From fd569ead31e3237724b372c383ef162e76c41a43 Mon Sep 17 00:00:00 2001
From: comfyanonymous <comfyanonymous@protonmail.com>
Date: Mon, 18 Aug 2025 22:35:47 -0400
Subject: [PATCH] Make the TextEncodeQwenImageEdit also set the ref latent.

If you don't want it to set the ref latent and want to use the
ReferenceLatent node with your custom latent instead just disconnect the
VAE.
---
 comfy_extras/nodes_qwen.py | 42 +++++++++++++++++++++++++++++++++++---
 1 file changed, 39 insertions(+), 3 deletions(-)

diff --git a/comfy_extras/nodes_qwen.py b/comfy_extras/nodes_qwen.py
index ab21acf32..b5088fae2 100644
--- a/comfy_extras/nodes_qwen.py
+++ b/comfy_extras/nodes_qwen.py
@@ -1,3 +1,25 @@
+import node_helpers
+import comfy.utils
+
+PREFERRED_QWENIMAGE_RESOLUTIONS = [
+    (672, 1568),
+    (688, 1504),
+    (720, 1456),
+    (752, 1392),
+    (800, 1328),
+    (832, 1248),
+    (880, 1184),
+    (944, 1104),
+    (1024, 1024),
+    (1104, 944),
+    (1184, 880),
+    (1248, 832),
+    (1328, 800),
+    (1392, 752),
+    (1456, 720),
+    (1504, 688),
+    (1568, 672),
+]
 
 
 class TextEncodeQwenImageEdit:
@@ -7,19 +29,33 @@ class TextEncodeQwenImageEdit:
             "clip": ("CLIP", ),
             "prompt": ("STRING", {"multiline": True, "dynamicPrompts": True}),
             },
-            "optional": {"image": ("IMAGE", ),}}
+            "optional": {"vae": ("VAE", ),
+                         "image": ("IMAGE", ),}}
+
     RETURN_TYPES = ("CONDITIONING",)
     FUNCTION = "encode"
 
     CATEGORY = "advanced/conditioning"
 
-    def encode(self, clip, prompt, image=None):
+    def encode(self, clip, prompt, vae=None, image=None):
+        ref_latent = None
         if image is None:
             images = []
         else:
             images = [image]
+            if vae is not None:
+                width = image.shape[2]
+                height = image.shape[1]
+                aspect_ratio = width / height
+                _, width, height = min((abs(aspect_ratio - w / h), w, h) for w, h in PREFERRED_QWENIMAGE_RESOLUTIONS)
+                image = comfy.utils.common_upscale(image.movedim(-1, 1), width, height, "lanczos", "center").movedim(1, -1)
+                ref_latent = vae.encode(image[:, :, :, :3])
+
         tokens = clip.tokenize(prompt, images=images)
-        return (clip.encode_from_tokens_scheduled(tokens), )
+        conditioning = clip.encode_from_tokens_scheduled(tokens)
+        if ref_latent is not None:
+            conditioning = node_helpers.conditioning_set_values(conditioning, {"reference_latents": [ref_latent]}, append=True)
+        return (conditioning, )
 
 
 NODE_CLASS_MAPPINGS = {