could you provide a demo for good text to image? #7

by - opened

That's my prompt:
"anime, hand-drawn and cel animation techniques, (fishing boat sailing on waves during storm), natural design, beautifully rendered and expressive rich colors, vibrant pastel colors, imaginative and fantastical landscapes, sharp attention to detail, realism and a strong sense of nostalgia and warmth, sharp attention to small details and textures, fantastical creatures, settings, depth and emotions emphasized and accentuated by lighting and shading, extremely high quality, incredibly high finite definition, high resolution, hand-drawn and cel animation techniques, anime --ar 3:2 --stylize 1000 --q 2"
And my result seems so bad

屏幕截图 2023-02-17 165838.png

So whta are the proper parameters?thx

changed discussion status to closed
changed discussion status to open

I think I trained this model on a standard 512 X 512 but you can try the demo on here when you click on the model card you should be able to work your prompts there

This is what I get with your prompt

Thx for your reply. The image generated by your inference API is good. But I wanna know what are the right parameters? I want to reproduce it in my colab notebook.

Ahh ok well I trained the model using TheLastBen's fast-DreamBooth notebook and I don’t remember the exact specs I ran it

I ran this pipeline on my own GPU, but I can't get the good result same as your demo. Why was that?What's wrong with my parameters?(I'm a new beginner)
屏幕截图 2023-02-20 160420.png

Im also curious which implementation works the best for this model!

Sign up or log in to comment