![]() Without watermarking, there is an ever-increasing risk that AI-generated images will end up in your datasets. With an universal watermarking system, AI-generated images can be excluded from all the pictures you scrapped, and your datasets will only consists of clean images. In the case of Stable Diffusion, the data consists of drawings made by human hands, or photos.Īll these pictures are usually not gathered manually, they are scraped through web-crawling sessions performed by computer programs that can't really make the difference between an "original" picture and an "AI-generated" one. To be trained, models have to be fed a large amount of data. Why? Your comparison doesn't not prove anything and is far fetched. I thought it might somehow add tracking info, but instead it just adds the word "StableDiffusionV1" to it, which will make AI generated images easier to filter out when training AI in the future. Safety_checker = StableDiffusionSafet圜om_pretrained(safety_model_id)Įdit: I removed the part about disabling the invisible watermark because it didn't do what I assumed. Safety_feature_extractor = om_pretrained(safety_model_id) Safety_model_id = "CompVis/stable-diffusion-safety-checker" In txt2img.py, find and delete the three separate lines (might be 26, 27, 28) that say:.Optional: Stopping the safety models from even loading to save Vram (thanks NotMyMain007) ![]() Replace it with this (make sure to keep the indenting the same as before):.X_checked_image, has_nsfw_concept = check_safety(x_samples_ddim) Open txt2img.py, and find the line (might be line 309) that says:.Open the "scripts" folder and make a backup copy of txt2img.py.They may also have been already removed in a fork.Īlso, where I say delete the lines, it might be better to just comment them out by putting a # in front of the line, to avoid changing the line numbers of the rest of the code (therefore making it easier to find the others mentioned). If using a fork (or a later version of it), the line number might be different so just search for the one I mention below. Note: This is assuming you are using the Stable-Diffusion repo from here.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |