A Photorealistic Image Generator, you just think about the prompt and GO, the rest is optional.

Usage Tip For Best Result :

Use natural language prompt.
Addition with Prompt Helper
(You can) but need no modifiers such as masterpiece, best quality, etc. jus refer to the first tip.
If you add Peoples LoRA (such as yourself or celebrities) make sure it has good trained quality to avoid bad images.
I have a few LoRA collections for you to test out (if you interested) that have been tested out good at Weight 1 at best accuracy so you dont have to raised or lowered its weight.

Notes :

Sometimes the result get a little bit noisy because of impact of SD 1.5 but give you nicer texture and tones on Human skin compare to more plasticky Flux output.
Please be aware that if you use style lora will have little effect to the generated image due to the initial images are based on Flux.
Retain composition when adding text to the image,
Compatible with most SD 1.5 LoRA, so your trained or favourite LoRA not going to waste.
It's not perfect, even Flux still give you weird anatomy on human but most of the time is very good.

Version Log :

201024 | Updated with Reworking Worklow with more detailed results.
081024 | Updated with adding Bokeh Lora created by @Hina Chen for more unique look
170924 | Updated with adding additional LoRA
140924 | Updated to less noise and more prominent SD 1.5 goodness is live

Related Posts