Generate and edit images and short videos using xAI Grok media APIs.
npx skills add https://github.com/X-manist/Cohmira --skill grok-media-x-manist
Please help me install this Agent Skill. Skill: grok-media Source: https://github.com/X-manist/Cohmira/tree/main/src/builtin-plugins/openmontage/.agents/skills/grok-media Command: npx skills add https://github.com/X-manist/Cohmira --skill grok-media-x-manist
POST /images/generations and single or multi-image editing via POST /images/edits, including prompt structuring guidance. - Video Generation: Supports text-to-video, image-to-video, and reference-image video through POST /videos/generations with async polling on GET /videos/{request_id}. - Pricing and Failure Handling: Documents per-image and per-second pricing, plus explicit handling of failed and expired polling states. - Use Case: A content team wants a short product video featuring a specific person and jacket. They submit a reference-image video request mapping <IMAGE_1> to the person and <IMAGE_2> to the jacket, then poll until the status is done and download the 720p clip. ## Quick Start Use the grok-media skill to generate a 720p reference-image video from these two image URLs with a slow push-in camera motion.High-intent search queries and answers about installing and using this skill.
Send a POST request to /images/generations with model, prompt, n, aspect_ratio, and resolution fields. Authenticate with an Authorization header containing Bearer $XAI_API_KEY against the base URL https://api.x.ai/v1.
Submit a POST to /videos/generations with a reference_images array of image URLs, then poll GET /videos/{request_id} until status is done. Reference placeholders like <IMAGE_1> in the prompt map each source image to a role in the video.
Image-to-video uses the image field to anchor the starting frame of the clip. Reference-image video uses reference_images to influence who or what appears without locking the first frame, which suits product placement and identity consistency.
Grok video outputs at 480p or 720p. Pricing is $0.05 per second at 480p and $0.07 per second at 720p, plus $0.002 per input image for image-conditioned requests.
The async polling endpoint can return failed or expired states after submission succeeds. Treat an expired status as a provider or runtime issue, preserve the endpoint, mode, and prompt summary in the error, and resubmit rather than silently switching providers.
Grok video is best treated as short-form generation, so it is a weak fit for long-form clips, workflows relying on deterministic seeds, or prompts with multiple scene changes. Keep prompts to one shot, one motion idea, and one emotional beat.
AgentPMTConvert PDF, Word, Excel, PowerPoint, HTML, images, and audio files into Markdown text.
ai-service-incubatorGenerate 1080x1920 MP4 short-form videos from scripts with subtitles, images, and music.
nuyoah-ai-worksGenerates Chinese portrait photography prompts and reshoots reference photo sets with new subjects.
memoirlabsGenerate text-to-video and image-guided videos via Hugging Face or Diffusers pipelines.