Future of Video Editing. How 'Vibe Editing' can fix audio, remove silences, add B-roll, correct eye contact, and even translate your video.

For anyone in Los Angeles, the third annual AI on the Lot is happening this May 28-29 in Culver City. It's the largest gathering of professionals interested in the AI film space. Tickets just went on sale! https://www.aionthelot.com/

Kling 2.0 has officially dropped, and while it's still early days, it looks like a meaningful step forward in realism, motion handling, and editability. The big addition is a new Multi-Elements Editor, which allows for swapping, adding, or removing parts of a shot—potentially a game-changer for those trying to build more complex or cinematic scenes. Prompt comprehension also seems improved, and character motion looks a little less robotic, though I’ll reserve judgment until I’ve had more hands-on time. Meanwhile, Veo 2—Google’s latest video model—is now accessible via Gemini for Advanced users. It can generate 8-second clips from text prompts with noticeably more realism and detail than earlier models. There's also a new feature called Whisk Animate that lets you bring static images to life—interesting, especially for storyboard-style workflows or quick visual tests. Plenty to dig into here. As always, curious to see how these tools perform in real workflows, not just in curated demos. More to come.

Hey AI Filmmakers, As a film editor who's had his hands on pretty much every AI video tool out there, I wanted to share my thoughts on Runway's new Gen 4 model that dropped last week. The Good, The Bad, and The Reality Let's be straightforward here. Runway Gen 4 brings some meaningful improvements to character and scene consistency across shots - something that's been a real pain point in AI video generation (VentureBeat). When characters shift appearance between frames or objects randomly change, it immediately breaks the illusion. Gen 4 does a better job handling this than their previous models. But here's where I'm disappointed: for a fourth-generation product from a company that just raised $308 million in funding (TechCrunch), I expected more substantial improvements in the actual quality of video generation. The model still routinely misinterprets prompts, creating outputs that technically "work" but miss the creative intent. How Does It Stack Up? When I compare Gen 4 to competitors like Kling, Hailuo, and Google's Veo-2, Runway is falling behind in a critical area: accurately producing what you ask for. These other models often require fewer attempts to get something that actually matches your prompt. Where Runway does shine is their control features. Their camera controls let you manipulate movement across six different axes (Stewart Gauld), giving you precise control over how your scene unfolds. But great controls paired with inconsistent generation quality is like having a professional steering wheel on a used car - the core performance still lets you down. My Advice for You For those of you working with these tools professionally: 1. Evaluate for your specific needs: Don't just go with Runway because it's the most talked about. Consider if its strengths align with your particular requirements. 2. Budget time for multiple attempts: When using Runway Gen 4, plan for several generation attempts to get results that actually match your creative vision. 3. Mix and match tools: Consider using Runway for specific technical shots where its controls excel, while using Kling or Hailuo for shots where accurate prompt interpretation matters more. 4. Be specific with prompts: I've found that extremely detailed prompts can help overcome some of Gen 4's interpretation issues, though it shouldn't be necessary. 5. Keep expectations realistic: Despite all the marketing hype, we're still in the early days of this technology. No model gets it right 100% of the time.