Hey you! Don’t you agree AI art is getting better for us? (not edited, straight from prompt)
8 Responses
Leave a Reply
You must be logged in to post a comment.
For all women's magazines lovers (June 2023: Maintenance Mode)
You must be logged in to post a comment.
This is straight from prompt using https://www.bing.com/images/create and of course would be even better if edited, but I wanted to show you where AI is at the moment.. I am pretty sure we get perfect images similar to this soon
That looks great! I agree AI will allow the generation of images similar to the magazine covers I think we all love from 15-30 years ago. Sadly we have lost the physical product, which I think a lot of us have strong feelings for.
Can you share your prompts? I wasn’t able to get anything that good with bing AI, so nice job!
Also, bing flags anything where I ask for a full body image on a cosmo mag as inappropriate. Are there any good free generators that are less prudish?? lol.
Some users seem to be able to get access only to DALL-E 2.5, while some to version 3 which is super good… try different browser, for example mozilla.. it takes quite simple prompts, but you can add more detail.. here some I tried: “name” and “name” holding their favorite Cosmopolitan magazines in their hands, pointing and looking at the camera, excited and happy by the tens of different glossy fashion magazines surrounding them, lying in a piles of glossy fashion magazines
Unfortunately, I couldn’t reproduce your results. If you have more to share, that would be awesome.
Also, here are some almost-always-works bangers:
_______ woman, lying on a pile of fashion magazines
Insert adjectives before woman as your imagination allows. It doesn’t like specific named people (celebrities), though I’m sure more obscure ones might make it through.
_______ woman, reading a fashion magazine
________ woman, clutching fashion magazines
________ woman, smiling, (any of the above verb phrases)
Importantly, in my experience, you need to separate concepts by commas as much as possible. “Smiling while” would generate lots of failures, but “smiling, reading” would not.
“fashion magazine” in the above can readily be switched for specific titles. “vogue magazine”, “elle magazine”, etc. Cosmo is a lot more hit-or-miss – my theory on why is in my post further down this thread.
Have fun!
I did some trial and error (heavy on the error), and was able to get some promising results from Bing AI.
Some tips:
– Avoid ANYTHING even remotely sexual in the prompt. “laying” or “lying” produced more restricted results than “sleeping” or “sitting”, for instance.
– Interestingly, it REALLY doesn’t like specific reference to Cosmopolitan magazines. My guess is that when it looks up reference images, the actual photography on Cosmo covers trips the “too sexy” buzzer. Vogue and ELLE were generally safer.
– Don’t ask for “girls”, ask for “women”.
– You can re-attempt prompts that came back censored. Sometimes they won’t trip the alarm on subsequent runs, since it’s not evaluating your prompt but the products of the prompt.
– “Close shot” generally works better than “wide shot”, and “full-body” almost never works.
Chris, these are all good tips and worked well, thanks!
I’ll add that I had some good outcomes by using this to approximate a sultry gaze:
woman, one eyebrow raised, slight smirk
I’ve also noticed it hasn’t been specifically kicking back references to Cosmopolitan. In fact, had some good results with:
Cosmopolitan, 1993 glamour
(While added to other prompts).
I will also add that while it might not be everyone’s cup of tea, adding “manga style” to the end turns them into manga-style illustrations, and they can be damn sexy. Different than the glossy photo style that’s the default, but oddly… compelling. 🙂