Flat Color - Style
Trained on images without visible lineart, flat colors, and little to no indication of depth.
ℹ️ LoRA work best when applied to the base models on which they are trained. Please read the About This Version on the appropriate base models and workflow/training information.
This is a small style LoRA I thought would be interesting to try with a v-pred model (noobai v-pred), for the reduced color bleeding and strong blacks in particular.
The effect is quite nice and easy to evaluate in training, so I've extended the dataset with videos in following versions for text-to-video models like Wan and Hunyuan, and it is what I am generally using to test LoRA training on new models now.
Recommended tags:
flat color, no lineart, blending, negative space, {{color}} backgroundDescription
Trained as a small experiment with https://github.com/tdrussell/diffusion-pipe
Gives a charming flat color style to Hunyuan t2v generations
FAQ
Comments (4)
How did you prepare the dataset for hunyuan? Just images, what kind of tagging did you use?
I beleave that he choose flat image and the captions describing the style of image, not its content.
@midiaplaay looks like images not videos, that's what I think too
Hello, yes as a small experiment the v1.0 [hunyuan] version is trained on just 20 images with booru style tagging from the dataset used to train the other versions on this model card.
In v2.0 [hunyuan] the training data is a relatively larger combination of:
- Images used from other versions this model card
- Images extracted as keyframes from several videos
- Short video clips ~40 frames each