This is the Double Exposure Embedding for SD 2.x.
It was trained with the v2.1 768 ema pruned model on 768px images of people, layered with a variety of surroundings. It's been shown to handle objects and animals good as well.
Place the file in the embeddings folder of your Automatic1111 installation.
Trigger the effect in the prompt by writing dblx.
Read more and get more prompts for sample images on the Hugging Face page.
Description
FAQ
Comments (17)
This is amazing!
Any plans to make the embedding compatible with sd 1.5?
Hi! Embeddings doesn't work as well with 1.5 as with 2.x. Any reason why you can't run 2.1? Anyway, I have also a model on hugging face called Double Exposure Diffusion. That's based on 1.5, but this embedding is much better resolution etc.
What are the prompts for the 2nd and 6th example images on the model page?
Oh! I have to check out why they didn’t come along.
great work，but 2.1 is too difficult to use, I haven't tried it successfully
solved……T_T
@ffffff Happy you got It working!
@joachim_s does it work with Huggingface current model 2.1 pruned ema ? I don't have the same hash as the one in the examples. Not sure if this is caused by the hash or by something I do wrong ... (but I use automatic111 and the inversion is correctly loaded, so weird...)
@jppaolim There are issues with replicating images at the same seed etc since the use of xformers. Not sure you know what this is but it helps with faster generation of images. So all I can really do here is to guide with styling through the settings and seeds to get something similar. And yes, you should mainly go by 2.1 ema pruned, but it works with any 2.1 based model :)
I have placed the file in the embeddings folder of Automatic1111 installation.
But cmd show: Textual inversion embeddings skipped: dblx.
How to fix that?
What pictures do you use as the data set? What are the training parameters？thank u so much if u tell me that
用什么模型组合呀？
What do you mean? It’s a finetuned embedding. Not a merge.
These are the best double-exposure, The vibrancy of the colours and detail is spot on.
Maybe the "Male" one could be a little fuller composition in the frame. Otherwise. Like I said.
Fun that you like it! I made it over a year ago now so it's already old 😀