Show HN: Open-source AI video editor

[drochetti]

https://github.com/fal-ai-community/video-starter-kit

Hey HN community! I'm one of the lead devs of this project at fal.ai and we created an open source lightweight video editor powered by the latest media AI models. The main goal was to tackle some challenges when dealing with complex media handling and encoding on the browser.

It all started as an internal experiment but as we tackled some of the issues it was clear there could be some value sharing it with the open source community.

Some of the key points and tech stack details:

- It uses IndexedDb, so all data is local (i.e. no auth, no cloud db)

- Multiple AI models for video, image, music and voice-over. APIs are provided by fal.ai

- Built with the typical React+Next.js, Shadcn front-end

- Used remotion.dev for the realtime video preview (this is such a great project, without it the codebase would be twice as large)

- File uploads so you can bring your own media by uploadthing.com

- ffmpeg for encoding the final video and also some ui tricks, like the audio waveform

We deployed a version of it and for now it's free to use. We do plan to add some rate limiting and a bring your own API Key next, but it's open source and I'm curious about what the community will build on top of it, or derive from it. Customize your own video app and if you do, please share.

If you have any questions, hit me up!

[satvikpendem]

I was just looking for something like this, is this similar to RunwayML and their editor?

[lastdong]

Thank you for open sourcing the project, the UI looks fantastic. I noticed that several comments have mentioned the potential to improve functionality by allowing users to replace the backend with local models, such as Hunyuan. Would you be open to considering this feature in the future? Additionally, if other developers were to contribute, do you think this could be a possibility?

[drochetti → lastdong]

Absolutely. The idea of being open source with a permissive license is that we're encouraging anyone to do whatever fits their use case.

You can replace anything, deploy on your own server, port it to other stacks... whatever brings value to you.

We're also open to PRs, cut an issue in the repo and we can get the conversation going.

[voiper1]

Wow, that's so fun. Just upload and generate assets super easily... queue the next one while it's working.

[Jayakumark]

Great release , where is the source code for FFMPEG library used?

[asadm]

this looks great! I would like standard timeline scrup, drag-drop etc.

Some eta on how long a generation will likely take.

[kfarr → asadm]

Second on timeline scrub (move timeline position depending on x position of click on timeline) and drag and drop asset from media into timeline. Also spacebar is usually a basic universal shortcut to play/pause

[drochetti → kfarr]

Drag and drop the media to the timeline, and drag the media along the timeline track is already supported.

We will keep improving the UI, including shortcuts. Thanks a lot for the feedback.

[drochetti → asadm]

Thanks for the feedback, eta would be great indeed. I'll look into it.

[oldgradstudent]

Cool project, I played with it a bit a few hours ago.

A small bug report: if you have both a still image and a video in the timeline, the video does not play in the preview.

[aykutkardas → oldgradstudent]

Glad you liked it. And thanks for reporting, I'll look into the issue.

[peppertree]

Open source part is just a thin layer of UI. Built to peddle their closed backend, which is built on others' open source work. Nice.

[drochetti → peppertree]

That's fair. But I know a lot of smart folks out there that have trouble building that "thin layer of UI". So if that helps them, mission accomplished.

Anyone can replace the AI layer with their own local models, other services... whatever suits your use case and preferences is fair game.

[echelon → peppertree]

Fal wants to sell compute, and since they're straggling behind the compute-as-an-API leaders they've attached their cart to the art, image, and video subset of the market.

They think they've got a leg up on their model partners. That the foundation model companies will mostly become generic copies of one another, servants of the compute layer. There are so many foundation video models now, and they'll battle it out over dwindling margins. Pika, Runway, Kling - they're all the same. And there's also growing open source foundation models.

The thing that stands in Fal's way is that the future of AI video for artists is local. Hunyuan and Comfy can run on desktop machines, and real artists hate the SaaS model and non local stuff. It doesn't look like we'll even need A100s to reach Pixar levels of quality.

The ones to watch in the art space are Comfy and Invoke. And Adobe.

Fal probably has a future in powering the API-driven YouTube Shorts slop, though there's probably an upper limit to the number of talking head startups.

But there's no way they win artists over to cloud. Sophisticated tools are going to be local. Any SaaS art tools that do achieve escape velocity will buy and manage their own compute and won't go through Fal as a reseller.

[echelon → bobxmax]

> This is just naive cope.

Seems to me like you're not watching a growing trend.

I'm deeply embedded in this community and comfy is the biggest thing in it.

https://trends.google.com/trends/explore?date=today%205-y&q=...

https://trends.google.com/trends/explore?date=today%205-y&q=...

https://trends.google.com/trends/explore?date=today%205-y&q=...

Everyone is tired of paying for generations they don't use and wants local.

Local isn't some perfect little fisher price UI that some product manager approved. Local can be mixed and matched and pushed to the limits.

Topaz Labs' run rate is more than Runway and Pika, so don't write off local tools.

[SrslyJosh]

Cool, I'd love to see more slop on YouTube.

[S0y → aykutkardas]

100% AI generated comment?