vargai 0.3.2 → 0.4.0-alpha2

package/docs/plan.md

@@ -1,66 +0,0 @@
-# varg ts-sdk migration plan
-
-## core
-
-- [ ] File (load from path, url, buffer)
-- [ ] File.toTemp() - save to temp file for editly integration
-- [ ] scene`` tagged template for composing elements in prompts
-- [ ] types (AspectRatio, Element, etc)
-
-## providers
-
-- [ ] @varg/fal (kling, nano-banana, lipsync)
-- [ ] @varg/elevenlabs (tts, music)
-- [ ] @varg/higgsfield (soul, characters)
-- [ ] @varg/heygen (talking heads)
-- [ ] @varg/openai (sora, gpt-image, dall-e)
-- [ ] @varg/replicate (birefnet, generic models)
-
-## generation
-
-- [ ] generateImage (text to image)
-- [ ] generateImage (image to image / edit)
-- [ ] generateVideo (text to video)
-- [ ] generateVideo (image to video)
-- [ ] generateSpeech (voiceover via elevenlabs)
-- [ ] generateMusic (music generation via elevenlabs)
-- [ ] generateLipsync (sync video to audio via fal)
-- [ ] generateTalkingHead (photo to talking video via heygen)
-- [ ] generateElement (character/item/style reference for consistent generation)
-
-## editing (use editly directly)
-
-users use editly npm package directly for video composition.
-we provide `File.toTemp()` to bridge ai outputs → editly paths.
-
-editly covers these py-sdk features:
-- [ ] crop_to_aspect_ratio → editly clips with resizeMode
-- [ ] mix_audio_with_ducking → editly audioNorm + audioTracks
-- [ ] apply_zoom_effect → editly zoomDirection on layers
-- [ ] create_ken_burns_effect → editly zoomDirection: "in" | "out"
-- [ ] create_slideshow → editly clips[] with transitions
-- [ ] concatenate_videos → editly clips[] with transitions
-- [ ] picture_in_picture → editly layers with position/size
-- [ ] blur_background_resize → editly resizeMode: "contain-blur"
-- [ ] 67 gl-transitions (fade, crossfade, wipe, cube, etc.)
-
-## custom effects (not in editly)
-
-- [ ] addCaptions (tiktok-style word-by-word) - custom fabric.js layer for editly
-- [ ] createSlider (before/after reveal animation)
-- [ ] createPushTransition (card push effect)
-- [ ] createSwipeAnimation (tinder-style card swipes)
-- [ ] createSplitScreen (side-by-side comparison)
-- [ ] createPackshot (end card with cta button + blinking cta)
-
-## image processing
-
-- [ ] removeBackground (via replicate birefnet)
-
-## notes
-
-- editly used directly (not wrapped)
-- File.toTemp() bridges ai outputs to editly string paths
-- providers are separate packages (@varg/fal, @varg/elevenlabs, etc)
-- generateElement returns { images, text } for use in prompts
-- scene`` template composes elements in prompts

package/docs/todo.md

@@ -1,14 +0,0 @@
-
-# TODO
-
-• [ ] fal
-• [ ] higgsfield
-• [ ] captions (exactly as in TikTok)
-• [ ] editing (simple concat of video frames)
-• [ ] voiceover via elevenlabs
-• [ ] music generation via elevenlabs
-• [ ] adding music on top of video
-• [ ] lypsync (via sync models over fal)
-• [ ] resize with blur: i.e. 9:16 is resided to 4:5 by adding sides with a blur on all video
-• [ ] Picture-in-the-picture: one video on top of another
-• [ ] fade / slideshow transitions