Kling AI Video is FINALLY Public | Impressions & Testing w/ Jack

Kling AI’s video generator has gone public globally, removing the earlier requirement for a Chinese phone number and shifting access to a straightforward website—an upgrade that matters because it turns a previously gated tool into something creators can test and iterate on immediately. Early impressions place Kling among the top-tier options available now, repeatedly described as a “true Sora competitor,” with particular strength in physics understanding and in animating people and animals in believable motion.

The workflow centers on a web interface with community galleries, where users can generate videos from either text prompts or uploaded images. Kling also adds image-to-video features that can animate specific subjects—such as a user’s own photo, a pet, or a famous painting—into short clips. The generator supports “video extensions,” letting users continue generation for longer outputs (up to roughly three minutes), though the default clip length is shorter (the tester cites defaults around five seconds, with additional limits such as 10 seconds depending on settings). The interface includes multiple modes, including a “high performance” option and a “high quality” mode that is “launching soon,” plus aspect ratio controls (notably 720p outputs and support for common formats like 16:9, portrait, and square).

Controls go beyond basic prompting. Users can specify camera movement types—horizontal, vertical, zoom, pan, tilt, and roll—and can apply negative prompts to steer results away from unwanted elements. A “creativity in relevance” slider appears to balance prompt adherence versus more natural motion, with the tester leaving it centered during early trials. Credits are metered: the account receives 66 credits every 24 hours, which the tester equates to about six generations, and there’s no immediate way to buy more credits yet. A subscription plan is expected, implying future monthly costs for heavier usage.

Hands-on testing produced mixed but often compelling results. Text-to-video attempts included a 1970s Bigfoot dancing at a disco and a drone-like shot over a candy/ice-cream/donut world. The drone sequence was praised for temporal consistency and “smoothness,” even if the exact candy-world look didn’t fully land. Image-to-video was where Kling stood out: a realistic burger-munching clip looked convincing enough to be described as “mind blown,” and a dog animation maintained the original aspect ratio while producing believable eye and head movement. Some image-to-video prompts showed “lost translation,” such as an octopus playing banjo turning into a different animal, suggesting prompt interpretation can falter—possibly when English prompts interact with internal language handling.

Community-made generations reinforced the pattern: Kling often excels at animating characters performing actions (people eating, animals interacting with objects, cinematic lens effects like fisheye), while occasional artifacts appear—especially in longer extensions or complex scenes involving hands, occlusions, or precise object interactions. Overall, the public release positions Kling as one of the most accessible and competitive video generators right now, with its strongest early edge in character motion and physics-adjacent realism, particularly for human and animal scenarios.