OpenMontage includes over 50 Python tools for video production. These are auto-discovered at runtime by the tool registry and grouped by category. Each tool declares its provider, runtime (LOCAL, API, or LOCAL_GPU), tier, and schemas. Selectors abstract families of providers for graceful fallback. Inspect the live set with this preflight command:
python -c "
from tools.tool_registry import registry
import json
registry.discover()
print(json.dumps(registry.provider_menu_summary(), indent=2))
"See the tool system for the BaseTool contract and selector pattern.
Analysis (4 tools)#
These tools extract structure and semantics from source footage or references.
transcriber(WhisperX): word-level transcription and speaker diarization.scene_detect: scene boundary detection via PySceneDetect + FFmpeg.frame_sampler: keyframe or uniform frame extraction with metadata.video_understand(CLIP/BLIP-2): semantic video understanding and embedding.
Audio (8 tools)#
Tools for voice, music, and post-processing.
elevenlabs_tts,google_tts,openai_tts,piper_tts: text-to-speech (ElevenLabs, Google Cloud, OpenAI, local Piper).tts_selector: ranks and routes among available TTS providers.music_gen: background music generation.audio_mixer: multi-track mixing and level balancing.audio_enhance: noise reduction and clarity processing.
Avatar (2 tools)#
talking_head(SadTalker/MuseTalk): animated talking-head video from audio and reference image.lip_sync(Wav2Lip): lip synchronization on existing video.
Enhancement (5 tools)#
Visual cleanup and quality upgrades.
upscale(Real-ESRGAN): resolution increase.bg_remove(rembg/U2Net): background removal.face_enhance: face detail improvement.face_restore(CodeFormer/GFPGAN): face restoration.color_grade(FFmpeg LUTs): color correction and grading.
Graphics (13 tools)#
Image and graphic asset creation.
flux_image,grok_image,google_imagen,openai_image,recraft_image,local_diffusion: image generation (fal.ai, xAI, Google, OpenAI, Recraft, local diffusion).pexels_image,pixabay_image: stock image retrieval.image_selector: routes image requests across providers.code_snippet: code block graphics.diagram_gen: diagram generation.math_animate(ManimCE): mathematical animation.image_gen(deprecated).
Subtitle (1 tool)#
subtitle_gen: SRT/VTT generation from timestamped text.
Video (18 tools)#
Video generation, stock retrieval, and editing primitives.
grok_video,heygen_video,higgsfield_video,veo_video,kling_video,runway_video,minimax_video: cloud video generation (xAI, HeyGen, Higgsfield, fal.ai Veo/Kling/MiniMax, Runway).wan_video,hunyuan_video,cogvideo_video,ltx_video_local,ltx_video_modal: local GPU video models.pexels_video,pixabay_video: stock video retrieval.video_selector: routes video requests.video_compose(FFmpeg): scene composition (dispatches to Remotion or HyperFrames when locked).video_stitch,video_trimmer: concat, trim, and basic assembly.
Composition runtime choice is locked at proposal and enforced by video_compose. See composition runtimes and provider guide for setup and selection details.