ResearchBenchmarksPairwise human preference on Prolific user study large-scale (test)Pairwise human preference on Prolific user study large-scale (test)Follow97.33Winning RatePrefGen77.923682.96188893.0382Dec 4, 2025Evaluation resultsMethodMethodLinksWinning RatePrefGenbaseline=ViPerbaseline=ViPer2025.1297.33PrefGenbaseline=InstantStylebaseline=InstantStyle2025.1290.67PrefGenbaseline=IP-Adapterbaseline=IP-Adapter2025.1288.67PrefGenbaseline=StyleAlignedbaseline=StyleAligned2025.1288.67PrefGenbaseline=EasyRefbaseline=EasyRef2025.1288PrefGenbaseline=Bagelbaseline=Bagel2025.1278.67