What It Does
Diffusae 2 brings generative AI directly into After Effects. It runs Stable Diffusion models locally on your GPU, letting you generate images from text prompts, edit video footage with AI, and create animations with frame-to-frame consistency. All processing happens on your machine, no cloud services or usage fees.
You can type a prompt to generate a texture, use InstructPix2Pix to transform existing footage (turn summer into winter, add cowboy hats to people), or apply ControlNet for guided image generation. The plugin includes a temporal consistency model that reduces flickering between frames, making it viable for actual animation work.
Key Features
Multiple model support. Load Stable Diffusion 1.5, SD 2.0, SDXL, LCM, Turbo models, or custom checkpoints from Civitai. The Models Manager handles installation and lets you switch between them.
InstructPix2Pix editing. Edit video layers by describing what you want changed. The model interprets your prompt and transforms the footage accordingly, useful for style transfer or adding elements that weren’t in the original shot.
Temporal consistency model. Reduces jitter between frames when generating animations. This addresses the common issue where generative AI produces inconsistent results from frame to frame.
ControlNet integration. Use depth maps, edge detection, or pose data to guide image generation. Gives you more control over composition and structure than text prompts alone.
Keyframeable prompt weights. Animate how strongly different parts of your prompt affect the output over time. Lets you morph between concepts or adjust influence dynamically.
Custom model support. Import checkpoint models from Civitai (except those requiring login). Useful if you’ve trained specific styles or need specialized models for character consistency.
Illusion Diffusion. Generate images that contain hidden patterns or shapes when viewed from different perspectives.
Local rendering. Everything runs on your GPU. No internet required after initial model downloads, no per-generation costs, and your footage never leaves your machine.
Who It’s For
Motion designers who want to integrate AI-generated elements without leaving After Effects. VFX artists looking to test concepts quickly or generate background textures. Anyone doing experimental work where you need to iterate on AI outputs within your normal compositing workflow.
The plugin works best if you already understand how Stable Diffusion models behave, what prompts produce useful results, and when AI generation makes sense versus traditional techniques.
Pricing
$99.99 for a single-user license (one-time purchase). A special upgrade price of $112.50 is available for users who own previous Baskl.ai products. No subscription, no monthly fees, no per-generation costs. A trial version is available to test hardware compatibility.
System requirements: Windows users need an NVIDIA GPU with 6GB+ VRAM (CPU fallback available but slow). Mac users need Apple Silicon (M1 or newer) running macOS 11.0 or later.