What you can make
- Background plates for titles and lower thirds, in 16:9, 9:16 or square, with room left for your text.
- Cut-out elements: objects, icons and props on a transparent background (OpenAI models), ready to animate on their own layer.
- Textures and overlays: light leaks, dust, grain and paper to blend over footage with Screen, Overlay or Multiply.
- Sky replacements and lower-third artwork without text, so the type stays live in EffectCraft.
Edit a frame of your comp
Use the current frame renders the active composition at the playhead and hands it to the model. Presets cover the common jobs:
- Remove background: a transparent PNG on OpenAI models, or the subject on pure green for Key Light on the others.
- Replace background or clean plate: remove a sign, a person or a rig and fill the gap, for a still background you can track or animate over.
- Relight or restyle a frame as a style frame or a look reference.
A still saved with Composition › Save Frame As › File… or rendered as a PNG from the Render Queue also lands in the panel as the image to edit. The download still happens as usual.
How it connects to EffectCraft
- In the editor, click AI tools in the top bar. The panel opens on the right; the app resizes to fit.
- Open the settings (gear icon), choose a provider and paste your API key. The key field links to the provider’s key page.
- Pick a preset or describe the image, choose an aspect ratio, and press Generate (or ⌘/Ctrl + Enter).
- Choose Add to comp. The image is imported into the Project panel and added to the active composition as a new layer at the current time. Import puts it in the Project panel only.
- Animate it like any footage layer: press P, S or T to reveal Position, Scale or Opacity, set keyframes, and apply effects from Effects & Presets.
Imported images are stored with your project in the browser, so they are still there after a reload.
Supported providers
| Provider | Default model | Get a key |
|---|---|---|
| OpenAI | gpt-image-2.5-flare | platform.openai.com |
| Google Gemini | gemini-3.1-flash-image | aistudio.google.com |
| OpenRouter | google/gemini-3.1-flash-image | openrouter.ai |
| Custom (OpenAI-compatible) | any model your API serves | your own endpoint |
OpenRouter gives one key access to many image models, including Gemini, GPT Image, FLUX, Seedream and Qwen Image. The custom option works with any service that implements OpenAI’s /images/generations and /images/edits endpoints and accepts browser requests.
Privacy: what goes where
| Data | Where it goes |
|---|---|
| Your API key | Stays in your browser. Sent only to the provider, in the request header. |
| Prompt and the frame you choose to edit | Sent from your browser straight to the provider you picked. |
| Your footage and the rest of your project | Never leaves your device. |
| effectcraft.run servers | Serve the static app files. They receive none of the above. |
The provider’s own terms and data policy apply to what you send them. Frames are scaled to 2048 pixels on the long side before upload to keep requests fast and within provider limits.
No key? Use EffectCraft’s own tools
Many jobs don’t need a model. Key Light (the Keylight-style keyer) pulls a green or blue screen, Roto Brush separates a subject across frames, Content-Aware Fill removes an object from a moving shot, and generators like Fractal Noise, Gradient Ramp and CC Particle World make backgrounds that move. All of them run on your device. See how to add effects to a video.
FAQ
Do I have to pay to use the AI tools?
You pay your AI provider for what you use, at their normal API prices. effectcraft.run charges nothing and adds no markup, because the request never passes through us.
Where is my API key stored?
In your browser only. By default it lasts for the current tab (sessionStorage). Tick “Remember my key on this device” to keep it in localStorage instead, and use “Forget all keys” in the panel settings to delete it.
Does the AI make the animation?
No. The models in the panel make still images. The motion comes from EffectCraft: you keyframe the generated layer’s position, scale and opacity, add effects such as Glow or Fractal Noise, and render. That keeps every move editable.
Can it generate video clips?
Not in this panel. It works with image models only. Generate a video elsewhere and import the file with File › Import › File…, or animate the stills the panel makes.
Which model should I pick?
For cut-out elements, an OpenAI GPT Image model with “Transparent background” ticked gives a PNG with real alpha. For backgrounds and frame edits, gemini-3.1-flash-image is quick and inexpensive and gemini-3-pro-image follows complex instructions better. “Load models” in the panel lists everything your key can use.
Why does a request fail with a CORS or network error?
The custom provider option needs an API that accepts requests from browsers. OpenAI, Google Gemini and OpenRouter all do. Some self-hosted or regional gateways don’t; use one of the built-in providers instead.