A world model and a photo effect answer different requests. A world model, in the two October 2026 announcements on this site, either rebuilds a place or simulates one. Most HiVideoStudio effects take a photo and play a template. Face Swap also takes a reference video. The three names on the late-October watchlist sit in three lanes. None of the new names is for sale here.
Three lanes
Echo-3 is the reconstruction lane. On 27 October 2026, a PlayCanvas post said a clip was generated by Echo-3 and rendered by the open-source PlayCanvas Engine. The quoted post says a phone video of a room became a 3D Gaussian splat, and the room was restyled. SpAItial's blog the same day says the capture is a few minutes on an iPhone and the result is metric, meaning distances match the real space. The engine is open source. The model weights are not in those posts.
Genie 4 is the simulation lane. DeepMind's post the same day describes a world that keeps going, with keyboard controls and text prompts that change weather, camera, or who is present, and with music, ambience, and dialogue generated with the picture. The preview went to trusted testers. TechCrunch reported a scene coherence of up to two minutes. DeepMind's limitation section says details can still drift. HiVideoStudio is not affiliated with Google or DeepMind. Face Swap is one to three character images and a reference video, 150 credits for 15 seconds at 768P.
Henge is the performance lane, and it is not a world model. Alisson Anjos's MIT studio, posted the same morning, takes a guide video and reference pictures on ComfyUI and MiniMax H3. Character Swap, Head Swap, Motion Transfer, and Sharpness are the modes in the README. The guide's frames lock the pose and the camera. Dot Henge is a search variant, not a domain this site registered. Henge is not for sale here.
The photo-effect lane is the catalog. One still, one template, a credit price shown before the render. Face Swap takes one to three character images plus a reference video: 150 credits, 15 seconds, 768P.
What the photo lane sells
Ozil Goal AI puts one photo into a fixed 15-second goal. The price is 50 credits. The performance does not come from a video you upload.
AI Zombie puts one photo into one of three 15-second scenes, with sound, at 45 credits. Survival Horror AI is a 15-second rainy flashlight scene at 50 credits. Thriller Dance is a 15-second dance at 50 credits, and it is silent. Hotel Lobby is the two-photo effect, an orange set, 90 credits for 15 seconds.
A couple zombie, a pumpkin head, a vampire, and the named song dances are planned and not for sale. Planning them did not add a world model. Face Swap is the guide-video recast on sale: one to three character images, 150 credits, 15 seconds at 768P.
If the search was "AI character swap video" or "AI head swap video from a photo," the sorting page is AI character swap video. If the search was the two world models against each other, the sorting page is Echo-3 and Genie 4.
What not to expect from a 15-second effect
A 15-second template does not hold a room at metric scale. It does not let you type a new camera move halfway through and keep the same simulation. A one-photo template does not take the dance you filmed. Face Swap takes a reference video and one to three character images, 150 credits for 15 seconds at 768P. The effect pages state the input, the length, and the credit cost.
The upload is JPEG, PNG, or WEBP, up to 10 MB, one person unless the effect is Hotel Lobby. Rules are in how to pick the input photo.
How to pick a lane
- Phone video of a real place, and you want a splat you can look around: read Echo-3. This site does not build it.
- A made-up place you want to steer, with generated sound: read Genie 4. This site does not host the preview.
- A clip you already shot, and up to three faces to place on it: open Face Swap. The price is 150 credits for 15 seconds at 768P.
- One photo and a scene this catalog already has: open that effect and read the credit price. That is the photo-template lane.
Quick answers
Are Echo-3 and Genie 4 video effects? No. Echo-3 in the cited posts reconstructs a capture. Genie 4 simulates a world. The effects catalog is a separate list of photo templates.
Is Henge a world model? No. It edits identity and motion inside a guide video. It does not scan a room and it does not steer a generated world.
Which photo effects are actually for sale? The indexed catalog. On the public 15-second table, Face Swap is 150 credits, Ozil Goal AI and Survival Horror AI and Thriller Dance are 50, AI Zombie is 45, and Hotel Lobby and Bulin 47 and Burning Bridges AI are 90. Planned templates have no price. Henge is not for sale.
Does this site generate any of the three new tools? No. The model pages and the Henge page are introductions. They are not for sale.
Sources
Echo-3: PlayCanvas, David Novotny, SpAItial. Genie 4: DeepMind and TechCrunch. Henge: Alissonerdx and the README. Prices from the public 15-second table in this repo. Last checked 2026-10-29.