thrifty_thor said: I think I'm getting pretty good at this.
Listen to kortanlein. pick a painter for the style, use a photo descripter to add realism. say women, not girls. does it sound like it could be a porn tag? a thesaurus will help you. use emotional and mood setting descriptors: silly, laughter, fun, shocked, surprised. don't get too physical descriptive of a person--that's a dead giveaway. describe their vocation or attitude. if you don't like the person, just regen the same prompt again. start slow and don't get too aggressive throw in words like friendly, safe, wholesome, design, historic, modern to add mood and balance a prompt with positive connotations. use words like motion or kinetic. good luck.
these are really good which program do you use for these ones? im looking for a free one better than bing
I did use Bing image generator for these. The challenge is tricking the AI to go where you want without breaking its rules. A lot of trial and error sadly.
i was struggling with getting good food wam with bing but its like you say just got to mess around with prompt. im getting better at it lol
I've found the more photo realistic the images are, the harder it is to get facial and hair coverage. The more artistic works/surreal achieve much better messiness.
This is remarkable stuff, although it's such a shame that the AI engine can reproduce the messy clothing and effects with such aplomb and yet is often a complete shambles with regard to faces.
Few things I discovered: - using -in the style of (Japanese artist)- will make most people Asian - using -white and black shiny dress- will produce a French maid outfit - using -white and black suit- will produce a student outfit with or without tie
Might stop for a bit until I come up with new ideas. I'll be back for sure.
kortanklein said: Few things I discovered: - using -in the style of (Japanese artist)- will make most people Asian - using -white and black shiny dress- will produce a French maid outfit - using -white and black suit- will produce a student outfit with or without tie
Might stop for a bit until I come up with new ideas. I'll be back for sure.
These are terrific. Would love to know your prompts for the whipped-cream textured/splattered/thrown effect, if you have a mind to share? Also I find it hard to get face-coverage!
- creamy pudding (best splat effect but a bit yellow) - Black Forest cake (which appears as a standard cream cake) - birthday cake - custard cream pie
The most important part is the tense used in the prompt. If you want the impact you can use -the moment when-, if you want after the impact you can use -has just been hit- or -has been hit-.
You can also not say anything about being hit with pies and just describe a game where people throw pies to win points - the AI will then figure out what happens (most of the time).
But ultimately it's a matter of luck and you just have to keep trying until you get the exact effect you want.
kortanklein said: Censor seems mostly gone now, really having fun with this. Even if it gets neutered again I have thousands of good images.
Next step is video AI
So much good stuff in this thread. Any way to get these up somewhere where it would be easier to view/download? I feel like there should be an archive for all this content.
Mortar said: Any way to get these up somewhere where it would be easier to view/download?
Doesnt UMD offer a membership subscription plan that allows downloading multiple images quicker and easier? I've been too cheap to try it but these threads may change my mind.
I'm also slightly concerned the Image feed on the forum will get flooded with these images, thats why im being quite reserved. I feel like these images are great but im slightly concerned that im pushing..real content creators images out the way.
kortanklein said: For the cake types I've had success with:
- creamy pudding (best splat effect but a bit yellow) - Black Forest cake (which appears as a standard cream cake) - birthday cake - custard cream pie
The most important part is the tense used in the prompt. If you want the impact you can use -the moment when-, if you want after the impact you can use -has just been hit- or -has been hit-.
You can also not say anything about being hit with pies and just describe a game where people throw pies to win points - the AI will then figure out what happens (most of the time).
But ultimately it's a matter of luck and you just have to keep trying until you get the exact effect you want.
Thank you! Excellent tips. I do quite enjoy yhe process of trying to 'think like the AI' to produce the result I'm looking for, and being rewarded with this completely unique image. It's quite the art.
I wish another filter would be added: all, female, male .... AI. Then you could explicitly search for it and wouldn't be bothered if you only want to see real content creators.
I tried my luck with Dall-E and I have to say, respect what you get out of it. I get comparatively kindergarten pictures out, with my few sentences.
ttauern said: I wish another filter would be added: all, female, male .... AI. Then you could explicitly search for it and wouldn't be bothered if you only want to see real content creators.
Mortar said: Any way to get these up somewhere where it would be easier to view/download?
Doesnt UMD offer a membership subscription plan that allows downloading multiple images quicker and easier? I've been too cheap to try it but these threads may change my mind.
The vip archive download does work well for the posts and replies that used zips to upload. It does not work well for small number of images in replies, or downloading all of the photos for this entire forum post. It works best for downloading entire user photo albums that have hundreds of photos.
ttauern said: I wish another filter would be added: all, female, male .... AI. Then you could explicitly search for it and wouldn't be bothered if you only want to see real content creators.
Having read that thread, which I hadn't seen before, and as someone who has absolutely no intention of basing images on real people, my first reaction is that I should probably give up posting AI images as the opposition to them here is only likely to increase...
On the other hand, I can't immediately see that the people being negative about AI images are producers of real-life WAM and it's the producers whose thoughts by far carry the most weight with me.
I've finally managed to achieve the ""submerged"" effect, it was a quite a challenge and it seems it doesn't really work well with clothes other than swimsuits. Are creamy butts in swimsuits ok? You can't do anything related to crawling but you can manage using -walks with effort-.
That was the last scenario I wanted to do so I can safely say I've reached my goal.
kortanklein said: I've finally managed to achieve the ""submerged"" effect, it was a quite a challenge and it seems it doesn't really work well with clothes other than swimsuits. Are creamy butts in swimsuits ok? You can't do anything related to crawling but you can manage using -walks with effort-.
That was the last scenario I wanted to do so I can safely say I've reached my goal.
These are really astounding! I've been trying to get the all-over look but usually only get a profile shot to generate.
kortanklein said: I've finally managed to achieve the ""submerged"" effect, it was a quite a challenge and it seems it doesn't really work well with clothes other than swimsuits. Are creamy butts in swimsuits ok? You can't do anything related to crawling but you can manage using -walks with effort-.
That was the last scenario I wanted to do so I can safely say I've reached my goal.
These are fantastic. I wonder what the 4 removed images looked like. I have a feeling like I would have liked those the most
Random assortment. Some from earlier generations than others. Themes vary. Let me know what you might want to see more of. Appreciate all input. All are just for fun.
I've been merrily generating images in Bing Image Creator only to find that it seems to be the case that I can't access any images I've generated apart from those that are shown in "Recent". Is that the case? If so, that's utterly ridiculous; SeaArt saves every image that you generate except for those that you delete.
Not only does it look like I've lost a lot of images that I wanted to keep (and share) because of not realising this, but I've also lost the prompts which generated them. It would have been good if BIC warned me of this when I first signed up or at least warned me when I first filled up the Recent list...
It also looks like BIC doesn't have a way of deleting generated images in the Recent list that are no good, which is also ridiculous. The images it can generate are fantastic, but the user interface is unfit for purpose.
If you look in your browser history, you will see the prompt in the title of the history link so can recover those. That said, you are correct in that it only retains the last 20 attempts generated on the site, you need to explicitly save them. I did see some one created a browser addin which automatically saves them but I can't find the link now. Will have a look for it.
messg said: If you look in your browser history, you will see the prompt in the title of the history link so can recover those. That said, you are correct in that it only retains the last 20 attempts generated on the site, you need to explicitly save them. I did see some one created a browser addin which automatically saves them but I can't find the link now. Will have a look for it.
Thanks, but unfortunately I have Firefox set to clear my browser history every time I close it. I had no idea that only the last 20 images were retained in Bing Image Creator.
The add-on sounds useful, so it would be great if you're able to find it.