dreambooth-docker
stable-diffusion-loopback-color-correction-script
dreambooth-docker | stable-diffusion-loopback-color-correction-script | |
---|---|---|
4 | 6 | |
132 | 28 | |
- | - | |
0.0 | 10.0 | |
about 1 year ago | over 1 year ago | |
Dockerfile | Python | |
- | MIT License |
Stars - the number of stars that a project has on GitHub. Growth - month over month growth in stars.
Activity is a relative number indicating how actively a project is being developed. Recent commits have higher weight than older ones.
For example, an activity of 9.0 indicates that a project is amongst the top 10% of the most actively developed projects that we are tracking.
dreambooth-docker
- Don't overpay for dreambooth training!
- Stable Diffusion links from around October 5, 2022 that I collected for further processing
-
Trained model produces exactly the same result as the original
For reference, I'm using this docker container, not sure if that is related: https://github.com/smy20011/dreambooth-docker
-
Want to use your own face ? Uploading a Tutorial today !
I'm using this Repo: https://github.com/smy20011/dreambooth-docker
stable-diffusion-loopback-color-correction-script
-
Loopback makes pictures lose contrast a lot
Hi, I have a problem with loopback that I struggle with for few days and I fail to fix it. When I use loopback, even just 5 steps, the pictures have progressively lower contrast until they are just pure gray. I have tried turning on/off the color correction in settings, I have tried this https://github.com/rewbs/stable-diffusion-loopback-color-correction-script script for color correction, I have tried using inpainting instead but nothing seems to help. Do you have any idea what may cause this, how to fix this? The weird thing is that when it generates, when I see the how it generated it looks fine but at the moment it saves it just loses contrast and changes. I am using the v4.5 model and DPM++ SDE sampling.
-
Black Muddy River by the Grateful Dead (a stable diffusion music video)
This video is the first thing I made that I wanted to share with more than just my family. I made most of it from the lyrics at 50 frames per verse. I interpolated from one verse to the next: “current_verse:1.0 AND next_verse:0.0” in 50 steps to “current_verse:0.0 AND next_verse:1.0” using a custom script I wrote for the automatic1111 repo. I started from the loopback color correction script (no longer needed! hurray for VAE!) and added code to allow an eight-parameter “move” that repositions each corner of the next image. It also interpolates between an initial move and an end move, so that the transitions are smooth.
- Stable Diffusion links from around October 5, 2022 that I collected for further processing
-
Good news: VAE prevents the loopback magenta skew!
Historically, loopbacks have suffered from a major colour skew problem, resulting in most UIs providing some kind of colour correction post processing step as an option. You can see some examples of the problem and workaround here. This didn't go away with the 1.5 model.
-
Experimenting with video, feedback would be great
As an aside, looks like this video is hitting the magenta/cyan skew issue that emerges when you do loopbacks with certain params. You could look into colour correction techniques to avoid this. See https://github.com/rewbs/stable-diffusion-loopback-color-correction-script for more info.
- User script to provide advanced colour correction options for img2img loopback for AUTOMATIC1111/stable-diffusion-webui
What are some alternatives?
efficient-dreambooth - [Moved to: https://github.com/smy20011/dreambooth-docker]
sd-parseq - Parameter sequencer for Stable Diffusion
AI-Horde - A crowdsourced distributed cluster for AI art and text generation
dreambooth-training-guide
stable-diffusion-webui - Stable Diffusion web UI
Stable-diffusion-webui-video
dreambooth-gui
Dreambooth-Stable-Diffusion - Implementation of Dreambooth (https://arxiv.org/abs/2208.12242) by way of Textual Inversion (https://arxiv.org/abs/2208.01618) for Stable Diffusion (https://arxiv.org/abs/2112.10752). Tweaks focused on training faces, objects, and styles.
fast-stable-diffusion - fast-stable-diffusion + DreamBooth
fast_Dreambooth_4_kaggle - a version of fast_Dreambooth by TheLastBen for kaggle notebook