I'm trying out some new prompts and stuff, and finally managed to get a bit of nudity on my ai. I must admit they're not fantastic, but I'm getting there slowly.
kemistry said: Clothed and in the office, tried to get a bit more life in the facial expressions
WetGownsLover said:
kemistry said: wet and messy
Where you making these images what are you using?
Its an SDXL model (Stable Diffusion).
Will you ever release some of your models? It'd be great to try out the chega mais pie style one you have on your 2023 album someday...
Im not sure how tbh, I dont want to put them on Civitai which is the main place and the models are 7GB so not something easy to share. The LoRA I made for the last post is 780MB which again is still big for sharing.
I'm trying out some new prompts and stuff, and finally managed to get a bit of nudity on my ai. I must admit they're not fantastic, but I'm getting there slowly.
It's been a while, so figured I'd upload a new batch of creations. I went through a stretch where I wasn't totally thrilled with a lot of my creations, so may be a couple here that have some imperfections, but I really liked the outfit/setting/expression, so I overlooked some issues that I normally wouldn't accept. I've been really happy with some recent images, so I've included some that I'm quite proud of. As usual, my personal preference leans towards formal wear, so there's lots of that, but there is some other clothing types mixed in. First, some wetlook images:
Got back into a Bing creation binge recently and decided to add a futuristic element to the game-show formula. I was getting some good results when I noticed a new option in the image details - resize.
This function will take the generated image and re-generate a higher definition version in 4:3 aspect ratio - 1792 X 1024 resolution.
It doesn't just change the aspect though, it incorporates all of the original image properties and creates an entirely new image with discernible similarities. Consequently entire assets or subjects will be altered or re-positioned.
Selecting this option will open the image in a new tab and commence the re-generation process, which typically takes 15-30 seconds to complete. You'll see a pale white preview window with subtle color changes.
It's not always successful however, so if it takes any more than a minute it's unlikely to produce a result. It doesn't consume boost credits, but will fail if you're at the prompt limit.
Perfect for desktop backgrounds or as screensaver images for sweet dreams.
I'm playing around with different AI generators, but as I'm looking for the most possible amount of realism I currently use Midjourney only because it's superior to the others.