You can now create video in Imaginations with two new models: Gemini Omni Flash and Luma Ray 3.2. Both let you turn a prompt, a start frame, or existing footage into video, so you can experiment with more visual styles without leaving your workspace.
New: Gemini Omni Flash and Luma Ray 3.2 are now available in the video model selector.
What each model does
Gemini Omni Flash
Gemini Omni Flash is a flexible video model that adapts to how you start:
From a start frame — turn a single image into video (image-to-video).
From a prompt only — generate video directly from text (text-to-video).
From reference images — supply up to 10 reference images to guide the look of your video (reference-to-video).
From an existing video — provide a source video plus a prompt to edit it.
Luma Ray 3.2
Luma Ray 3.2 focuses on generating video from a start frame or a text prompt:
From a start frame — animate a single image (image-to-video).
From a prompt only — generate video from text (text-to-video), with a choice of six aspect ratios so you can match the format to your channel.
How to use it
Open Imaginations and start a new video generation.
In the model selector, choose Gemini Omni Flash or Luma Ray 3.2.
Add your inputs: a text prompt, a start frame, reference images, or a source video, depending on the result you want.
If you're using Luma Ray 3.2 with a prompt only, pick your aspect ratio.
Generate, then review and refine your results.
Tips
Use reference images with Gemini Omni Flash when you want the video to stay close to a particular visual style — you can supply up to 10.
When you generate from reference images, remember to include at least one image before you start. The generation won't run without one.
As with all Imaginations output, these models get you a strong draft. Add your finishing touches to land exactly on brand.