Realism 1.5 model by training and Dare merge.
Recommend:
Euler a
CFG:7
Negative prompt
(nsfw:0.7),(worst quality:1.5),(low quality:1.5),(normal quality:1.5),lowres,watermark, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, jpeg artifacts, signature, username, blurry, artist name, colorless,(deformed:1.3)
Description
FAQ
Comments (6)
Amazing how already bad SD 1.5 models are comparing to FLUX, SD3 or even old SDXL quality.
Realism? where?
I'm not quite sure what the last dude was trying to say but this model does what I want it to and does it quite wonderfully. Thank you
A wonderful combination of good photorealism and powerful versatility. This model is really good, very flexible and responsive, the pictures are expressive, reflects artistic styles well.
My advice to the author is to place more sample images on the model's page so that people better understand its capabilities.
Great cool job! Thanks!
Thanks, i add some showcase from your config
@GuardSkill It's an honor for me! :)
It feels like using slightly better version of vanilla SD1.5. Almost every image with messed hands and feet, ignoring backgrounds, 50% of cases will be Asian face.
Actually even demo image has mutated hands, needless to say more.