hi ! Thanks again for this video. the best way to get variations of an image is to copy paste the custom prompt that dall-e generates by himself, because this is the actual prompt used by dall-e for this image, not the one we provided. It's like if openai thought that we needed an assistant to write more detailed prompts. Or the point is to generate different prompts for more variety of styles, more creativity, as a starting point. But I don't understand what's this seed thing is. What it is from openai perspective? how it works?
Hey there! yea that totally make sense. I will try that out for variations. As for seed, it is a unique id that you can get chatgpt to assign to every image generated. So if you want to keep some similarities of your next generation, using the id can help keep it consistent. But I also realise that when chatgpt generates that 4 image grid, all 4 images are tagged to the same seed id. So the trick is to get chatgpt to generate 1 image at a time then use the seed id for your next prompt for more consistency. I might just do another video on that. Anyways thanks for your comment!
