metadata

tags:
  - text-to-image
  - lora
  - diffusers
  - template:diffusion-lora
widget:
  - text: >-
      a puppy and a cute mouse staring at each othera puppy and a cute mouse
      staring at each other
    output:
      url: images/ComfyUI_00176_.png
  - text: siamese cats playing
    output:
      url: images/ComfyUI_00175_.png
  - text: butterflies fluttering
    output:
      url: images/ComfyUI_00169_.png
  - text: coneflower and Snow leopard; wax lace broccoli afremov hues shades hues
    output:
      url: images/ComfyUI_00166_.png
  - text: >-
      The rose and the wishing Well , hyperdetailed, subtractive lighting,
      romanticism, surrealism, vivid and vibrant
    output:
      url: images/ComfyUI_00117_.png
  - text: >-
      before us we see a man sitting at a table. he is handsome, rugged, and
      looks ready for anything.
    output:
      url: images/ComfyUI_00100_.png
  - text: two dogs chasing a chicken
    output:
      url: images/ComfyUI_00096_.png
  - text: a man walking a  tiger
    output:
      url: images/ComfyUI_00078_.png
base_model: stabilityai/stable-diffusion-3.5-large
instance_prompt: impastoch
license: apache-2.0

thick-paint

Prompt
a puppy and a cute mouse staring at each othera puppy and a cute mouse staring at each other

Prompt
coneflower and Snow leopard; wax lace broccoli afremov hues shades hues

Prompt
The rose and the wishing Well , hyperdetailed, subtractive lighting, romanticism, surrealism, vivid and vibrant

Prompt
before us we see a man sitting at a table. he is handsome, rugged, and looks ready for anything.

Model description

This is a style lora, designed to give your stable diffusion 3.5 large generations a more organic feel. Images can have painted canvas textures with a thin coating of paint at times, but will mostly display a visible look as if the image were created by applying thick paint to the canvas with a pallet knife.

Because this is a style lora, images from short prompts will show more of the effect than images from longer prompts.

Suggested lora weight settings: 1 for short prompts to 3 for fairly long prompts.

The images in the following examples were created in ComfyUI with the following settings:

EmptySD3LatentImage Node: Width: 1024 Height: 1024

Ksampler node: Steps: 32 Cfg: 3.5 Sampler: Euler Scheduler: Simple denoise: 1.0

ModelSamplingSD3 Node: Shift: 2.0

Lora Loader node: Lora Weight: 1.0 Clip weight: 1.0

NOTE: if you are using the latest version of ComfyUI, use this <lora:thick-paint-lora.safetensors:1.0> at the start of your prompt rather than the trigger word

Trigger words

You should use impastoch to trigger the image generation.

Download model

Weights for this model are available in Safetensors format.

Download them in the Files & versions tab.