better results with textural inversions
seems like textural inversion is better than dreambooth for general style models like this.
I haven't done any extensive testing but in general I found dreambooth has much more coherent generations than plain textural inversion. Although with careful tuning textual inversion could give you more variety of results.
Some sample for this style here : https://twitter.com/originalmaderix/status/1575484821874167809
I haven't done any extensive testing but in general I found dreambooth has much more coherent generations than plain textural inversion. Although with careful tuning textual inversion could give you more variety of results.
Some sample for this style here : https://twitter.com/originalmaderix/status/1575484821874167809
ahh i see thanks for explaining, there's a lot more possibility with dreambooth while TI is stubborn