
This is a LoRA for generating fantasy artwork featuring dragons and enhances the default mash-up between rendered/drawn/statue/western/eastern dragons of Flux (which is remarkably similar to that of the good old SD 1.x days) to a more professional and consistent high quality digital artwork style. It was trained on a selection of high-quality, high-resolution non-anthropomorphic (or what some folks call "feral") western dragons from various artists.
The dataset is mostly the same as that used for PonyDragon with some additions and a more poetic and detailed captioning that Flux seems to like so much.
Against my own intuition, I've followed the advice to use a fairly low LoRA rank and trained it at just 512 pixels and it seems to work pretty well, actually. Still, it is a quite experimental LoRA and there's definitely room for improvements, especially when it comes to the details.
NOTE: If you have downloaded the model before 2024/09/09, you should re-download it, since the old version had incorrect hashes in the metadata that collided with those of several other LoRAs created with ai-toolkit due to a bug in the trainer.
描述:
Initial experimental release.
训练词语:
名称: fluxdragon-v1.safetensors
大小 (KB): 42081
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success