agentleFS
mikeOnBreezemikeOnBreeze/cc-crossbeam/.claude/skills/fal-ai/SKILL.md

fal-ai

Skill297 starsChanged 9 months agoMITView on GitHub

This skill enables AI video generation from images AND text-to-speech voiceover generation using Fal.ai's API. Use this skill when the user asks to (1) generate videos from images (image-to-video), or (2) generate voiceovers/narration from text (text-to-speech via ElevenLabs). Works seamlessly with the nano-banana skill for image-to-video workflows. IMPORTANT: Check references/ for latest models and pricing - AI models change frequently.

  • Reads credentials
  • Installs packages

6 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 5 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Test with Veo Fast: - Verify quality improvement at $0.15/sec

    line 45
  2. Distinctive:

    When the user describes what they're looking for, match to these voices:

    line 142
  3. Distinctive:

    Models evolve rapidly: Check reference docs dates.

    line 213
  4. Distinctive:

    Video is expensive: Always be aware of costs. Iterate cheap, render expensive.

    line 214
  5. Distinctive:

    TTS is cheap: Run voice casting calls (~$0.02 for 3 samples) before committing to full narration

    line 215

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Fal.ai Media Generation

Generate AI videos from images and AI voiceovers from text using Fal.ai's API.

Capabilities:

  • Image-to-Video: Generate video clips from static images (see below)
  • Text-to-Speech: Generate voiceovers using ElevenLabs voices (see TTS section)

Prerequisites

This skill requires a FAL_API_KEY in the project's .env or .env.local file:

FAL_API_KEY=your_api_key_here

To obtain an API key, visit: https://fal.ai/dashboard/keys

Primary Capability: Image-to-Video

Generate video clips from static images using scripts/image_to_video.py:

python scripts/image_to_video.py image.png \
  --prompt "camera slowly pans right, gentle motion"

Parameters:

  • image (required): Path to input image or URL
  • --prompt, -p: Motion/action description for the video
  • --model, -m: Video model to use (default: kling)
  • --output, -o: Output directory or file path (default: current directory)
  • --duration, -d: Video duration in seconds (model-dependent, typically 5-10)

Model Selection Strategy

Video generation is expensive. Follow this workflow:

  1. Iterate with cheap models (Wan/Kling) - Perfect your prompts at $0.05-0.07/sec
  2. Test with Veo Fast - Verify quality improvement at $0.15/sec
  3. Final render with Veo Standard - Premium output at $0.40/sec

Quick Model Reference (January 2026)

FlagModelPrice/sec5-sec CostBest For
wanWan 2.5$0.05$0.25Cheapest iteration
klingKling 2.5 Turbo Pro$0.07$0.35Best value (default)
veo-fastVeo 3.1 Fast$0.15$0.75Quality test
veoVeo 3.1 Standard$0.40$2.00Premium final

For detailed model comparison, strengths/weaknesses, and latest updates, see references/api_reference.md.

Usage Examples

Basic video generation (default Kling model):

python scripts/image_to_video.py photo.png \
  --prompt "gentle breeze moves the leaves, soft lighting"

Budget iteration with Wan:

python scripts/image_to_video.py photo.png \
  --prompt "camera zooms in slowly" \
  --model wan

Premium render with Veo:

python scripts/image_to_video.py photo.png \
  --prompt "cinematic dolly shot, dramatic lighting" \
  --model veo \
  --output final_video.mp4

Specify duration:

python scripts/image_to_video.py photo.png \
  --prompt "waves crash on the shore" \
  --duration 10

Workflow with Nano Banana

Generate an image, then create a video from it:

# Step 1: Generate image with Nano Banana
python .claude/skills/nano-banana/scripts/generate_image.py \
  "a serene Japanese garden with cherry blossoms" \
  --output garden.png

# Step 2: Iterate with Kling (default, $0.35 for 5 sec)
python skills/fal-ai/scripts/image_to_video.py \
  garden.png \
  --prompt "gentle breeze moves cherry blossom petals, camera slowly pans right"

# Step 3: Final render with Veo when satisfied ($2.00 for 5 sec)
python skills/fal-ai/scripts/image_to_video.py \
  garden.png \
  --prompt "gentle breeze moves cherry blossom petals, camera slowly pans right" \
  --model veo

Text-to-Speech Voiceovers

Generate AI voiceovers using ElevenLabs voices via scripts/text_to_speech.py:

python scripts/text_to_speech.py "Your text here" --voice george

Parameters:

  • text (required): Text to convert to speech
  • --voice, -v: Voice name (see casting guide below)
  • --model, -m: TTS model (default: eleven-v3)
  • --output, -o: Output directory or file path
  • --stability: Emotion control 0-1 (lower = more emotion)
  • --similarity: Voice matching 0-1
  • --style: Expression exaggeration 0-1
  • --speed: Speaking pace 0.7-1.2
  • --list-voices: Show all available voices

TTS Model Reference (January 2026)

FlagModelPrice/1K charsBest For
eleven-v3ElevenLabs Eleven v3$0.10Latest, audio tags [whispers] etc.
turboElevenLabs Turbo v2.5$0.05Fast iteration, low latency
multilingualMultilingual v2$0.10Best stability

Voice Casting Quick Reference

When the user describes what they're looking for, match to these voices:

Female Voices:

VoiceBest For
rachelNarration, explainers, tutorials (calm, warm)
ariaConversational, podcasts (engaging, social)
sarahCorporate, professional (clear, neutral)
lauraMarketing, launches (upbeat, energetic)
charlottePremium brands (British, elegant)
lilyWellness, calm content (soft, gentle)

Male Voices:

VoiceBest For
georgeDocumentaries, serious narration (British, authoritative)
charlieCasual explainers (natural, relaxed)
rogerTrailers, announcements (deep, commanding)
ericNews-style, corporate (professional, clear)
chrisBrand voices, ads (warm, trustworthy)
brianEducational, history (mature, wise)

For full casting descriptions and parameter presets, see references/tts_reference.md.

TTS Usage Examples

Basic voiceover:

python scripts/text_to_speech.py "Welcome to our product demo." --voice george

Voice casting (run multiple to compare):

TEXT="Introducing the future of productivity."
python scripts/text_to_speech.py "$TEXT" --voice george -o casting_george.mp3
python scripts/text_to_speech.py "$TEXT" --voice eric -o casting_eric.mp3
python scripts/text_to_speech.py "$TEXT" --voice chris -o casting_chris.mp3

Documentary style (authoritative, slower):

python scripts/text_to_speech.py "In the depths of the ocean..." \
  --voice george --stability 0.65 --speed 0.95

Conversational style (more emotion):

python scripts/text_to_speech.py "Hey, check this out!" \
  --voice aria --stability 0.4 --style 0.3

With audio tags (eleven-v3 only):

python scripts/text_to_speech.py "[whispers] This is a secret..." --voice rachel

Resources

scripts/

  • image_to_video.py - Image-to-video generation script
  • text_to_speech.py - Text-to-speech voiceover script
  • requirements.txt - Python dependencies (install with pip install -r requirements.txt)

references/

  • api_reference.md - Video model comparison, pricing, best practices
  • tts_reference.md - Voice casting guide, parameter presets, TTS best practices

Notes

  • Models evolve rapidly: Check reference docs dates. If >1 month old, research latest models on Fal.ai before generating
  • Video is expensive: Always be aware of costs. Iterate cheap, render expensive.
  • TTS is cheap: Run voice casting calls (~$0.02 for 3 samples) before committing to full narration
  • Queue-based API: Generation takes time. Scripts show progress updates.
  • Output formats: Videos = MP4, Audio = MP3
  • Duration limits: Video models typically support 5-10 seconds. Check api_reference.md.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.