DALL·E 3

[bobvan2]

Agreed this is why alot of render artists are losing motivation..

[Little Sally]

Agreed this is why alot of render artists are losing motivation..

I can see why you'd think that way.

My view is from what I've seen so far, that because it's obviously very easy to produce imagery like this, it will soon lose its novelty value and become commonplace.

In other words, boring, until the next fad comes along.

So have fun while ye may!

[foreverlurk]

Agreed this is why alot of render artists are losing motivation..

IMHO, I don't think AI is quite there yet. An experienced render artist has so much more freedom and flexibility than those AIs. I'm just showing what can be done, with a LOT of frustrations (+75% of renders blocked), to bypass the TWO layers of politically correct filtering in place by Microsoft. Also, I'm pretty sure those will get patched in the coming weeks, so after that it'll be even harder. I'm not just talking about the nudity, it's anything relating with female characters, ie. you can't have them sad, have them scared, locked in a cage or grabbed (even playfully). Maybe having them shrunk will be blocked, too? With all the negative press around AI, I'm sure those barriers aren't going anywhere soon, barriers that artists don't have.

Anyways, it's not my goal to make any real artist depressed or unmotivated.

[Prof Sai]

instead of spamming this thread, maybe.

Nah, just put your name in the title and own it. Post everything you do here. (Mirrors to other sites are fine, you never know what will vanish.)

With all the negative press around AI, I'm sure those barriers aren't going anywhere soon, barriers that artists don't have.

Just use an offline renderer instead. You might need more post processing for a while, but the privacy and lack of censorship is worth it.

Anyways, it's not my goal to make any real artist depressed or unmotivated.

Ta! I really have to focus.

[kira123]

With all the negative press around AI, I'm sure those barriers aren't going anywhere soon, barriers that artists don't have.

Just use an offline renderer instead. You might need more post processing for a while, but the privacy and lack of censorship is worth it.

What offline renderer do you recommend?

[Prof Sai]

There are two free options that I use:

For starting out, DiffusionBee is best:

https://diffusionbee.com/download_v2

It says it is a "Technical Preview", but it is rock solid on my machine and everyone on the Discord recommends it. Don't bother with the 1.7 version, use 2.2. There is a waitlist for the windows version, but no clear timeline. Intel chips are not recommended,

I also use https://drawthings.ai/
This can even run on an iPhone somehow. It can render at higher resolution than DiffusionBee, but I find the more advanced features harder to use. It is also noticeably slower.

Both programs can be quirky, with pose control not working well on either. I get best results using it as if it were Google Photos, and then making montages in an art program. (Pixelmator, Gimp, photoshop.).

[kira123]

Yeah, I don't have a Mac, nor anything Apple-related. So, got any for Windows and Android?

[Prof Sai]

I'm afraid not. I can find some Reddit posts with people talking about stuff, but it is not clear to me what they are saying.

[kira123]

Dang...

[Elana]

An easy start for windows would probably be
easy diffusion
https://easydiffusion.github.io

and for a wide collection of different models to use with this
have a look at the model collections here (you probably want something based on stable diffusion Xl, so it can guess text parts or stable diffusion 1.5 just because it has the widest range of actual different models available)
https://civitai.com

[foreverlurk]

Just use an offline renderer instead. You might need more post processing for a while, but the privacy and lack of censorship is worth it.

Thanks, but I'm already using Stable Diffusion locally. It's a PITA sometimes, had to relearn the UI (ComfyUI) when I started using SDXL. Still, DALL-E just works out of the box and is a much simpler experience, even with all the hurdles. That being said, they run DALL-E on huge datacenters, whereas SDXL runs on my machine - so let's keep expectations in check.

BTW I've started sharing my AI stuff on DeviantArt, if anyone cares:
https://for3v3rlurk.deviantart.com/

I'll still post here and maybe elsewhere (I've heard rumors of size content purges on DA?!).

[Little Sally]

BTW I've started sharing my AI stuff on DeviantArt, if anyone cares:
https://for3v3rlurk.deviantart.com/

I'll still post here and maybe elsewhere (I've heard rumors of size content purges on DA?!).

It might be a good idea to clearly state you're using AI when creating your DA pics (under each one individually I mean).

There have been some issues with artists complaining about AI created work, and DA did advise AI users to do that.

I'm not one of those who've complained about it btw, just so you know. And I haven't heard any rumours about size related work being a problem either, though blatant explicit porn is always likely to attract unwanted attention. Your submissions certainly aren't in that category. Hope that's helpful.

Edit: I see on closer inspection you do state it's AI created work (might have been more obvious in bolder text though). So all's well.

[foreverlurk]

It might be a good idea to clearly state you're using AI when creating your DA pics (under each one individually I mean).

You're right, I did write it in my bio and under each picture there is the mention:

"Created using AI tools"

Maybe I should tag all my pictures too, and put it in the name of the galleries. I don't want to mislead people or pretend I didn't use AI (in fact sometimes I used DALL-E first, then pass through SD to fix some details, then GIMP).

Thanks for the feedback.

[Little Sally]

It might be a good idea to clearly state you're using AI when creating your DA pics (under each one individually I mean).

You're right, I did write it in my bio and under each picture there is the mention:

"Created using AI tools"

Maybe I should tag all my pictures too, and put it in the name of the galleries. I don't want to mislead people or pretend I didn't use AI (in fact sometimes I used DALL-E first, then pass through SD to fix some details, then GIMP).

Thanks for the feedback.

Yes, labelling pics and galleries clearly as AI is a good idea, and then there's no excuse for anyone to bitch about it.

Your page looks good btw, so good luck with the project!

[bobvan2]

I was not speaking for myself. It's a hot topic on Daz render forum. I guess the biggest factor being A I characters look more lifelike. Forever and Sally also bring up valid points.

[Prof Sai]

I was not speaking for myself. It's a hot topic on Daz render forum. I guess the biggest factor being A I characters look more lifelike. Forever and Sally also bring up valid points.

I think that someone will find a way to combine the two technologies, using AI to build on Daz and add the realism and life that Daz lacks, while retaining the far superior control of CGI.

Using AI as "Throw some random ideas at me." is also valuable.

[bobvan2]

I was not speaking for myself. It's a hot topic on Daz render forum. I guess the biggest factor being A I characters look more lifelike. Forever and Sally also bring up valid points.

I think that someone will find a way to combine the two technologies, using AI to build on Daz and add the realism and life that Daz lacks, while retaining the far superior control of CGI.

Using AI as "Throw some random ideas at me." is also valuable.

Who has been doing it for bit. He does one offs. In a story context could be more challenging. I am sure it will get there. Nice to see what folks are coming up with all the same.

[Pacific1982]

Am getting some interesting results with prompts saying “Giant troll and modern young woman.”

[Pacific1982]

But my attempts at shrunken women give me results that look like little porcelain dolls.

[foreverlurk]

But my attempts at shrunken women give me results that look like little porcelain dolls.

What prompt did you use? I can help.

Another thing I'm trying out is comics... turns out DALL-E can create some cohesion between scenes, keeping the same characters, clothes, etc. here's a quick mockup I did where a female spellcaster shrinks herself:

[kira123]

I tried Easy Diffusion, told them to do a pic of a woman grabbed by a giant hand in anime style. All I ended up with are pics of women in a scrambled mess. I gave up. =_=

[foreverlurk]

I tried Easy Diffusion, told them to do a pic of a woman grabbed by a giant hand in anime style. All I ended up with are pics of women in a scrambled mess. I gave up. =_=

They all have their own strength and weaknesses... Stable Diffusion doesn't understand size difference as well as DALL-E. But it's much better at full realism and photography. Also, no filtering.

DALL-E's understanding of natural language prompts is crazy good. But you have to use workarounds to do more kinky stuff.

Midjourney has yet to release their next iteration, keeping an eye on that.

[foreverlurk]

So many prompts blocked since yesterday... everything is NSFW, and mentioning "woman" is ban-worthy.

[Prof Sai]

So many prompts blocked since yesterday... everything is NSFW, and mentioning "woman" is ban-worthy.

Try uterus bearing person.

[kira123]

So many prompts blocked since yesterday... everything is NSFW, and mentioning "woman" is ban-worthy.

Classic Current-Year Microsoft...