From 5ea20c114872bd0a6530fcf905a248fd20f1f7d0 Mon Sep 17 00:00:00 2001
From: leejet <leejet714@gmail.com>
Date: Tue, 27 Jan 2026 00:56:04 +0800
Subject: [PATCH] auto detect z-image-omni

---
 model.cpp            | 14 +++++++++++++-
 model.h              |  3 ++-
 stable-diffusion.cpp |  7 +++++--
 3 files changed, 20 insertions(+), 4 deletions(-)

diff --git a/model.cpp b/model.cpp
index 253dd25..4a6fad8 100644
--- a/model.cpp
+++ b/model.cpp
@@ -1039,6 +1039,8 @@ SDVersion ModelLoader::get_sd_version() {
     bool is_xl                       = false;
     bool is_flux                     = false;
     bool is_flux2                    = false;
+    bool is_z_image                  = false;
+    bool is_z_image_omni             = false;
     bool has_single_block_47         = false;
     bool is_wan                      = false;
     int64_t patch_embedding_channels = 0;
@@ -1071,7 +1073,10 @@ SDVersion ModelLoader::get_sd_version() {
                 return VERSION_OVIS_IMAGE;
             }
             if (tensor_storage.name.find("model.diffusion_model.cap_embedder.0.weight") != std::string::npos) {
-                return VERSION_Z_IMAGE;
+                is_z_image = true;
+            }
+            if (tensor_storage.name.find("model.diffusion_model.siglip_embedder.0.weight") != std::string::npos) {
+                is_z_image_omni = true;
             }
             if (tensor_storage.name.find("model.diffusion_model.blocks.0.cross_attn.norm_k.weight") != std::string::npos) {
                 is_wan = true;
@@ -1174,6 +1179,13 @@ SDVersion ModelLoader::get_sd_version() {
         return VERSION_FLUX2_KLEIN;
     }
 
+    if (is_z_image) {
+        if (is_z_image_omni) {
+            return VERSION_Z_IMAGE_OMNI;
+        }
+        return VERSION_Z_IMAGE;
+    }
+
     if (token_embedding_weight.ne[0] == 768) {
         if (is_inpaint) {
             return VERSION_SD1_INPAINT;
diff --git a/model.h b/model.h
index e16ac3a..553e159 100644
--- a/model.h
+++ b/model.h
@@ -48,6 +48,7 @@ enum SDVersion {
     VERSION_FLUX2,
     VERSION_FLUX2_KLEIN,
     VERSION_Z_IMAGE,
+    VERSION_Z_IMAGE_OMNI,
     VERSION_OVIS_IMAGE,
     VERSION_COUNT,
 };
@@ -123,7 +124,7 @@ static inline bool sd_version_is_qwen_image(SDVersion version) {
 }
 
 static inline bool sd_version_is_z_image(SDVersion version) {
-    if (version == VERSION_Z_IMAGE) {
+    if (version == VERSION_Z_IMAGE || version == VERSION_Z_IMAGE_OMNI) {
         return true;
     }
     return false;
diff --git a/stable-diffusion.cpp b/stable-diffusion.cpp
index ea5d5e1..1eb1c80 100644
--- a/stable-diffusion.cpp
+++ b/stable-diffusion.cpp
@@ -51,6 +51,7 @@ const char* model_version_to_str[] = {
     "Flux.2",
     "Flux.2 klein",
     "Z-Image",
+    "Z-Image-Omni",
     "Ovis Image",
 };
 
@@ -3713,8 +3714,10 @@ sd_image_t* generate_image(sd_ctx_t* sd_ctx, const sd_img_gen_params_t* sd_img_g
         ggml_tensor* latent = sd_ctx->sd->encode_first_stage(work_ctx, img);
         ref_latents.push_back(latent);
 
-        auto clip_vision_output = sd_ctx->sd->get_clip_vision_output(work_ctx, *ref_images[i], false, -2);
-        ref_clip_feats.push_back(clip_vision_output);
+        if (sd_ctx->sd->version == VERSION_Z_IMAGE_OMNI) {
+            auto clip_vision_output = sd_ctx->sd->get_clip_vision_output(work_ctx, *ref_images[i], false, -2);
+            ref_clip_feats.push_back(clip_vision_output);
+        }
     }
 
     if (sd_img_gen_params->init_image.data != nullptr || sd_img_gen_params->ref_images_count > 0) {