deforum-for-automatic1111-webui VS sd_dreambooth_extension

Deforum https://github.com/deforum-art/deforum-for-automatic1111-webui

might be deforum (project) (webui extension) or img2img loopback wave (webui extension). it's definitely something with img2img though. that's clear from how it's warping.

Thanks for the workflow. To simplify the process I created a PR that hopefully will make it a bit faster, specially if you're working in rented cloud GPU's.

The sequence after the train with the flashing background is the only exception (1:03 to 1:24), that part was made using the Deforum notebook: https://deforum.github.io/

ive gotten better results applying some color correction to my loops using the same way as they do in deforum https://github.com/deforum-art/deforum-for-automatic1111-webui/blob/automatic1111-webui/scripts/deforum_helpers/colors.py

🔸 Deforum extension for Automatic1111 (Local Install): https://github.com/deforum-art/deforum-for-automatic1111-webui

Deforum plugin settings files: render, coloration. The black and white canny edges + scribble-mode mid-stage rendered video. You can check the prompts keyframing file to see if I got the characters names right. Despite my best efforts, some of the characters appearances were off, I think it was either the character names were incorrectly spelled (I was referring to the Touhou wiki) or they had been underrepresented in the dataset or there was overlap between their naming and some other franchises. There is also the issue of random unrelated characters appearing when the prompt contains no characters and instead is purely abstract, I guess it relates to this specific model retraining.

I am trying to use StableDiffusion via AUTOMATIC1111 with the Dreambooth extension

When I first attempted SD training, I was very frustrated. It wasn't until I found this obscure forum thread on Github that I actually started producing great results with Dreambooth. Because I have such satisfactory results, I'm very reluctant to beat my brains against LoRa and its related training techniques. I gave up trying to train TI embeddings a long time ago. And I never figured out how to train or how to use hypernetworks. I've only been able to get good results with Dreambooth directly because of that thread I linked above. I make LoRas by extracting them from Dreambooth-trained checkpoints. And I have no idea if I'm doing the extractions the right way or not.

Getting the same with automatic1111 and sd_dreambooth extension. Check out more here in the issues log: https://github.com/d8ahazard/sd_dreambooth_extension/issues/1266

It's several moths old and many things have changed. But the spreadsheet available through this thread on Github has been indispensable for me when I train Dreambooth models. I'm astounded no one talks about it. I bring it up all the time. The research presented there should be continued. I'd love to see similar research done for SD v2.1.

Training rate is paramount. Read this Github thread.

https://github.com/d8ahazard/sd_dreambooth_extension/discussions/547/ (in depth training principles understanding)

sd_dreambooth_extension https://github.com/d8ahazard/sd_dreambooth_extension.git main 926ae204 Fri Mar 31 15:12:45 2023 unknown

also the Dreambooth extension supports API (https://github.com/d8ahazard/sd_dreambooth_extension/blob/main/scripts/api.py) so i'm not sure where do you get those news :/