Sunset Riders: the Dynamite Ribbons, an experiment with AI (1 Viewer)

Watt Khan

Club Regular
Joined
Jul 6, 2021
Been playing around a bit with realistic AI image generation when I thought of the idea of making a small Zako project out of it. Using the Realistic Vision V2.0 model, I've generated nine images based on the prompt description of the Zakos in Sunset Riders. Exactly nine because that is the number of times that these girls appear in the game. In the Arcade game, these girls are presented uniformly as cute albeit inept criminals. I've opted to add more individuality to these girls and weave a sort of narrative around them:​

Sunset Riders: the Dynamite Ribbons

In the lawless American frontier, the eccentric Englishman Richard Rose has created a criminal web built upon hierarchies and organizations. This enforced organization is evident in the criminal sisterhood known as the Dynamite Ribbon Girls. Composed entirely of blonde cowgirls, these sisters in crime are primarily known for wielding the good old boom stick, creating explosions both in the destructive and sensual sense. Although often associated with explosive nitroglycerine, Richard Rose values these girls more for their ability to sway and manipulate those influential people in society. Their combat abilities, though, leave a lot to be desired. Endangering themselves and their comrades, these cowgirls are often careless when handling their TNT explosives during firefights. To compensate, these girls would rely on the Dynamite Ribbon signature of playing coy to extract themselves out of sticky situations. The average lawman would stiffen and blush unable to handle the feminine charms of these cowgirls, giving enough time for the average male mook to shoot them right where they stand. Despite their notoriety, the Dynamite Ribbons only consist of nine members spread across Richard Rose's criminal empire.​

Stage 1.jpg

Although not quite veterans, Amber and Eryn have seen their fair share of action with lawmen and the usual fat gullible official. These two Dynamite Ribbons have been given the short end of the stick being assigned under the penny-pinching Simon Greedwell. Most other criminal heads are usually very liberal when distributing stolen loot; Greedwell, on the other hand, wouldn't spare a dime if his mother's life depended on it. Frustrated with this whole affair, Amber and Eryn are forced to look into alternative ways of satisfying their material wants. Fortunately for Eryn, the area that she is assigned to is close to the home of a corrupt sheriff that she has wrapped around her finger. Donning the tightest and shortest pair of pink hot pants in her wardrobe, Eryn is confident that she could suck the sheriff dry for all he is worth. Amber is in a less favourable spot, unable to contact and leech on her victims whom she calls her "little darlings."; she is forced to find extra earnings from amongst her ilks. It isn't a bad arrangement by any means, Amber quite enjoys keeping company with the boys during the night, and the arising conversations, surprisingly, could be pretty thoughtful.​

Stage 3.jpg

A trio of Dynamite Ribbons also operates in the criminal-infested town of Bloody Gulch, governed by the tyrannical mass murderer Dark Horse. As the oldest Dynamite Ribbon in town, Harper acts as the default leader of the trio. She walks with an air of self-confidence much warranted, given her instrumental role in the downfall of Bloody Gulch. Playing the role of mistress to the elites of Bloody Gulch, Harper has tricked the mayor into surrendering the town to Richard Rose. The uniforms of these girls would, understandably, offend the sensibilities of respectable women. However, Emilin's choice of dress is the most daring out of all the Dynamite Ribbons. Unlike Harper, Emilin despises the hypocritical facade of polite society, preferring to gallivant with the dregs and plebs of Bloody Gulch. Promising irresistible favors, Emilin has successfully swayed numerous law enforcers and vigilante groups to join the Criminal fold.

Millie is perhaps the most unique out of the Dynamite Ribbons due to her taboo preference for women over men. The women of Bloody Gulch, as in any other town, are the purveyors of gossip, and such idle talks can be quite indispensable for the intelligent criminal. Millie's job is to get intimate with the barmaids, burlesque dancers, and farmer's daughters, coaxing them to reveal any delicate information potentially useful to Richard Rose's criminal enterprise. Millie is often teased by the rest of the Dynamite Ribbon girls because she is not as well-endowed as them; this is a source of major insecurity for poor Millie. However, Millie does afford a smile every time she hears her criminal sisters complain of chaffing on their breasts.​

Stage 8.jpg

At the foot of the mountain which supports Richard Rose's mansion, the Dynamite Ribbons, and Rose's other bandit groups have established their main headquarters. Here, we can find the founder and head honcho of the Dynamite Ribbons, Victoria. Although age has slightly withered her beauty, Victoria is still respected as an important matriarch figure by Richard Rose's wider gang. Victoria no longer "takes to the field" like her sisters. She retains the more important role of coordinating the group's courtesan infiltration efforts. Always by her side, Anne acts as a drill sergeant of sorts and recruiter to the Dynamite Ribbon group. This is a post she has inherited from the more senior Harper. Having recently recruited the still very young Eve and Beckie, Anne is determined to shape these girls up, making them worthy of the Dynamite Ribbon fame.

Just reaching twenty, Eve is perhaps too overzealous in trying to prove to Anne that she has mastered the Dynamite Ribbon art of seduction. Despite her best efforts, however, Eve could never really pin a man down once she got his initial attention. Anne has to remind Eve constantly that it takes more than just physical attraction to be a successful Dynamite Ribbon. On the other hand, the problem with the barely eighteen-year-old Beckie is that she is just too shy in all the things that she does. Despite these issues, Anne is confident that these young ladies would one day become successful official members of the Dynamite Ribbon group. She can't help but smile thinking about the bright future of their small criminal sisterhood.

Unfortunately, Anne's optimism would soon prove to be misplaced. Word is spreading in the criminal underworld that a group of expert gunslingers known as the Sunset Riders are hellbent on the destruction of Richard Rose and his goons. With the incoming onslaught, the days of the Dynamite Ribbons are perhaps numbered. Will the girls be able to soften the hearts of the Sunset Riders as they have done with previous lawmen? Or will they receive swift and brutal justice? Only time will tell...​
 
Last edited by a moderator:

Weoooo

Master of this Domain
Joined
Dec 3, 2010
Man, I didn't realize that there were only 9 in the game. Somehow that adds this real intense layer to their vulnerability; there's not endless amounts.
 

KeiBree

Master of this Domain
Joined
Jul 27, 2012
These are amazing :)
Can you post the prompts you used? I'm interested in experimenting with this but don't know where to start and wouldn't mind having an example I can de-construct :)
 

Watt Khan

Club Regular
Joined
Jul 6, 2021
These are amazing :)
Can you post the prompts you used? I'm interested in experimenting with this but don't know where to start and wouldn't mind having an example I can de-construct :)

Sure!

For the Sunset Riders Zako:-

Prompt: Fit caucasian blonde woman, long hair, naughty face, royal blue shirt, long sleeve, cleavage, pink cowboy hat, pink shorts, full body, shapely legs

Negative prompt: midriff (for the standard zako look)

For Linda (in red colours) :-

Prompt:
Fit sexy woman, slightly muscular, Caucasian model, Beautiful Face, Seductive, Scarlet red tight spandex cat suit,
sleeveless, legs tight fit, tight fit legs, leg covered, leg covered, V-neck, light Chestnut hair, Big Curly Hair, Belt, belt, belt, bracelet, wall background, 3/4 shot

Negative prompt: Midriff, bikini, bare legs, naked legs, sleeves, arm sleeves, shiny,

For the Ginger Runark Doll:-

Prompt: Fit sexy woman, Caucasian, Beautiful Face, hot face, buttoned dark fuschia buttoned power suit, low-cut, cleavage, sleeves, short skirt, Ginger hair, Long straight curled Hair, 3/4 shot

Negative Prompt: Midriff, bikini, trousers, shirt, latex

A few tips:

To add more variety to the faces, try adding nationality to the ethnic descriptors ex: British caucasian, French Caucasian etc. Try being specific with your colours to get better results like Fuschia, Royal Blue etc. You can add background prompts like: canyon background or town background. Careful though, I find that adding backgrounds could heavily influence the generated image and not for the better.

At its current state, AI generation is not good with specifics. So even though I mention pink shorts and royal blue shirt, the output image might be a cowgirl with pink shirt and blue shorts, instead. Just keep regenerating till you get the intended results. Also it might help if you keep hammering a single description. Hence, the "belt, belt, belt" in the Linda prompt.

Hope this helps!
 
Last edited:

horatiojones80

Content Creator
Joined
Jun 22, 2011
I just got stable-diffusion up and running on one of my computers. I find I can generate standing portraits of women just fine, but anything other than that looks like a body horror mess. I cannot for the life of me get it to generate an image of a woman lying down normally. So my idea of having a woman lying face down (dead or not) and just doing the rest in photoshop is pretty much moot. Have you had any luck with this? I'm running it on a laptop with an intel chip so I had to disable floating point arithmetic to get it to work,
 

Watt Khan

Club Regular
Joined
Jul 6, 2021
I just got stable-diffusion up and running on one of my computers. I find I can generate standing portraits of women just fine, but anything other than that looks like a body horror mess. I cannot for the life of me get it to generate an image of a woman lying down normally. So my idea of having a woman lying face down (dead or not) and just doing the rest in photoshop is pretty much moot. Have you had any luck with this? I'm running it on a laptop with an intel chip so I had to disable floating point arithmetic to get it to work,

Unfortunately, that's just an inherent limitation of the current AI models available to the public. They're good with creating portrait/pin-up style images, but they falter with anything more complex. I've tried my hand with creating some zako "death poses"; unsurprisingly, the only decent batch are the portrait style images:


american-caucasian-model-unconscious-on-ground-dead-medium-sized-breasts-c...eductive-low-cut-...jpg
american-caucasian-model-sleeping-medium-sized-breasts-closed-eyes-royal-b...-cleavage-thick-l...jpg
 

Weoooo

Master of this Domain
Joined
Dec 3, 2010
Unfortunately, that's just an inherent limitation of the current AI models available to the public. They're good with creating portrait/pin-up style images, but they falter with anything more complex. I've tried my hand with creating some zako "death poses"; unsurprisingly, the only decent batch are the portrait style images:


View attachment 145574View attachment 145575

A lot of the 'death' pictures I've seen have a real photomanip quality to them, where it's very clear they're based on pinup poses but the expression is changed. It is, oddly, something that charmed me from being openly hostile to AI stuff to seeing the way it can express people's creativity; trying to make something assemble the thing you want leads to expression, too. I'm still not, like... happy about it, but I'm not openly upset at the idea anymore

Which is, I'm sure, the most important thing people could've ever heard.
 

Cobra02

Club Regular
Joined
Jun 24, 2016
Unfortunately, that's just an inherent limitation of the current AI models available to the public. They're good with creating portrait/pin-up style images, but they falter with anything more complex. I've tried my hand with creating some zako "death poses"; unsurprisingly, the only decent batch are the portrait style images:


View attachment 145574View attachment 145575
You can try ControlNet if you are using Automatic1111. or you can start tweaking with img to img to get some decent results. i got kinda acceptable results. Not always but this helps.
 

horatiojones80

Content Creator
Joined
Jun 22, 2011
A lot of the 'death' pictures I've seen have a real photomanip quality to them, where it's very clear they're based on pinup poses but the expression is changed. It is, oddly, something that charmed me from being openly hostile to AI stuff to seeing the way it can express people's creativity; trying to make something assemble the thing you want leads to expression, too. I'm still not, like... happy about it, but I'm not openly upset at the idea anymore

Which is, I'm sure, the most important thing people could've ever heard.
I'd love AI if I could get it to automate some of the drudgery involved with making art and spirte art. Like taking a photo and producing a clean sketch outline of the main subject with no background, or taking a sprite sheet and changing something about every image in it. Like taking a character and making her fatter, or taking an image and making the subject face another way. It's totally possible but the model just needs to be trained the right way and the stable diffusion model is just not. It's trained to blatantly rip off existing art, top google searches and stock images to generate a boring portrait photo that only sometimes matches your specifications. I don't have the hardware to train a model either. I can't run this on my gaming PC becasue it has an AMD card (AMD: not even once) so I'm running this on my laptop. Still I managed to generate a few reference pictures that I could trace over and use in future projects.

A properly trained AI model could be used to take an image and just spit out DOOM sprites from it. How fucking cool would that be? I could pump out so many Zako doom mods from that since the only thing holding me back is the spriting.
 

Attachments

  • 00011-926228739.png
    00011-926228739.png
    586.3 KB · Views: 89
  • 00000-3701314337.png
    00000-3701314337.png
    529.6 KB · Views: 103
  • 00020-3186930505.png
    00020-3186930505.png
    399.4 KB · Views: 96

Watt Khan

Club Regular
Joined
Jul 6, 2021
A lot of the 'death' pictures I've seen have a real photomanip quality to them, where it's very clear they're based on pinup poses but the expression is changed. It is, oddly, something that charmed me from being openly hostile to AI stuff to seeing the way it can express people's creativity; trying to make something assemble the thing you want leads to expression, too. I'm still not, like... happy about it, but I'm not openly upset at the idea anymore

Which is, I'm sure, the most important thing people could've ever heard.

As an on and off again artist, I absolutely get where you're coming from. It can be quite depressing seeing people just churn out (often times, very monotonous looking) AI generated art. Weirdly enough though, I don't really feel that strongly about AI generating realistic images as opposed to stylized pieces. You're never realistically ever going to get the perfect model to cosplay as your favourite female mook from some obscure 90s beat em up game.

I've always liked the idea of physically designing zako based costumes, in fact, tailoring for figurine models is something that I absolutely would love to pursue in the near future. But when it comes to having actual breathing human models involved, AI is the only option which satisfies that itch that I have of wanting to see zakos in "real life" form.

I'd love AI if I could get it to automate some of the drudgery involved with making art and spirte art. Like taking a photo and producing a clean sketch outline of the main subject with no background, or taking a sprite sheet and changing something about every image in it. Like taking a character and making her fatter, or taking an image and making the subject face another way. It's totally possible but the model just needs to be trained the right way and the stable diffusion model is just not. It's trained to blatantly rip off existing art, top google searches and stock images to generate a boring portrait photo that only sometimes matches your specifications. I don't have the hardware to train a model either. I can't run this on my gaming PC becasue it has an AMD card (AMD: not even once) so I'm running this on my laptop. Still I managed to generate a few reference pictures that I could trace over and use in future projects.

A properly trained AI model could be used to take an image and just spit out DOOM sprites from it. How fucking cool would that be? I could pump out so many Zako doom mods from that since the only thing holding me back is the spriting.

Having dabbled in sprite making and having close friends who are talented sprite artists, using the word "automate" sort of sucks the heart and artistry behind the whole enterprise. I don't see the artistic process as any sort of drudge, on the contrary, it can be quite rewarding seeing visible improvements in your work. But hey, that's just me and this is certainly not the proper place to discuss the ethics or philosophy of art.
 
Last edited:

Weoooo

Master of this Domain
Joined
Dec 3, 2010
I think the realistic stuff doesn't bother me, either. Feels more like a photomanip rather than robbing someone else's style, I guess.

Electra from SoR3, Fox from Undercover Cops, Swallow from... that one game, and I'm not sure about the last one. Silent Dragon?
 

Watt Khan

Club Regular
Joined
Jul 6, 2021
I think the realistic stuff doesn't bother me, either. Feels more like a photomanip rather than robbing someone else's style, I guess.

Electra from SoR3, Fox from Undercover Cops, Swallow from... that one game, and I'm not sure about the last one. Silent Dragon?

lol, I think it's hilarious that neither of us could remember the name of that game where Swallow is from. Same goes for the last zako and I generated her based entirely off of memory. In retrospect, I didn't get her design 100% correct. She's from a game called The Young Boss, Jjang

That electra looks stunning!

Prompt? :P

There ya go 😁:

Fit sexy woman, french caucasian, buxom, Beautiful Face, seductive, hot face, perfect body, thin waist, hour glass body, strapless royal blue one-piece, low-cut, cleavage, short gloves, blonde hair, long straight curled hair, long hair, garters, blue tall boots, back alley background, afternoon, sunset, close up shot, 3/4 shot

Interestingly, I notice that certain prompts sometimes no longer yield similar results. I cannot recreate the images of those two skimpy zakos from the last batch using the old prompts. I guess this has something to do with AI models consistently updating its method/approach to machine learning. It's kinda a shame, but fascinating at the same time. Certain images are literally relics of an older AI model configuration.
 

horatiojones80

Content Creator
Joined
Jun 22, 2011
As an on and off again artist, I absolutely get where you're coming from. It can be quite depressing seeing people just churn out (often times, very monotonous looking) AI generated art. Weirdly enough though, I don't really feel that strongly about AI generating realistic images as opposed to stylized pieces. You're never realistically ever going to get the perfect model to cosplay as your favourite female mook from some obscure 90s beat em up game.

I've always liked the idea of physically designing zako based costumes, in fact, tailoring for figurine models is something that I absolutely would love to pursue in the near future. But when it comes to having actual breathing human models involved, AI is the only option which satisfies that itch that I have of wanting to see zakos in "real life" form.



Having dabbled in sprite making and having close friends who are talented sprite artists, using the word "automate" sort of sucks the heart and artistry behind the whole enterprise. I don't see the artistic process as any sort of drudge, on the contrary, it can be quite rewarding seeing visible improvements in your work. But hey, that's just me and this is certainly not the proper place to discuss the ethics or philosophy of art.
Yeah but what I do is basically sampling any way. I take existing photos, remove the background, trace over them in another layer, fill in any missing parts of limbs then shade them, then turn them into sprites. If something could make that process easier anyway for my purposes which are purely not for profit, but for making mods, that'd be awesome. I'm sure manually drawing things (which I suck at) is a lot of fun, but does anyone actually have fun in DAZ 3D studio? Because when I tried to use that program I wanted to pull my fucking hair out.
 

Users who are viewing this thread

Top


Are you 18 or older?

This website requires you to be 18 years of age or older. Please verify your age to view the content, or click Exit to leave.