feat: add krea2 support

README.md

@@ -15,6 +15,7 @@ API and command-line option may change frequently.***
 
 ## 🔥Important News
 
+* **2026/06/25** 🚀 stable-diffusion.cpp now supports **Krea2**
 * **2026/06/04** 🚀 stable-diffusion.cpp now supports **Ideogram4**
 * **2026/05/31** 🚀 stable-diffusion.cpp now supports **PiD**
 * **2026/05/27** 🚀 stable-diffusion.cpp now supports **Lens**
@@ -51,6 +52,7 @@ API and command-line option may change frequently.***
     - [Anima](./docs/anima.md)
     - [ERNIE-Image](./docs/ernie_image.md)
     - [Boogu Image](./docs/boogu_image.md)
+    - [Krea2](./docs/krea2.md)
     - [HiDream-O1-Image](./docs/hidream_o1_image.md)
     - [Ideogram4](./docs/ideogram4.md)
   - Image Edit Models

docs/krea2.md

@@ -0,0 +1,27 @@
+# How to Use
+
+Krea2 uses a Krea2 diffusion transformer, the Wan2.1 VAE, and Qwen3-VL 4B as the LLM text encoder.
+
+## Download weights
+
+- Download Krea2 Raw
+    - safetensors: https://huggingface.co/krea/Krea-2-Raw/tree/main
+    - gguf: https://huggingface.co/realrebelai/KREA-2_GGUFs/tree/main/BASE
+- Download Krea2 Turbo
+    - safetensors: https://huggingface.co/krea/Krea-2-Turbo/tree/main
+    - gguf: https://huggingface.co/realrebelai/KREA-2_GGUFs/tree/main/TURBO
+- Download vae
+    - safetensors: https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/blob/main/split_files/vae/wan_2.1_vae.safetensors
+- Download Qwen3-VL 4B
+    - safetensors: https://huggingface.co/Comfy-Org/Krea-2/tree/main/text_encoders
+    - gguf: https://huggingface.co/Qwen/Qwen3-VL-4B-Instruct-GGUF/tree/main
+
+## Examples
+
+### Krea2
+
+```
+.\bin\Release\sd-cli.exe --diffusion-model ..\..\ComfyUI\models\diffusion_models\Krea-2-Raw-Q8_0.gguf --llm ..\..\ComfyUI\models\text_encoders\Qwen3-VL-4B-Instruct-Q4_K_M.gguf --vae ..\..\ComfyUI\models\vae\wan_2.1_vae.safetensors -p "a lovely cat holding a sign says 'krea2.cpp'"  --diffusion-fa -v --offload-to-cpu
+```
+
+<img width="256" alt="Krea2 Raw example" src="../assets/krea2/example.png" />

src/conditioning/conditioner.hpp

@@ -1518,7 +1518,7 @@ struct LLMEmbedder : public Conditioner {
             arch = LLM::LLMArch::GPT_OSS_20B;
         } else if (sd_version_is_pid(version)) {
             arch = LLM::LLMArch::GEMMA2_2B;
-        } else if (sd_version_is_ideogram4(version) || sd_version_is_boogu_image(version)) {
+        } else if (sd_version_is_ideogram4(version) || sd_version_is_boogu_image(version) || sd_version_is_krea2(version)) {
             arch = LLM::LLMArch::QWEN3_VL;
         } else if (sd_version_is_z_image(version) || version == VERSION_OVIS_IMAGE || version == VERSION_FLUX2_KLEIN) {
             arch = LLM::LLMArch::QWEN3;
@@ -1837,6 +1837,17 @@ struct LLMEmbedder : public Conditioner {
                 prompt_attn_range.second = static_cast<int>(prompt.size());
                 prompt += "<|im_end|>\n";
             }
+        } else if (sd_version_is_krea2(version)) {
+            prompt_template_encode_start_idx = 34;
+            out_layers                       = {2, 5, 8, 11, 14, 17, 20, 23, 26, 29, 32, 35};
+
+            prompt = "<|im_start|>system\nDescribe the image by detailing the color, shape, size, texture, quantity, text, spatial relationships of the objects and background:<|im_end|>\n<|im_start|>user\n";
+
+            prompt_attn_range.first = static_cast<int>(prompt.size());
+            prompt += conditioner_params.text;
+            prompt_attn_range.second = static_cast<int>(prompt.size());
+
+            prompt += "<|im_end|>\n<|im_start|>assistant\n";
         } else if (sd_version_is_longcat(version)) {
             spell_quotes = true;
 

src/core/ggml_extend.hpp

@@ -1382,7 +1382,16 @@ __STATIC_INLINE__ ggml_tensor* ggml_ext_attention_ext(ggml_context* ctx,
             if (!ggml_backend_supports_op(backend, kqv)) {
                 kqv = nullptr;
             } else {
-                kqv = ggml_view_3d(ctx, kqv, d_head, n_head, L_q, kqv->nb[1], kqv->nb[2], 0);
+                kqv = ggml_view_4d(ctx,
+                                   kqv,
+                                   d_head,
+                                   n_head,
+                                   L_q,
+                                   N,
+                                   kqv->nb[1],
+                                   kqv->nb[2],
+                                   kqv->nb[1] * n_head,
+                                   0);
             }
         }
     }

src/model.h

@@ -49,6 +49,7 @@ enum SDVersion {
     VERSION_LONGCAT,
     VERSION_PID,
     VERSION_IDEOGRAM4,
+    VERSION_KREA2,
     VERSION_ESRGAN,
     VERSION_COUNT,
 };
@@ -186,6 +187,13 @@ static inline bool sd_version_is_ideogram4(SDVersion version) {
     return false;
 }
 
+static inline bool sd_version_is_krea2(SDVersion version) {
+    if (version == VERSION_KREA2) {
+        return true;
+    }
+    return false;
+}
+
 static inline bool sd_version_uses_flux_vae(SDVersion version) {
     if (sd_version_is_flux(version) || sd_version_is_z_image(version) || sd_version_is_boogu_image(version) || sd_version_is_longcat(version)) {
         return true;
@@ -226,7 +234,8 @@ static inline bool sd_version_is_dit(SDVersion version) {
         sd_version_is_lens(version) ||
         sd_version_is_longcat(version) ||
         sd_version_is_pid(version) ||
-        sd_version_is_ideogram4(version)) {
+        sd_version_is_ideogram4(version) ||
+        sd_version_is_krea2(version)) {
         return true;
     }
     return false;