Krea 2 Enhancer LoRA
Enhancer LoRA for the new Krea 2 text-to-image model.
Designed to improve image quality across any style, including realistic, cinematic, anime, illustration, fantasy, product, portrait, and stylized images. It helps add cleaner detail, sharper textures, better lighting, and a more polished final look while keeping your original prompt style intact.
A workflow is included to help you get started quickly.
All example images were created using that workflow.
Description
FAQ
Comments (20)
💕👍🏻
Thanks a lot to the author(s) of this lora which looks amazing ! It would be cool to have some sort of with/without lora comparison in the showcase images :)
it would also be cool if the people receiving high-quality, specialized assets entirely for free took the five minutes to test it themselves, rather than treating open-source creators like a retail customer service department. If you want a comparison, make one and post it in the review section.
@beydi1 Weirdge reponse, lol. They were just making a suggestion, they didn't sound entitled or anything like that. They didn't demand anything.
You are aggro for no reason.
@kossan To be fair, the average AI bro is pretty obnoxiously entitled.
@fox23vang226 Ye, that is fair. And I agree.
But that doesn't really justify flaming someone who isn't, does it? <.<
After work today I'll do that
I tried to post a bunch of comparisons but its only showing 2 of them.. no idea why. CivitAI is so darn buggy!
@vrgamedevgirl Thanks a lot, you are very kind ! And np for the bug, I will test your lora either way ;)
Thanks for the lora and getting me inspired by this workflow too. Using a few chained samplers myself but that makes more sense like this
This actually works, thanks. So many other LoRA do absolutely nothing or just trash the image.
Great work. Thank you. Amazing lora.
Doesn't work for me, makes images softer
Did u use my workflow?
@vrgamedevgirl ill try it later
@fox23vang226 Ok. I'm thinking it might be the workflow, otherwise if you could send me examples of with and without the lora on discord we can troubleshoot.
my discord name is vrgamedevgirl
Workflow and LORA are on point. Great job! ^^
Something is wrong with WF.
At least not when creating portraits, but more general plan, cinematic images.
For example, as many times as I tried, there were errors. "In the Middle Ages, a Lithuanian pagan priest, dressed in white, with a white hood on his head, with his hand outstretched, points to something far away in the fog. The priest stands in a dense oak forest. cinematic image."
Constantly either problems with fingers, or with clothes (there are some pieces of sleeves hanging in the air above the sleeves).
Or both of those errors.
When I tried the same lore with a regular 8-step WF, Er_Sde sampler, everything was fine.
So I guess something is wrong with the WF settings, because the lore works fine.
you can change the sampler settings in the subgraph. Krea 2 is still pretty new and very new when i made the workflow so settings are all still being tested to find the best ones.
Your welcome to change the settings to fit your needs inside the subgraph.
@vrgamedevgirl Samplers i change. But with "FlowMatch ..." node i dont know what to do... I dont know any parameters...