Add AI prompt-to-image generation and image-to-prompt to our site. Stable Diffusion. -- 2
Budget: $250 – $750 USD
Would like to add two features to our site: Prompt-to-image generation and Image-to-prompt, powered by cloud GPU and opensource ai models such as Stable diffusion.
We will be using TensorDock (https://www.tensordock.com) for cloud GPU. They will have an API for us next week. Host nodes API: https://marketplace.tensordock.com/api/v0/client/deploy/hostnodes
Here is a tutorial article with steps similar to what we will do. https://towardsdatascience.com/launch-a-web-api-for-stable-diffusion-under-45-seconds-bbd88cfe41d8
Right now, we have a preliminary website https://ainsfw.com written in PHP with Laravel, that allows users to upload ai Generated images that are NSFW.
The two features we are interested in adding to the site are:
1: Feature that allows users to create AI images using models like Stable Diffusion, and have features like up-scaling with GFP-GAN model (https://github.com/TencentARC/GFPGAN) or similar using cloud GPU from runpod.io (or better option). User would input a prompt on our homepage @ aiNSFW.com and select a few key parameters (more advanced parameters in the future). Request would go to cloud GPU (tensorDock) running ai image model, and image would be returned to our site and displayed for user to save to profile or try again. This would be the main image generation (allow certain # of free generations/day/user) and have a paid tier for more use. We could include more models in future.
Would like users to have the standard options for image generation such as seed #, steps #, iterations, width, height, cfg.
AUTOMATIC1111 has developed a great WEBUI for stable diffusion with lots of features for reference. https://github.com/AUTOMATIC1111/stable-diffusion-webui/
2: Ability to use Image-to-Prompt model such as CLIP Interrogator by @pharmapsychotic
https://github.com/pharmapsychotic/clip-interrogator
This would be used to input user’s uploaded image and get the most likely prompt keywords that were used to make that image. These keywords could be used to automatically recommend #categories for that image as well as screen the image for banned terms (child, young, ect) and flag such image to admin for review.
Please let me know if you feel you can make this happen and give a price quote for the above. Let me know if you need any further details. Thank you for your time, and look forward to creating a unique place on the web.
We will be using TensorDock (https://www.tensordock.com) for cloud GPU. They will have an API for us next week. Host nodes API: https://marketplace.tensordock.com/api/v0/client/deploy/hostnodes
Here is a tutorial article with steps similar to what we will do. https://towardsdatascience.com/launch-a-web-api-for-stable-diffusion-under-45-seconds-bbd88cfe41d8
Right now, we have a preliminary website https://ainsfw.com written in PHP with Laravel, that allows users to upload ai Generated images that are NSFW.
The two features we are interested in adding to the site are:
1: Feature that allows users to create AI images using models like Stable Diffusion, and have features like up-scaling with GFP-GAN model (https://github.com/TencentARC/GFPGAN) or similar using cloud GPU from runpod.io (or better option). User would input a prompt on our homepage @ aiNSFW.com and select a few key parameters (more advanced parameters in the future). Request would go to cloud GPU (tensorDock) running ai image model, and image would be returned to our site and displayed for user to save to profile or try again. This would be the main image generation (allow certain # of free generations/day/user) and have a paid tier for more use. We could include more models in future.
Would like users to have the standard options for image generation such as seed #, steps #, iterations, width, height, cfg.
AUTOMATIC1111 has developed a great WEBUI for stable diffusion with lots of features for reference. https://github.com/AUTOMATIC1111/stable-diffusion-webui/
2: Ability to use Image-to-Prompt model such as CLIP Interrogator by @pharmapsychotic
https://github.com/pharmapsychotic/clip-interrogator
This would be used to input user’s uploaded image and get the most likely prompt keywords that were used to make that image. These keywords could be used to automatically recommend #categories for that image as well as screen the image for banned terms (child, young, ect) and flag such image to admin for review.
Please let me know if you feel you can make this happen and give a price quote for the above. Let me know if you need any further details. Thank you for your time, and look forward to creating a unique place on the web.