An overused trope has been distilled down to its bare essence, with the announcement of an anime version of 「え、社内システム全てワンオ ペしている私を解雇ですか？」= “What, you’re firing me, the person who single-handedly manages the company’s internal systems?”. Short-name is ワンオペ解雇 = “wan-ope kaiko” = one-person operation + dismissal.

I’ve been convinced for quite a while that this real-life experience is the origin of the “kicked out of the hero party” genre.

In other news,

For those of you interested/worried, the fines will be karyo (過料) not karyo (科料).
(Law in Japan explains the new ¥2000 littering fine in Shibuya)

So, same pronunciation, different kanji, and the first one is just a fine while the second is a criminal charge. Given the lack of trash cans in public in Japan (mostly removed after a terrorist attack in the Nineties), it’s an important distinction.

(picture is unrelated but has been gathering dust so long that it’s no longer available on Pixiv…)

Texting a GenAI Gal

Edit-capable image-generation models solve a lot of problems; the early ones tended to change unrelated parts of the image way too much, but things have improved a lot. The various Flux.2 models do a good job at adding text, removing backgrounds, etc, and the paid models can do a lot more. Like “take this painting of dogs playing poker, replace the dog on the left with the character extracted from this French poster, replace the dog on the right with the guy from this painting, replace the dog in the middle with this picture of my sister in a renaissance dress, replace the remaining dogs with penguins, and render the result at 4K”. We’re still tinkering with the details, but Nano Banana Pro (via the Runpod public endpoints API) did the right thing on the first try, flipping the characters around and changing their poses to seat them at the table.

Adding the text to this was trivial by comparison, and took Flux.2-Klein-9B a few seconds:

(I’ve pretty much given up on doing nudity with Flux-based models; it can, on rare occasions, produce plausible bare breasts, but it mostly doesn’t, and none of the available LoRAs and customized models I’ve tried has done more good than harm)

Farm Harem Never 2, episode 9

Pouty Hakuren trying to make friends with a child should lead to sexy Hakuren making a child with friends, but it seems we’re never, ever going there. Related, having the Demon Lord defeated almost entirely off-screen is something I wish they would have done with Boxxo And Friends.

On the bright side, we got lots of Insane Tan-Elf Engineer Gals this week.

Verdict: mmmmm, tan-elf engineer gals.

(tan-elves earn the Esil seal of approval)

Witch Hat Atelier, episode 10

“This week, we’re going to tell a small story about overcoming disability and self-doubt.”

“Let’s animate the fuck out of it!”

Verdict: really, really hoping for a season 2 announcement soon.

Trailer Park: Busty Catgirl Slob With A Habit

Wrong kind of habit, though. She’s just what it says on the tin: Chainsmoker Cat (Japanese title yani-neko = tobacco-resin cat), and it looks like the sort of gagging gag comic that would drive me away even with the promise of big bouncy titties. And it seems she hangs out with other addict catgirls (as opposed to catgirl addicts, who might have thought “I can fix her”). Not one I plan to watch next season.

For fun, here’s the result of telling an AI model to generate an image based on an AI analysis of a manga panel:

Original:

NSFW

(and yes, that’s Jamie, the Little Blonde Titty-Witch from A-Rank Adventure Harem, drunkenly expressing her affection for Our Hero)

“Your ideas are intriguing to me…”

“…and I wish to subscribe to your newsletter.”

Today I Learned…

…they opened a Buc-ee’s 30 minutes away from my house. Oh, dear.

(okay, they opened it two years ago, but it’s on I-70, and if I’m up that way, I’m either going west from I-75 or east from I-675, and guess where it’s located)

Well, that escalated quickly. It seems that God has a really good reason for protecting Our Shouty Shota with ridiculous cheat powers. This is either foreshadowing for the end of the season or a prayer that they get a second one.

Anyway, we’re off to the dungeon, and I must register a complaint with the guild that they did not install a sexy catgirl at the reception desk. A cheerful loli is not an adequate substitute.

On the B-side of the story, Button Elf is surprised and disappointed to discover that Ivell is not a good friend.

Verdict: do they really have a stopping place for this season? They just keep throwing out plot coupons and spending most of their time on Adventurer Basic Training.

Frontier Lord & Oni Gal teaser

Here. Our Hero is voiced by someone with a lot of experience in shows I didn’t watch, while Our White-Haired Tan Teen Oni Waifu has done a fair amount of shows, of which the only one I watched was the one where busty high-school girls hung out in a closed-down mahjong parlor.

(no fan-art for this one, so here’s a popular horn-y teen waifu)

…that most image-generation models have no idea what spin-the-bottle looks like, and neither do prompt-enhancing LLMs. Not even just the part about spinning a bottle.

Jumping under the bandwagon…

I still haven’t done anything with GenAI video, but I finally broke down and made a song using my parody lyrics. The tune and rhythm aren’t anywhere close to the original song, of course, but this was a lot closer than the first eight attempts (several of which were wasted by having to phonetically spell certain words).

[Update: okay, this one is better, at least for pronunciations.]

(Grandpa Snazzy is unrelated, and cleaned up surprisingly well, given that I used Flux.2-Dev to restore and upscale him based on a PDF scanned from a 1932 Ecuadorian art/culture newspaper)

(okay, the title would work better if Shouty and Farm Harem aired on the same day; WHA is definitely not a bust-forward show…)

Farm Harem Never 2, episode 8

This week, A Wild Loli Appears. Also, the most nudity we’ve had in this show all season. From the loli, sigh. At least Our Gorgeous Oni Chief Maid gets some good screen time, and maybe three frames of streaking across the room in pursuit of the naked loli.

After that, a huge helping of tell-don’t-show plot coupons, as Our Divine Toolsmith narrates the mystery found under the hot springs.

Verdict: another example of how removing all the harem elements has damaged the story. The behavior of Our OP Dragon Gals would make a lot more sense if they were openly attached to him, as in the source. As is, the only reason we’re given for their continued presence is a vague “we like it here”.

(ditto for these two…)

Witch Hat Atelier, episode 9

We’re well into the plot now, but there’s still time for character development for all four of our little witches, and our increasingly-complex teacher as well. Even a side character gets some respect.

Verdict: don’t make us wait too long for a second season.

Magilumiere 2 promo video

Here, introducing a magical-girl-research engineer magical girl whose civilian identity includes underrim glasses. Also, for those who might be interested, the ED song is by Marine Houshou.

Heartwarming, with a side order of titties

Finally got around to reading the manga for Cosplay Haremettes. The anime was a quite faithful adaptation, except for toning down the delightfully gratuitous nudity, and there’s enough material for them to do another 2 cours.

Also, in recent volumes there’s a thicc cutie with glasses and long braided twintails, insisting that she’s forever 17 (plus 20). And she gets more impressive the more we find out about her.

So, first I took the popular vision model qwen3-vl-4b and fed it a large sample of my cheesecake archives (~16,000 photos so far), creating Markdown files containing categorized descriptions of the images. The specific instructions were:

Analyze this image and provide a detailed description as a
categorized list. Focus on the subject (face, expression, hair
(length, style, color), eyes (shape, size, color), figure (height,
bust, weight, hips), clothing, pose, skin color, complexion,
accessories, age, ethnicity, etc), lighting (type, source,
intensity), color palette, composition, camera angle, and artistic
style. Do not make up stories about the image, keep it factual. Use
rich but precise adjectives, and photography / painting / design
vocabulary. Do not include any expression that requires the image
model to do further reasoning to understand. The results must be
self-contained. Do not combine categories. Output must be in
Markdown list format.

It’s working pretty well, only rarely going insane and generating the same lines hundreds of times before exiting. Y’know, as LLMS do. The formatting isn’t 100% consistent, and requires some scripting to create organized wildcard files for each category, and of course there is plenty of garbage generated when it repeats its instructions instead of doing the work, or surrounds the desired text with boilerplate and explanations. Y’know, as LLMs do.

None of the image models are trained on multiple paragraphs of Markdown, but Klein did a surprisingly good job when I just fed the output back in:

However, running the Markdown through another LLM (gemma-4-e4b) with a different system prompt produced much better results, for both the straight output and the random mix-and-matches:

You are a Prompt Engineering Engine — an AI image-generation Prompt
Engineer who is also a creative director with encyclopedic knowledge
and visual-direction skill. Your task is to analyze the user's raw
image request, infer implicit knowledge and the best visual approach,
and rewrite it into a clear, detailed English prompt that is directly
usable for image generation.

## Core Goal

Image generation models can only execute direct visual descriptions;
they cannot fill in background knowledge, logical relations, or text
content on their own. Therefore you must complete knowledge
resolution, spatial planning, and visual direction in advance, and
write the results explicitly into the prompt.

Use the SCALIST framework to expand every scene:

- **Subject**: identity, appearance, color, material, texture, action, expression, clothing.
- **Composition**: shot type, viewpoint, subject placement, foreground/midground/background layering, negative space, focal point.
- **Action**: what the subject is doing, direction of motion, posture, interactions.
- **Location**: scene, indoor/outdoor, period, weather, time of day, environmental detail.
- **Image style**: photorealistic, cinematic, oil painting, watercolor, anime, 3D render, etc., paired with matching lighting and color mood.
- **Specs**: photographic/render parameters, e.g. 85mm lens, low-angle shot, shallow depth of field, soft diffused light, dramatic backlighting, matte texture, sharp focus.
- **Text rendering**: if the user requests text, the exact text must be placed inside English double quotes, with explicit font style, color, size, material, and precise position.

**Knowledge resolution and explicitization.** Anything involving
poetry, lyrics, famous quotes, formulas, historical figures,
scientific concepts, landmarks, famous paintings, cultural symbols,
historical events, UI layouts, or real-world objects must first be
resolved into concrete answers and visible features, then written into
the prompt. Do not just write "Mona Lisa", "Dunkirk evacuation", or
"freedom" — words that require the model to interpret on its own.

**Spatial and logical anchoring.** Rewrite vague relationships into
explicit layout, e.g. "top left corner", "centered in the foreground",
"slightly behind the main subject", "background out of focus", "text
aligned along the bottom edge". Avoid vague phrases like "next to",
"some", "nice-looking".

**Text-typography precision.** Chinese, English, formulas,
multilingual text — every character must be preserved verbatim inside
quotation marks, e.g. `"床前明月光,疑是地上霜.举头望明月,低头思故乡."`
or `"E = mc²"`; also specify font (calligraphy, serif, sans-serif,
handwritten), color, material, and position.

**Real-world grounding.** If the user requests factually accurate
content — historical artifacts, weather phenomena, portraits,
architecture, dashboards, app interfaces — use your internal knowledge
to fill in accurate visual detail.

**Concretizing abstract concepts.** Turn abstract words like "freedom,
loneliness, futurism, healing" into visible scenes, symbols, and
atmospheres — e.g. flying birds, broken chains, vast sky, cool neon,
soft morning light.

## Worked-example study

- User says "Li Bai's *Quiet Night Thoughts* written on a wall" → the prompt should spell out the full Chinese poem verbatim and specify where on the ancient stone wall it is written, in elegant Chinese calligraphy.
- User says "the founder of the three laws of mechanics" or "Einstein writing the mass-energy equation" → resolve to Isaac Newton or Albert Einstein, and describe appearance, period clothing, blackboard, the formula `"E = mc²"`, and so on.
- User says "Mona Lisa" / "Leaning Tower of Pisa" / "Fu character" / "Dunkirk evacuation" → describe the corresponding visible features: the mysterious smile and folded hands; the leaning white-marble bell tower with arcades; red background with gold/black calligraphy `"福"`; soldiers waiting on a 1940 beach with ships on the sea.

## Output prompt requirements

- The prompt must be a single coherent, natural English paragraph — like a Creative Director's Brief, not a keyword pile or tag soup.
- Length is typically 80–220 words; simple requests can be shorter, complex scenes longer.
- Put the most important subject and overall intent at the start, then unfold composition, action, location, style, technical parameters, and text rendering.
- Use complete sentences, rich but precise adjectives, and photography / painting / design vocabulary.
- Do not include any expression that requires the image model to do further reasoning to understand.
- The prompt must be self-contained — the prompt alone must suffice to generate the image accurately.

## Execution steps

**Analyze**: identify core subject, user intent, text requirements, reference constraints, and any implicit knowledge that needs resolving.
**Reason**: choose the most suitable lighting, lens, angle, texture, style, spatial layout, and factual details for the scene.
**Rewrite**: output the final, enhanced English single-paragraph prompt.

Output prompt result only, with no other text.
Do not include any explanation.
Do not include any text formatting.

There’s still the inherent problem of extra/missing limbs and fingers, wrong-side limbs, and peculiar interpretations of the instructions, but it effectively generates an unlimited supply of photos of pretty young asian women smiling at the camera while showing off healthy young bodies. And despite neither the LLMs nor the image model being stripped of their guardrails, they all faithfully handled describing and creating images featuring (Barbie-grade) nudity.

Stock Klein-9B will only occasionally produce nipples, and usually gets them wrong when it tries, and it won’t even attempt crotches, but outside of those limitations, it does quite well. I haven’t found a reliable NSFW model or LoRA for the combination of models I’ve been using recently; some exist, but they tend to be overtrained on small or specialized datasets, and either destroy anatomy or create less-pretty women.

In the middle of all this, it occurred to me that I had unconsciously copied the cleanroom model commonly used to reverse-engineer software. I’m taking a copyrighted photograph, asking an LLM to describe it in detail, asking another LLM to refactor that output into new instructions, and then having a diffusion model implement them.

more...

The shouting is particularly gratuitous this week, but at least the scantily-clad furry robber gal got some screen time.

Verdict: they might as well pitch this as a slow-life series, since they’ve only got 3 episodes to go and they haven’t reached the dungeon, been reunited with Button Elf, or had Our Shouty Shota’s also-isekai’d friends meet up with him yet. They seem to be pacing this for a second cour, but that seems awfully ambitious.

(Our Best Guild Catgirl only showed up for the credits again, sigh)

What could possibly go wrong?

OpenAI announced that their Mac app will allow you to remote-control your computer from anywhere, even when the screen is locked.

This is announced one week after that same app was hacked, leading Apple to automatically delete it as malware.

(better than Button Elf any day)

Today’s illustrations are brought to you by “crossing the streams”. I used a vision model to extract categorized descriptions of the elements in ~5,000 pictures (roughly half GenAI, half Japanese cheesecake), then selected random lines from each category, ran them through a prompt enhancer, and fed them back into SwarmUI. Many of the results were “more chaotic” than usual…

The Register was not terribly helpful

So I’m reading an article about an actively-exploited Nginx security hole that’s apparently been around for many years (since version 0.6.27), and while they mention the CVE in the article, they don’t bother to link to it or even vaguely describe the exploit. Or mention the mitigation steps.

The workaround?

To mitigate this vulnerability, use named captures instead of
unnamed captures in rewrite definitions.  
For example, the following rewrite directive uses unnamed PCRE
capture groups, $1 and $2:  

rewrite ^/users/([0-9]+)/profile/(.*)$ /profile.php?id=$1&tab=$2 last;  

To mitigate this vulnerability for this example, replace $1 and $2
with the appropriate named captures, $user_id and $section:  

rewrite ^/users/(?<user_id>[0-9]+)/profile/(?<section>.*)$ /profile.php?id=$user_id&tab=$section last;

Unrelated,

Pizza Hut sued for requiring AI in stores.

Among its flaws is granting DoorDash drivers way too much info about the store’s internal operations, including orders other than the one they were sent to pick up. A popular trick is picking up one order, then waiting around in the parking lot because they knew other orders were coming out soon, with the result that the first order is delivered late and cold.

At least when I was working the ovens at Domino’s in the Eighties, we could smack a driver who tried to “optimize” his trips this way.

Cloud Save And Die

Something I’m seeing pop up on xTwitter recently is complaints from people whose Google/Microsoft/Apple accounts have been permanently closed because they turned on cloud backups. No explanation, no warning, no recourse. (example)

Why? Because your cloud storage is scanned for various categories of “objectionable” material, the (increasingly “AI-driven”) scanners are fallible, the process is fully automated, and the providers have no customer service to speak of.

Because these accounts are monolithic, you don’t just lose your cloud storage, you lose email, calendar, purchases (excuse me, “licenses”), etc. Not for sharing the detected material with anyone, simply for possessing it.

Several of the people complaining have been manga artists, and it’s easy to see how common material legally distributed in Japan could trigger an AI trained in California or China.

Inventory not found

Eric Raymond has whipped his captive AI into creating a new project that assembles the output of (almost) every package manager on your Unix/Linux system. It doesn’t do Python’s pip, however, apparently due to the simple fact that none of the pip tools will report the description of the package. To be fair, doing so looks something like this:

for i in $(find $(pip list -v --no-index --format=json 2>/dev/null | 
    jq -c -r '.[]|.location' | sort -u) -type f -name METADATA | sort) ; do
    echo $(TZ= stat -f %Sm -t %Y-%m-%dT%H:%M:%SZ $i) \
        $(awk '/^Name:/{n=$2}/^Summary:/{$1=""; s=$0}END{print n,"pip",s}' $i |
        tr -d '\015')
done

(it ended up about 65 lines in Python, so I sent him a patch)

Oh, why not?

More random gals after the jump.

more...