Artflow AI

[Ithilan]

Hello everyone living in our science fiction world, where AI soon will take over, or at least be deeply integrated in our lives.

In the art world it has no doubt already taken a center stage, for good and bad. But in terms of creativity I find it a gift and particularly for the kind of hobbies most here perhaps engage in. So I wanted to create this topic to talk about the particular Art AI called Artflow.ai that I found about a year ago myself and have promoted a bit on the BG discord back then too.

It has improved a lot since and though im not the best at prompting it, it has created some artwork for a lot of my characters that I am very satisfied with. A number of other people on the server have had amazing results with it, so I thought it might be in the interest of others too to toy around with and realise some creative ideas they have.

I have generated about 2500 pictures with it and deleted perhaps 1800 of those as they had weird distorted faces, hands or multiple ears and too many fingers, so patience seems a key to success from my experience and persistence with prompts to see its many interpretations till it gets it right.

Here is an example of some pictures it has generated for my characters:

Shandril Brightmantle

Sigrid Grímsdóttir

Shiera Ravenscroft

Neira Ethemille

And recently I have been experiementing with their relatively new video studio as well, seeing what it is capable of, my creativity is perhaps not the greatest in the world, so feel free to cringe hard but heres an attempt of a brief video to give a bit of life to one of the above pictures:

I think the potential of tools such as this is amazing in a narrative sense for our little niche community. Whether it is someone as artistically void of skill as me, looking for a character portrait, or perhaps DMs or others wanting to create pictures for their campaigns, settings, monsters and villains. Or short videos and the like, I thought it was appropriate to make a topic here and showcase some of its abilities (in my unimaginative hands ) and hope others will find as much joy from it as I personally have, for BG related content or otherwise.

Feel free to share your experiences with this or other AIs, tips for prompts or ideas. My own are perhaps very stereotypical D&D styled inputs such as bard, cleric, knight, armor descriptions, details about skin complexion, eye colors, hair and the setting or action they are making, if it is an angled shot or portrait style, the time of day, background and so forth. Many of the style options for this particular AI, will yield varying results depending on inputs as well.

Lastly ill say that I have paid absolutely nothing for using this AI, for picture or video creation, but the videos do use a currency and has payment options, I know nothing about those or what benefits it may give, but as a free tool it has plenty of possibilities for my own needs and over time it seems you earn their respective credit regardless and that is how I have made the brief video above.

I hope it may be a help to enhance someone elses story telling or immersion, have fun!

[Aspect of Sorrow]

Been in this sphere since around 2016 and have seen the most improvement coming out of Midjourney, though they have a leg up on the competition with Jim Keller's experience there are open source derivatives out there that are fast moving into the space based on GAN/Stable Diffusion forks. Original training models were unethical but the since created and approved content helps the human positive reinforcement element to course correct. I presently take elements from failed result sets and add them into layers to help produce the final output.

Stable Diffusion integration also exists for a number of products such as Photoshop and Gimp, one such example on Gimp is OpenVino AI Plugin with a video of its usage here.

[Steve]

AI art generation is amazing. And yes, AI will eventually take over the world, starting with replacing RL players in games with AI players for those RL players. lol.

But nothing will ever replace learning how to draw yourself, and making your own images from your own imagination! Everyone can do it. If anyone put as much time into learning how to draw as they do into gaming, you WILL get a positive result.

[JustAnotherGuy]

After seeing it here, I've been playing with artflow.ai. I like it, but it's a bit odd. For one, they don't know what "Short beard", "Trimmed beard", "Short Facial Hair", or "Trimmed Facial Hair" is. I keep getting a beard like a dwarf.

On the other hand, typos make things interesting.

For instance, when you leave the "d" off in "beard" and instead put "Add a short bear to the face" . . .


I'm continuing to play with it, and am getting closer to what I want.

[Hullack]

Been in this sphere since around 2016 and have seen the most improvement coming out of Midjourney, though they have a leg up on the competition with Jim Keller's experience there are open source derivatives out there that are fast moving into the space based on GAN/Stable Diffusion forks.

I tried that in the past but it was always too overloaded to let non-paid users generate any images. I stumbled on this one that let me in right away and seems promising: https://app.leonardo.ai/

[Ithilan]

I presently take elements from failed result sets and add them into layers to help produce the final output.

Stable Diffusion integration also exists for a number of products such as Photoshop and Gimp, one such example on Gimp is OpenVino AI Plugin with a video of its usage here.

I do as well to some extend, but some times the abbominations the AI create simply need to be purged with the delete button. There are a few plugins for GIMP i used myself to retouch images and heal sections, removing the mutated sixth finger or the tiara I did not want etc. but that almost requires a guide on its own

AI art generation is amazing. And yes, AI will eventually take over the world, starting with replacing RL players in games with AI players for those RL players. lol.

But nothing will ever replace learning how to draw yourself, and making your own images from your own imagination! Everyone can do it. If anyone put as much time into learning how to draw as they do into gaming, you WILL get a positive result.

I do agree with you Steve and im full of admiration for the many talented people both in and outside of this community, that have artistic skills in any measure. But sadly I dont have the time or ability myself, so ill settle for this and submit to our AI overlords prematurely.

After seeing it here, I've been playing with artflow.ai. I like it, but it's a bit odd. For one, they don't know what "Short beard", "Trimmed beard", "Short Facial Hair", or "Trimmed Facial Hair" is. I keep getting a beard like a dwarf.

On the other hand, typos make things interesting.

For instance, when you leave the "d" off in "beard" and instead put "Add a short bear to the face" . . .


I'm continuing to play with it, and am getting closer to what I want.

You can get some hilarious results at times, I tried to have it portray a Rashemi belonging to the Owlbear Lodge, it did not compute, nor does it understand what many weapons are I have experienced or specific objects people are holding, unless you are Sirion prompting it to hold a fish

Meet Jandar of the Owlbear Lodge (sorry for doing you dirty like this Alcreed):

[Ithilan]

I tried that in the past but it was always too overloaded to let non-paid users generate any images. I stumbled on this one that let me in right away and seems promising: https://app.leonardo.ai/

That looks really interesting, ill have to toy around with this AI too, thanks for sharing it.

[Deathgrowl]

Merenia Aujiraesthyr, bronze dragon disciple:


Obahzk Elf-skinner, blackguard of Gruumsh

[Tekill]

I have to be a party pooper and be the voice of that brings a bit of disharmony to our new AI Utopia.

I find some of these portraits, especially of the young heroines, to be too perfect. It is off putting to me as it strikes me as very unrealistic.
Perfectly symmetrical faces
Utterly flawless delicate skin (not so much as a freckle)
Professionally applied make-up
Hair fun colors, with every hair, eyebrow and eyelash perfectly in place.

This stereotypical ideal perspective on beauty is offensive because it completely lacks any REAL character at all.

Not a spec of dirt, zero wear and tear - nothing worn, no sign of fatigue or stress. These are supposed to be hardened, experienced, warriors and rogues and mages. But when I see them I am thinking, spoiled entitled trust fund kids playing dress up.

I have to admit, the orc looks pretty cool.

[Steve]

Not a spec of dirt, zero wear and tear - nothing worn, no sign of fatigue or stress. These are supposed to be hardened, experienced, warriors and rogues and mages. But when I see them I am thinking, spoiled entitled trust fund kids playing dress up.

Tekill, meet Fantasy.

[Tekill]

Not a spec of dirt, zero wear and tear - nothing worn, no sign of fatigue or stress. These are supposed to be hardened, experienced, warriors and rogues and mages. But when I see them I am thinking, spoiled entitled trust fund kids playing dress up.

Tekill, meet Fantasy.

I am not saying all beauty is bad taste. But a character has to have character- if it wants to be good fantasy.

[renshouj]

Hidden: show

I have to be a party pooper and be the voice of that brings a bit of disharmony to our new AI Utopia.

I find some of these portraits, especially of the young heroines, to be too perfect. It is off putting to me as it strikes me as very unrealistic.
Perfectly symmetrical faces
Utterly flawless delicate skin (not so much as a freckle)
Professionally applied make-up
Hair fun colors, with every hair, eyebrow and eyelash perfectly in place.

This stereotypical ideal perspective on beauty is offensive because it completely lacks any REAL character at all.

Not a spec of dirt, zero wear and tear - nothing worn, no sign of fatigue or stress. These are supposed to be hardened, experienced, warriors and rogues and mages. But when I see them I am thinking, spoiled entitled trust fund kids playing dress up.

I have to admit, the orc looks pretty cool.

I can't lie, I quite agree with this, Tekill. Everything looks very "pristine", and I personally prefer my characters more rugged.
Though, I also understand how hard it can be to find a portrait or image of your character that you feel truly fits them, and AI has been an amazing tool to make that happen. So I very much don't judge those that use AI to find what truly resonates with them, and steve isn't lying when he says that sometimes fantasy is like this, YA novels where the protagonist always has perfect hair. I'm okay with it

[Ithilan]

I dont disagree with any of you at all, but I do think thats a greater reflection of player designs than the AIs capabilities. And like Steve says, fantasy in general. My general character designs are very uninspiring in that regard, but you can absolutely use AI to make more "normal" people too, instead of something from a super model catalouge of Faerûn ^^




Or an ugly necromancer witch on her way to lichhood if you so desire


Also super cool pictures DG! Really like the scales on Merenia's face.

[Tekill]

Yeah those pictures are very cool.
Okay I believe the bridge of peace has just been built between humankind and AI. AI has given us neat art.
We shall look at this as being a sign that hence forth and forevermore there will be no more issues between mankind and AI. All shall prosper!
Yaaaaaay!

[athornforyourheart]

This seems like a peaceful conversation so I'll join in. I think AI art is amazing but has a long way to go. I believe the dall-e 4.0 is releasing soon or has and that is the next gen of AI Art but that often requires you to pay for credits to access the really great deep rooted AI Art stuff. I have an account and purchased a few credits to check it out and it yields promising results. I often find the free AI art applications or the ones that take very small monthly payments often produce the same results. And that is an issue I have with some AI art is that at this point, I can look at an image and tell by its design that its AI, so it loses that homefeel rugged feel that I think tekill is talking about.

I have a storage of AI art that I've been collecting for my projects. These are licensee to me but Ill share a few of them here. I find that most AI art is best when used as a foundation and not just a done piece. At least for now, until it gets better. I think what AoS posted above is a good example. Where you don't use the AI art to produce everything, but various pieces to that of which you can edit in an image editing program. I know not everyone has the capability or skill to edit things so easily in photoshop but I think eventually, we wont have to? But that might be a long ways off.

I wont get into the discussion about how AI art doesn't compare to actually drawn out concept artists because in a way that is like comparing apples to pears (not oranges), because there are similarities. I've seen videos of actual concept artists take AI art and draw on it and make some awesome stuff. Anywhere, here are some that I have in my AI folder:

Hidden: show