Runway Aleph and JSON Ads | 15 AI Film News Updates
This past week was crazy. In this article and video, I’ll talk about 15 of the latest AI Film News headlines. Buckle Up.
In the world of AI Film News, this past week was full of new updates and AI tools. From Runway’s Aleph tools to Luma’s update to Modify, and even HunYuan’s newest release, there is so much that you need to know about.
Below is both a video and an article diving deep into the latest AI Film News. Check it out, and I hope you enjoy it. :)
AI Film News Video
Below is the video highlighting the latest AI Film News. We cover tons in the video, but if you want even more updates, check out the article below or subscriber to our newsletter to get it sent straight to you.
The Biggest AI Film News Updates Yet
Here’s your complete breakdown of the most important AI film news and tools of the week. Make sure you ceck out the video above too for some tutorials on how to practically use the latest updates.
1. Runway Aleph: Edit Reality with Text Prompts
Runway’s newest model, Aleph, just dropped—and it might be the most practical and powerful AI filmmaking tool we’ve seen yet.
Aleph allows creators to modify live-action or AI-generated footage with simple, natural language. Instead of clunky frame-by-frame rotoscoping or style transfers, Aleph enables what Runway calls “multi-task generation,” where you can:
Add or remove objects
Transform lighting, time of day, environment, or weather
Generate new angles and “coverage” of a scene
Modify wardrobe, style, or background
Replace characters using uploaded reference images
Original Video
Output Video
Aleph isn’t perfect—especially when it comes to fast motion—but it’s a major leap for directors who want to reshape their footage after the fact. It’s currently available via Generate Video > Gen 4 > Chat Mode inside Runway for CPP users.
2. Luma’s Modify Video with Instructions
Just after Aleph was released, Luma Labs announced their own natural language video editor: Modify Video with Instructions.
Original Video
Output Video
The tool lets you upload footage and prompt changes using everyday language. In early side-by-side testing, Luma had a tendency to over-transform footage or miss subtle details, but it’s a sign that major players are racing to turn text into fully controllable post-production workflows.
Expect comparisons between Aleph and Luma to become a regular part of the creative process.
3. Midjourney’s Start & End Frame for Video Generation
Midjourney released a major update to its video generation system—adding start frame and end frame controls for more deliberate storytelling.
Start Frame
End Frame
Output
Now you can:
Define a continuous, stylistically aligned motion shot
Add looping functionality
Extend videos shot-by-shot using manual prompts and matching keyframes
While transitions still require careful styling, this update makes Midjourney one of the most usable AI video generators for creators who want precision and continuity.
4. JSON Prompting Goes Viral After IKEA Ad
A viral IKEA ad, created using Google’s Veo 3, made waves online for its use of JSON prompting—a structured format that gives AI very specific, step-by-step scene instructions.
Unlike freeform prompts, JSON makes outputs:
More consistent and repeatable
Easier to control with logic and formatting
Better aligned to brand and narrative goals
Check out the custom GPT we created for you to use when creating a JSON prompt.
🎯 Why It Matters:
If you want your AI video tool to act like a commercial production assistant, JSON is the secret weapon. We even made a free JSON prompt converter to help you try this technique yourself.
5. Runway and IMAX bring AI Films to the Big Screen
AI filmmaking is officially going mainstream.
Runway announced a partnership with IMAX to screen 10 AI-generated films from its AI Film Festival (AIFF) at 10 U.S. theaters from August 17–20, including:
Definitely check it out if you are in one of these areas!!
6. Kling Elements Update
Kling’s powerful “Elements” feature got a big upgrade in version 1.6.
Now you can upload multiple references (e.g. wardrobe, environment, objects), select specific regions like face, costume, or background, and even generate outputs with better stylistic consistency
The main downside is that this update is currently only in Kling 1.6, not 2.1. You’ll need to use the Kling website directly (not Freepik or Krea) to access it.
7. Higgsfield “Steal” Chrome Extension: Copy Any Image
Higgsfield released a Chrome extension called “Steal” that lets you right-click any image on the web and instantly apply its lighting, composition, and framing to your own AI-generated characters or scenes.
It’s a deceptively powerful shortcut for filmmakers and prompt artists who want to match reference photos without writing complex camera descriptions.
Bonus chaos: some generations randomly include sound effects or music.
8. Wan 2.2 for ComfyUI: Open-Source Powerhouse
The open-source video model Wan 2.2 is now available inside ComfyUI, and it’s seriously powerful. Wan 2.2 offers better aesthetics, text rendering, camera control, and temporal consistency
You’ll of course need some strong hardware (or offloading) to run it smoothly—it’s heavy. But Lora training is already available, and compatibility with 2.1 tools is strong.
9. Ideogram “Character” Model for Consistent Faces
Ideogram now lets you train a character from a single image. Upload a photo, enter a prompt, and it’ll generate the same person from multiple angles.
✅ Pros:
Photorealistic and consistent
Great for repeat characters or multi-shot storytelling
Works well with multiple perspectives
⚠️ Cons:
Still struggles with facial expressions
Emotion accuracy isn’t quite there yet
It’s free to try and probably the most accurate cloud based character reference tool when it comes to visual continuity.
10. OpenAI GPT-5 Is Coming Soon
Sam Altman confirmed on X that GPT-5 is “coming very soon,” likely in August.
This new version is expected to combine GPT-4 reasoning, GPT-4o performance, as well as Operator-style agent tools
Early testers say it’s smarter about when to use what, potentially giving users the best balance of speed, quality, and context.
Could it reshape how we generate and structure scripts or pre-production materials? We’ll see.
11. YouTube Shorts Now Has AI Video Tools
YouTube Shorts is getting a new wave of AI tools, including:
Convert any image into a 6-second video (powered by Veo 2)
Improved “Dream Screen” features
Watermarked labels via DeepMind’s SynthID
YouTube is signaling that creators can use AI—but only if it’s properly labeled. Expect more clarity around AI monetization soon.
12. Hunyuan World Model 1.0: Open 3D Scene Generation
Tencent released Hunyuan World 1.0, an open-source model for generating 3D worlds from either text or image prompts.
Key Features:
Compatible with standard graphics pipelines (VR, games, simulations)
Can export as mesh files
Combines panoramic generation with hierarchical 3D scene layout
Part of a broader open-source push, including Hunyuan3D, HunyuanVideo, and Hunyuan-A13B
Access requires a China-compatible login (for now), but it’s a huge deal for game devs and immersive worldbuilders.
13. AI Films of the Week
Simon Meyer’s Miniature Car Commercial. Simon won our previous AI Advertising commercial and is an incredibly talented AI Filmmaker and creative.
Check out the ad and let us know what you think!
Wonderful work on this, Simon!
14. CR Meetups & Events
We have Curious Refuge AI Filmmaking meetups happening all over the world! blow are pictures from the recent meet up in Toronto. It looks like an absolutely blast!
Here are some of the meet ups coming up. Check out our AI Film Events page for a full list.
Denver – August 2
Seattle – August 6
Kuala Lumpur – August 16 (Runway Meetup)
Lagos, Nigeria – September 13
Africa’s first AI film festival: NAIFF!
15. Comedy sketch submissions
Don’t forget that we are only one week away from the submission deadline for our AI Sketch comedy competition.
We are incredibly excited to see all of your submissions and cannot wait to share your films with the world!
As always, reaach out if you have any questions!
AI Film News Sent to You!
Fill out the form below to get AI Film News sent straight to your inbox every week! We would love to help you stay update on the latest news. We make it to where you don;t have to search, but you get weekly update emails and video sent straight to you so that you are up to date on all of the latest AI Video tool updates.
I hope you enjoyed this weekly rundown of the AI Film news, and you every have any questions or want a specific tutorial make sure to leave a comment or shoot us a message!