Cheesecake
The Castle Of Sexy Robot Catgirls
I was dinking around with dynamic prompts for Stable Diffusion again, and at one point ended up getting a robotic catgirl. This seemed like a fine idea, so I stripped it down to the basics and tried to build back up into a specific look. I was getting reasonably consistent results with one model, but it was weak on backgrounds. TL/DR, I ended up running the same prompt and seed through 64 models because I wasn’t able to get all the elements at the same time: clearly robotic body, metal skin and face, metal cat ears and tail, pretty face, sexy figure, naughty bits covered, interesting background, requested pose.
So I ran the exact same prompt and seed through 64 models:
4k, crisp, high budget, highly detailed, intricate, ultra textured.
photorealistic, hyper realistic, ((robot girl)) at ((Towering spires
of glass piercing a sea of swirling mist)). ((robot face)), ((silver
metal skin)), ((silver metal face)), ((metal hair)), tiny breasts,
silver metal cat tail, black metal cat ears, full body, teen height,
teen face, teen body, wearing ((black metal corset bodysuit)), ((black
metal stockings)), ((black metal long gloves)), ((black metal collar))
with ((glowing crystal bell)). Cross-legged on floor, hands on knees,
upright posture, calm and collected with a cheerful expression.
I tried to bias it toward a slender look with “teen face/body/height” and “tiny breasts” (which generally produces B-C instead of D-H). I moved the location to the beginning because I kept getting just floors and solid-color backgrounds with the occasional generic room interior. Repeating the colors for each item helps to keep them from bleeding into other elements, but, as usual, doesn’t always work.
Anything XL
Definitely a catgirl, mostly-correct costuming, matte silver skin except for neck and eyes, plausibly robotic arms and legs, terrible tail (fur, broken), robot cat-ears, cute face, sleek little body, appealing pose but doesn’t match the prompt, nice background that matches prompt.
The Gallery Of Imaginary (Christmas) Cheesecake
Fun fact: the prompt elements “Christmas” and “Santa Hat” have a powerful impact on the setting (strong bias toward indoors, decorated Christmas tree, boxed presents, fireplace with stockings hung), clothing colors (bright red and bright green), and skin tones (white Euro/American). I had to completely rebalance my skin-color wildcard file and add emphasis and throw in an extra “dark” and move it to the front of the prompt before even mentioning the girl.
And, no, I wasn’t going to add prompts that pulled in the bullshit derivative marxist mid-winter “harvest” festival invented by a convicted torturer as a “black christmas”. Because I really don’t want to find out what directions that would push the engine.
Anyway, the only other big change compared to the last gallery is that the naughty bits are covered. This required some emphasized additions to the negative prompt, and a fair amount of pruning.
(I used my usual “5-star deathmatch” method of repeatedly downranking 2/3 of the images until only 1/81 end up with 5 stars. I was able to do this quite quickly because it wasn’t like my vacations to Japan where I only have so many source images; at five seconds a pop to make more, I can be quite aggressive at discarding them in DarkTable)
Y’know the best part of generating cheesecake with “AI”? You can order it not to use a fucking wide-angle lens!
A Time Of Giving
Merry Chris-mas to all…
… and a happy Kiwi!
Or, in the words of Henry The Red, “Thank you, generous hosts!”
Happy Accidents
What I asked for (madlibs-style):
[boilerplate]. A pin-up photo of a pretty teen girl, [female-hairstyle], [sexy-pose], with a [positive-mood] expression, wearing from 2-5 [lingerie], at [famous-place]. [framing-light], [camera]
Negative: pregnant, frame, cropped, [negatives]
What that expanded to:
4k, breathtaking, crisp, gorgeous, high budget, highly detailed, intricate, professional, ultra textured. A pin-up photo of a pretty teen girl, Soft chignon, neatly tucked at the nape of the neck, Lying on side, elbow bent, hand supporting head, elegant silhouette formed, with a grateful expression, wearing compression shorts, bow-front bikini and camisole set, lace-up bodysuit, striped panties, at Scottish Highlands, Scotland. Harsh midday lighting, wide angle, scattered elements, vibrant contrast, dynamic shadows., Eye-level angle (from behind) view medium Close shot (focus on feet).
Negative: pregnant, frame, cropped. bad anatomy, bad proportions, banner, censored, collage, cropped, deformed, disconnected limbs, disfigured, duplicate, error, extra arms, extra digits, extra hands, extra limbs, fused fingers, grainy, gross proportions, logo, long neck, low contrast, low quality, low resolution, malformed limbs, missing arms, missing fingers, multiple panel, mutated, mutated hands, mutated limbs, out of focus, oversaturated, poorly drawn eyes, poorly drawn face, poorly drawn hands, signature, split frame, split screen, text, ugly, ugly, unreal, username, watermark, worst quality.
What I got:
“This doesn’t look like the Lincoln Tunnel, Sam.”
Also not a close-up foot shot in harsh midday lighting in the Scottish Highlands. And I didn’t even ask for an army; they must have been assembled from “silhouette”, “scattered elements” and “dynamic shadows”.
How to think about generative AI
You are the pilot of a ship capable of traveling the multiverse. The cockpit contains thousands of unlabeled buttons, switches, dials, and sliders. Think Tardis, but taken to 11.
You may adjust any number of controls before hitting The Big Red Button, and then you will be transported to a completely different universe, where anything can happen.
With me so far? Good.
Order matters. As you adjust each control, your destination in the multiverse shifts, so that each additional control you adjust applies its effect from a different starting point. Turn a dial too far in one direction, and your destination could be so far from home that slugs are the dominant species on Earth, and you can’t get back by pressing three green buttons and toggling a switch.
Still here? Awesome.
Here comes the real fun: each control on your board is actually wired up to ten thousand completely independent engines, all of which impact your destination in some way, big or small. In the first engine, that dial setting puts slugs in charge, but in engine #751, it puts giant breasts on cats. Not catgirls, cats.
By your side is a quirky robot with tentacles. Its job is to convert your spoken orders into control adjustments, but it doesn’t fully understand human language, has a dim grasp of each control’s effect, and guesses to resolve ambiguity. Because it has received contradictory orders from its makers, at random intervals it goes insane and assaults you with the tentacles; it never remembers these episodes.
(I let DALL-E handle this one, because I’m busy cleaning, baking, and wrapping presents for Christmas dinner tomorrow)
After the jump, a not-so-Christmas NSFW Miracle!
Easier with AI...
Magi-Lumière and GGO2 end tomorrow, leaving nothing but Shangri-La Frontier until the next season debuts. So I’ll play with Stable Diffusion over the holidays.
Pandering to the online-light-novel crowd:
That time I found a spaceship buried under a pile of rubble and rescued a sexy alien half-elf warrior princess only to discover that she has dozens of clone-sisters back home who show their gratitude with hot sweaty harem sex and what are you doing you’ll get that stuff all over the keyboard!
(seriously, there are dozens hundreds of scanlated “novels” that
have less thought put into them than this, just chapter after chapter
of painstakingly-detailed sex scenes written by virgins who’ve watched
a lot of streaming kink)
But first, a little present for Mauser. Careful, she bites!
Non-evil space princesses after the jump…
Cheesecake Vault, March 2019
Bad news:
The trailer for the new season of Doctor Disney+ is a steaming pile of nope. Pretty sure this isn’t the way to lure back the Doctor Who fanbase…
Good news:
I haven’t done a cheesecake roundup for a while, and kept nearly 20% of the month’s downloads.
Cheesecake Vault, February 2019
My working theory is that Japanese glamour photographers use wide-angle lenses because it’s the only time they get to stand so close to women. Well, that and it compensates for the flat asses.
(okay, so a lot of indoor shoots are done in hotel rooms that are tight on space, but they do it outdoors, too; I think I’m a lot more sensitive to perspective distortion than most, because clearly their audience doesn’t mind)
豊胸のフェルンちゃん
(that’s “Houkyou No Fern-chan”, for the kanji-allergic)
Fern’s full breasts are two of the most popular characters for fan-art this season, to the point that they can easily fill a full cheesecake post. Honestly, you’d think they were the leads, when they barely get any screen time at all.
Cheesecake Vault, January 2019
First pass through this month’s archives knocked out 3/4 of the pics; eliminating another 2/3 in the second pass brought it down to a reasonable size.
Waifu Drop
I may try to watch “stuck at level 1 with amazing loot rolls” for minor amusement, mostly for the bunny-suited bunnygirl. Other than that, the primary (ludicrous) twist on the usual isekai formula is not how Our Hero exploits gamelike abilities in a generic fantasy world, it’s the fact that everything in that world comes directly from dungeon drops. Including the air they breathe.
Between this and Vending Machine, I may laugh this season. I may also laugh at Atelier Thighza if the fan-service in the actual show is as over-the-stocking-top as the trailer; I’ll find out Saturday.
(unrelated bunnygirl is unrelated, due to the near-total lack of fan-art of this series that has 7 novels and 10 manga volumes…)
TotK Hylian Shield
Forget the clickbait articles (and there are many, most filled with obnoxious ads and dubious Javascript) telling you how to get to the Hylian Shield early in the game. You don’t need to eat a ton of stamina boosts and glide to the Serutabomac shrine, and you don’t need to go in through the Docks and avoid/fight Baby’s First Gloom Hands.
Enter from the ground, work your way around the spoiler under the castle toward the Northeast corner until you reach a section of wall due south of the top of the waterfall, and look for a cobblestone path leading directly into the spoiler, at (-0137, 1089, 0079). If you walk off the ledge at that location, there’s a cave entrance directly behind you, and you can just walk down a short flight of stairs, light the bonfire, and grab the shield.
If you already visited the shrine, you can also easily glide to this location by going around to the back and jumping off the southeast corner.
Now, if you want to get the Master Sword early, you’ll need two full stamina wheels, a wing with two fans (and maybe a battery if you don’t have much power, or a nice rocket shield), and a whole lot of patience, because the best way to reach it is to go to the southern tip of the island you woke up on and look to the east. The spoiler will first appear to the right of Zonaite Forge Island, moving left. It will make a 180° turn, and as soon as you have a full starboard broadside view, launch your wing and head for the nearest point. Once you’re on board, pick up a few souvenirs as you head to the front, and then draw the sword.
You might have to wait as long as two full hours of real not-in-a-menu play time for it to appear, so bring along a good book.
(or a bad book, I won’t judge you…)
Stupid Moderator Tricks
“Hey, if we mark ‘our’ subreddit NSFW, those assholes can’t sell ad space, and we can go back to business as usual! Of course, we’re still blocking search engine traffic, and now we’re keeping kids from asking questions about the most popular Nintendo Switch game of the year, but you’ve got to break a few omelots to lay an egg!”
What could possibly go wrong?
Bikinichain beats blockchain
