This page shows examples of pairs of compressed images for which SSIM disagrees with all other metrics about which image is best. Up to you to judge if it made a good call or not!
 |  |  |
| avif 444 s6 q59 SSIMULACRA_2_modelD=41.68495861 AIMOS_interpolated=0.4409 SSIM=0.992429 PSNR-Y=30.398672 Butteraugli_3norm= 4.308322 VMAF-NEG=74.367529 DSSIM=0.00525879 PSNR-HVS=32.97703 | Reference | cavif s10 q20 SSIMULACRA_2_modelD=45.88944308 AIMOS_interpolated=0.4924 SSIM=0.987779 PSNR-Y=32.227277 Butteraugli_3norm= 3.140061 VMAF-NEG=78.745224 DSSIM=0.00499962 PSNR-HVS=35.17468 |
 |  |  |
| avif 444 s6 q55 SSIMULACRA_2_modelD=35.36956302 AIMOS_interpolated=0.3242 SSIM=0.990404 PSNR-Y=31.633310 Butteraugli_3norm= 3.453294 VMAF-NEG=69.895568 DSSIM=0.00608157 PSNR-HVS=34.86369 | Reference | cavif s10 q26 SSIMULACRA_2_modelD=44.07503892 AIMOS_interpolated=0.4611 SSIM=0.984547 PSNR-Y=32.922997 Butteraugli_3norm= 3.010663 VMAF-NEG=74.982310 DSSIM=0.00586770 PSNR-HVS=36.15981 |
 |  |  |
| avif 444 s6 q47 SSIMULACRA_2_modelD=61.67765925 AIMOS_interpolated=0.6095 SSIM=0.989569 PSNR-Y=38.396573 Butteraugli_3norm= 2.180791 VMAF-NEG=82.763069 DSSIM=0.00283024 PSNR-HVS=38.32709 | Reference | cavif s4 q32 SSIMULACRA_2_modelD=61.97225068 AIMOS_interpolated=0.6317 SSIM=0.985870 PSNR-Y=39.852586 Butteraugli_3norm= 1.925736 VMAF-NEG=85.302878 DSSIM=0.00277737 PSNR-HVS=38.91443 |
 |  |  |
| avif 444 s6 q59 SSIMULACRA_2_modelD=49.05282435 AIMOS_interpolated=0.4632 SSIM=0.991657 PSNR-Y=29.600617 Butteraugli_3norm= 4.039972 VMAF-NEG=68.891865 DSSIM=0.00467196 PSNR-HVS=30.77344 | Reference | cavif s10 q24 SSIMULACRA_2_modelD=55.73060843 AIMOS_interpolated=0.5518 SSIM=0.986407 PSNR-Y=31.680735 Butteraugli_3norm= 2.968070 VMAF-NEG=77.757137 DSSIM=0.00464125 PSNR-HVS=34.27098 |