Stable Cascade: 3 Stage Image Generation Model for ComfyUI

Stable Cascade is a 3-stage image generation model by Stability AI. Stage C generates low-res latents, Stage B upscales, Stage A VAE decodes to pixels.

S

Stable Cascade

Text-to-ImageMulti-StageOpen Source

Three-stage image generation model by Stability AI. Stage C generates a low-resolution latent, Stage B upscales it, and Stage A VAE decodes to pixel space. Also supports ControlNets (canny, inpainting) for guided generation.

Hugging FaceGitHubPaper
DeveloperStability AI
Release Date2024-03
Architecture3-stage: Stage C (DiT) → Stage B (DiT upscaler) → Stage A (VAE)
LicenseStable Cascade License (see HF repo)
Supported ControlNetsCanny, Inpainting (via stage_c controlnet)
VRAM Requirement16 GB+ for full pipeline

Guides and workflows related to this model series.

No articles found.