Communitygithub.com

FilippTrigub/abra

>- Animate a still image into a short video clip using fal.ai's LTX-2.3 Fast image-to-video model in the cloud. No GPU required - runs entirely on fal.ai serverless infrastructure.

abra とは?

abra is a Claude Code agent skill that >- Animate a still image into a short video clip using fal.ai's LTX-2.3 Fast image-to-video model in the cloud. No GPU required - runs entirely on fal.ai serverless infrastructure.

対応✓Claude Code~Codex CLI~Cursor
npx skills add https://github.com/FilippTrigub/abra/tree/HEAD/skills/animate-image

お気に入りのAIに質問する

このエージェントスキルを事前に読み込んだ状態で新しいチャットを開きます。

ドキュメント

liven — Image to Video (Cloud)

Brings still images to life using the LTX-2.3 Fast video diffusion model running on fal.ai's cloud infrastructure. Provide an image and a motion prompt; the model generates a short video clip (6-20 seconds depending on settings).

The skill directory (where this SKILL.md lives) is referred to as $SKILL_DIR below.

Cloud-based. No local GPU required. Requires FAL_API_KEY environment variable. Pricing: $0.04/s for 1080p, $0.08/s for 1440p, $0.16/s for 4K.


When to Use

Use this skill when the user wants to:

  • Animate a photo or illustration into a short video without local GPU
  • Create cinematic motion from a still (e.g. slow zoom, pan, light flicker)
  • Generate social media video content from product shots or portraits
  • Use higher resolutions (up to 4K) without local VRAM constraints

Setup (first run only)

cd "$SKILL_DIR" && uv sync

Set the FAL_API_KEY environment variable:

export FAL_API_KEY="your-api-key-here"

Agent Workflow

1. Ask the user

Before I animate the image(s), I need to know:

💬 Motion prompt  (required)
   Describe the motion you want, e.g.:
   "slow cinematic push-in, golden hour light, subtle camera drift"

🎬 Clip settings
   - duration     — video length in seconds (default: 6, options: 6, 8, 10, 12, 14, 16, 18, 20)
                    Note: durations >10s only support 25 FPS and 1080p
   - resolution   — output quality (default: 1080p, options: 1080p, 1440p, 2160p/4K)
   - aspect_ratio — video aspect ratio (default: auto, options: auto, 16:9, 9:16)
   - fps          — frames per second (default: 25, options: 24, 25, 48, 50)
   - generate_audio — generate AI audio to match video (default: true)

📁 Input / output directories  (default: ./input and ./output)

Wait for user response before proceeding.

2. Edit config.json

Write or update $SKILL_DIR/config.json based on the user's choices.

3. Run

cd "$SKILL_DIR" && uv run python scripts/img2vid.py --config config.json

4. Report results

Tell the user:

  • Output file paths
  • Video duration and resolution
  • Cost estimate (based on duration and resolution)
  • Direct download URLs (videos are hosted on fal.media temporarily)

Config Reference

KeyValuesDefaultDescription
input_dirpath./inputFolder containing source images
output_dirpath./outputDestination folder
promptstring(required)Motion description
durationinteger6Video duration in seconds (6-20)
resolutionstring1080pOutput resolution: 1080p, 1440p, 2160p
aspect_ratiostringautoAspect ratio: auto, 16:9, 9:16
fpsinteger25Frames per second: 24, 25, 48, 50
generate_audiobooleantrueGenerate AI audio for the video
end_image_urlstringnullOptional end frame for transition videos

Common Invocations

# Animate with a motion prompt (uses defaults: 6s, 1080p, 25fps)
cd "$SKILL_DIR" && uv run python scripts/img2vid.py \
  --prompt "slow cinematic push-in, golden hour light"

# Generate a longer 10-second video
cd "$SKILL_DIR" && uv run python scripts/img2vid.py \
  --prompt "gentle wind in the trees" --duration 10

# Create a vertical video for social media
cd "$SKILL_DIR" && uv run python scripts/img2vid.py \
  --prompt "subtle camera movement" --aspect-ratio 9:16

# 4K quality (higher cost: $0.16/s)
cd "$SKILL_DIR" && uv run python scripts/img2vid.py \
  --prompt "cinematic zoom out" --resolution 2160p --duration 8

# Higher frame rate for smoother motion
cd "$SKILL_DIR" && uv run python scripts/img2vid.py \
  --prompt "fast action sequence" --fps 48

# Disable audio generation
cd "$SKILL_DIR" && uv run python scripts/img2vid.py \
  --prompt "silent meditation scene" --no-audio

# Create transition between two images
cd "$SKILL_DIR" && uv run python scripts/img2vid.py \
  --prompt "smooth morphing transition" \
  --end-image /path/to/end_image.jpg

Output

Each input image produces one .mp4 clip in output_dir, named <image_stem>.mp4. The video is also uploaded temporarily to fal.media CDN and a download URL is provided in the console output.

Video metadata (resolution, fps, duration) is printed after generation.


Error Handling

  • Missing FAL_API_KEY → clear error with setup instructions
  • Invalid image format → error with supported formats (PNG, JPEG, WebP, AVIF, HEIF)
  • API errors → fal.ai error message propagated
  • Network issues → retry logic with clear error messages

Pricing Reference

ResolutionPrice per second
1080p$0.04
1440p$0.08
2160p (4K)$0.16

Examples:

  • 6s video at 1080p = $0.24
  • 10s video at 1080p = $0.40
  • 10s video at 4K = $1.60

Individual skills in this repo

This repo contains 7 individual skills — each has its own dedicated page.

FilippTrigub/abra

>- Auto-describe and caption images using a local vision-language model. Writes a JSON sidecar per image containing a one-sentence description, a suggested Instagram caption with hashtags, and detected content tags.

FilippTrigub/abra

>- Animated caption pipeline. Use this skill when the user wants to burn word-by-word animated captions into videos — using Whisper for transcription and pycaps for rendering. Supports default minimalist style or a futuristic CSS theme with alternating gold/magenta glowing words.

FilippTrigub/abra

>- Cut videos into segments, rearrange them, and produce an output video with a specific cuts-per-second rate. Uses MoviePy for video manipulation. Prioritizes audio transcription for timestamped cutting, falls back to adaptive scene detection.

FilippTrigub/abra

>- Edit a region of a video using a text prompt via Wan2.1-VACE inpainting. Supports two modes: background (auto-segment via rembg) or region (rectangle defined by fractions). Requires a CUDA GPU with at least 8 GB free VRAM.

FilippTrigub/abra

>- Video enhancement pipeline. Use this skill when the user wants to sharpen, colour grade, warm, or normalise the audio of videos — including presets for natural, cinematic, or vivid looks.

FilippTrigub/abra

Generate videos from text or images using Higgsfield's multi-model cloud platform. Supports kling, seedance, dop, and dop-preview models. Uses a small preset layer for common creative styles. Auto-detects text-to-video or image-to-video based on input. No GPU required.

FilippTrigub/abra

>- Remove the background from every frame of a video using AI (BiRefNet-general via rembg). Outputs transparent-background video or composites onto a solid colour or image. Requires a CUDA GPU with at least 3 GB free VRAM.

関連スキル