Ok, so I think that part of my problem is that I am not doing the prompts right. Can you give me an example of how you would put in prompts, like in midjourney?
Look at this whole thread, as you start to go down it youāll see I include a ton of prompts for all types of needs. This should help a lot:
Keep any questions about those prompts to this thread to keep it organized.
Thatās so close, so did seeing the prompts help you? You can also āuploadā a reference image. Right after you type /imagine you can paste a link to any image online then type the prompt. So it would look like this:
/imagine https://canada1.discourse-cdn.com/flex007/uploads/getupearlier/original/1X/df00dc39fd360db763a5f5a401ebf53ddf213067.jpeg a huge metal ball of some kind standing in the middle of it, in the style of detailed character design, wandering eye, future tech, serge marshennikov, primordial creatures, technological art, neo-op --s 750 --v 5.2
Iād be willing to bet the results are almost identical
It does help. I see that I have to be a bit more descriptive with it.
I shared that with my Destiny clan and they loved it.
I will keep practicing! Thanks!
Wild. Yes! Hey please feel free to invite any of them in here to ask questions. Iām really interested in offering up this type of info here. Having other like minded people in here would be great.
I asked it to do a realistic girls face, HR Giger style art, ultra realistic, fujifilm, high resolution, and this is what I got. Still working on trying to use more descriptive terms to get exactly what I want.
And thatās why sometimes DALL-E is better because itās a conversation. You can just prompt and then mold it in real time via conversation. Like I did here to get this image of ābrain scansā:
I do like ChatGPT as well as midjourney. I find ChatGPT is easier for doing logos. I can take those logos and open them in illustrator, convert them to vector and add text and other cool stuff. So, Iām thinking for now Iāll stick with those two and keep working on getting better with it.
These have become my 2 go to. I goof off with Adobe Firefly but itās political correctness is so awful it just gives the most ridiculous output. If I need a still photo of something that has nothing to do with people, generally, itās great.
Working on prompts, coming along. This one came out pretty cool.
I have been batting AI on text since I started here. The 2 things I wanted most which were human likeness and cool text treatments were huge disappointments. I still havenāt seen that much improvement. It literally always shoots out typos.
Yea, I try and leave out asking to do text if I can help it.
Itās funny though, on X(Twitter) I can spot all the AI generated images a mile away now, because of the messed up text.
This is an interesting article about āThe Steparound Approachā which I have been doing intuitively since day 1:
Prompt Engineering Technique Known As The Step-Around Is Gaining Steam As Generative AI Becomes Less Forthright
The Step-Around Prompt And How It Has Come To Be
By the phrasing of being a said-to-be āstep-around promptā, the gist is that you purposely word your prompt to evade being detected as having provided a question or matter considered sensitive in nature or consideration. You are trying to dig deeper into the AI and not get snagged at the front porch. Most of todayās generative AI apps are rigged to immediately reject a prompt that the AI maker has already decided is unfit for answering.
Not everyone likes this kind of āBig Brotherā automated prescreening and likens this path to censoring userās questions. If you are interested in the ongoing heated debate about when and why modern-day generative AI should or should not refuse to answer a question, see my analysis at the link here. Significant issues underlying AI ethics and AI law come into play in this contentious matter. Please know we are in the early days of generative AI adoption and the AI Wild West might inevitably be rejiggered by ethical precepts or additional laws and regulations.
@Imax68 my man hit me with that logo thing, I want to give it a shot
Hey Mike, it didnāt work out. Customer didnāt want to go through with the job.
Iām looking into on how to sell AI art online. I guess I need to learn how to use aspect ratios so the image has the higher resolutions. Any suggestions?
I have used Midjourney for high resolutions but not 300DPI poster print size. More like very good 4K wallpaper size. I havenāt gotten anything close out of DALL-E but I also wonder, given how anyone can now make pretty amazing images if thereās a true marketplace for it?
I would also assume thereās some AI up-res type service.
I would think itās all been cheapened? Let me know what youāve looked into, Iām interested too.