The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface. The fastest local inference engine in the world.
Fix VAEEncodeForInpaint to support WAN VAE tuple downscale_ratio (#11572)
rattus6207f86c18d2
diff --git a/nodes.py b/nodes.pyindex 56b74ebe3..1aa391f4a 100644--- a/nodes.py+++ b/nodes.py@@ -378,14 +378,15 @@ class VAEEncodeForInpaint:CATEGORY = "latent/inpaint"def encode(self, vae, pixels, mask, grow_mask_by=6):- x = (pixels.shape[1] // vae.downscale_ratio) * vae.downscale_ratio- y = (pixels.shape[2] // vae.downscale_ratio) * vae.downscale_ratio+ downscale_ratio = vae.spacial_compression_encode()+ x = (pixels.shape[1] // downscale_ratio) * downscale_ratio+ y = (pixels.shape[2] // downscale_ratio) * downscale_ratiomask = torch.nn.functional.interpolate(mask.reshape((-1, 1, mask.shape[-2], mask.shape[-1])), size=(pixels.shape[1], pixels.shape[2]), mode="bilinear")pixels = pixels.clone()if pixels.shape[1] != x or pixels.shape[2] != y:- x_offset = (pixels.shape[1] % vae.downscale_ratio) // 2- y_offset = (pixels.shape[2] % vae.downscale_ratio) // 2+ x_offset = (pixels.shape[1] % downscale_ratio) // 2+ y_offset = (pixels.shape[2] % downscale_ratio) // 2pixels = pixels[:,x_offset:x + x_offset, y_offset:y + y_offset,:]mask = mask[:,:,x_offset:x + x_offset, y_offset:y + y_offset]