QueyFramesQueyFrames
gemini aikeyframesaudioblogearly access

gemini ai · video and images

A Gemini AI video generator with the rest of the shot attached.

QueyFrames runs Google’s Gemini models for video, images and text — but not one prompt box at a time. Each one is a node on a canvas, so the still you generated becomes the first frame of the clip, the script you wrote becomes the voiceover, and the beats you marked re-time the animation over the top.

Every call is server-side and metered in credits. There is no API key to paste and nothing about your account reaches the browser.

sign up for early access→common questions

What Gemini does on the canvas

Four of the node types call Gemini directly, and a fifth — the chat sidebar — uses it to build the graph.

  • video · omni-flash

    text to video · image to video

    Gemini omni-flash generates 3–10 second clips at 720p, 16:9 or 9:16. Prompt it cold for text-to-video, or wire a still in as the first frame for image-to-video. Extra images ride along as references, each one addressable from the prompt by position.

  • image

    gemini image generation

    Prompt-driven stills with model, aspect ratio and size selects. Publish the result and it feeds the video node's first frame, the render node's backdrop, or the timeline as a still clip.

  • llm

    gemini text

    Text generation with a model select and a free-text or list output mode. Wire up to six inputs and place each one into the prompt by name, so a prompt is assembled from the graph rather than retyped.

  • chat agent

    gemini, in the sidebar

    Ask for the graph and it builds it: adds nodes, writes their prompts, sets keyframes, connects them and lays them out. It runs the same connection validation your pointer does, so it can't wire anything a hand couldn't.

Two things the prompt has to carry

Gemini’s video request has no field for timing and no field for what an image is for — an image is known only by its position in the request. QueyFrames writes both into the prompt for you.

  • timing

    Wired keyframes are numbered across every input slot and placed individually — {keyframe1}, {keyframe2} — so you can write "then {keyframe2} and hold" mid-sentence. A keyframe with no description falls back to the properties it moves, or to a cut point.

  • references

    Images are sent in a fixed order — first frame, then references — with an "image 1 / image 2" preamble, and the prompt names each one as {image1}, {image2}. Order is the contract, so reordering never silently re-points a reference.

Gemini AI video generation — common questions

Which Gemini model generates the video?
Video comes from Gemini omni-flash, Google's video generation model. Clips run 3 to 10 seconds at 720p and 24fps, in 16:9 or 9:16 — the model chooses the duration from the prompt.
Can Gemini generate video from an image?
Yes. Wire an image node into the video node's first image slot and the still becomes the opening frame, which is image-to-video generation. Add more images and they ride along as references you can name in the prompt.
Does it also generate images?
Yes — the image node calls Gemini image generation with a prompt plus model, aspect ratio and size selects. Its output feeds the video node as a first frame, so you can art-direct a still first and animate it second.
What else runs on Gemini here?
Text generation for the llm and list nodes, the structured render plans behind the Remotion node, brand color palettes, and the chat agent in the sidebar that can build the canvas for you.
Do I need a Gemini API key?
No. Every Gemini call is made server-side against a QueyFrames key and metered in credits on your account, so nothing reaches the browser and there is no key to manage.

Keep reading

  • QueyFrames — AI video generator on a node canvas

    The AI video generator the whole canvas is built around — video, images, audio and text on one graph.

  • AI keyframe animation for AI video

    What a keyframe means here: a timing every node reads, so motion, cuts and audio land on the same beat.

  • AI music, voice and sound effects generator

    The audio half — AI music, AI voice generation and sound effects, all publishing timings the video nodes read.

  • AI video generation guides and workflows

    Guides and workflow notes — prompting for pacing, image to video, and keeping audio and picture on the same beat.

Generate a still, animate it, cut it to the beat.

QueyFrames is in early access. Create an account to join the list.

sign up for early access→

QueyFrames — AI video generator by QuokkaQuery

overviewgemini aikeyframesaudioblogtermsprivacysign up