Learn to create better images with AI!
Check out our crash course in prompt engineering & AI art generation!

shehabelkadi posted about 1 month ago
28 views 0 comments

Image was created using proprietary SD 1.5 model with a proprietary Text Encoder using a proprietary adapter for UNET and CLIP attention. The prompt is taken from the FLUX section here at PromptHero.

FYI: this is not image to image or IPAdapter use, this is straight-up use of SD 1.5 text to image with a proprietary adapter developed by me that uses a LORA extracted from SDXL and SD3.5 with CLIP using GEMMA and attention alignment technique inspired by ELLA+EMMA paper by TencentQQGYLab... It has been 9 months of hard work and a lot of time on a GTX 1650 4gb vRAM.

Read more...