Qualitative Results
April 8, 2024 ยท View on GitHub
Iconography Style
Case: Pandas
Prompt: a panda rowing a boat in a pond.
Style: iconography
Preview:
| (a) Sample raster image with Stable Diffusion | (b) Convert raster image to a vector via LIVE | (c) VectorFusion: Fine tune by LSDS |
Script:
python run_painterly_render.py -c vectorfusion.yaml -pt "a panda rowing a boat in a pond. minimal flat 2d vector icon. lineal color. on a white background. trending on artstation" -save_step 50 -respath ./workdir/pandas -d 294308
Case: Starbucks
Prompt: A Starbucks coffee.
Style: iconography
Preview:
| (a) Sample raster image with Stable Diffusion | (b) Convert raster image to a vector via LIVE | (c) VectorFusion: Fine tune by LSDS |
Script:
python run_painterly_render.py -c vectorfusion.yaml -pt "A Starbucks coffee. minimal flat 2d vector icon. lineal color. on a white background. trending on artstation" -save_step 50 -respath ./workdir/starbucks -d 367022
Case: Sloth
Prompt: A smiling sloth wearing a leather jacket, a cowboy hat and a kilt.
Style: iconography
Preview:
| (a) Sample raster image with Stable Diffusion | (b) Convert raster image to a vector via LIVE | (c) VectorFusion: Fine tune by LSDS |
Script:
python run_painterly_render.py -c vectorfusion.yaml -pt "A smiling sloth wearing a leather jacket, a cowboy hat and a kilt. minimal flat 2d vector icon. lineal color. on a white background. trending on artstation" -save_step 50 -respath ./workdir/sloth -d 280328
Case: Owl
Prompt: an owl standing on a wire.
Style: iconography
Preview:
| (a) Sample raster image with Stable Diffusion | (b) Convert raster image to a vector via LIVE | (c) VectorFusion: Fine tune by LSDS |
Script:
python run_painterly_render.py -c vectorfusion.yaml -pt "an owl standing on a wire. minimal flat 2d vector icon. lineal color. on a white background. trending on artstation" -save_step 50 -respath ./workdir/owl -d 857581
Case: Train
Prompt: a train.
Style: iconography
Preview:
| (a) Sample raster image with Stable Diffusion | (b) Convert raster image to a vector via LIVE | (c) VectorFusion: Fine tune by LSDS |
Script:
python run_painterly_render.py -c vectorfusion.yaml -pt "a train. minimal flat 2d vector icon. lineal color. on a white background. trending on artstation" -save_step 50 -respath ./workdir/train -d 857581
Case: Fire
Prompt: fire.
Style: iconography
Preview:
| (a) Sample raster image with Stable Diffusion | (b) Convert raster image to a vector via LIVE | (c) VectorFusion: Fine tune by LSDS |
Script:
python run_painterly_render.py -c vectorfusion.yaml -pt "fire. minimal flat 2d vector icon. lineal color. on a white background. trending on artstation" -save_step 50 -respath ./workdir/fire -d 1452
Pixel-Art Style
Case: Guitar
Prompt: Electric guitar.
Style: Pixel-Art
Preview:
| (a) Sample raster image with Stable Diffusion | (b) Convert raster image to a vector via LIVE | (c) VectorFusion: Fine tune by LSDS |
Script:
python run_painterly_render.py -c vectorfusion.yaml -pt "Electric guitar. pixel art. trending on artstation" -save_step 50 -respath ./workdir/guitar -update "style=pixelart" -d 428484
Case: Hamburger
Prompt: A delicious hamburger.
Style: Pixel-Art
Preview:
| (a) Sample raster image with Stable Diffusion | (b) Convert raster image to a vector via LIVE | (c) VectorFusion: Fine tune by LSDS |
Script:
python run_painterly_render.py -c vectorfusion.yaml -pt "A delicious hamburger. pixel art. trending on artstation" -save_step 50 -respath ./workdir/hamburger -update "style=pixelart" -d 499578
Case: Pikachu
Prompt: Pikachu.
Style: Pixel-Art
Preview:
| (a) Sample raster image with Stable Diffusion | (b) Convert raster image to a vector via LIVE | (c) VectorFusion: Fine tune by LSDS |
Script:
python run_painterly_render.py -c vectorfusion.yaml -pt "A delicious hamburger. pixel art. trending on artstation" -save_step 50 -respath ./workdir/hamburger -update "style=pixelart" -d 499578
Sketch Style
Case: Dragon
Prompt: watercolor painting of a firebreathing dragon.
Style: Sketch
Preview:
| SVG initialization | VectorFusion fine-tune 500 step | VectorFusion fine-tune 1500 step |
Script:
python run_painterly_render.py -c vectorfusion.yaml -pt "watercolor painting of a firebreathing dragon. minimal 2d line drawing. trending on artstation" -save_step 50 -respath ./workdir/dragon-sketch -update "style=sketch num_segments=5 radius=0.5 sds.num_iter=1500" -d 947593
Case: The Eiffel Tower
Prompt: The Eiffel Tower.
Style: Sketch
Preview:
| SVG initialization | VectorFusion fine-tune 500 step | VectorFusion fine-tune 1500 step |
Script:
CUDA_VISIBLE_DEVICES=0 python run_painterly_render.py -c vectorfusion.yaml -pt "The Eiffel Tower. minimal 2d line drawing. trending on artstation" -save_step 50 -respath ./workdir/EiffelTower-sketch -update "style=sketch skip_live=True num_segments=5 radius=0.5 sds.num_iter=1500" -d 965058
Case: Temple
Prompt: A 3D rendering of a temple.
Style: Sketch
Preview:
| SVG initialization | VectorFusion fine-tune 500 step | VectorFusion fine-tune 1500 step |
Script:
python run_painterly_render.py -c vectorfusion.yaml -pt "A 3D rendering of a temple. minimal 2d line drawing. trending on artstation" -save_step 50 -respath ./workdir/temple-sketch -update "style=sketch num_segments=5 radius=0.5 sds.num_iter=1500" -d 809385