Merge branch 'ltx-95-latents-yiyi' of github.com:huggingface/diffusers into ltx-95-latents-yiyi

yiyixuxu · yiyixuxu · commit 6f7e83755bd2 · 2025-03-10T12:31:05.000+01:00
diff --git a/src/diffusers/pipelines/ltx/pipeline_ltx_condition.py b/src/diffusers/pipelines/ltx/pipeline_ltx_condition.py
@@ -475,6 +475,7 @@ def _prepare_video_ids(latents: torch.Tensor, scale_factor: int = 32, scale_fact
             torch.arange(0, num_frames, patch_size_t, device=device),
             torch.arange(0, height, patch_size, device=device),
             torch.arange(0, width, patch_size, device=device),
+            indexing="ij",
         )
         latent_sample_coords = torch.stack(latent_sample_coords, dim=0)
         latent_coords = latent_sample_coords.unsqueeze(0).repeat(batch_size, 1, 1, 1, 1)

Original file line number	Diff line number	Diff line change
`@@ -475,6 +475,7 @@ def _prepare_video_ids(latents: torch.Tensor, scale_factor: int = 32, scale_fact`
`475`	`475`	`torch.arange(0, num_frames, patch_size_t, device=device),`
`476`	`476`	`torch.arange(0, height, patch_size, device=device),`
`477`	`477`	`torch.arange(0, width, patch_size, device=device),`
	`478`	`+ indexing="ij",`
`478`	`479`	`)`
`479`	`480`	`latent_sample_coords = torch.stack(latent_sample_coords, dim=0)`
`480`	`481`	`latent_coords = latent_sample_coords.unsqueeze(0).repeat(batch_size, 1, 1, 1, 1)`