Generating these images is very expensive in terms of machine power, storage, and especially GPU, which are very expensive and scarce.
Our service is cheaper and faster because we use our own API and we don't have to pay to use theirs. Also, we don't put requests in a queue - we use an entire GPU instance to create your pictures. Finally, because it's cheaper and faster for us to generate pictures, we have tested our models with thousands of pictures and gotten the best results.
Because this is a complex process involving many machine resources, I hope the technology will continue improving. Each time, the generation of the images will be faster and cheaper.
Stable diffusion v1.5 with Dreambooth. We use Python to work with pictures cropping and transformation.
Getting the images for free is a complicated task but doable. SelfieAI simplifies this process, hiding to the user the complexity of dealing with models, prompts, steps, seeds, guidance scale, samplers, etc. This is a good tutorial to get the pictures for free here
Check your spam folder or contact us if the problem persists <a id={styles.link} href="mailto:photos@selfieai.net"> mailto:photos@selfieai.net </a> mailto:photos@selfieai.net