I'm a artist, designer & engineer.
I'm currently focusing on Stable Diffusion text-to-image and would like to build a career on my Stable Diffusion skills, either as an artist, as an engineer or both.
At the moment this "LORA" was created, I used the following platforms to run SDXL :
https://www.mage.space/u/johnslegers - I have almost 10000 images here, all publicly available, with the prompts I used to create them
https://creator.nightcafe.studio/u/johnslegers - I created almost 1000 images here, all publicly available, with the prompts I used to create them
https://musavir.ai/ - This platform doesn't allow me to share my prompts & output
I use the following social media to share selections of the work I've created on these platforms :
https://www.facebook.com/johnslegersart - This is my main platform, where I post most of my content
https://www.linkedin.com/in/johnslegers/ - I use this platform mostly for business related networking
The "LORA" is trained on 500+ hand-picked works I made in Nightcafé, Mage or Musavir, each manually captioned.
The "LORA" allows you to more easily produce AI art in the style of the artworks I've created with SDXL on the three platforms mentioned.
The "LORA" is officially available at three places :
Description
FAQ
Comments (8)
I want to thank you for making this available on tensor dot art/models/653647357902198498 🙏
Thanks. If you like this LORA, you may want to check out my 2 other LORAs as well.
All 3 LORAs are available on Tensor, as well as Epic Diffusion (my 1.x checkpoint).
@johnslegers I will. Thank you 🙏
The example images are quite varied in style so it's really hard to figure out what this Lora is trained for exactly. Also the prompts in the example images are like the whole kitchen sink of art styles so it's hard to know what this lora is doing if anything. Maybe some explanation?
I trained this LORA on 500+ of my works, each of which was originally produced with Mage, Nightcafé or Musavir, using one of their base SDXL models. The training images had been manually captioned and consist mostly of cyberpunk, steampunk, dieselpunk & goth themed images. Styles include black & white lithography, black & white charcoal, digital painting, oil painting & photorealism.
I have tested & compared the output of various prompts for the same seed, both with an without this LORA, and my experience is that it tends to alter the output in various ways, from making an image more grungy or more creepy to making it more sexy. Overall, though, I'd argue it does exactly what it is supposed to do : make it more easier to output content that looks like the overall style of my content as posted @ https://www.facebook.com/johnslegersart/.
Care to share generation data?
https://www.facebook.com/photo/?fbid=122126030912075369&set=a.122095126250075369
https://www.facebook.com/photo?fbid=122124589550075369&set=pcb.122124589856075369
Not yet. I'm planning to create a custom private model based on those, and use her as an AI avatar / character.
Maybe I'll share details at a later time, but for the time being those are among the few pics I'd like to keep the prompts private for.
I hope you understand...
Beautiful work This new world of AI art generation is so thrilling and amazing. My mind is continuously spinning with all the new things I am able to learn. Thank you for being a part of that by sharing your insight, lora and prompts!