My original intention was to post all of the images I've generated that I think are worth sharing in a single thread, so that anyone who's interested can easily see them all together in one place.

Unfortunately, however, I've found that adding posts to a thread doesn't necessarily mean that the "Last Post" timestamp is updated. Instead, a note is sometimes added to the bottom of the message which says "this post won't bump the thread to the top". I don't know what the criteria for this to happen are (please tell me if you know!) but when it happens, of course, it means that people won't be aware that I've added more images to the thread unless they go into it and scroll all the way down to check. Obviously most people won't do that because the "Last Post" timestamp hasn't changed and it's tedious to keep checking just in case a new post has been added.

Therefore, unless there's some way of getting around what I've outlined above, I'll have to start a new thread whenever I'm posting more images. If you haven't seen any of my previous images, the first thread I posted (in which I explain the rationale behind what I'm doing) can be found here: https://umd.net/groups/glamour-ai-wetlook-featuring-skirts-dres

Anyway, I've been producing a lot of new images since my last post, but have been holding fire on sharing them pending finding out if I can do anything about the "Last Post" timestamp issue. I haven't been able to get any info on that, so I'm going to crack on.

As stated in the subject, these images were produced using the model called "Dream Photo AD 1.5". There are a lot of images in this update and a fair bit of similarity (even I am becoming tired of the AI engine's fondness for pleated skirts!). Some of you may notice a bit more experimentation with the backgrounds in this update, where I'm exploring a few themes. I haven't gone overboard with that because the backgrounds are blurry anyway, but I'm trying to do a little bit to improve the variety and/or atmosphere of the images.

I hope you find at least some of the images enjoyable

Absolutely stunning artwork. Thank you for posting.

The changes you made to the backgrounds are incredible. I like the fantasy look. It is quite nice to see variety and the randomness. My favorite aspect of generating images is seeing what randomly comes out. It seems like you need randomness to have creativity, and it is very difficult for humans to be random.

Thank you. My images are pretty much the result of trial and error, with a massive number of images having to be rejected because they're not right in one way or another. The AI engine is both remarkable and frustrating in equal measure!

I suppose I feel a little bit of a fraud in some respects, because the real effort is being put in by mFeelzGood and others who are using real images to train their own models, rather than just utilising existing models. I'm not sure I'll ever have the time and/or capability to delve into AI image generation to that extent.

I'll just keep doing what I'm doing and try to vary things up enough to hopefully keep it interesting, both for myself and for others.

I wish the backgrounds were a bit more realistic and that they weren't so blurred, but it looks like I'm powerless to do anything about the blurring as it's the nature of the beast for the kind of images I'm producing.

Some of the things the AI does are absolutely remarkable, whilst others are so ridiculous and feel like they should be simple to avoid, such as putting hands where there should be feet and giving the people in the images extra fingers/arms/legs and even extra heads on occasions! I gather that the AI engines aren't very good at counting and that this is apparently understandable, but I haven't looked into it too much.

There is definitely a tinge of excitement on every generation run, but most end with disappointment as you've no doubt frequently found yourself! Patience is an absolute given for doing this!

These are incredible. Among the best I've seen!