This time it's Rebecca as mentioned, still choosing Schnell, but this time I only trained the LoRA using natural language captions from CogVLM and also reduced the batch size.NOTE: The hires fix models don’t seem to work well with Flux Schnell. After extensive testing, I found that 4xUltrasharpV1.0 seems to have the least issues, but it still feels quite poor compared to using it on SD1.5 and SDXL. Hopefully, there will be a dedicated hires fix model for Flux Dev and Schnell in the future. you like my content, support me at:<a target="_blank" rel="ugc" href="https://ko-fi.com/duchaiten">https://ko-fi.com/duchaiten</a>Patreon in here: <a target="_blank" rel="ugc" href="https://linktr.ee/Duc_Haiten">https://linktr.ee/Duc_Haiten</a>If you want to use my models to create unlimited images at a small cost without having to install anything, <a target="_blank" rel="ugc" href="http://mage.space">mage.space</a> will be a good choice, i have some models that are exclusively available on there.

Files with SHA256: 5dfc6470c6fb44b900826b49053e24b22a4f9c87ac2fa143cf39a96af3fd4506