Research index
Research profiles
Every tool we track, grouped by what it actually does. 13 of 13 carry published research; anything unresearched is marked Not researched rather than filled in with estimates.
Text & image to video
- Kling AI Moderately supported Complex physical movement and holding a character's face steady across a shot.
- Higgsfield Not established Camera-move control and stylised cinematic shots, with access to several models in one place.
- Runway Gen-4.5 Moderately supported Creative and editorial work — generative editing, motion brush and in-context controls.
- PixVerse Limited evidence Fast short-form output for TikTok, Reels and YouTube Shorts on a low monthly budget.
- Luma Dream Machine Limited evidence Naturalistic motion and image-to-video from a single still.
- Vidu Limited evidence Reference-to-video — feeding it a character or object and keeping it consistent.
- Google Flow Limited evidence Scene-by-scene filmmaking with Veo, if you are already inside Google's AI subscription.
- Sora 2 Not established Discontinued. OpenAI closed the Sora app and web in April 2026; the API closes 24 September 2026.
Avatars & presenters
Upscaling & restoration
- Topaz Video AI Moderately supported Upscaling, denoising and frame interpolation on footage you already have.
Editing & assembly
- InVideo AI Limited evidence Turning a script or prompt into a finished edit with stock footage, subtitles and voiceover.
- CapCut Not established Free general-purpose editing with AI helpers, strongest for short-form social.