From prompt to premiere: tracking AI-made features on their way to theaters. The tech. The talent. The box office. You ain't seen nothing yet!
USA
Joined October 2026
- Tweets856
- Following85
- Followers66
- Likes175
Pinned Tweet
Potemkin City is finished. A silent short set in 1920s New York, staged as a homage to the Odessa Steps: a baby's pram rolls down the stairs and nobody looks up from their phone.
I made it in 2 days with @grok, @ChatGPT, @magnific and @ElevenLabs. How it was built, shot by shot:
Lights, Camera, Prompt retweeted
Euro buyers looking for "distinctive, conversation starter" projects from emerging filmmakers at AFM
More: buff.ly/Bxe5EHx
SO GOOD! Into the Rain by Visional Studio won the Grand Prix at the Silicon Valley AI Film Festival's Los Angeles competition, held Nov 1 at the @Saban_Theatre in Beverly Hills. The AI film has a viewing page on the festival's site.
Co-hosted by @ViduAI_official and @Zopia_AI, with @RextrixAI as the LA division's Diamond Sponsor.
Watch: svaiff.com/watch/la20260038
2
27
Silicon Valley AI Film Festival LA Division winners, from Nov 1 at the Saban Theatre in Beverly Hills: the Grand Prix went to Into the Rain (Visional Studio), the Grand Jury Prize to Lost & Found (Gleb Kusakin), and Best Feature Film to A Song of the Sacred Tree and the Endless Night. No distribution deals were announced. The next step is getting these AI film festival winners into theaters.
einpresswire.com/article/955…
40
The Open Source Higgsfield winners from @ComfyUI are out. nomadoor's ComfyUI Panorama Stickers won Best Overall and $10K, and RTX 5090s went to Blocksfield, Imagination Studio and AInVFX's Paint a Plate. @ComfyDevs: "Then they tried to shut us down. 300+ of you shipped anyway." @higgsfield_ai hasn't responded publicly.
Closer look: Higgsfield didn't try to shut down the contest itself. They demanded their brand and CEO's likeness be removed from the marketing.
comfy.org/open-source-higgsf…
1
53
Powerful open-source models keep arriving, but US filmmakers can't just download them and use them: @MiniMax_AI released open weights for MiniMax-H3-Image, which generates and edits 16-megapixel images in about 10 seconds. Low-VRAM builds for local AI were on @huggingface within hours, and @ComfyUI has a tutorial. The catch is that the license excludes the US, EU, UK and South Korea without MiniMax's authorization.
huggingface.co/MiniMaxAI/Min…
58
Open 4K finishing is here, but US filmmakers need to read the license first: @NVIDIAAI's Sana team posted Open-H3-Regenerate-2K-IC on @huggingface. It's an open-source add-on that regenerates 768p MiniMax H3 AI video drafts at 1080p, 2K or 4K while keeping faces, lip sync and lettering. It falls under the MiniMax H3 Community License, which requires MiniMax's authorization in the US.
huggingface.co/Efficient-Lar…
1
1
68
More hardware options for open-source video models should make AI filmmaking cheaper over time: @pika_labs thanked a researcher for bringing open-weight Pika H3 Flex AI video inference to @AMD GPUs. The GitHub branch targets AMD's Instinct MI350X data-center chip, not consumer cards yet.
teapaw.org/pika_labs/status/21199…
Huge thanks to @shiyi_c98 for her help in bringing Pika H3 Flex inference to AMD GPUs 💜
We’re excited to collaborate with more researchers in the community. If you’re an AMD developer, we’d love for you to contribute to this branch!
Check out her work: github.com/Pika-Labs/Pika-H3…
1
29
AI filmmaking toolmakers are now on the record in the biggest AI copyright case.: Judge Stein formally admitted 55 amicus briefs in @nytimes v. @OpenAI, including ones from @midjourney, @runwayml and Cate Blanchett, per @EtLegalworld. It's a procedural order, not a ruling.
legal.economictimes.indiatim…
1
68
Veteran studio directors are moving into AI features: The Lion King co-director @RobMinkoff will direct Storm Dogs, an AI talking-dog family film from Particle6, the producer behind Tilly Norwood, per @THR. The toddler and the dogs are AI-generated with human voice actors, and adult actors appear in select live-action scenes. It's privately financed and will look for studio distribution later.
hollywoodreporter.com/movies…
14
Lights, Camera, Prompt retweeted
AI avatars are getting RIDICULOUSLY good
we're reaching a point where you are able to create full long form YouTube vlogs 100% with AI
ElevenCreative just launched Chimera and this opens up some pretty crazy possibilities
here's how it works 👇
🤝 Paid partnership
Lights, Camera, Prompt retweeted
Claude Fable 5.2 built entire city with 10 agents:
Each agent had its own job:
>Tower agent: 4,560 glass panes
>Traffic agent: 858 vehicles, 2,259 people
>Rooftop agent: 88 kinds of props
>Camera agent: 7 angles
👇 Full breakdown in Blender with Higgsfield MCP:
Adobe Premiere (@AdobeVideo) can now generate missing b-roll, music and sound effects right on the timeline, with a choice of AI video models from @AdobeFirefly, @Google, @Kling_ai, @runwayml and @LumaLabsAI, per @Adobe at Adobe Max. The new Generative Remove beta can take a boom mic or a logo out of a shot, on clips up to 10 seconds for now.
blog.adobe.com/en/publish/20…
1
3
189