From 990a7ae7f20df5a5092200fad129007f84252ae0 Mon Sep 17 00:00:00 2001
From: comfyanonymous <121283862+comfyanonymous@users.noreply.github.com>
Date: Mon, 18 May 2026 20:01:43 -0700
Subject: [PATCH] Initial work to make downscale_ratio_temporal work. (#13972)

---
 comfy/sample.py                      | 12 ++++++++++--
 comfy_extras/nodes_custom_sampler.py |  6 ++++--
 nodes.py                             |  3 ++-
 3 files changed, 16 insertions(+), 5 deletions(-)

diff --git a/comfy/sample.py b/comfy/sample.py
index 653829582..2be0cae5f 100644
--- a/comfy/sample.py
+++ b/comfy/sample.py
@@ -37,11 +37,12 @@ def prepare_noise(latent_image, seed, noise_inds=None):
 
     return noises
 
-def fix_empty_latent_channels(model, latent_image, downscale_ratio_spacial=None):
+def fix_empty_latent_channels(model, latent_image, downscale_ratio_spacial=None, downscale_ratio_temporal=None):
     if latent_image.is_nested:
         return latent_image
     latent_format = model.get_model_object("latent_format") #Resize the empty latent image so it has the right number of channels
-    if torch.count_nonzero(latent_image) == 0:
+    is_empty = torch.count_nonzero(latent_image) == 0
+    if is_empty:
         if latent_format.latent_channels != latent_image.shape[1]:
             latent_image = comfy.utils.repeat_to_batch_size(latent_image, latent_format.latent_channels, dim=1)
         if downscale_ratio_spacial is not None:
@@ -51,6 +52,13 @@ def fix_empty_latent_channels(model, latent_image, downscale_ratio_spacial=None)
 
     if latent_format.latent_dimensions == 3 and latent_image.ndim == 4:
         latent_image = latent_image.unsqueeze(2)
+
+    if is_empty and downscale_ratio_temporal is not None:
+        if downscale_ratio_temporal != latent_format.temporal_downscale_ratio:
+            ratio = downscale_ratio_temporal / latent_format.temporal_downscale_ratio
+            new_t = max(1, round(latent_image.shape[2] * ratio))
+            latent_image = comfy.utils.repeat_to_batch_size(latent_image, new_t, dim=2)
+
     return latent_image
 
 def prepare_sampling(model, noise_shape, positive, negative, noise_mask):
diff --git a/comfy_extras/nodes_custom_sampler.py b/comfy_extras/nodes_custom_sampler.py
index c67145d2d..02fb9385f 100644
--- a/comfy_extras/nodes_custom_sampler.py
+++ b/comfy_extras/nodes_custom_sampler.py
@@ -750,7 +750,7 @@ class SamplerCustom(io.ComfyNode):
         latent = latent_image
         latent_image = latent["samples"]
         latent = latent.copy()
-        latent_image = comfy.sample.fix_empty_latent_channels(model, latent_image, latent.get("downscale_ratio_spacial", None))
+        latent_image = comfy.sample.fix_empty_latent_channels(model, latent_image, latent.get("downscale_ratio_spacial", None), latent.get("downscale_ratio_temporal", None))
         latent["samples"] = latent_image
 
         if not add_noise:
@@ -770,6 +770,7 @@ class SamplerCustom(io.ComfyNode):
 
         out = latent.copy()
         out.pop("downscale_ratio_spacial", None)
+        out.pop("downscale_ratio_temporal", None)
         out["samples"] = samples
         if "x0" in x0_output:
             x0_out = model.model.process_latent_out(x0_output["x0"].cpu())
@@ -949,7 +950,7 @@ class SamplerCustomAdvanced(io.ComfyNode):
         latent = latent_image
         latent_image = latent["samples"]
         latent = latent.copy()
-        latent_image = comfy.sample.fix_empty_latent_channels(guider.model_patcher, latent_image, latent.get("downscale_ratio_spacial", None))
+        latent_image = comfy.sample.fix_empty_latent_channels(guider.model_patcher, latent_image, latent.get("downscale_ratio_spacial", None), latent.get("downscale_ratio_temporal", None))
         latent["samples"] = latent_image
 
         noise_mask = None
@@ -965,6 +966,7 @@ class SamplerCustomAdvanced(io.ComfyNode):
 
         out = latent.copy()
         out.pop("downscale_ratio_spacial", None)
+        out.pop("downscale_ratio_temporal", None)
         out["samples"] = samples
         if "x0" in x0_output:
             x0_out = guider.model_patcher.model.process_latent_out(x0_output["x0"].cpu())
diff --git a/nodes.py b/nodes.py
index 374217eea..42fb8fd56 100644
--- a/nodes.py
+++ b/nodes.py
@@ -1524,7 +1524,7 @@ class SetLatentNoiseMask:
 
 def common_ksampler(model, seed, steps, cfg, sampler_name, scheduler, positive, negative, latent, denoise=1.0, disable_noise=False, start_step=None, last_step=None, force_full_denoise=False):
     latent_image = latent["samples"]
-    latent_image = comfy.sample.fix_empty_latent_channels(model, latent_image, latent.get("downscale_ratio_spacial", None))
+    latent_image = comfy.sample.fix_empty_latent_channels(model, latent_image, latent.get("downscale_ratio_spacial", None), latent.get("downscale_ratio_temporal", None))
 
     if disable_noise:
         noise = torch.zeros(latent_image.size(), dtype=latent_image.dtype, layout=latent_image.layout, device="cpu")
@@ -1543,6 +1543,7 @@ def common_ksampler(model, seed, steps, cfg, sampler_name, scheduler, positive,
                                   force_full_denoise=force_full_denoise, noise_mask=noise_mask, callback=callback, disable_pbar=disable_pbar, seed=seed)
     out = latent.copy()
     out.pop("downscale_ratio_spacial", None)
+    out.pop("downscale_ratio_temporal", None)
     out["samples"] = samples
     return (out, )