AI Studio media generation
Use video references, TTS voices, and supported attachment limits in Clico AI Studio.
Clico AI Studio routes media requests to the appropriate Gemini model. Choose the generation type first, then set a preferred model and output parameters when needed.
For the complete workflow, see Get started with AI Studio. For conversational tasks, see Chat in AI Studio.
Attachments
You can attach up to 14 files to a chat, including up to 3 videos. An attached video has a role toggle:
- Edit source uses one video as the source for editing or extending it. Only one source video can be attached to a request.
- Video reference uses videos to guide a new generation. All attached videos must use the same role; edit sources and references cannot be mixed in one request.
When you add multiple videos to a new request, Clico treats them as references. Reference clips must be 3 seconds or shorter. Gemini Omni supports up to 3 reference videos for clips up to 10 seconds, and up to 1 reference video for 11–15 second clips. For first/last-frame generation, use two images; those clips are limited to 10 seconds. Omni accepts up to 6 image references per generation, subject to Clico's overall attachment limit.
Each message can also include up to 2 audio files and 2 documents. Gemini's inline Interactions request limit is 20 MB; Clico keeps a safety margin and limits the complete inline request to 18 MB to leave room for prompts and tool metadata. Compress or remove attached media if a request exceeds that limit.
Video settings
Clico's default video duration is 5 seconds. Available durations depend on the generation type and references. Fifteen-second clips support at most one reference video; extensions are limited to 3–10 seconds per turn and 30 seconds total.
Text-to-speech voices
The voice picker searches Google's prebuilt voice catalog and your saved voice remixes. Choose a voice for single-speaker speech, or ask Studio Chat for a dialogue with two speakers. Multi-speaker output supports up to two prebuilt voices; saved custom voices are available for single-speaker generation.
Image resolution
Nano Banana 2 Lite outputs at 1K. When it is selected, Clico limits the resolution picker to 1K and resets incompatible 2K/4K selections. Other image models expose the resolutions they support.
Music settings
For Lyria 3.5, choose a genre, vocal mode (automatic, vocals, or instrumental), and an approximate target length. These choices are added as prompt guidance; the Gemini API does not expose them as strict generation parameters, so the resulting duration is not exact. Lyria 3 Clip always produces a 30-second clip, so Clico locks its length setting to 30 seconds. Lyria 3.5 can use up to 10 attached images as music references.