From 196bb895fd0af69ff424f86252ce73f8b02bef7c Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?St=C3=A9phane=20du=20Hamel?= <stephduh@live.fr>
Date: Fri, 12 Dec 2025 02:39:09 +0100
Subject: [PATCH] patch size consistent with Flux1

---
 flux.hpp             |  1 -
 stable-diffusion.cpp | 11 +----------
 vae.hpp              |  4 ++--
 3 files changed, 3 insertions(+), 13 deletions(-)

diff --git a/flux.hpp b/flux.hpp
index fc30987..df3c4c8 100644
--- a/flux.hpp
+++ b/flux.hpp
@@ -1309,7 +1309,6 @@ namespace Flux {
             } else if (sd_version_is_longcat(version)) {
                 flux_params.context_in_dim = 3584;
                 flux_params.vec_in_dim     = 0;
-                flux_params.patch_size     = 1;
             }
             for (auto pair : tensor_storage_map) {
                 std::string tensor_name = pair.first;
diff --git a/stable-diffusion.cpp b/stable-diffusion.cpp
index 1e8f04a..f89e426 100644
--- a/stable-diffusion.cpp
+++ b/stable-diffusion.cpp
@@ -1346,12 +1346,6 @@ public:
                     latent_rgb_bias = flux2_latent_rgb_bias;
                     patch_sz               = 2;
                 }
-            } else if (dim == 64) {
-                if (sd_version_is_flux(version) || sd_version_is_z_image(version) || sd_version_is_longcat(version)) {
-                    latent_rgb_proj = flux_latent_rgb_proj;
-                    latent_rgb_bias = flux_latent_rgb_bias;
-                    patch_sz        = 2;
-                }
             } else if (dim == 48) {
                 if (sd_version_is_wan(version)) {
                     latent_rgb_proj = wan_22_latent_rgb_proj;
@@ -1916,7 +1910,7 @@ public:
         int vae_scale_factor = 8;
         if (version == VERSION_WAN2_2_TI2V) {
             vae_scale_factor = 16;
-        } else if (sd_version_is_flux2(version) || sd_version_is_longcat(version)) {
+        } else if (sd_version_is_flux2(version)) {
             vae_scale_factor = 16;
         } else if (version == VERSION_CHROMA_RADIANCE) {
             vae_scale_factor = 1;
@@ -1945,8 +1939,6 @@ public:
                 latent_channel = 3;
             } else if (sd_version_is_flux2(version)) {
                 latent_channel = 128;
-            } else if (sd_version_is_longcat(version)) {
-                latent_channel = 64;
             } else {
                 latent_channel = 16;
             }
@@ -2247,7 +2239,6 @@ public:
             sd_version_is_qwen_image(version) ||
             sd_version_is_wan(version) ||
             sd_version_is_flux2(version) ||
-            sd_version_is_longcat(version) ||
             version == VERSION_CHROMA_RADIANCE) {
             latent = vae_output;
         } else if (version == VERSION_SD1_PIX2PIX) {
diff --git a/vae.hpp b/vae.hpp
index 740a565..ad5db1b 100644
--- a/vae.hpp
+++ b/vae.hpp
@@ -553,7 +553,7 @@ public:
 
     struct ggml_tensor* decode(GGMLRunnerContext* ctx, struct ggml_tensor* z) {
         // z: [N, z_channels, h, w]
-        if (sd_version_is_flux2(version) || sd_version_is_longcat(version)) {
+        if (sd_version_is_flux2(version)) {
             // [N, C*p*p, h, w] -> [N, C, h*p, w*p]
             int64_t p = 2;
 
@@ -592,7 +592,7 @@ public:
             auto quant_conv = std::dynamic_pointer_cast<Conv2d>(blocks["quant_conv"]);
             z               = quant_conv->forward(ctx, z);  // [N, 2*embed_dim, h/8, w/8]
         }
-        if (sd_version_is_flux2(version) || sd_version_is_longcat(version)) {
+        if (sd_version_is_flux2(version)) {
             z = ggml_ext_chunk(ctx->ggml_ctx, z, 2, 2)[0];
 
             // [N, C, H, W] -> [N, C*p*p, H/p, W/p]