Is it possible to generate similar captions using EveryDream tools for the images for which we have written prompts manually instead of using any tool??
Hey, I haven’t had a chance to do an extensive and conclusive deep dive on that yet. As mentioned in the write up, so far we know for sure that different captions have significant impact on the results. To figure out that relationship steps one is to gain control over the captioning process and that’s what this post is about
Is it possible to generate similar captions using EveryDream tools for the images for which we have written prompts manually instead of using any tool??
What are the results of the testing? How did any of this impact the training?
Hey, I haven’t had a chance to do an extensive and conclusive deep dive on that yet. As mentioned in the write up, so far we know for sure that different captions have significant impact on the results. To figure out that relationship steps one is to gain control over the captioning process and that’s what this post is about