readme

README.md

@@ -1,15 +1,111 @@
----
-{}
----
-Bokeh 1.0 Medium
-Bokeh Image
-
-Features
-We have pushed text-to-image models to new heights in terms of fidelity. Compared to models with similar capabilities like Flux1.1pro ultra raw/Recraft v3 raw, Bokeh medium demonstrates leading advantages in high fidelity and detail preservation, with capabilities in certain domains that rival human photographers.
-The model supports various prompt formats, with photography terminology equally applicable to Bokeh.
-We have evaluated the community's needs and training costs for fine-tuning this model. You can easily perform downstream training tasks (Finetune, LoRA, Lycoris, and academic research) on this base model.
-The official version of Bokeh medium is coming soon with compatible ControlNet support in development.
-
-Contact
-Website: https://tensor.art https://tusiart.com
-Developed by: TensorArt
+# Bokeh 3.5 Medium
+<div align="center">
+<img src="ad2.jpg" alt="00205_" width="620"/>
+</div> 
+
+Bokeh 3.5 Medium is a **Continue-training** model built upon the **stable diffusion 3.5 medium** foundation, further refined using a **500W high-resolution open-source dataset** with rigorous **aesthetic curation**. This ensures outstanding image quality, fine detail preservation, and enhanced controllability.
+
+This model is released under the Stability Community License.
+For more details, visit [Tensor.Art](https://tensor.art) or [TusiArt](https://tusiart.com) to explore additional resources and useful information.
+
+## Overview
+
+- **Continue-training on SD3.5M**, leveraging a large-scale **500W high-resolution dataset**, carefully curated for aesthetic quality.
+- **Supports hybrid short/long caption training** for enhanced natural language understanding.
+  - **Short Captions:** Focus on core image features.
+  - **Long Captions:** Provide broader scene context and atmospheric details.
+- **Recommended Resolutions:**  
+  `1920x1024`, `1728x1152`, `1152x1728`, `1280x1664`, `1440x1440`
+- **Best Quality Training Resolution:** `1440x1440`
+- **Supports LoRA fine-tuning.**
+
+## Advantages
+
+### 🖼️ High-Quality Image Generation
+- **State-of-the-art visual fidelity** with improved detail extraction and **aesthetic consistency**.
+- **Enhanced resolution support** up to **200W pixels**, ensuring highly detailed image outputs.
+- **Carefully curated dataset** ensures better composition, lighting, and overall artistic appeal.
+
+### 🎯 Powerful Custom Fine-Tuning
+- **Exceptional LoRA training support**, making it highly effective for:
+  - Photography
+  - 3D Rendering
+  - Illustration
+  - Concept Art
+
+### ⚡ Efficient Inference & Training
+- **Low hardware requirements for inference:**
+  - **Medium model:** 9GB VRAM (without T5)
+  - **Full weights inference:** 16GB VRAM (suitable for local deployment)
+- **LoRA fine-tuning VRAM requirement:** 12GB - 32GB
+
+## Known Issues
+
+- **Potential human anatomy inconsistencies.**
+- **Limited ability to generate photorealistic images.**
+- **Some concepts may suffer from aesthetic quality issues.**
+
+
+## Prompting Guide
+
+### Use a structured prompt combining:
+- **Main subject** (e.g., `"Close-up of a macaw"`)  
+- **Detailed features** (e.g., `"vivid feathers, sharp beak"`)  
+- **Background environment** (e.g., `"dimly lit environment"`)  
+- **Atmospheric description** (e.g., `"soft warm lighting, cinematic mood"`)  
+
+### Best Practices:
+- **Avoid overly complex prompts**, as the model already has strong text encoding. Overloading details can cause **T5 hallucination artifacts**, reducing image quality.  
+- **Do not use excessively short prompts** (e.g., single words or 2-3 tokens) unless combined with **LoRA or Image2Image (i2i)** techniques.  
+- **Avoid mixing too many unrelated concepts**, as this can lead to visual distortions and unwanted artifacts.  
+- **Optimal token length:** **30-70 tokens**.  
+
+### Negative Prompting
+- **Negative prompts strongly influence image quality.**  
+- Ensure they **do not contradict the main subject** to avoid degrading the output.  
+
+
+
+## Example Output
+Using diffusers：
+```python
+import torch
+from diffusers import StableDiffusion3Pipeline
+
+pipe = StableDiffusion3Pipeline.from_pretrained("/mnt/share/pcm_outputs/bokeh_3.5_medium", torch_dtype=torch.bfloat16)
+pipe = pipe.to("cuda")
+
+image = pipe(
+    "Close-up of a macaw, dimly lit environment",
+    num_inference_steps=28,
+    guidance_scale=4,
+    height=1920,
+    width=1024,
+).images[0]
+image.save("macaw.jpg")
+```
+Using comfyui:
+To use this workflow in **ComfyUI**, download the JSON file and load it:
+
+[Download Workflow](bk_workflow.json)
+
+## Recommended Training Configuration
+
+For **LoRA fine-tuning**, the following tools and settings are recommended:
+
+### 🔧 Training Tools
+- **Kohya_ss:** [GitHub Repository](https://github.com/bmaltais/kohya_ss.git)
+- **Simple Tuner:** [GitHub Repository](https://github.com/bghira/SimpleTuner)
+
+### ⚙️ Suggested Training Settings
+```bash
+--Resolution 1440x1440
+--t5xxl_max_token_length 154
+--optimizer_type AdamW8bit
+--mmdit_lr 1e-4
+--text_encoder_lr 5e-5
+```
+
+## Contact
+* Website: https://tensor.art  https://tusiart.com
+* Developed by: TensorArt

bk_workflow.json

@@ -0,0 +1,683 @@
+{
+  "last_node_id": 63,
+  "last_link_id": 258,
+  "nodes": [
+    {
+      "id": 30,
+      "type": "CLIPTextEncodeSD3",
+      "pos": [
+        518,
+        -261
+      ],
+      "size": [
+        258.8465881347656,
+        200
+      ],
+      "flags": {
+        "collapsed": true
+      },
+      "order": 8,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "clip",
+          "type": "CLIP",
+          "link": 258
+        },
+        {
+          "name": "clip_g",
+          "type": "STRING",
+          "widget": {
+            "name": "clip_g"
+          },
+          "link": 222
+        },
+        {
+          "name": "clip_l",
+          "type": "STRING",
+          "widget": {
+            "name": "clip_l"
+          },
+          "link": 223
+        }
+      ],
+      "outputs": [
+        {
+          "name": "CONDITIONING",
+          "type": "CONDITIONING",
+          "links": [
+            236
+          ],
+          "slot_index": 0
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "CLIPTextEncodeSD3"
+      },
+      "widgets_values": [
+        "anime,render,cartoon,3d,bad structure",
+        "anime,render,cartoon,3d,bad structure",
+        "",
+        "none",
+        true,
+        true,
+        true
+      ]
+    },
+    {
+      "id": 32,
+      "type": "CLIPTextEncodeSD3",
+      "pos": [
+        518,
+        -319
+      ],
+      "size": [
+        262.4820556640625,
+        190
+      ],
+      "flags": {
+        "collapsed": true
+      },
+      "order": 9,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "clip",
+          "type": "CLIP",
+          "link": 257
+        },
+        {
+          "name": "clip_g",
+          "type": "STRING",
+          "widget": {
+            "name": "clip_g"
+          },
+          "link": 122
+        },
+        {
+          "name": "clip_l",
+          "type": "STRING",
+          "widget": {
+            "name": "clip_l"
+          },
+          "link": 123
+        },
+        {
+          "name": "t5xxl",
+          "type": "STRING",
+          "widget": {
+            "name": "t5xxl"
+          },
+          "link": 152
+        }
+      ],
+      "outputs": [
+        {
+          "name": "CONDITIONING",
+          "type": "CONDITIONING",
+          "links": [
+            237
+          ],
+          "slot_index": 0
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "CLIPTextEncodeSD3"
+      },
+      "widgets_values": [
+        "Close-up of a macaw, dimly lit environment",
+        "Close-up of a macaw, dimly lit environment",
+        "Close-up of a macaw, dimly lit environment",
+        "empty_prompt",
+        true,
+        true,
+        true
+      ]
+    },
+    {
+      "id": 17,
+      "type": "PrimitiveNode",
+      "pos": [
+        487,
+        -196
+      ],
+      "size": [
+        263.0819396972656,
+        82
+      ],
+      "flags": {},
+      "order": 0,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "INT",
+          "type": "INT",
+          "links": [
+            21
+          ],
+          "slot_index": 0
+        }
+      ],
+      "title": "seed",
+      "properties": {
+        "Run widget replace on values": false
+      },
+      "widgets_values": [
+        220636977427261,
+        "randomize"
+      ]
+    },
+    {
+      "id": 52,
+      "type": "PrimitiveNode",
+      "pos": [
+        218,
+        -94
+      ],
+      "size": [
+        210,
+        151.54025268554688
+      ],
+      "flags": {
+        "collapsed": false
+      },
+      "order": 1,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "STRING",
+          "type": "STRING",
+          "links": [
+            222,
+            223
+          ],
+          "slot_index": 0
+        }
+      ],
+      "title": "Negative_prompt",
+      "properties": {
+        "Run widget replace on values": false
+      },
+      "widgets_values": [
+        "anime,render,cartoon,3d,bad structure"
+      ]
+    },
+    {
+      "id": 10,
+      "type": "TripleCLIPLoader",
+      "pos": [
+        -98,
+        -265
+      ],
+      "size": [
+        521.9664916992188,
+        120.35124206542969
+      ],
+      "flags": {},
+      "order": 2,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "CLIP",
+          "type": "CLIP",
+          "links": [
+            257,
+            258
+          ],
+          "slot_index": 0
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "TripleCLIPLoader"
+      },
+      "widgets_values": [
+        "bokeh_clip_g.safetensors",
+        "bokeh_clip_l.safetensors",
+        "t5xxl_fp16.safetensors"
+      ]
+    },
+    {
+      "id": 62,
+      "type": "Note",
+      "pos": [
+        -101,
+        107
+      ],
+      "size": [
+        304.94696044921875,
+        114.46440887451172
+      ],
+      "flags": {},
+      "order": 3,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [],
+      "properties": {},
+      "widgets_values": [
+        "Do not enter overly complex prompt words, which will cause serious degradation of image performance,you can use emotional and atmospheric cues to improve the picture quality"
+      ],
+      "color": "#432",
+      "bgcolor": "#653"
+    },
+    {
+      "id": 19,
+      "type": "PreviewImage",
+      "pos": [
+        1334,
+        -468
+      ],
+      "size": [
+        539.6785278320312,
+        669.1779174804688
+      ],
+      "flags": {
+        "collapsed": false
+      },
+      "order": 12,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "images",
+          "type": "IMAGE",
+          "link": 24
+        }
+      ],
+      "outputs": [],
+      "properties": {
+        "Node name for S&R": "PreviewImage"
+      },
+      "widgets_values": []
+    },
+    {
+      "id": 18,
+      "type": "VAEDecode",
+      "pos": [
+        1091,
+        -413
+      ],
+      "size": [
+        200.854736328125,
+        50.05826187133789
+      ],
+      "flags": {},
+      "order": 11,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "samples",
+          "type": "LATENT",
+          "link": 22
+        },
+        {
+          "name": "vae",
+          "type": "VAE",
+          "link": 23
+        }
+      ],
+      "outputs": [
+        {
+          "name": "IMAGE",
+          "type": "IMAGE",
+          "links": [
+            24
+          ],
+          "slot_index": 0
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "VAEDecode"
+      },
+      "widgets_values": []
+    },
+    {
+      "id": 59,
+      "type": "Note",
+      "pos": [
+        487,
+        105
+      ],
+      "size": [
+        249.3325653076172,
+        104.7717514038086
+      ],
+      "flags": {},
+      "order": 4,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [],
+      "properties": {},
+      "widgets_values": [
+        "1920x1024 1728x1152 1152x1728 1280x1664 1440x1440"
+      ],
+      "color": "#432",
+      "bgcolor": "#653"
+    },
+    {
+      "id": 5,
+      "type": "EmptyLatentImage",
+      "pos": [
+        475,
+        -54
+      ],
+      "size": [
+        266.8973388671875,
+        116.21234893798828
+      ],
+      "flags": {},
+      "order": 5,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "LATENT",
+          "type": "LATENT",
+          "links": [
+            243
+          ],
+          "slot_index": 0
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "EmptyLatentImage"
+      },
+      "widgets_values": [
+        1280,
+        1664,
+        1
+      ]
+    },
+    {
+      "id": 12,
+      "type": "KSampler",
+      "pos": [
+        771,
+        -418
+      ],
+      "size": [
+        279.5604553222656,
+        258
+      ],
+      "flags": {
+        "collapsed": false
+      },
+      "order": 10,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "model",
+          "type": "MODEL",
+          "link": 256
+        },
+        {
+          "name": "positive",
+          "type": "CONDITIONING",
+          "link": 237
+        },
+        {
+          "name": "negative",
+          "type": "CONDITIONING",
+          "link": 236
+        },
+        {
+          "name": "latent_image",
+          "type": "LATENT",
+          "link": 243
+        },
+        {
+          "name": "seed",
+          "type": "INT",
+          "widget": {
+            "name": "seed"
+          },
+          "link": 21
+        }
+      ],
+      "outputs": [
+        {
+          "name": "LATENT",
+          "type": "LATENT",
+          "links": [
+            22
+          ],
+          "slot_index": 0
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "KSampler"
+      },
+      "widgets_values": [
+        220636977427261,
+        "randomize",
+        28,
+        4,
+        "dpmpp_2m",
+        "sgm_uniform",
+        1
+      ]
+    },
+    {
+      "id": 20,
+      "type": "PrimitiveNode",
+      "pos": [
+        -102,
+        -94
+      ],
+      "size": [
+        306.11773681640625,
+        152.6992950439453
+      ],
+      "flags": {
+        "collapsed": false
+      },
+      "order": 6,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "STRING",
+          "type": "STRING",
+          "links": [
+            122,
+            123,
+            152
+          ],
+          "slot_index": 0
+        }
+      ],
+      "title": "Positive_prompt",
+      "properties": {
+        "Run widget replace on values": false
+      },
+      "widgets_values": [
+        "Close-up of a macaw, dimly lit environment"
+      ]
+    },
+    {
+      "id": 13,
+      "type": "CheckpointLoaderSimple",
+      "pos": [
+        -95,
+        -414
+      ],
+      "size": [
+        510.9742431640625,
+        107.2224349975586
+      ],
+      "flags": {},
+      "order": 7,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "MODEL",
+          "type": "MODEL",
+          "links": [
+            256
+          ],
+          "slot_index": 0
+        },
+        {
+          "name": "CLIP",
+          "type": "CLIP",
+          "links": [],
+          "slot_index": 1
+        },
+        {
+          "name": "VAE",
+          "type": "VAE",
+          "links": [
+            23
+          ],
+          "slot_index": 2
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "CheckpointLoaderSimple"
+      },
+      "widgets_values": [
+        "bokeh_3.5_medium.safetensors"
+      ]
+    }
+  ],
+  "links": [
+    [
+      21,
+      17,
+      0,
+      12,
+      4,
+      "INT"
+    ],
+    [
+      22,
+      12,
+      0,
+      18,
+      0,
+      "LATENT"
+    ],
+    [
+      23,
+      13,
+      2,
+      18,
+      1,
+      "VAE"
+    ],
+    [
+      24,
+      18,
+      0,
+      19,
+      0,
+      "IMAGE"
+    ],
+    [
+      122,
+      20,
+      0,
+      32,
+      1,
+      "STRING"
+    ],
+    [
+      123,
+      20,
+      0,
+      32,
+      2,
+      "STRING"
+    ],
+    [
+      152,
+      20,
+      0,
+      32,
+      3,
+      "STRING"
+    ],
+    [
+      222,
+      52,
+      0,
+      30,
+      1,
+      "STRING"
+    ],
+    [
+      223,
+      52,
+      0,
+      30,
+      2,
+      "STRING"
+    ],
+    [
+      236,
+      30,
+      0,
+      12,
+      2,
+      "CONDITIONING"
+    ],
+    [
+      237,
+      32,
+      0,
+      12,
+      1,
+      "CONDITIONING"
+    ],
+    [
+      243,
+      5,
+      0,
+      12,
+      3,
+      "LATENT"
+    ],
+    [
+      256,
+      13,
+      0,
+      12,
+      0,
+      "MODEL"
+    ],
+    [
+      257,
+      10,
+      0,
+      32,
+      0,
+      "CLIP"
+    ],
+    [
+      258,
+      10,
+      0,
+      30,
+      0,
+      "CLIP"
+    ]
+  ],
+  "groups": [],
+  "config": {},
+  "extra": {
+    "ds": {
+      "scale": 0.7247295000000004,
+      "offset": [
+        833.5543134270109,
+        652.7487494515917
+      ]
+    },
+    "VHS_latentpreview": false,
+    "VHS_latentpreviewrate": 0,
+    "VHS_MetadataImage": true,
+    "VHS_KeepIntermediate": true
+  },
+  "version": 0.4
+}