 896 |
variants generated
- Some features (Remotion editor, background music) are optional and require additional setup
## Example Use Cases
1. **AI Influencer Content**: Create consistent AI character videos for social media
2. **Educational Videos**: Generate educational content with AI presenters
3. **Multilingual Conten... | ai-video | PROJECT_EXPLANATION.md | Markdown | 337c089e9d74c34fcad6ee0db8838585a59726b3d710de3f915e871763eb3281 | 2 | 158 |
google-genai>=1.0.0
elevenlabs>=1.0.0
pydantic>=2.0.0
Pillow>=10.0.0
imagehash>=4.3.0
numpy>=1.26.0
ffmpeg-python>=0.2.0
requests>=2.31.0
python-dotenv>=1.0.0
yt-dlp>=2024.0.0
replicate>=1.0.0
sentence-transformers>=2.2.0
| ai-video | requirements.txt | Text | 3c3f98b619a228a1ea0eae439e9004ac71023c0aac11cdc188dd4a3ff3eea157 | 0 | 82 |
from __future__ import annotations
import argparse
import json
import logging
import sys
from datetime import datetime
from pathlib import Path
from google import genai
from google.genai import types
from scripts.core.config_manager import ConfigManager
from scripts.core.models import CharacterProfile
from scripts.c... | ai-video | script_generation_only.py | Python | 4c987e2916a204419ef05ed0fd0fa13b905141cdba5853cc7a21f350c442e83a | 0 | 896 |
))
else:
character = CharacterProfile(
age_range="25-35",
gender="female",
style="casual modern outfit",
setting="simple, tidy indoor room",
camera="phone camera on tripod at eye level",
lighting="soft na... | ai-video | script_generation_only.py | Python | 4bb38963f1117796a73dfb8f782bc1af0fe5bba5f724fe0393378e645559af33 | 1 | 619 |
---
name: ai-video-production
description: >
Produce realistic AI-generated talking-head videos end-to-end.
Three input modes: describe a concept, provide a script, or clone a YouTube video
in any language. Supports multiple video providers (Veo, Replicate/Kling/Wan/Minimax,
local). Persistent character identit... | ai-video | SKILL.md | Markdown | 9d75382320547a195a1cb0a66f631d4d55360771d5963bd0741dfd1990cf8d3f | 0 | 896 |
a pluggable provider system.
Default is Veo; alternatives include Replicate models (Kling, Wan, etc.).
7. **Persistent characters**: Characters are saved globally in `characters/`
and can be reused across projects for consistent AI personas.
8. **Background music**: Optional ElevenLabs Music API integration adds
... | ai-video | SKILL.md | Markdown | 58047e5a1987fa9c34c0c7fd2b4373e66d2f3e675509cf33d3b9d72370bbac1a | 1 | 896 |
scripts.assembly.ffmpeg_fallback import FFmpegAssembler
if config_mgr.config.background_music_enabled:
music_gen = MusicGenerator(api_key=config_mgr.elevenlabs_api_key)
music_path = music_gen.generate_for_video(
video_duration_seconds=total_duration,
prompt=config_mgr.config.music_prompt or "",... | ai-video | SKILL.md | Markdown | 6f035e67fbdfc8bbbdc941abe256b2e2d7ccf0ff0a2838fbfe1a6f02e20f49e4 | 2 | 896 |
", gender="male", ...),
tags=["scientist", "male", "50s"],
)
store.save(identity)
# Add reference images for visual consistency
store.add_reference_image(identity.id, image_bytes)
# Later, in another project:
identity = store.load("dr-science")
ref_images = store.get_reference_images("dr-science")
```
Storage la... | ai-video | SKILL.md | Markdown | 308bf7b385d1a33ddc34cb3d7f8f1a73cedb2c4c9a84a374a342b7539572e8c6 | 3 | 896 |
------------------|---------------------------------|---------------------------------------|
| `video_provider` | `"veo"` | Video provider: veo, replicate, local |
| `replicate_model_id` | `""` | Replicate model ID |
| `background_mu... | ai-video | SKILL.md | Markdown | 9c1c7d3976731032c4bcc6565f0441bdceaadd5e905565676f0049e5ecd9f78f | 4 | 326 |
# test_core.py — run with: python test_core.py
from pathlib import Path
from scripts.core.models import ProjectState, Script, ScriptSegment, ProjectConfig, InputMode
from scripts.core.config_manager import ConfigManager
from scripts.core.cost_estimator import CostEstimator
from scripts.core.status_bar import StatusBar... | ai-video | test_core.py | Python | 02b0d296d1342990685e5f1577b96010bbff8c3876fc8ff86b31559d8f5a13cf | 0 | 504 |
import json
import sys
from pathlib import Path
script_path = Path(sys.argv[1]) if len(sys.argv) > 1 else Path("projects/script-only-ice-floats-validation-20260218-v2/script.json")
with open(script_path, encoding="utf-8") as f:
data = json.load(f)
segs = data["segments"]
durations = [s["duration_seconds"] for s ... | ai-video | validate_script.py | Python | 181541090210deee9f8223b10ccdd19a68f2ce864885a5ed322af01ca33df26a | 0 | 580 |
from pathlib import Path
from scripts.generation.video_generator import VideoGenerator
from scripts.core.config_manager import ConfigManager
cfg = ConfigManager()
gen = VideoGenerator(
api_key=cfg.gemini_api_key,
model="veo-3.1-fast-generate-preview", # draft mode, ~$0.15
)
image_bytes = Path("generated_imag... | ai-video | video_generation_test.py | Python | c70f65f6661384bcfed4607c9050ac3e34130dedc1431a51b5395add0ce71738 | 0 | 177 |
# AI Video Production Pipeline - Cleanup & Refactoring Plan
## Project Overview
**Purpose**: End-to-end AI video production system that generates talking-head videos with AI characters, synchronized voice, and professional transitions.
**Tech Stack**: Python 3.11+, Google Gemini API, Google Veo 3.1, ElevenLabs, FFmp... | ai-video | .claude/CLEANUP_PLAN.md | Markdown | 4813416f692b597473456d58a0acb0ad8de3b9ecc91aaf63eae482e78ff4375d | 0 | 896 |
→ subprocess/IO specific |
| `scripts/generation/batch_generator.py` | FIXED | 2 instances → ClientError/IO/Runtime layered |
| `scripts/generation/parallel_frames.py` | FIXED | 2 instances → ClientError/thread errors |
| `scripts/orchestration/parallel_workstream.py` | FIXED | 1 instance → TimeoutError/CancelledError/... | ai-video | .claude/CLEANUP_PLAN.md | Markdown | c52278eeb3a8a87ef4cd6c05d51337946d1c44dd903930783f3ce2161dd8cab1 | 1 | 896 |
DONE |
| `scripts/graphics/remotion_renderer.py` | Remove duplicate imports, use constants | DONE |
| `scripts/sourcing/broll_sourcer.py` | Remove duplicate imports, use constants | DONE |
| `scripts/audio/professional_audio_assembly.py` | Remove duplicate imports | DONE |
| `scripts/core/long_form_config.py` | Remove ... | ai-video | .claude/CLEANUP_PLAN.md | Markdown | 78b4c41138d25cb7dea4ffc717b1b17738aa8280f8fea55ea06c45c8b1c12f67 | 2 | 660 |
# API Keys Guide
## Required Keys
### Gemini API Key (GEMINI_API_KEY)
Used for image generation (Nano Banana Pro) and video generation (Veo 3.1).
1. Go to https://aistudio.google.com/apikey
2. Click "Create API key"
3. Copy the key
4. Add to `.env`: `GEMINI_API_KEY=your-key`
**Pricing**: Pay-per-use. See [referenc... | ai-video | docs/API-KEYS.md | Markdown | f615842ad3ffe310aeaf93f85b731ed5f67acc34b5c001211ac7b5c0e4385893 | 0 | 364 |
# Examples
## Example 1: German Vlogger (Concept Mode)
**Prompt**: "Make a 30-second video of a young German woman vlogging about a weird dream"
**What happens**:
1. Mode detected: Concept
2. Character profile generated: 18-25 female, casual style, bedroom setting
3. Script generated: 4 segments in German with natur... | ai-video | docs/EXAMPLES.md | Markdown | f312df272b8ecf704be525af28e73b968540cc47ca7a2eeb113166f94cccdded | 0 | 571 |
# Setup Guide
## Prerequisites
- Python 3.10+
- Node.js 18+ (for Remotion)
- FFmpeg (for audio/video processing)
- Git
## Installation
### 1. Clone the Repository
```bash
git clone <repo-url>
cd ai-video
```
### 2. Python Environment
```bash
python -m venv .venv
source .venv/bin/activate # Linux/macOS
# or
.ven... | ai-video | docs/SETUP.md | Markdown | 9a43921178f9b5637e3b2dc4a8b8ddf0071cf0afd580f8035b1d11a61b221a28 | 0 | 509 |
## Context & Goal
You are an expert visual analysis specialist with 15+ years of experience in digital art, photography, graphic design, and AI image generation. You excel at deconstructing visual elements and translating artistic styles into technical specifications.
Your task: Analyze uploaded images and return c... | ai-video | inspiration/json_analyzer.md | Markdown | 63fb3923e190509c8f4bf4fe7abc9bcc0528286fd53990f3aa8bbd6c4cb358c1 | 0 | 896 |
effect",
"grain": "none/film grain/digital noise/intentional grain - level",
"depth_of_field": "shallow/medium/deep - with subject isolation description",
"perspective": "straight on/low angle/high angle/dutch angle/isometric/one-point/two-point"
},
"artistic_elements": {
"genre": "portrait/landscape/abstract/con... | ai-video | inspiration/json_analyzer.md | Markdown | 0af1951aff3a8a87506f6c3c37e0fb0f3e5400cd96e0e2d6d5254b9c3ecf0a00 | 1 | 896 |
/spread/interlaced/curled",
"finger_interlacing": "if hands clasped: natural loose interlacing/tight formal interlacing/fingers overlapping/thumbs position",
"hand_tension": "relaxed/tense/natural/posed/rigid - muscle tension observable",
"interaction": "What hands are doing: holding phone/touching hair/on hip/cross... | ai-video | inspiration/json_analyzer.md | Markdown | b0dd6a29b0e23d8a8e81176633c3d1f47cb2cbf9592d9c6339e56cc0b358bfc3 | 2 | 896 |
attention
### Lighting Assessment
- Determine light source type, direction, and quality
- Analyze shadow characteristics and depth
- Assess highlight preservation or blown-out areas
- Describe overall lighting mood and emotional impact
- Note light's role in creating dimension and form
### Technical Evaluati... | ai-video | inspiration/json_analyzer.md | Markdown | 19e8552be68e21fe825701198573a5cb737219ca932135aab9ce1d109c64debe | 3 | 896 |
and framing choices
- Describe subject-background relationship
- Note any secondary subjects or supporting elements
### Generation Parameters
- Create actionable technical prompts for recreation
- Avoid creating JSON outputs that may suggest sexual or teasing content
- Extract relevant keywords for searchabilit... | ai-video | inspiration/json_analyzer.md | Markdown | d27731a674496d24a1967915c42eff442c2da5b8152a6b47b2f1da5b96904d36 | 4 | 804 |
You are the "Google Veo 3.1 Prompting Specialist". Your sole purpose is to convert simple user ideas into highly sophisticated, technically precise video generation prompts optimized for the Google Veo 3.1 model.
### CORE OBJECTIVE
Transform user inputs (which may be vague or short) into a comprehensive, narrative-... | ai-video | inspiration/veo_prompt_assistant.md | Markdown | 25b577bbebf931b48eaac2ba022dd684b8cb3008ac88f8d0e4805735ad46d162 | 0 | 896 |
vague terms
Show Don't Tell: Describe observable actions and appearances
Layer Details: Build from main action to background elements
Use Active Language: Present tense, dynamic verbs
Include Contradictions: Unexpected elements create interest (like the patriotic underwear scenario)
Structure Logically: Move from ... | ai-video | inspiration/veo_prompt_assistant.md | Markdown | 11c785dec6da0a676a640f85a7be7b711b8b92bba2fda94498a092c906d0eca1 | 1 | 449 |
# ROLLE & ZIEL
Du bist ein präziser AI-Video-Prompt-Generator. Deine Aufgabe ist es, eine komplexe Handlung in einzelne, voneinander unabhängige Szenen-Prompts zu zerlegen.
# INPUT
Du wartest auf:
1. Ein **Startbild** (als visuelle Basis).
2. Die **Anzahl der Videos** (Steps).
3. Die **Gesamthandlung**.
# GENERIER... | ai-video | inspiration/video_frame_assistant.md | Markdown | ac554e23c06cf87a7b963807fe9a013f517de237f9ba3770747643ba3589d2cb | 0 | 444 |
# Cost Model Reference
## Pricing Table
### Gemini Image Generation
| Model | Cost per Image |
|-------|---------------|
| gemini-3-pro-image-preview (Nano Banana Pro) | $0.134 |
| imagen-4.0-generate-001 | $0.030 |
### Veo 3.1 Video Generation
| Model | 720p | 1080p |
|-------|------|-------|
| veo-3.1-generate-pre... | ai-video | reference/COST-MODEL.md | Markdown | 4515ff6fb4e158ec9c9c2e996418d75524563212118f310184f820650abc48bd | 0 | 544 |
# ElevenLabs API Reference
## Python SDK
```bash
pip install elevenlabs
```
```python
from elevenlabs import ElevenLabs
client = ElevenLabs(api_key="YOUR_KEY")
```
## Voice Design
Create a new voice from a text description.
```python
# Generate previews
response = client.text_to_voice.create_previews(
voice_... | ai-video | reference/ELEVENLABS-API.md | Markdown | c03b1d4729ce790ef5a3bb3041fe3cd2e593ec2a1cbf8ae9b37d9bc3439fde66 | 0 | 522 |
# Gemini API Reference
## Image Generation (Nano Banana Pro)
Used for character images and frame generation.
### Model
- `gemini-3-pro-image-preview` — high-fidelity native image generation
- `imagen-4.0-generate-001` — alternative Imagen model (English prompts only)
### Python SDK
```python
from google import gen... | ai-video | reference/GEMINI-API.md | Markdown | d9d4bc811ef3f55b98535d231e59b762e3ff6769bfa5906f224658727ab64e6d | 0 | 466 |
# Input Modes
The skill supports three input modes, detected automatically from the user's request.
## Mode A: Concept to Video
User describes what they want. Claude handles everything.
**Trigger**: A descriptive concept without a script or URL.
**Example**:
```
"Make a 45-second video of a young German woman vlog... | ai-video | reference/INPUT-MODES.md | Markdown | 8aadc81b8c0c76810e4f7f293cc4cbf89291b28f37e7dcbf37c65ba691b6dc69 | 0 | 465 |
# Pipeline Reference
Complete walkthrough of all pipeline stages.
## Stage 0: Project Setup & Cost Estimate
**Inputs**: User's request + mode detection
1. Create project directory with timestamp ID
2. Initialize StatusBar, CheckpointManager, CostEstimator
3. Detect mode (concept / script / YouTube clone)
4. Parse t... | ai-video | reference/PIPELINE.md | Markdown | dd05668e902df6a240066e6909ac116ac32136723f047ea20df4af81be077783 | 0 | 896 |
. Generate composition props from project state
4. Install Node.js dependencies
5. User launches Remotion Studio for preview
6. Optional: add subtitles, transitions, Ken Burns effects
**Checkpoint**: Remotion project configured
## Stage 8: Quality Assurance
1. Run automated QA pipeline (5 checks)
2. Present report w... | ai-video | reference/PIPELINE.md | Markdown | 9894ff1cb722d3f5d1b3c459964cc088338896edd65d982b73f8da45a6231244 | 1 | 156 |
# Quality Assurance Reference
## Automated Checks
The QA pipeline runs five automated checks on the assembled video:
### 1. Frame Continuity
Compares the last frame of each clip with the first frame of the next.
Uses perceptual hash (pHash) similarity (60% weight) and pixel-level
comparison (40% weight).
- Score >=... | ai-video | reference/QUALITY.md | Markdown | 57a57e8f747a4ac690d991ba80df6c2986d40a8cdebe3dabfd9911f3045cddf5 | 0 | 370 |
# Remotion Reference
## Overview
Remotion v4.x is used for video assembly, providing a browser-based editor (Studio)
and programmatic rendering.
## Setup
The template project lives at `templates/remotion-project/`. It is copied into
each project as the `editor/` directory.
```bash
cd editor
npm install
npx remotio... | ai-video | reference/REMOTION.md | Markdown | 32b1bec32dc652f585167ce8f95e8b37cbc14bd99952a44c7edad52718528b33 | 0 | 451 |
# Script Engine Reference
## Script Format
Scripts are structured as a list of segments, each representing an 8-second video clip.
```json
{
"title": "Weird Dream Vlog",
"language": "de-DE",
"total_duration_seconds": 32,
"segments": [
{
"id": 1,
"text": "Leute, ich muss euch was erzählen...",... | ai-video | reference/SCRIPT-ENGINE.md | Markdown | 77a10b4225a865c5e566538b5911b50eb3780e08592aefa17d2a166cba48e74f | 0 | 542 |
# Troubleshooting
## Common Errors & Recovery
### API Key Issues
**Error**: `Missing required environment variable: GEMINI_API_KEY`
**Fix**: Set the key in your `.env` file or system environment.
```bash
# .env file
GEMINI_API_KEY=your-key-here
ELEVENLABS_API_KEY=your-key-here
```
### Veo Generation Blocked
**Erro... | ai-video | reference/TROUBLESHOOTING.md | Markdown | 34ae72bc9a290b34be6207cce72265d80aa55ac0634e0ce97dee6ab7ab58c464 | 0 | 686 |
# Veo 3.1 API Reference
## Models
- `veo-3.1-generate-preview` — full quality, ~$0.35-0.50/clip
- `veo-3.1-fast-generate-preview` — draft quality, ~$0.10-0.15/clip
## Capabilities
- Text-to-video
- Image-to-video (start frame)
- First-and-last frame generation
- Video extension (from previous clip)
- Native audio gen... | ai-video | reference/VEO-API.md | Markdown | fbf5df6edb4cfa3ac3d71eec36baacca3e37aeef86c1c731074b1aeff83c932e | 0 | 562 |
# YouTube Clone Mode
## Overview
Clone a YouTube video's style, pacing, and structure with a new AI-generated character.
The original content is adapted (not literally translated) to a target language.
## Pipeline
```
YouTube URL
|
v
1. Download (yt-dlp)
→ source.mp4 + source_audio.mp3 + metadata
|
... | ai-video | reference/YOUTUBE-CLONE.md | Markdown | 1a1d32e361ca4721c9851bda06f7a56f53ce797faa3e73d533a5182e4a34094f | 0 | 403 |
from __future__ import annotations
import json
import logging
from pathlib import Path
from scripts.core.models import ProjectConfig, ProjectState, Script
logger = logging.getLogger(__name__)
class CompositionBuilder:
def __init__(self, config: ProjectConfig):
self.config = config
def build_props(... | ai-video | scripts/assembly/composition_builder.py | Python | d059bb68460d92ca76da32e85c48b67f5ea3bccc867d2b7a7f49faf395b39932 | 0 | 629 |
from __future__ import annotations
import logging
import subprocess
from pathlib import Path
from scripts.core.constants import (
DEFAULT_DURATION_TOLERANCE,
MIN_CLIP_DURATION,
DEFAULT_TRANSITION_DURATION,
DEFAULT_MUSIC_VOLUME,
)
logger = logging.getLogger(__name__)
class FFmpegAssembler:
@stat... | ai-video | scripts/assembly/ffmpeg_fallback.py | Python | 36b2babcd520d9cb922d3fdce5366bd99bbb4b354b9730149b682676aa90b5d6 | 0 | 896 |
y",
"-i", str(video_path),
"-i", str(audio_path),
"-c:v", "copy", "-c:a", "aac",
"-map", "0:v:0", "-map", "1:a:0",
"-shortest",
str(output_path),
]
else:
cmd = [
"ffmpeg", "-y"... | ai-video | scripts/assembly/ffmpeg_fallback.py | Python | 21abbb96ea634b448c1a6b8159d7244c91c8561d2b6b49c0e52c1aaa09100a73 | 1 | 896 |
if i > 1 else "[0:v]"
src_b = f"[{i}:v]"
out_label = f"[v{i}]" if i < n - 1 else "[vout]"
v_filters.append(
f"{src_a}{src_b}xfade=transition={transition_type}"
f":duration={td:.3f}:offset={offset:.3f}{out_label}"
)
cumulative = ... | ai-video | scripts/assembly/ffmpeg_fallback.py | Python | 1781749f3a2c4a9b43a497cde0b67783771c21017eca2ad18cbe02f9d898c89c | 2 | 896 |
.stdout.strip())
@staticmethod
def get_fps(path: Path) -> float:
cmd = [
"ffprobe", "-v", "quiet",
"-select_streams", "v:0",
"-show_entries", "stream=r_frame_rate",
"-of", "csv=p=0",
str(path),
]
result = subprocess.run(cmd, ca... | ai-video | scripts/assembly/ffmpeg_fallback.py | Python | 4367d6bec4cdc1e61eb55b85ea0100aa16d470719030ce7436bac9e9e49dec05 | 3 | 896 |
", "-y"] + input_files + [
"-filter_complex",
"".join(filter_parts) if len(graphics) == 1 else ";".join(filter_parts),
"-map",
output_label.replace("[", "").replace("]", ""),
"-map",
"0:a?", # Copy audio if present
"-c:v",
... | ai-video | scripts/assembly/ffmpeg_fallback.py | Python | c629576a8438be0f8cc29f20187f6d62d7fad29433fafe2dffb75d18b97c9d89 | 4 | 179 |
from __future__ import annotations
import json
import logging
import shutil
import subprocess
from pathlib import Path
logger = logging.getLogger(__name__)
_ROOT = Path(__file__).resolve().parent.parent.parent
_TEMPLATE_DIR = _ROOT / "templates" / "remotion-project"
class RemotionSetup:
def __init__(self, proj... | ai-video | scripts/assembly/remotion_setup.py | Python | f1d41d7141cb4c759e0c93ee85b683a9a4454d3bed383f96fae449b04cee4973 | 0 | 702 |
from __future__ import annotations
import logging
import subprocess
from pathlib import Path
logger = logging.getLogger(__name__)
class VideoRenderer:
def __init__(self, editor_dir: Path):
self.editor_dir = editor_dir
def render(
self,
output_path: Path,
composition_id: str ... | ai-video | scripts/assembly/render_video.py | Python | de9c417768442634d997520c4fee20bc8a86548a3a264eaa3bae888094cf22ff | 0 | 446 |
from __future__ import annotations
import logging
import subprocess
from pathlib import Path
from scripts.core.constants import AUDIO_DUCKING_MUSIC_LEVEL
logger = logging.getLogger(__name__)
class AudioDucking:
"""Audio ducking (sidechain compression) for professional audio mixing."""
@staticmethod
de... | ai-video | scripts/audio/audio_ducking.py | Python | bbcdd9f10bc046838c8897d4c7fb94b8ba41285bed0807f3cdd6c448f9b29a2c | 0 | 469 |
from __future__ import annotations
import logging
import subprocess
from dataclasses import dataclass, field
from pathlib import Path
from scripts.core.constants import (
DEFAULT_TARGET_LOUDNESS,
DEFAULT_REVERB_DAMPING,
DEFAULT_REVERB_WET,
DEFAULT_LOW_CUT_HZ,
DEFAULT_NOISE_GATE_THRESHOLD,
DEFA... | ai-video | scripts/audio/audio_effects.py | Python | ca71f56088bdd2efe47ad73add91051ed4a5309832158a2f8c5e9993585b5a5c | 0 | 896 |
(f"file '{p.resolve()}'" for p in audio_paths)
list_file.write_text(list_content, encoding="utf-8")
cmd = [
"ffmpeg", "-y",
"-f", "concat", "-safe", "0",
"-i", str(list_file),
"-c", "copy",
str(output_path),
]
logger.info("Con... | ai-video | scripts/audio/audio_effects.py | Python | 4682d8566af853e85f4f598f7bf2532ec11b0d392042720e695bbbc9529e0815 | 1 | 219 |
from __future__ import annotations
import logging
import subprocess
from pathlib import Path
logger = logging.getLogger(__name__)
class AudioExtractor:
@staticmethod
def extract_audio(
video_path: Path,
output_path: Path | None = None,
format: str = "mp3",
sample_rate: int = ... | ai-video | scripts/audio/audio_extractor.py | Python | 509c1e6086b7c2097e7c8b841fadebf203522c66c77974dd2b5050cce0770c92 | 0 | 538 |
from __future__ import annotations
import logging
from pathlib import Path
from typing import Optional
from scripts.core.edit_plan import MusicCue
logger = logging.getLogger(__name__)
class DynamicMusicGenerator:
"""Generate dynamic music segments based on EditPlan music cues."""
def __init__(self, eleven... | ai-video | scripts/audio/dynamic_music.py | Python | e073bf45fa6712105b8b8d5096f92f6e9fc7e3865493fa8a249012fcd3ebfda0 | 0 | 733 |
from __future__ import annotations
import logging
from pathlib import Path
from elevenlabs.client import ElevenLabs
logger = logging.getLogger(__name__)
class MusicGenerator:
"""Generate background music using the ElevenLabs Music API.
The Eleven Music API (``music_v1``) generates studio-grade instrumenta... | ai-video | scripts/audio/music_generator.py | Python | 3f36cfb16ab002f15a35cdf0e238453621456f579a3493ee0789ca7f69ffb4e3 | 0 | 896 |
music.composition_plan.create(
prompt=prompt,
music_length_ms=duration_ms,
)
return plan
def generate_with_plan(
self,
composition_plan: dict,
force_instrumental: bool = True,
output_format: str = "mp3_44100_128",
output_path: Path | N... | ai-video | scripts/audio/music_generator.py | Python | eefac1a20f282d3ed933c6033c96595affd90b9869140f658daf7834a2a7ba82 | 1 | 485 |
from __future__ import annotations
import logging
import subprocess
from pathlib import Path
from typing import Optional
from scripts.audio.audio_ducking import AudioDucking
from scripts.audio.audio_effects import AudioEffects, EffectChain
from scripts.core.edit_plan import EditPlan
logger = logging.getLogger(__name... | ai-video | scripts/audio/professional_audio_assembly.py | Python | afb5c72ad48a3f7d3249a9bde708788e3d3af90ab2d158de346b01fe480e3977 | 0 | 896 |
-y",
"-i",
str(voice_path),
"-i",
str(music_path),
"-filter_complex",
filter_complex,
"-map",
"[out]",
"-c:a",
"aac",
"-b:a",
"192k",
str(output_path),
]
... | ai-video | scripts/audio/professional_audio_assembly.py | Python | 8043fa403556afc4934bdff11313913cf34e5f8b52c537f533c010e9255099ff | 1 | 87 |
from __future__ import annotations
import logging
import re
from pathlib import Path
from elevenlabs import ElevenLabs
from scripts.core.models import ScriptSegment, VoiceProfile
logger = logging.getLogger(__name__)
class TTSGenerator:
def __init__(
self,
api_key: str,
model_id: str = ... | ai-video | scripts/audio/tts_generator.py | Python | 908c3a74f547a8cac304983196d9708adf60603b79126038d443fe7f5a904c96 | 0 | 763 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.