Visual storytelling with AI media generation

MiniMax H3 Max and Wan 3.0 are now available for turning a still image into a moving shot, and both generate sound along with the picture. H3 Max follows a prompt unusually closely, so a shot you describe in detail comes back looking like what you asked for. Wan 3.0 stretches to 30 seconds, far past the short loops most models allow, and costs a fraction of what the premium models charge for comparable quality. Both let you supply an end frame when the action needs to land on a specific composition.

GPT Image 2.5 Flare and Seedream 5.0 Pro are now in the model list. Flare is the one to reach for when words have to appear in the picture — signs, labels, packaging, poster headlines come back legible instead of garbled, so put them in quotes in your prompt. It can also put a subject on a transparent background, which gives you a character or prop you can drop straight onto another shot without cutting it out. Seedream 5.0 Pro is the photographic one: strong realism at 2K, and it prompts natively in many languages, which often gives more authentic detail for culturally specific scenes.

Every assistant and specialist in LORY now runs on GPT-5.6 Luna, the fastest model in OpenAI's GPT-5.6 family. Chat, story writing, character and location design, storyboarding, media generation, voice selection and reference gathering all use it. Expect quicker replies in chat and shorter waits on every playbook run.

A new Referrals page lets you share your unique link, watch your invited friends sign up, and earn rebates on every pack they buy. Friends who join through your link get a larger welcome pack, and you get an instant signup token plus a percentage rebate on each of their paid pack purchases for a period. Check the referrals page to see the program rules and earning examples.