first commit

README.md

@@ -0,0 +1,72 @@
+---
+license: other
+license_name: stabilityai-ai-community
+license_link: LICENSE.md
+language:
+- en
+base_model:
+- stabilityai/stable-diffusion-3.5-medium
+pipeline_tag: text-to-image
+---
+
+<div align="center">
+**TensorArt Stable Diffusion 3.5 Medium ControlNet Depth**
+
+# <img src="./assets/showcase.png"/>
+</div>
+
+# With SD3.5M
+```python
+import torch
+from diffusers import StableDiffusion3ControlNetPipeline
+from diffusers.models import SD3ControlNetModel
+from diffusers.utils import load_image
+
+controlnet = SD3ControlNetModel.from_pretrained("tensorart/SD3.5M-Controlnet-Depth")
+pipe = StableDiffusion3ControlNetPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-3-medium",
+    controlnet=controlnet
+)
+pipe.to("cuda", torch.float16)
+
+control_image = load_image("https://huggingface.co/tensorart/SD3.5M-Controlnet-Depth/resolve/main/images/depth.png")
+prompt = "A cyberpunk man"
+negative_prompt = "low quality, worst quality, deformed, distorted, disfigured, motion smear, motion artifacts, fused fingers, bad anatomy, weird hand, ugly, monochrome"
+
+image = pipe(
+    prompt, 
+    num_inference_steps=30,
+    negative_prompt=negative_prompt, 
+    control_image=control_image, 
+    guidance_scale=4.5,
+).images[0]
+image.save('image.jpg')
+```
+
+# With TensorArt's SD3.5M Turbo
+```python
+import torch
+from diffusers import StableDiffusion3ControlNetPipeline
+from diffusers.models import SD3ControlNetModel
+from diffusers.utils import load_image
+
+controlnet = SD3ControlNetModel.from_pretrained("tensorart/SD3.5M-Controlnet-Depth")
+pipe = StableDiffusion3ControlNetPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-3-medium",
+    controlnet=controlnet
+)
+pipe.to("cuda", torch.float16)
+
+control_image = load_image("https://huggingface.co/tensorart/SD3.5M-Controlnet-Depth/resolve/main/images/depth.png")
+prompt = "A cyberpunk man"
+negative_prompt = "low quality, worst quality, deformed, distorted, disfigured, motion smear, motion artifacts, fused fingers, bad anatomy, weird hand, ugly, monochrome"
+
+image = pipe(
+    prompt, 
+    num_inference_steps=8,
+    negative_prompt=negative_prompt, 
+    control_image=control_image, 
+    guidance_scale=1.5
+).images[0]
+image.save('image.jpg')
+```

config.json

@@ -0,0 +1,33 @@
+{
+  "_class_name": "SD3ControlNetModel",
+  "_diffusers_version": "0.32.0.dev0",
+  "_name_or_path": "stabilityai/stable-diffusion-3.5-medium",
+  "attention_head_dim": 64,
+  "caption_projection_dim": 1536,
+  "dual_attention_layers": [
+    0,
+    1,
+    2,
+    3,
+    4,
+    5,
+    6,
+    7,
+    8,
+    9,
+    10,
+    11,
+    12
+  ],
+  "extra_conditioning_channels": 0,
+  "in_channels": 16,
+  "joint_attention_dim": 4096,
+  "num_attention_heads": 24,
+  "num_layers": 23,
+  "out_channels": 16,
+  "patch_size": 2,
+  "pooled_projection_dim": 2048,
+  "pos_embed_max_size": 384,
+  "qk_norm": "rms_norm",
+  "sample_size": 128
+}

diffusion_pytorch_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f768adb86fa5585afd4fc9a301317d703c69921630278e1f8095364c6f71a306
+size 4930003400