KI-generiert / AI-generated: Dieser Beitrag wurde mit KI erstellt und vor der Veröffentlichung nicht substanziell von einer menschlichen Redaktion geprüft.

xAI adds reference-guided 1080p video generation to Grok Imagine

xAI has expanded Grok Imagine Video 1.5 from image animation into a broader video-generation system with text-to-video, up to seven image references, native 1080p output, and API support. The July 31 update matters because it moves Grok Imagine closer to production-style creative workflows where teams need repeatable characters, products, locations, and audio cues rather than one-off clips.

Key takeaways

• Grok Imagine Video 1.5 now supports text-to-video, image-to-video, and reference-to-video.
• xAI says native 1080p is generally available on Grok.com Imagine, iOS, and Android.
• The API model grok-imagine-video-1.5 supports image references, text-to-video, and 1080p generation.
• Voice reference support exists, but xAI says it is available on request rather than broadly self-serve.
• Vercel AI Gateway lists the model with up to seven uploaded images and resolution controls, but flags video generation as a cost-sensitive workflow.

Grok Imagine moves beyond a starting image

The earlier Grok Imagine Video 1.5 release focused on animating an existing image with synchronized audio. The new release adds prompt-only generation, so a creator can describe a shot without first supplying a still frame. xAI says text-to-video combines its image generation and image-to-video systems, while native 1080p is now supported for both text-to-video and image-to-video.

That changes the model's role. It is no longer only a quick image animator for social clips; it can start scenes from text, preserve identity from reference material, and produce higher-resolution output for teams that need assets beyond mobile previews.

References target character and scene consistency

The strongest workflow change is multi-reference control. xAI says each reference image can lock a different element in place, such as a face, product, or location. The release supports up to seven references per generation, which gives creators a more direct path to holding a character while changing the setting, holding the setting while changing the subject, or keeping both while changing the action.

Voice consistency is more limited. xAI describes a workflow where a character image and voice reference hold across scenes, but the API section says voice reference support is available on request. That distinction matters for developers: image references are part of the live API story, while voice references may require separate access.

API access makes this a developer workflow

The official xAI article says image references, text-to-video, and native 1080p are live in the xAI API under grok-imagine-video-1.5. Its sample request uses client.video.generate, reference image URLs, duration, aspect ratio, and resolution parameters.

Vercel AI Gateway also lists Grok Imagine Video 1.5 as available through its model catalog. The listing shows a playground with image uploads, duration controls, 480p, 720p, and 1080p resolution options, plus provider pricing fields. That gives developers a second route for testing the model without wiring directly to xAI first.

Limits to check before scaling

The release is useful, but teams should treat it like a generation system with operational costs and access boundaries. Vercel notes that video generation can take minutes for longer or higher-resolution clips, and that billing varies by duration and resolution. It also says zero data retention is not currently supported for this model through AI Gateway.

Teams using brand assets, likenesses, or voice references should also review xAI's terms and their own consent workflow. Reference-guided generation is powerful precisely because it can preserve recognizable subjects across scenes, which raises approval, rights, and review requirements for commercial use.

The practical next step is a small benchmark: test short clips at lower resolution, measure prompt adherence and identity consistency, then move only the strongest prompts to 1080p.

Source check

• xAI's primary announcement confirms the July 31 update, text-to-video, native 1080p, multi-reference support, and API model name.
• xAI developer release notes list the July 31 grok-imagine-video-1.5 modality update.
• Vercel AI Gateway's model page corroborates availability through AI Gateway, up to seven images, resolution controls, and operational caveats.