An Unbelievable Week for AI Video
The pace of change is relentless
Happy Friday! This week was not as productive as I’d hoped because I spent so much time watching demos of all the advancements in AI video. It’s been only 5 weeks since I wrote that the pace of change is unprecedented and even though I cover this stuff passionately, I was stunned by this weeks advances.
On Monday, Runway rolled out ‘Aleph’, their new cinematic Model that let’s take any image or video, and change the camera angle, the lighting, the scene, even the season. It’s like having multiple camera operators and a VFX team in your pocket and it’s mind blowing. Here’s a quick video I threw together.
Not to be outdone, Luma Labs released a similar feature on Wednesday that will let you restyle your footage with a single prompt. While I haven’t played with it yet, here’s a bit of their home page.
In June, when I was writing about the Seedance model (made by TikTok parent company Bytedance), I said that “all the major advances in AI Video were coming out of China” and that trend continues. Alibaba just released their Wan 2.2 video model which does things I’ve never seen before. Just go check out the promo site.
On Thursday, Krea announced a new image model that merges their technology with Flux Kontext with even more realistic results.
FLUX.1 Krea is a state of the art open source text to image model created with an emphasis on aesthetics and photorealism. Unlike other models trained on massive datasets, FLUX.1 Krea was trained on hand curated, high-quality training data to deliver the best results. Try it for free on Krea or download the weights and run it on your own hardware.
It’s currently free to test drive and I spent most of yesterday afternoon playing with it. Check it out
If you’re like me , your LinkedIn feed is probably exploding with those exploding Ikea box ads. While it looks complicated, it’s just a long JSON prompt that you put into Google’s VEO 3 model. You can copy and paste this and edit the details if you keep the syntax. Thanks to Seb Jeffries for this prompt.
{"description": "Cinematic shot of a sunlit Scandinavian bedroom. A sealed IKEA box trembles, opens, and flatpack furniture assembles rapidly into a serene, styled room highlighted by a yellow IKEA throw on the bed. No text.",
"style": "cinematic",
"camera": "fixed wide angle",
"lighting": "natural warm with cool accents",
"room": "Scandinavian bedroom",
"elements": [
"IKEA box (logo visible)",
"bed with yellow throw",
"bedside tables",
"lamps",
"wardrobe",
"shelves",
"mirror",
"art",
"rug",
"curtains",
"reading chair",
"plants"
],
"motion": "box opens, furniture assembles precisely and rapidly",
"ending": "calm, modern space with yellow IKEA accent",
"text": "none",
"keywords": [
"16:9",
"IKEA",
"Scandinavian",
"fast assembly",
"no text",
"warm & cool tones"
]
}
So what does all this mean? These tools are incredibly democratic and accessible and now anyone with a vision can tell their story. However, for those of us in the creative professions, the bar has been raised and when anyone can make an exploding box Ikea ad, it’s not that amazing. For me, I’m going to focus on my writing as a key value proposition. It’s not worth blowing through video credits until I have a solid idea.
That’s it for today, thanks for reading, keep in touch, do amazing things, stay creative, consider sending this newsletter to a friend…or even buy me a coffee.