Dreambooth-Stable-Diffusion

To put things in perspective, the dataset it's trained on is ~240TB and Stability has over ~4000 Nvidia A100 (which is much faster than a 1080ti). Without those ingredients, you're highly unlikely to get a model that's worth using (it'll produce mostly useless outputs). That argument also makes little sense when you consider that the model is a couple gigabytes itself, it can't memorize 240TB of data, so it "learned". But if you want to create custom versions of SD, you can always try out dreambooth: https://github.com/XavierXiao/Dreambooth-Stable-Diffusion, that one is actually feasible without spending millions of dollars on GPUs.

trained on is ~240TB and Stability has over ~4000 Nvidia A100 (which is much faster than a 1080ti). Without those ingredients, you're highly unlikely to get a model that's worth using (it'll produce mostly useless outputs). That argument also makes little sense when you consider that the model is a couple gigabytes itself, it can't memorize 240TB of data, so it "learned". But if you want to create custom versions of SD, you can always try out dreambooth: https://github.com/XavierXiao/Dreambooth-Stable-Diffusion, that one is actually feasible without spending millions of dollars on GPUs.

To put things in perspective, the dataset it's trained on is ~240TB and Stability has over ~4000 Nvidia A100 (which is much faster than a 1080ti). Without those ingredients, you're highly unlikely to get a model that's worth using (it'll produce mostly useless outputs). That argument also makes little sense when you consider that the model is a couple gigabytes itself, it can't memorize 240TB of data, so it "learned". But if you want to create custom versions of SD, you can always try out dreambooth: https://github.com/XavierXiao/Dreambooth-Stable-Diffusion, that one is actually feasible without spending millions of dollars on GPUs.

To put things in perspective, the dataset it's trained on is ~240TB and Stability has over ~4000 Nvidia A100 (which is much faster than a 1080ti). Without those ingredients, you're highly unlikely to get a model that's worth using (it'll produce mostly useless outputs). That argument also makes little sense when you consider that the model is a couple gigabytes itself, it can't memorize 240TB of data, so it "learned". But if you want to create custom versions of SD, you can always try out dreambooth: https://github.com/XavierXiao/Dreambooth-Stable-Diffusion, that one is actually feasible without spending millions of dollars on GPUs.

Wow, and Ruiz et al published the work only around 4 months ago. Who did the first open source, was it https://github.com/XavierXiao/Dreambooth-Stable-Diffusion?

If you want to train the model, you can try Dreambooth-Stable-Diffusion. https://github.com/XavierXiao/Dreambooth-Stable-Diffusion

Dreambooth and Texual inversion is already here, and it's been just over a month since Stable Diffusion was released, so I'd bet on sooner rather than later. https://github.com/XavierXiao/Dreambooth-Stable-Diffusion https://textual-inversion.github.io/