Table 2 An ablation study comparing various conditions injected into the PDANet (Ours) model
From: A novel flexible identity-net with diffusion models for painting-style generation
CLIP Score \(\uparrow\) | LPIPS \(\downarrow\) | FID \(\downarrow\) | VTC \(\uparrow\) | ASL \(\uparrow\) | AP \(\uparrow\) | Creativeness \(\uparrow\) | |
|---|---|---|---|---|---|---|---|
Baseline | 0.8288 | 0.5185 | 1015 | 3.0% | 3.0% | 4.5% | 7.6% |
+indentitynet | 0.8081 | 0.6229 | 1270 | 31.8% | 27.3% | 21.2% | 13.6% |
+stylenet | 0.8506 | 0.4799 | 1976 | 3.0% | 4.6% | 10.6% | 19.7% |
Ours | 0.8844 | 0.4763 | 774 | 62.2% | 65.1% | 63.7% | 59.1% |
