The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface. The fastest local inference engine in the world.
Improve context window resizing for SCAIL2 (CORE-286) (#14394)
Barish Ozbay91187c58d946
diff --git a/comfy/model_base.py b/comfy/model_base.pyindex 2289e0812..ab4a11022 100644--- a/comfy/model_base.py+++ b/comfy/model_base.py@@ -1816,7 +1816,24 @@ class WAN21_SCAIL2(WAN21_SCAIL):def resize_cond_for_context_window(self, cond_key, cond_value, window, x_in, device, retain_index_list=[]):if cond_key in ("sam_latents", "pose_latents"):- return comfy.context_windows.slice_cond(cond_value, window, x_in, device, temporal_dim=2, temporal_offset=1)+ # Return sliced view omitting retain_index_list+ return comfy.context_windows.slice_cond(cond_value, window, x_in, device, temporal_dim=2, temporal_offset=0)+ if cond_key == "ref_mask_latents" and hasattr(cond_value, "cond") and isinstance(cond_value.cond, torch.Tensor):+ # The ref mask is just a single frame padded with frames of zeros, so just grab the first frames for all windows+ full_ref_mask = cond_value.cond+ video_frame_count = x_in.shape[2]+ if full_ref_mask.shape[2] != video_frame_count + 1:+ return None+ window_length = len(window.index_list)++ # Account for the causal anchor frame if it exists+ anchor_index = getattr(window, "causal_anchor_index", None)+ if anchor_index is not None and anchor_index >= 0:+ window_length += 1++ window_ref_mask = full_ref_mask[:, :, :window_length + 1].to(device)+ return cond_value._copy_with(window_ref_mask)+return super().resize_cond_for_context_window(cond_key, cond_value, window, x_in, device, retain_index_list=retain_index_list)def concat_cond(self, **kwargs):