+1
I'm getting the results I was getting a year ago with similar prompts. It's not just the issue of photorealistic quality or getting the prompts to work right, I also get quality artefacts I have not been seeing for over 6 months. It seems the previous model was custom-trained, and that this change lost all the ability and quality that have been acquired. Personally, I have seen zero improvement in any area over the last one. And it's not a matter of prompt length.
+1 I'm getting the results I was getting a year ago with similar prompts. It's not just the issue of photorealistic quality or getting the prompts to work right, I also get quality artefacts I have not been seeing for over 6 months. It seems the previous model was custom-trained, and that this change lost all the ability and quality that have been acquired. Personally, I have seen zero improvement in any area over the last one. And it's not a matter of prompt length.