I have been experimenting heavily with flux style trainings, using this Alex Ross artwork dataset as the basis for these tests. The goal is finding the ideal formula for flux style trainings. I would love to see what all of you create with this style, this is what keeps me going....please post your results 🙏! All feedback is welcome and appreciated.Outputs on v2 are much more consistent than v1 while still retaining the prompting magic of flux.<h2 id="version-2.0-4qa17obs7">Version 2.0</h2>This improved version was trained on 27 manually captioned images of Alex Ross artwork at 2,000 steps.Trigger word <code>alex ross style painting</code><h2 id="version-1.0-3h6od6fwo">Version 1.0</h2>Trained on 116 automatically captioned images of Alex Ross artwork @ 2,600 steps.Trigger word: <code>alex ross style</code>

Files with SHA256: 33d65be24f6d1e8f9705ca28931c79689c9683792845056c14da1ec5d89fb462