Stable Diffusion AI Image Generator

1681718658232.jpeg

Was trying for a green eyed Hercule Poirot. I guess the image database only has brown eyed ones. What do I type to get what I want. I tried 'Hercule Poirot with green eyes' to generate above.

He looks like Sean Connery although the mustache not bad.
 
LetsDance said:
View attachment 1509677

Was trying for a green eyed Hercule Poirot. I guess the image database only has brown eyed ones. What do I type to get what I want. I tried 'Hercule Poirot with green eyes' to generate above.

He looks like Sean Connery although the mustache not bad.
Click to expand...
Prompting alone often doesn't give the results one wants, more often than not it requires inpainting, or some other kind of fix or manipulation (Photoshop \ Affinity Photo \ Gimp \ Photopea, etc) to get the desired output. If you're dead set on achieving it with just prompting, use weights to give importance to certain attributes.

If you have Automatic1111 you can try to change the eye colour using IMG2IMG to send it to "inpaint sketch" and paint over the eyes with a green brush.

But yeah, personally, prompts alone are definitely not enough for me to do the things I want to do, which is why there are wonderful toys like ControlNet, IMG2IMG, Upscaling, Inpainting, Outpainting, and so on...
 
I tried 'Beautiful blonde elf woman topless' -one result, but the image too beautiful for words. I would show it here, but it features some nudity. I will try those weights, thanks for the tip.

www.howtogeek.com

How to Write an Awesome Stable Diffusion Prompt

Stable Diffusion, a popular AI art generator, requires text prompts to make an image. Sometimes it does an amazing job and generates exactly what you want with a vague prompt. Other times, you get suboptimal outputs. Here are some tips and tricks to get ideal results.
www.howtogeek.com www.howtogeek.com
 
So apparently using PyTorch2 alone doesn't increase speed all that much, but using the optimisation "--opt-sdp-attention" instead of xformers with Torch2 can double speeds from what I've seen around online. It also lowers VRAM usage?

github.com

Optimizations

Stable Diffusion web UI. Contribute to AUTOMATIC1111/stable-diffusion-webui development by creating an account on GitHub.
github.com github.com
 
Photobashed some of my landscape (seascape) photos with some stock images, ran it through SD with ControlNet and got quite a nice digital painting from it, tweaked it a bit with IMG2IMG, inpainting and upscaling.

On The Water.jpg
 
For those who want to read an in-depth article regarding ML and CUDA supremacy.

Has a section on PyTorch2.

www.semianalysis.com

How Nvidia’s CUDA Monopoly In Machine Learning Is Breaking - OpenAI Triton And PyTorch 2.0

Over the last decade, the landscape of machine learning software development has undergone significant changes. Many frameworks have come and gone, but most have relied heavily on leveraging Nvidia's CUDA and performed best on Nvidia GPUs. However, with the arrival of PyTorch 2.0 and OpenAI's...
www.semianalysis.com www.semianalysis.com
 
You must log in or register to reply here.
Back
Top
Sign up to the MyBroadband newsletter
X