Boxxo Or Bust 2, last two episodes

I never felt motivated to watch last week’s grimdark horseshit episode, but I might as well embrace the sunk-cost fallacy and get to the end. Here goes…

Shoe Fetishist acquired his peculiar hobby because he was looking for the boots that turned up as the prize in the recent competition. He’d bought them for his childhood friend and adventuring partner, only to wake up after losing a fight to find her mutilated corpse, with legs and boots missing. Yeah, that’s a great story that fits right in with slapstick comedy and lighthearted cheesecake.

Naturally they find the villain responsible without any difficulty, and he and his pals come right out and confess while twirling their metaphorical Snidely Whiplash mustaches. It’s “not well written”.

Basically both episodes are full of near-constant whiplash and expository lumps, right up to the final post-credits scene in which they announce that they’ll continue telling the wrong story in a future season.

Verdict: yeah, no; the people responsible for this mess don’t deserve any eyeballs for another season.

Speaking of ‘bust’,

Miri Hanai in a cowkini:

Tag-a-licious

One thing that’s annoyed me with my blog is that I gave up manually tagging posts with keywords several thousand posts ago, and now I sometimes have trouble finding the quote or reference I’m looking for. So I’m letting an LLM chew on the problem. Specifically, seed-oss-36b, which has been doing a bang-up job inventing color-coordinated cheesecake pin-up costumes and poses, without the annoying wokescold moralization that Google and OpenAI’s offline models ram past your tonsils and gag you with just for using the word “sexy”.

The code is trivial; all the fun is in the system prompt and the user prompt:

Examining the “reasoning”, it clearly separates the TOML metadata header from the body content and parses both as requested, and creates output in the correct format. I don’t necessarily want to use all the tags, but by writing them out to a separate file for each blog post, I can go through with another script and pull out the most common and the most distinctive tags.

When it finishes. Which at ~2 minutes/post on the Mac Mini is going to take about a week. The 4090 would be faster… if the model fit. 20GB on a 24GB card is a bit tight.

Tagged ‘delicious’

Let’s see what sort of poses and costumes seed-oss-36B has been giving me to use in my wildcards. I started out at the usual 1024x1600 resolution, but since I’m making pin-ups, I quickly switched to a Playboy-centerfold aspect ratio at 896x1920.

But not before it fabricated a waifu pic I found so tasty that I not only ran it through the refiner and upscaler, I then ran it through the commercial Topaz Photo AI tool for even more cleanup and a 4x upscale (that did not run me out of memory!), bringing the final result to a silly-resolution 9216x14400.

I might even print her out as a real poster; she could hang on the not-Zoom-visible wall in my office:

(10 megapixels, 132 megapixels)

more...

I’d rather find a second waifu than write about evil right now…

(I found her in a bulk-wildcard-testing run where I set it to use the default sampler/scheduler, CFG=3.5, Steps=37, which took 1.5 minutes/image; cranking the steps up to 60 and using Heun++2/Beta improved quality at the cost of taking 6.75 minutes; refining it at 40% and upscaling it by 2.25x with 4xNomosUniDAT_bokeh_jpg eliminated the artifacts and significantly improved detail, but took 25.75 minutes. Worth it?)

(trying to do a 4x upscale ran out of memory after 45 minutes…)

Fiverr lays off 30% of employees…

…to focus on AI. Meanwhile, Fiverr freelancers are making a bundle cleaning up after AI.

🎶 🎶 It’s the cirrrrrrcle of liiiiiiife 🎶 🎶

Wow, Your Forma was worse than I thought…

Some reviewers made a fuss about the chemistry between the two leads being due to their voice actors being married in real life. Not any more.

Maybe she wasn’t willing to do the Sylphy voice in bed… 😁

(official art from the light novels, where she’s a busty little chibi; I’ve already used up all the fan-art, and I didn’t have any luck with the LoRAs)

Now that’s a knife dock!

OWC is selling a Thunderbolt 5 dock with 3x USB-A 3.2 ports, 1x USB-C 3.2 port, 4x Thunderbolt 5 ports (1 used for upstream), 1x 2.5Gb Ethernet port, and 2x 10Gb Ethernet ports with link aggregation support. For those who need their Mac Mini to have multiple 8K monitors and serious NAS bandwidth.

It’d be kind of wasted on me, so I’m thinking I’ll settle for the CalDigit Thunderbolt 4 dock: 4x Thunderbolt 4 (1 upstream), 4x USB-A 3.2.

Out of context

“Reasoning” models tend to do better at prompts like “give me a list of N unique action poses”, except when they dive up their own assholes and start generating reams of text about the so-called thinking process they’re following faking. I’ve had them spend 10+ minutes filling up all available context with busywork, without ever returning a single result. If they do return something, the list is often non-unique or cut off well before N.

But the context is the real problem, because if you only got N/2 results, you’d like to coax it to finish the job, but there’s a good chance that the “thinking” has consumed most of the context, and most models do not support purging context in the middle of generating a response.

It would also be useful to copy the setup of a known working chat and generate a completely different kind of list. I definitely don’t need the context of a search for poses and lighting setups when I ask for festive holiday lingerie ideas.

You can’t just reload the model with larger context and continue the chat. You can’t fork the chat into a new one; that preserves the context, too. What you want is to wipe all answers and context and preserve the settings and your prompts, so you don’t have to set up the chat again. In LM Studio, this functionality doesn’t seem to exist.

So I wrote a quick Perl script to load the JSON conversation history and remove everything but your prompts.

#!/usr/bin/env perl
use 5.020;
use strict;
use warnings;
use JSON;

open(my $In, $ARGV[0]);
my $json = join(' ', <$In>);
close($In);
my $conv = decode_json($json);

# remove token count
$conv->{tokenCount} = 0;
# rename
$conv->{name} = "SCRUBBED " . defined $conv->{name} ? $conv->{name} : "";
# remove all messages not from user.
my $out = [];
foreach my $msg (@{$conv->{messages}}) {
    push(@$out, $msg) if $msg->{versions}->[0]->{role} eq "user";
}
$conv->{messages} = $out;
print encode_json($conv);

This dumps a fresh JSON conversation file to STDOUT, which can be given any filename and dropped into ~/.lmstudio/conversations.

Why Perl? Because I could write it faster than I could explain the specs to a coding LLM, and I didn’t need multiple debugging passes.

(BTW, 32K context seems to be a comfortable margin for a 100-element list; 16K can be a bit tight)

Digits and Shadows

It’s kind of amazing how the “state of the art” keeps advancing in AI without fixing any of the known problems. LLMs hallucinate because they must, and image-diffusion engines can’t count fingers or recognize anatomical impossibilities because they don’t use anatomy. All the alleged coherence you hear about in AI output is either sheer accident or aggressive post-processing.

Let’s examine a few types of Qwen Image failures:

“Thumbs, how do they work?”

Trying to put your own hand into a position seen in a generated image can be awkward, painful, or simply impossible if it’s on the wrong side.

more...

Call Of The Night 2, episode 11

Well, something just happened. With one episode to go, they really can’t deeply explore this new development, but at least everyone will live to see another Night.

Verdict: plot, with occasional animation. You know the conflict is over when they drop last season’s OP song into the mix.

Kaiju No. 8 2, episode 9

Our Mighty Tsuntail is back! Only in a supportive supporting-character kind of way for now, but this is a combination “must get stronger” and “everything’s better with friends” episode, in which Our Monstering Hero reconnects with his comrades and his childhood crush, while the world goes mad with the results of Naughty Number Nine’s latest efforts.

Dungeon Chibis rewatch

Crush-chan is less annoying the second time through. Naginata Gal is Best Girl by a huge margin. If new seasons keep sucking, I’ll be doing a lot of rewatches like this for a while.

Speaking of which, the three seasons of Dog Days never got a US Bluray release, but subbed Bluray rips are still torrentable. This is another show where everyone involved clearly loved what they were making.

(“pet me, you stupid hero”)

There would need to be explanations…

…if these showed up in the background of a Zoom meeting:

My office just got bigger

The one at home, that is, which was overcrowded with the old Ikea 31-inch-deep tables I brought from my old place and finally replaced. I didn’t use them as desks in my California house, but the desk set I had there went to Goodwill instead of onto the truck, so the tables were pressed into service.

I hate to just hand out a “that’s what she said” opportunity, but it’s amazing the difference seven inches makes. The room just feels so much bigger with 24-inch-deep tables. Also Ikea, but the tops are solid bamboo plywood instead of the old veneer-over-particle-board. They’re also 7 inches longer than the old tables, so there’s more room to fit between the legs. (coughcough)

I also picked up two bamboo side tables (1, 2) for a printer stand and a spare workspace. All of it had typical Ikea assembly hardware, but the bamboo is head-and-shoulders above their usual construction material. Pity they didn’t have the matching bamboo monitor stand in stock, because I’d have bought two of those to free up even more desk space.

Sadly,

My M2 MacBook Air simply does not support 2 external monitors at once. Can’t be done without buying a DisplayLink adapter and installing their software driver (remember DisplayLink? Turns out that’s still a thing). So I can either use the really nice dual HDR portable monitors as one display with a 1-inch gap in the middle, or move them over to the M4 Mini. Since I want the HDR displays for photo editing, that means moving all my photo archives and workflows over sooner rather than later, but not today; between emptying the office, scrubbing it floor to ceiling, hanging pictures, assembling furniture, and moving everything back last night, I’m just a tiny bit tired and sore.

The day Google AI Overview actually worked

I feel it necessary to call this out, because it’s been wrong so many times that I usually automatically scroll past it. I only read it today because it was 6:30 AM and I hadn’t had any caffeine yet.

Anyway, after upgrading to MacOS “Sequoia” recently, I discovered the annoying new system-wide Ctrl-Enter keyboard shortcut. Because my fingers have decades of training in Emacs keybindings in the shell, I often repeat the previous command by hitting “Ctrl-P, Enter”, but my pinky often stays on the Ctrl key. Now Ctrl-Enter pops up a “contextual menu” everywhere, unless you disable it in Settings -> Keyboard -> Keyboard Shortcuts -> Keyboard -> “Show Contextual Menu”.

I was astonished that Google actually gave me a correct answer for once. Shame they didn’t credit the web site they stole it from.

(actually, everything on this particular panel should be disabled, IMHO, and quite a few others as well)

Qwen Image LoRAs not ripe yet

I’ve yet to find one that works as advertised. Either they visibly degrade image quality, or they just do… nothing at all.

My Cyber Princess Waifu continues to amuse, though.

more...

(we now resume our regularly scheduled randomblogging, until the next slaughter of innocents by Left-wing domestic terrorists)

Boxxo Or Bust 2, episode 11

This is really not a good week for the show to go grimdark, revealing Our Shoe Fetishist’s bloody, revenge-fueled path. I’m just not going to watch it right now.

Here’s something wholesome instead:

Warner Bros wants to be the Broadcom of streaming

CEO insists HBO Max should be priced like diamonds. Six customers are so much easier to support than six million.

Not a fan

I bought a new unmanaged 10Gb switch, because my existing one only had 2 10Gb ports with the rest being 2.5Gb. Three years ago, the price of a switch with 8 10Gb RJ45 ports was a bit too spicy, but now it’s just annoying.

What really hurts is when a desktop switch has a fan loud enough to be heard from the next room. And I need it to be in the same room. For comparison, I barely notice the Synology NAS that’s on the same desk, even when it’s doing a RAID scrub. For the sake of my sanity and productivity, I guess I’ll be building the switch a padded crib…

New Monitor(s)

The new monitor is an “ASUS ZenScreen Duo OLED 360° Foldable Portable Monitor (MQ149CD)”, which is quite a mouthful, but the TL/DR is that it’s two 1920x1200 HDR panels that can be used separately or together, portrait or landscape. To use it as two separate monitors, you need to use two cables, but despite the one-inch gap between the panels, you can still get good use out of it as a single 1920x2400 display.

I’m using the 1/4”-by-20 tripod screw to mount it higher than the kickstand, and despite all claims of portability, yeah, you have to plug it in to drive it properly (1 of the 3 USB-C ports is power-only). Right now I’ve just got it hooked up with HDMI, because my current dock was low on free USB-C ports, but once I get things rearranged I’ll separate them, because Lightroom’s window layout is a bit inflexible, and the one-inch gap is bad for photo editing. Better to use the application’s “secondary display” support.

This does make me want a vertical 20+ inch 4K HDR portable monitor, though. I just don’t want it to be from a company I’ve never heard of before, which is what’s all over Amazon.

“Shared memory and fancy NPUs are no match for a good graphics card at your side, kid.”

While I’m discussing new toys, I’ll mention that I also picked up a refurbished M4 Pro Mac Mini, which in theory has much faster “AI” performance than my M2 MacBook Air. With 64 GB of RAM, it can run mid-sized offline LLMs at a decent pace, and even do quick image captioning if I downscale raw images to a sensible resolution.

What it can’t do is more than very basic image generation. It’s not the memory, it’s the cores: not enough of them, and not fast enough. This matches what I’ve heard about the insanely-pricy Mac Studio and the merely-way-overpriced systems built on the AMD Ryzen AI Max+ 395. Their performance comparisons are based entirely on running models that simply don’t fit into a consumer graphics card’s VRAM. “Our toaster is much faster at running this 48 GB model than a 24 GB Nvidia 4090!”

The Mini is a huge speedup for Lightroom and Photoshop, though, which made it worthwhile for me. Any “AI” playability is a bonus.

(I have never seen a human being hold a camera this way)

Related, “Dear Draw Things programmers”

Come back when you discover the concept of legibility. Dipshits.

Accidental Disney Princess

I was testing a few more LLM-enhanced dynamic prompts, and one in particular stood out: a series of very consistent images of a fresh-faced young gal who could make Bambi’s woodland friends sit up and beg.

4K resolution, crisp and highly detailed, create an illustration that exudes a high-budget aesthetic. Depict an average height, lovely ethnic Egyptian woman who is 18 years old with a petite figure. She has deep blue eyes, lobed ears, a straight nose, a wide chin, an U-shaped jaw, dimpled cheeks, and a flat forehead. Her heart-shaped face highlights her gentle features. Her skin is healthy and alabaster white, with festive holiday makeup that complements her almond-shaped eyes and full lips. Her hair is steel gray, styled in a cybernetic pixie cut with metallic edges and glowing circuit-like patterns. The woman has a happy expression as she stands with one arm raised, bathed in luminous waves of light. The background features a lush jungle filled with singing flora, glittering with raindrops under the radiant colors of a rainbow sun. Subtle dawn light filters through, creating an aerial perspective with open framing and pastel tones that evoke a gentle awakening. The composition highlights her graceful form against the vibrant greenery, capturing a dreamlike atmosphere.

Sadly, the prompt that worked so well in Qwen Image was… less successful with other models that don’t have an embedded LLM to parse complete sentences sensibly, and failed completely with a few that couldn’t handle the combination of the Beta scheduler and the Heun++ 2 sampler (I may redo those with their defaults).

It’s not that the others all failed, they just didn’t produce consistent characters who looked like my princess and/or lived in her enchanted forest.

more...

Everyone on the Left, raise your hand if you’re not a murderous savage. Now look around your peer group and count the lack of hands, then come over and rejoin civilization. There might still be time.

Those who didn’t raise your hands? Do not expect civilization to protect you any more.

(no cheeky anime picture, out of respect for the victims)

(…that’s not another season of Frieren or Apothecary Diaries)

The World’s Strongest Rearguard anime

I consider this to be one of the best cheat-power isekai light novel series. Our Ass-Guardian Hero reincarnates into a fantasy world that basically forces all the newcomers into dungeon crawls without any real choice or training, but while he’s being railroaded by a hot chick with big boobs, he accidentally chooses a rare class. Everyone else is trying to be things like “valkyrie”, “soldier”, “forager”, “tennis coach”, etc, but hearing that parties are always looking for someone to take the rearguard position, he just writes down “rearguard”.

And he gains the ability to buff any party member who is in front of him. It should come as no surprise that he ends up with a party of gorgeous gals, but it might be surprising that he is a complete gentleman who doesn’t so much as stare at the asses he’s guarding. He’s not oblivious to their (mostly-underage) charms, but he never, ever goes there.

What makes it work is that despite him being potentially and eventually overpowered, he and his party have to work for every fight, and their trust and teamwork keep them alive just as much as their skills. His power gives them an edge, but his decency and kindness have just as much of an impact on the world.

Add to that that every not-quite-haremette has a distinct personality and goal, and you have something that I’m begging the animators not to screw up.

The buried lede in this article is that book nine is finally coming out after a four-year wait. Book 8 wrapped up a major story arc, but left a lot of character arcs unresolved, so woo-hoo!

(Our Hero is in the background, with His Formerly Abusive Former Manager on the left, Our Terrifying Teen Swordsgal on the right, and Our Best Girl Rescue Lizardgal in the middle; and the butterflies are seriously bad news, with their presence suggesting two cours)

Just-Outta-Time Logistics

I bought an HDR monitor on Amazon, which was supposed to arrive Monday. At 4pm, they announced that for unknown reasons it never left an Ohio warehouse, and it would instead arrive Tuesday. Three hours later, that message disappeared, and the order status reverted to arriving-Monday-but-not-shipped-yet. It stayed like that until after 10pm, when it went to “delayed, not shipped” again, with no further explanation.

In the wee hours of the morning, Amazon’s entire order-history system apparently went offline, and all status was unavailable.

As of 8:30 AM, it is allegedly across town, waiting to get put on a truck. Unless they erase that status update.

Unrelated,

FedEx grudgingly coughed up an approximate delivery time for Tuesday’s signature-required package (not from Amazon), which is good, since I have to pick up my niece after school, and I’d hate to have to go chase down the package at their nearest office after the truck returned to base.

Amusingly, when I asked for details, they offered two confirmation methods: SMS to my (correct!) cellphone number, or email to a masked address that I didn’t immediately recognize. It would have reached me anyway, since I own the domain, but it took me a moment to figure out that they’d decided that the email address associated with my physical address was the one I’d created specifically for B&H Photo.

Because they shipped me a CompactFlash Express card (fastest storage for my 4k-video-capable Sony A7SIII) via FedEx a while back.

Mistral Small is gaaaaaaaay…

Given the following picture of Yuzuha Hongo:

And the relatively-capable vision-enabled LLM mistral-small-3.2, we get the following caption and tag set that contain a rather serious error:

A young woman with long brown hair, straight hair style, and a slim body type is standing in a bathroom. She has fair skin, small breasts, and is wearing a pink floral bikini top with thin straps. She has a slight smile on her face and is looking directly at the camera. The bathroom features white tiles, a shower head, and a towel rack with towels hanging on it. Water droplets are visible on her skin.

Tags: young woman, long brown hair, straight hair style, slim body type, fair skin, small breasts, pink floral bikini top, thin straps, slight smile, direct gaze, white tiles, shower head, towel rack, hanging towels, water droplets

(ran it a few more times, and got completely different results every time; sometimes it correctly handled her boobs (coughcough), other times it decided her nipples were visible, several times it hallucinated a towel rack with towels, occasionally it remembered to mention “Asian facial features”, etc, etc; honestly, I trust the output so little that I’m tempted to have my script loop over the same picture N times and keep only the tags that show up at least 50% of the time; I’m also going to borrow the styling of the Qwen prompt-enhancer below)

Enhanced dynamic prompting

With my new still-in-progress dynamic prompt collection, I’m turning this:

__style/photo__ of a __var/prettygal__ with __skin/_normal__ and __hair/_normal__, with a __mood/_normal__ expression. __pose/sexy__. Her location is __place/_normal__. __scene__.

into prompts like this:

4k, crisp, high budget, highly detailed, intricate, ultra textured. Create a photograph of a average height, beautiful, ethnic American, adolescent woman with supple figure, Steel gray eyes, softly-curled lobe ears, Angular Nose, V-Shaped Chin, Strong Jaw, Angular Cheeks, Angular Forehead, Triangle face shape, Natural “no-makeup” makeup with healthy Sun-Kissed-Tan skin and Natural Gray hair, Low twist with braids, creating texture and detail, with a warm expression. Standing with weight on one leg, other foot planted flat, arm crossed over chest, wrist aligned, chin up. Her location is Grand Baie, Mauritius. Rainy glow, close framing, layered reflection, muted palette, melancholic ambiance.

Then using LM Studio’s Python API to pass it through qwen2.5-7b-instruct, with this (taken from the demo code in their repo):

And getting enhanced prompts like this:

4K resolution, crisp and highly detailed, with a high budget aesthetic that emphasizes intricate texture. Create a photograph of an average-height, beautiful ethnic American adolescent woman with a supple figure. Her steel gray eyes sparkle softly, and her ears are gently curved. She has an angular nose, a V-shaped chin, a strong jaw, angular cheeks, and an angular forehead, forming a triangle face shape. Her natural “no-makeup” makeup highlights healthy sun-kissed tan skin, while her naturally gray hair is styled in a low twist with braids that add texture and detail. She stands with weight on one leg, the other foot planted firmly, and her arm is crossed over her chest with wrists aligned. Her chin is lifted slightly, giving a warm expression. The setting is Grand Baie, Mauritius. The photo captures a rainy glow, using close framing to create layered reflections in the background, maintaining a muted palette that evokes a melancholic ambiance.

Now, let’s see what sort of picture that produces…

more...

So it turns out that not only does conflict build character, but it works the other way around, too. The Power Of Friendship wins the trial, and in the process we get a pretty-well-incorporated flashback that fills in some more details on Our Mighty Tsuntail’s mom. Pity she wasn’t there for it, but at least she shows up for a minute or two, continuing the offscreen path of “I must get stronger”.

Verdict: if this weren’t so good, I’d really hate the trope abuse.

(“🎶 1. cut a hole in a box 🎶”; unrelated, of course)

Power sucks

My MacBook Air is running out of power while plugged in, despite not even being warm to the touch. The Anker dock, on the other hand, is blistering hot. The CPU is barely ticking over, and Apple’s battery monitor even claims “No Apps Using Significant Energy”.

The culprit, of course, is “AI”. I’ve got LM Studio running a prompt-enhancement prompt with qwen2.5-7b-instruct, taking the output of my dynamic-prompt-generation script and improving it (which it’s actually pretty good at). I had to find the much-higher-wattage original Apple charger, which I’ve never needed to use before. It turns out there’s an “energy impact” field reported in Activity Monitor, but it doesn’t charge GPU/NPU usage to the correct app.

Apple’s so excited about moving fast and breaking things with “AI” that they can’t spare the engineering talent to update their tools to even report how much power LLMs are using on their products.

In fact, when you sort Activity Monitor by “energy impact”, LM Studio is at the top despite having a “score” that’s less than 10% of what Activity Monitor claims for itself. The only field that seems to be honest is “12 hr power”, which puts LM Studio an order of magnitude above the rest. But only because it’s been running for several hours, sigh.

(like an iceberg (or “Apple Intelligence”), 90% of Froppy is underwater…)

More “stimulating” covers…

“We want YOU to join the Star League!”

more...

Our Great (Hot) Detective’s plan is revealed, as Our Wannabe-Sucker Hero confronts her and lays out the evidence. Unfortunately he didn’t plan for the aftermath.

Verdict: it was nice to see Nazuna go Maximum Effort and prove that Anko/Kyoko’s vampire-fighting tactics only really worked when her opponent wasn’t serious, neatly subverting the waif-fu girlboss stereotype that’s so common. She can’t beat a vampire in a straight-up fight, no matter how clever she is.

Text generation in Qwen Image

The majority of my failed attempts to generate text have been with relatively-rare words (“harem”, “Hornblower”, etc), so I was a bit surprised when it failed 90% of the time at the word Babies (as in Sugar), while Baby always worked. Then I found an article explaining that the way they got improved text generation in both Latin scripts and Hanzi/Kanji was to explicitly train it with tons of images of words rendered in specific fonts.

In other words, it will almost never produce correct results for words outside of the training set; it would need to integrate a text-rendering engine the way the big boys do. Given that they already bolted a full LLM into it to parse the prompts, that’s doable, but they’d probably have to open up their license to use something like Freetype or ImageMagick. This may be something where I end up creating automated vocabulary LoRAs with a Python script, or asking a high-end LLM to generate masking templates for the image generator and ImageMagick command-lines to composite the text onto the results.

Anyway, I was thinking about that social-media “boys don’t read” kerfluffle, and decided the thing to do was create some things that might stimulate their sense-a-wunda…

BTW, this is not what I had in mind when I asked for “glowing alien symbols”…

more...