This is insanely COOOOL
video-shotcraft just crossed 11K stars and it's an open-source skill that turns Claude Code or Codex into a full film studio
point your agent at your product and it storyboards, animates, sound-designs, and renders a cinematic promo video no video editor, no agency, no budget
what it does:
- 157 shot recipe cards covering entrances, product close-ups, title reveals, beat-synced cuts, and more
- 214 motion previews you can browse and pick before you start
- 2.5D camera moves, real page captures, film-grade SFX all composed as a Remotion project
- ships with Ink Press, a validated 36-second promo template you can swap your product into immediately
setup (under 2 min):
step 1: tell your agent
Install this skill for me: https://t.co/8ssABGYVcJ
step 2: point it at your product
describe your app, drop in screenshots and copy
step 3: let it render
it outputs a finished 1920×1080 30fps MP4 with cuts, captions, and sound design baked in
important:
> Apache-2.0, completely open source
> built on Remotion (free for individuals and small teams; companies check Remotion's license)
> runs locally your product screenshots and copy never leave your machine
> also check out video-talkcraft, the narration-video sibling by the same author
bookmark this before it hits 100K stars
This is insanely COOOOL
video-shotcraft just crossed 11K stars and it's an open-source skill that turns Claude Code or Codex into a full film studio
point your agent at your product and it storyboards, animates, sound-designs, and renders a cinematic promo video no video editor, no agency, no budget
what it does:
- 157 shot recipe cards covering entrances, product close-ups, title reveals, beat-synced cuts, and more
- 214 motion previews you can browse and pick before you start
- 2.5D camera moves, real page captures, film-grade SFX all composed as a Remotion project
- ships with Ink Press, a validated 36-second promo template you can swap your product into immediately
setup (under 2 min):
step 1: tell your agent
Install this skill for me: https://t.co/8ssABGYVcJ
step 2: point it at your product
describe your app, drop in screenshots and copy
step 3: let it render
it outputs a finished 1920×1080 30fps MP4 with cuts, captions, and sound design baked in
important:
> Apache-2.0, completely open source
> built on Remotion (free for individuals and small teams; companies check Remotion's license)
> runs locally your product screenshots and copy never leave your machine
> also check out video-talkcraft, the narration-video sibling by the same author
bookmark this before it hits 100K stars
You can use UNCENSORED VIDEO AI MODEL for totally free locally
Sulphur 2 is at 101K downloads and 2.1K likes on Hugging Face
uncensored local video model based on LTX 2.3, per its card. text to video and image to video, plus the other LTX 2.3 formats, and you run the weights yourself. official inference isn't out yet
what it does:
- text to video
- image to video
- other LTX 2.3 formats, per the card
- public weights, not gated
- ComfyUI workflows in the repo, plus a local prompt enhancer
how to get it:
step 1: go to https://t.co/x1n4of7ioy
step 2: download one dev build
sulphur_dev_fp8mixed is 29GB, sulphur_dev_bf16 is 46GB
step 3: grab the distill lora from the same repo
card says start with one dev build plus that lora. workflows still say sulphur_final, so use the lora or the full model, not both at once
step 4: load the t2v or i2v workflow from the workflows folder in ComfyUI
important:
> LTX-2 community license, not MIT
> companies with $10M+ annual revenue need a paid commercial license from Lightricks
> uncensored is the card's word. the license still bans non-consensual deepfakes, anything involving minors, and bypassing LTX safety filters
> official inference is coming soon, so this is a local download, not a site you just open
> better setup notes are also still coming, per the card
bookmark it before official inference lands
You can use UNCENSORED VIDEO AI MODEL for totally free locally
Sulphur 2 is at 101K downloads and 2.1K likes on Hugging Face
uncensored local video model based on LTX 2.3, per its card. text to video and image to video, plus the other LTX 2.3 formats, and you run the weights yourself. official inference isn't out yet
what it does:
- text to video
- image to video
- other LTX 2.3 formats, per the card
- public weights, not gated
- ComfyUI workflows in the repo, plus a local prompt enhancer
how to get it:
step 1: go to https://t.co/x1n4of7ioy
step 2: download one dev build
sulphur_dev_fp8mixed is 29GB, sulphur_dev_bf16 is 46GB
step 3: grab the distill lora from the same repo
card says start with one dev build plus that lora. workflows still say sulphur_final, so use the lora or the full model, not both at once
step 4: load the t2v or i2v workflow from the workflows folder in ComfyUI
important:
> LTX-2 community license, not MIT
> companies with $10M+ annual revenue need a paid commercial license from Lightricks
> uncensored is the card's word. the license still bans non-consensual deepfakes, anything involving minors, and bypassing LTX safety filters
> official inference is coming soon, so this is a local download, not a site you just open
> better setup notes are also still coming, per the card
bookmark it before official inference lands