Decart released an update this month enabling real-time character generation and motion at high frame rates, with upgraded models globally.

The company positions the feature for streamers who need animated avatars driven by arbitrary input images in live broadcasts and low latency.

Real-time generation and streamer focus

Decart's engine renders characters at 30 fps, accepting any uploaded or captured image as a starting reference for motion in real time.

Animations aim to follow user input continuously, producing smooth motion even for atypical subjects such as a penguin character during live streams.

Latency is reported around 1 second, which the company describes as acceptable for most live streaming scenarios and interactive broadcasts with minor buffering.

Quality and current limitations

Image fidelity has improved significantly compared with earlier public iterations, but artifacts remain visible on character outlines and textures under fast motion.

Render softness and general blurriness affect peripheral details, which can reduce perceived sharpness on streamed avatars at higher resolutions and scaling settings.

Eye tracking shows the most frequent failures, producing a gaze that sometimes looks slightly swollen or misaligned with head motion.

Availability and outlook

The update is currently distributed free of charge, with no immediate paid tier announced for the new generation pipeline across supported platforms.

Decart likens the release to earlier moves by other vendors, noting similarities to Krea's rollout approximately two years ago in approach.

Independent testing and broader adoption will determine whether the quality meets streaming professionals' requirements under sustained real-world load and low-bandwidth conditions.

According to CGIT_Vines, early demonstrations highlight the platform's real-time capabilities and remaining visual artifacts observed in community tests.