v1: 5 checkpoints, samples, workflow, model card

.gitattributes

@@ -33,3 +33,24 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+samples/best_grid.jpg filter=lfs diff=lfs merge=lfs -text
+samples/checkpoint_grid.jpg filter=lfs diff=lfs merge=lfs -text
+samples/step1200_portrait_s10.png filter=lfs diff=lfs merge=lfs -text
+samples/step1200_portrait_s15.png filter=lfs diff=lfs merge=lfs -text
+samples/step1200_scene_s10.png filter=lfs diff=lfs merge=lfs -text
+samples/step1200_scene_s15.png filter=lfs diff=lfs merge=lfs -text
+samples/step1600_portrait_s10.png filter=lfs diff=lfs merge=lfs -text
+samples/step1600_portrait_s15.png filter=lfs diff=lfs merge=lfs -text
+samples/step1600_scene_s10.png filter=lfs diff=lfs merge=lfs -text
+samples/step1600_scene_s15.png filter=lfs diff=lfs merge=lfs -text
+samples/step2000_portrait_s10.png filter=lfs diff=lfs merge=lfs -text
+samples/step2000_portrait_s15.png filter=lfs diff=lfs merge=lfs -text
+samples/step2000_portrait_s20.png filter=lfs diff=lfs merge=lfs -text
+samples/step2000_scene_s10.png filter=lfs diff=lfs merge=lfs -text
+samples/step2000_scene_s15.png filter=lfs diff=lfs merge=lfs -text
+samples/step2000_scene_s20.png filter=lfs diff=lfs merge=lfs -text
+samples/step400_portrait_s10.png filter=lfs diff=lfs merge=lfs -text
+samples/step400_scene_s10.png filter=lfs diff=lfs merge=lfs -text
+samples/step800_portrait_s10.png filter=lfs diff=lfs merge=lfs -text
+samples/step800_scene_s10.png filter=lfs diff=lfs merge=lfs -text
+samples/strength_grid.jpg filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,101 @@
+---
+base_model: black-forest-labs/FLUX.1-dev
+tags:
+  - flux
+  - lora
+  - text-to-image
+  - style
+  - kohya-ss
+license: other
+license_name: flux-1-dev-non-commercial-license
+license_link: https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md
+instance_prompt: sty1ref
+pipeline_tag: text-to-image
+---
+
+# sty1ref — FLUX.1-dev style LoRA (v1)
+
+A stylised portrait-illustration LoRA: hard-edged planar shading, flat colour blocking
+across the face, visible geometric facets in skin and hair, muted palette.
+
+**Trigger word: `sty1ref`** — put it at the front of the prompt.
+
+## Recommended settings
+
+| Setting | Value |
+|---|---|
+| Checkpoint | `sty1ref-step00002000.safetensors` |
+| **LoRA strength** | **1.4 – 1.6** (not 1.0 — see below) |
+| Base | `flux1-dev` (fp8_e4m3fn is fine) |
+| Sampler / scheduler | euler / beta, 28 steps |
+| FluxGuidance | 3.0 |
+| CFG | 1.0 (Flux dev is CFG-distilled; negatives do nothing) |
+
+## Read this before using it
+
+**Strength 1.0 is too weak.** This LoRA is undertrained at unit strength — a prompt
+rendered at 1.0 comes back looking like base Flux with a light stylisation pass. The
+faceted planar shading only appears clearly from about 1.4 upward. 1.5 is the sweet
+spot; 2.0 works but muddies the midtones.
+
+**It is portrait-biased.** All 12 training images are head-and-shoulders portraits, so
+the trigger has only ever co-occurred with faces. Consequences:
+
+- Portraits of unseen subjects: works well at 1.5.
+- Scenes without people (streets, landscapes, objects): the output becomes *painterly*
+  but does not pick up the hard faceted planes. At 1.0 it is essentially unstyled.
+
+If you need this texture on arbitrary scenes, retrain with non-portrait references in
+the same style — full figures, architecture, objects, landscapes. That is the fix; no
+strength value substitutes for it.
+
+## Checkpoints
+
+| File | Steps | Notes |
+|---|---|---|
+| `sty1ref-step00000400.safetensors` | 400 | barely stylised |
+| `sty1ref-step00000800.safetensors` | 800 | faint |
+| `sty1ref-step00001200.safetensors` | 1200 | usable at 1.5 |
+| `sty1ref-step00001600.safetensors` | 1600 | close second |
+| `sty1ref-step00002000.safetensors` | 2000 | **recommended** |
+
+`samples/` holds two renders per checkpoint at a fixed seed (42) — one portrait of an
+unseen subject, one people-free scene — plus the comparison grids.
+
+## Training recipe
+
+kohya-ss `sd-scripts` (sd3 branch), `flux_train_network.py`, on one RTX 5090 (32 GB),
+~1 hour for 2000 steps.
+
+```
+--network_module networks.lora_flux --network_dim 24 --network_alpha 24
+--network_train_unet_only
+--optimizer_type adamw8bit --learning_rate 1e-4
+--lr_scheduler constant_with_warmup --lr_warmup_steps 40
+--max_train_steps 2000 --save_every_n_steps 400
+--gradient_checkpointing --mixed_precision bf16 --fp8_base --sdpa --highvram
+--timestep_sampling shift --discrete_flow_shift 3.1582
+--model_prediction_type raw --guidance_scale 1.0 --loss_type l2
+```
+
+Dataset: 12 images, aspect-ratio bucketing at 1024 base with `bucket_no_upscale`,
+`keep_tokens = 1`. Captions describe **content only** (subject, clothing, framing,
+background) so that everything constant across the set collapses onto the trigger token
+rather than scattering across style adjectives.
+
+Known cause of the weak transfer: 12 images at lr 1e-4 for 2000 steps undercooks this
+style. A v2 should use more images and/or lr 2e-4.
+
+## Dataset
+
+`kirusanth08/sty1ref-dataset` (private) — the 12 cleaned images and their caption files.
+
+## Usage (ComfyUI)
+
+`workflow/workflow_api.json` is a working API-format graph at the recommended settings.
+`workflow/workflow.json` drags onto the canvas.
+
+## Licence
+
+Inherits the FLUX.1-dev non-commercial licence. The training references were collected
+from the web and are not owned by the author of this LoRA; treat outputs accordingly.

sty1ref-step00000400.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dc3af6670ee57338fa8ab2a22e8d335da3d65af10456102463c25535c9a8c7dd
+size 229855896

sty1ref-step00000800.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:363df8326d2e6fdac7f5c7dc177c5969df22f8fe760662f06fc29e09961dc210
+size 229855896

sty1ref-step00001200.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:68a3aef8bff4230340ff408bf829129b9fdc1354b145fe296d684f9827e92ddf
+size 229855904

sty1ref-step00001600.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:14d820ccf154ebba0a7c6279715c3a1948b7a143dc6cf50aa12b2c1121f31c46
+size 229855896

sty1ref-step00002000.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:39b232035417eedf2b657e5f422f889d7f98d4c56e27494f5b8a33cf5bfa38a7
+size 229855904

workflow/workflow.json

@@ -0,0 +1,524 @@
+{
+ "last_node_id": 11,
+ "last_link_id": 12,
+ "nodes": [
+  {
+   "id": 1,
+   "type": "UNETLoader",
+   "pos": [
+    40,
+    40
+   ],
+   "size": [
+    380,
+    82
+   ],
+   "flags": {},
+   "order": 0,
+   "mode": 0,
+   "inputs": [],
+   "outputs": [
+    {
+     "name": "MODEL",
+     "type": "MODEL",
+     "links": [
+      1
+     ],
+     "slot_index": 0
+    }
+   ],
+   "properties": {
+    "Node name for S&R": "UNETLoader"
+   },
+   "widgets_values": [
+    "flux1-dev.safetensors",
+    "fp8_e4m3fn"
+   ],
+   "title": "Load Flux dev UNET"
+  },
+  {
+   "id": 2,
+   "type": "LoraLoaderModelOnly",
+   "pos": [
+    40,
+    180
+   ],
+   "size": [
+    380,
+    82
+   ],
+   "flags": {},
+   "order": 0,
+   "mode": 0,
+   "inputs": [
+    {
+     "name": "model",
+     "type": "MODEL",
+     "link": 1
+    }
+   ],
+   "outputs": [
+    {
+     "name": "MODEL",
+     "type": "MODEL",
+     "links": [
+      2
+     ],
+     "slot_index": 0
+    }
+   ],
+   "properties": {
+    "Node name for S&R": "LoraLoaderModelOnly"
+   },
+   "widgets_values": [
+    "sk3tchref.safetensors",
+    1.0
+   ],
+   "title": "Style LoRA  (strength 0.8-1.0)"
+  },
+  {
+   "id": 3,
+   "type": "DualCLIPLoader",
+   "pos": [
+    40,
+    320
+   ],
+   "size": [
+    380,
+    130
+   ],
+   "flags": {},
+   "order": 0,
+   "mode": 0,
+   "inputs": [],
+   "outputs": [
+    {
+     "name": "CLIP",
+     "type": "CLIP",
+     "links": [
+      3
+     ],
+     "slot_index": 0
+    }
+   ],
+   "properties": {
+    "Node name for S&R": "DualCLIPLoader"
+   },
+   "widgets_values": [
+    "clip_l.safetensors",
+    "t5xxl_fp16.safetensors",
+    "flux",
+    "default"
+   ],
+   "title": "Text encoders"
+  },
+  {
+   "id": 4,
+   "type": "VAELoader",
+   "pos": [
+    40,
+    500
+   ],
+   "size": [
+    380,
+    58
+   ],
+   "flags": {},
+   "order": 0,
+   "mode": 0,
+   "inputs": [],
+   "outputs": [
+    {
+     "name": "VAE",
+     "type": "VAE",
+     "links": [
+      5
+     ],
+     "slot_index": 0
+    }
+   ],
+   "properties": {
+    "Node name for S&R": "VAELoader"
+   },
+   "widgets_values": [
+    "ae.safetensors"
+   ],
+   "title": "Flux VAE"
+  },
+  {
+   "id": 5,
+   "type": "CLIPTextEncode",
+   "pos": [
+    470,
+    40
+   ],
+   "size": [
+    520,
+    240
+   ],
+   "flags": {},
+   "order": 0,
+   "mode": 0,
+   "inputs": [
+    {
+     "name": "clip",
+     "type": "CLIP",
+     "link": 3
+    }
+   ],
+   "outputs": [
+    {
+     "name": "CONDITIONING",
+     "type": "CONDITIONING",
+     "links": [
+      6,
+      7
+     ],
+     "slot_index": 0
+    }
+   ],
+   "properties": {
+    "Node name for S&R": "CLIPTextEncode"
+   },
+   "widgets_values": [
+    "sk3tchref, a full body character illustration of a heavy-set adult man with deep dark brown skin, thick dark beard with grey streaks, short silver-grey buzz cut, gold hoop earring and thin gold chain, slouched and disinterested, holding a half-empty wine bottle low in one hand, wearing a faded olive-green oversized delivery jacket with cracked white TOONCITY lettering on the chest, grey undershirt, baggy dark work trousers, scuffed brown boots and a crumpled canvas bag over one shoulder, hand-painted character concept art, broad flat brush shapes, hard-edged colour blocking, muted desaturated palette, flat off-white background, thin grey shadow smear under the feet"
+   ],
+   "title": "Prompt  -  must start with sk3tchref"
+  },
+  {
+   "id": 6,
+   "type": "FluxGuidance",
+   "pos": [
+    470,
+    330
+   ],
+   "size": [
+    520,
+    58
+   ],
+   "flags": {},
+   "order": 0,
+   "mode": 0,
+   "inputs": [
+    {
+     "name": "conditioning",
+     "type": "CONDITIONING",
+     "link": 6
+    }
+   ],
+   "outputs": [
+    {
+     "name": "CONDITIONING",
+     "type": "CONDITIONING",
+     "links": [
+      8
+     ],
+     "slot_index": 0
+    }
+   ],
+   "properties": {
+    "Node name for S&R": "FluxGuidance"
+   },
+   "widgets_values": [
+    3.0
+   ],
+   "title": "FluxGuidance (adherence knob)"
+  },
+  {
+   "id": 7,
+   "type": "ConditioningZeroOut",
+   "pos": [
+    470,
+    430
+   ],
+   "size": [
+    520,
+    40
+   ],
+   "flags": {},
+   "order": 0,
+   "mode": 0,
+   "inputs": [
+    {
+     "name": "conditioning",
+     "type": "CONDITIONING",
+     "link": 7
+    }
+   ],
+   "outputs": [
+    {
+     "name": "CONDITIONING",
+     "type": "CONDITIONING",
+     "links": [
+      9
+     ],
+     "slot_index": 0
+    }
+   ],
+   "properties": {
+    "Node name for S&R": "ConditioningZeroOut"
+   },
+   "widgets_values": [],
+   "title": "Empty negative (Flux dev is CFG-distilled)"
+  },
+  {
+   "id": 8,
+   "type": "EmptySD3LatentImage",
+   "pos": [
+    470,
+    520
+   ],
+   "size": [
+    520,
+    106
+   ],
+   "flags": {},
+   "order": 0,
+   "mode": 0,
+   "inputs": [],
+   "outputs": [
+    {
+     "name": "LATENT",
+     "type": "LATENT",
+     "links": [
+      10
+     ],
+     "slot_index": 0
+    }
+   ],
+   "properties": {
+    "Node name for S&R": "EmptySD3LatentImage"
+   },
+   "widgets_values": [
+    832,
+    1216,
+    1
+   ],
+   "title": "Canvas"
+  },
+  {
+   "id": 9,
+   "type": "KSampler",
+   "pos": [
+    1040,
+    40
+   ],
+   "size": [
+    380,
+    262
+   ],
+   "flags": {},
+   "order": 0,
+   "mode": 0,
+   "inputs": [
+    {
+     "name": "model",
+     "type": "MODEL",
+     "link": 2
+    },
+    {
+     "name": "positive",
+     "type": "CONDITIONING",
+     "link": 8
+    },
+    {
+     "name": "negative",
+     "type": "CONDITIONING",
+     "link": 9
+    },
+    {
+     "name": "latent_image",
+     "type": "LATENT",
+     "link": 10
+    }
+   ],
+   "outputs": [
+    {
+     "name": "LATENT",
+     "type": "LATENT",
+     "links": [
+      11
+     ],
+     "slot_index": 0
+    }
+   ],
+   "properties": {
+    "Node name for S&R": "KSampler"
+   },
+   "widgets_values": [
+    4242,
+    "randomize",
+    28,
+    1.0,
+    "euler",
+    "beta",
+    1.0
+   ],
+   "title": "KSampler  -  cfg MUST stay 1.0"
+  },
+  {
+   "id": 10,
+   "type": "VAEDecode",
+   "pos": [
+    1040,
+    350
+   ],
+   "size": [
+    380,
+    46
+   ],
+   "flags": {},
+   "order": 0,
+   "mode": 0,
+   "inputs": [
+    {
+     "name": "samples",
+     "type": "LATENT",
+     "link": 11
+    },
+    {
+     "name": "vae",
+     "type": "VAE",
+     "link": 5
+    }
+   ],
+   "outputs": [
+    {
+     "name": "IMAGE",
+     "type": "IMAGE",
+     "links": [
+      12
+     ],
+     "slot_index": 0
+    }
+   ],
+   "properties": {
+    "Node name for S&R": "VAEDecode"
+   },
+   "widgets_values": []
+  },
+  {
+   "id": 11,
+   "type": "SaveImage",
+   "pos": [
+    1040,
+    440
+   ],
+   "size": [
+    440,
+    450
+   ],
+   "flags": {},
+   "order": 0,
+   "mode": 0,
+   "inputs": [
+    {
+     "name": "images",
+     "type": "IMAGE",
+     "link": 12
+    }
+   ],
+   "outputs": [],
+   "properties": {
+    "Node name for S&R": "SaveImage"
+   },
+   "widgets_values": [
+    "sk3tchref"
+   ]
+  }
+ ],
+ "links": [
+  [
+   1,
+   1,
+   0,
+   2,
+   0,
+   "MODEL"
+  ],
+  [
+   2,
+   2,
+   0,
+   9,
+   0,
+   "MODEL"
+  ],
+  [
+   3,
+   3,
+   0,
+   5,
+   0,
+   "CLIP"
+  ],
+  [
+   5,
+   4,
+   0,
+   10,
+   1,
+   "VAE"
+  ],
+  [
+   6,
+   5,
+   0,
+   6,
+   0,
+   "CONDITIONING"
+  ],
+  [
+   7,
+   5,
+   0,
+   7,
+   0,
+   "CONDITIONING"
+  ],
+  [
+   8,
+   6,
+   0,
+   9,
+   1,
+   "CONDITIONING"
+  ],
+  [
+   9,
+   7,
+   0,
+   9,
+   2,
+   "CONDITIONING"
+  ],
+  [
+   10,
+   8,
+   0,
+   9,
+   3,
+   "LATENT"
+  ],
+  [
+   11,
+   9,
+   0,
+   10,
+   0,
+   "LATENT"
+  ],
+  [
+   12,
+   10,
+   0,
+   11,
+   0,
+   "IMAGE"
+  ]
+ ],
+ "groups": [],
+ "config": {},
+ "extra": {},
+ "version": 0.4
+}

workflow/workflow_api.json

@@ -0,0 +1,121 @@
+{
+ "1": {
+  "class_type": "UNETLoader",
+  "inputs": {
+   "unet_name": "flux1-dev.safetensors",
+   "weight_dtype": "fp8_e4m3fn"
+  }
+ },
+ "1b": {
+  "class_type": "LoraLoaderModelOnly",
+  "inputs": {
+   "model": [
+    "1",
+    0
+   ],
+   "lora_name": "sty1ref-step00002000.safetensors",
+   "strength_model": 1.5
+  }
+ },
+ "2": {
+  "class_type": "DualCLIPLoader",
+  "inputs": {
+   "clip_name1": "clip_l.safetensors",
+   "clip_name2": "t5xxl_fp16.safetensors",
+   "type": "flux"
+  }
+ },
+ "3": {
+  "class_type": "VAELoader",
+  "inputs": {
+   "vae_name": "ae.safetensors"
+  }
+ },
+ "4": {
+  "class_type": "CLIPTextEncode",
+  "inputs": {
+   "clip": [
+    "2",
+    0
+   ],
+   "text": "sty1ref, an old fisherman with a thick white beard and a navy knitted cap, facing the viewer, muted grey green background"
+  }
+ },
+ "5": {
+  "class_type": "FluxGuidance",
+  "inputs": {
+   "conditioning": [
+    "4",
+    0
+   ],
+   "guidance": 3.0
+  }
+ },
+ "11": {
+  "class_type": "ConditioningZeroOut",
+  "inputs": {
+   "conditioning": [
+    "4",
+    0
+   ]
+  }
+ },
+ "12": {
+  "class_type": "EmptySD3LatentImage",
+  "inputs": {
+   "width": 896,
+   "height": 1152,
+   "batch_size": 1
+  }
+ },
+ "13": {
+  "class_type": "KSampler",
+  "inputs": {
+   "model": [
+    "1b",
+    0
+   ],
+   "positive": [
+    "5",
+    0
+   ],
+   "negative": [
+    "11",
+    0
+   ],
+   "latent_image": [
+    "12",
+    0
+   ],
+   "seed": 42,
+   "steps": 28,
+   "cfg": 1.0,
+   "sampler_name": "euler",
+   "scheduler": "beta",
+   "denoise": 1.0
+  }
+ },
+ "14": {
+  "class_type": "VAEDecode",
+  "inputs": {
+   "samples": [
+    "13",
+    0
+   ],
+   "vae": [
+    "3",
+    0
+   ]
+  }
+ },
+ "15": {
+  "class_type": "SaveImage",
+  "inputs": {
+   "images": [
+    "14",
+    0
+   ],
+   "filename_prefix": "sty1ref"
+  }
+ }
+}