[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / s / t / u / v / vg / vm / vmg / vr / vrpg / vst / w / wg] [i / ic] [r9k / s4s / vip] [cm / hm / lgbt / y] [3 / aco / adv / an / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / pw / qst / sci / soc / sp / tg / toy / trv / tv / vp / vt / wsg / wsr / x / xs] [Settings] [Search] [Mobile] [Home]
Board
Settings Mobile Home
/g/ - Technology


Thread archived.
You cannot reply anymore.


[Advertise on 4chan]


File: CS civwife.png (1.56 MB, 981x985)
1.56 MB PNG
Previous Thread: >>109111831

>Links:
>DALL-E 3 and GPT-Image 1/1.5
https://www.bing.com/images/create
https://designer.microsoft.com/image-creator
https://copilot.microsoft.com/
>GPT-Image 2
https://chatgpt.com/
>Imagen 4 and Nano Banana (Pro/2)
https://gemini.google.com/app
https://labs.google/fx/tools/flow
https://aistudio.google.com/prompts/new_image
>Midjourney [PAID]
https://www.midjourney.com/home
>Meta AI (txt2img based on MJ):
https://ai.meta.com/
>NovelAI [PAID]
https://novelai.net/image
>Perchance
https://perchance.org/ai-text-to-image-generator
>AI videos
https://pixverse.ai/
https://hailuoai.video/

>Instructions and download for BIC EDIT script:
https://rentry.org/bicedit

>Prompt Generator
https://tipseason.com/dalle-prompt-generator

>DALL-E 3 image generation guide
https://files.catbox.moe/o4yihf.png
https://files.catbox.moe/320rm1.png
https://files.catbox.moe/8gge6o.png
https://files.catbox.moe/dcokjf.png

>Perchance generator list
https://pastebin.com/XRGqaW4W

>Related Boards
>>>/vg/vig
>>>/co/dall-e
>>>/trash/aibf
>>>/aco/dall
>>>/tg/slop

Sometimes the slop is just intensely beautiful.
>>
>>109140139
whoops, forgot to delate imagegen 4 entirely since it's apparently Kill.
>>
>>109140143
Oh well, it's fine. We can delete it whenever it will be, I might be the only one autistic enough to care. Thanks for the bake. I've linked it back in the previous thread.
>>
Anyways, anyone else using Capcut to gen stuff and having most of the gens use the default slopper instead of GPT image 2?
It can still make me stuff like comics and the trademark gpt_image 2 areoale so it's entirely on the frontend.
Worked fine a few days ago but not it's being shit for no reason. So far only telling it to "regenerate using GPT image 2" several times seems to work.
>>
>>
File: GPT image 1.png (2.22 MB, 1122x1402)
2.22 MB PNG
An atompunk hyper-realistic portrait of a comely young woman (pale skin complexion, doll-like face, tied light golden brown hair, skinny, blue eyes, rockabilly fashion, headphone) sitting in a 60's diner, sunset. Slightly blushing, smiling softly to viewer. The art style blends semi-realistic anime aesthetics with high-end digital painting, clean lines, detailed textures, metallic hues and neon accents, dynamic angle. Peaceful&intimate atmosphere.
>>
File: GPT image 2.png (2.34 MB, 1122x1402)
2.34 MB PNG
A steampunk hyper-realistic portrait of a comely young woman (pale skin complexion, doll-like face, tied light golden brown hair, skinny, blue eyes, copper goggles) standing inside a retrofuturist laboratory. The art style blends semi-realistic anime aesthetics with high-end digital painting, clean lines, detailed textures, sepia tones and metallic hues, dynamic angle. masterpiece.
>>
File: GPT image 3.png (2.4 MB, 1122x1402)
2.4 MB PNG
A post-apocalyptic hyper-realistic portrait of a comely young woman (pale skin complexion, doll-like face, tied light golden brown hair, skinny, blue eyes) distressed camo attire) roaming around the ghost town. The art style blends semi-realistic anime aesthetics with high-end digital painting, bold lines, detailed textures, earth tones, sepia, and vivid accents. Nostalgic yet dramatic atmosphere.
>>
File: GPT image 4.png (2.3 MB, 1122x1402)
2.3 MB PNG
A dieselpunk hyper-realistic portrait of a comely young woman (pale skin complexion, doll-like face, tied light golden brown hair, skinny, blue eyes, aviator goggles and leather clothing) standing near 40's retrofuturist interceptor aircraft at aerodrome. The art style blends semi-realistic anime aesthetics with high-end digital painting, clean lines, detailed textures, earthy tones and metallic hues, dynamic angle.
>>
File: GPT image 5.png (2.51 MB, 1122x1402)
2.51 MB PNG
A solarpunk hyper-realistic portrait of a comely young woman (pale skin complexion, doll-like face, tied light golden brown hair, skinny, blue eyes, baggy dungarees, headgear) gardening exotic plants in her forest near her crawler vehicle, sunny day at summer. The art style blends semi-realistic anime aesthetics with high-end digital painting, clean lines, featuring subtle shading, detailed textures, bright and saturated colors. Optimistic atmosphere.
>>
Is it just me, or do they have the same face?
>>
File: GPT image 6.png (2.24 MB, 1536x1024)
2.24 MB PNG
Vintage noir art meets 1980s anime, dramatic chiaroscuro lighting, visible expressive brushstrokes, painted cel-shaded texture. A detailed Caucasian man in tactical outfit, intense focused expression, holding a Glock 17 handgun with both hands, POV directly from the front of the handgun’s muzzle, extreme foreground barrel distortion, smoky alley atmosphere, rain-slick reflections, high contrast shadows, retro anime linework, cinematic composition, moody suspense, film grain, muted teal and amber palette.
>>
File: GPT image 7.png (2.45 MB, 1536x1024)
2.45 MB PNG
Serene winter boreal forest at sunset, snow-laden ancient evergreens in the foreground and midground, distant jagged mountains silhouetted against the fading light, sun setting behind the peaks casting a soft glow across the snow. Bucolic, peaceful, dream-like and nostalgic atmosphere. Lo-fi sumi-e style with expressive, layered, calligraphic brushstrokes in the distinctive ethereal style of Yoshitaka Amano, minimalist composition, soft muted pastel palette of lavenders, pale pinks, sage greens and cool blues, strong confident black ink outlines, delicate linework, atmospheric depth, highly artistic, tranquil and magical
>>
>>109140365
Well, description of the characters is the same and the basic style is the same "hyper-realistic portrait", "semi-realistic anime aesthetics with high-end digital painting". But yeah, gpt-image-2 isn't immune from same facing. I was noticing the same with the skunkgirls in the previous thread. I think it just has less embedded beautification in comparison to 1.5.
>>
>>109121520
>>109139083
>>109140204
Nice.
>>
>>
>>
File: Jun 26, 2026, 12_26_10 PM.png (2.6 MB, 1050x1498)
2.6 MB PNG
>>
Anyone still uses DALLE?
>>
>>109136569
>...without a reference image that is already huge...

So I should attempt to make a template I can alter.
A laboratory experiment in progress...
>>
>>109141361
For a quick job, it could have turned out worse.
>>
>>
>>
File: stupid.jpg (1.3 MB, 1402x1122)
1.3 MB JPG
>>
File: 202606261437.jpg (445 KB, 896x1200)
445 KB JPG
>>109141688
Quite round.
>>109141361
>template
hm...
>>
File: OC by karepack, mutated.jpg (441 KB, 768x1376)
441 KB JPG
>Standing view of character from <image_0>. Her bust has grown unrealistically gigantic, similar to or larger than the body proportions from <image_1>.
idk about Gemini the app but it works in Flow
>>
I'm seeing the upscale feature failing a lot on Flow today.
>>
>>109142437
wdym, I don't think it has ever worked in a usable way
>>
>>109142133
Math doesn't check out.
>>
>>109142526
I'm not saying it's good/bad, I mean that it literally says "Upscaling Failed!", as a red error message, over and over again and it doesn't give me the upscaled generation; on my end that was a quite rare/uncommon error before today.
>>
>>
>>109142716
What model, out of curiosity? It looks like it's neither Gemini/NBP/2 nor ChatGPT/image-2. It's nice, anyway.
>>
File: kwabl.jpg (107 KB, 792x748)
107 KB JPG
>>109142771
>>
>>
>>
File: catto.jpg (1009 KB, 888x1184)
1009 KB JPG
Fagposters where u at
>>
File: blondie.jpg (821 KB, 888x1184)
821 KB JPG
>>
>>
new the wolf male would show up over in other boards posting and being a troll. comical it cant just troll here.
>>
I don't think AI is ready to take any jobs.
>>
I shouldn't let the AI write the script like this, but it's amusing to see what it comes up with. It has a bias about how a bee hive would behave.
>>
File: pistol.jpg (598 KB, 896x1200)
598 KB JPG
>>
File: 1753305630729138.jpg (2.45 MB, 3392x5056)
2.45 MB JPG
>>
File: 1762102740604569.jpg (1.45 MB, 3392x5056)
1.45 MB JPG
>>
>>
>>109140244
pls respond
>>
File: lul.jpg (296 KB, 1024x1024)
296 KB JPG
I'm gonna miss this ridiculous crap so much.
>>
>>109146333
are they finally shutting dalle down?
>>
>>109141361
>>
>>109146463
Yes, Bing Image Creator is the last one standing, and it's already slated to be replaced. Some users already can't access it.
>>
>>
>>109146470
My hypothesis that made me stick to mostly off-angle shots, is that the image generators might be losing dimensionality by relying on full front and full back shots. Even if not intentionally coded to do it, it's not hard to turn a full front image into washboard girl.
>>
>>
>>109142771
it's gpt-image-2
>>
>>109147412
Thanks. It weirdly doesn't get identified by the verifier. Did you strip the SynthID watermark? I understand it's possible, but the only solution I've seen was paid.
>>
the furthest I got not getting policy filtered
>>
>>
>>109147602
I don't do anything to the images. I make them in flowith. Maybe flowith strips the watermark. I set the quality low, maybe the low quality model doesn't have a watermark?
>>
>>109146905
>losing dimensionality by relying on full front
That makes sense. I had a similar thought some time later, hence the side view. Straight rear is especially meaningless when it isn't showing anything.
>>
>chinfag is back, this time "trying" to be more subtle
>>
>>109148325
>chinfag
Who?
>>
>>109148494
ignore him. tons of people on 4chan have mental illness and make up fake "super villains" in their heads to stalk and villain because their spergtism got triggered over a word or , or something the poster did that somehow sullied the mentaltism's reality so bad he went into batsy who stalks mode.

don't even bother trying to understand it and just mock it for the bullshit it is.
everyone here on this site is coocoo for cocoa puffs and in various tiers and degrees from harmless to oh shit do not post when they are show up at your home and fuck your corpse deranged. eveh had killers and others post in the past.
>>
>>109147872
The same is true for Capcut's gpt-image-2 gens, I've just checked it out of curiosity. Weird.
>>109148325
I second the "who?". The term gives me literally one result before today (and it was back in 2021, pre-de3).
>>109148605
>tons of people on 4chan have mental illness and make up fake "super villains" in their heads to stalk and villain because their spergtism got triggered over a word or , or something the poster did that somehow sullied the mentaltism's reality so bad he went into batsy who stalks mode.
I guess it's pointless to even ask whether you have any degree of self awareness.
>>
>>
>>
>>
>>109148827
right back at you and your very obvious lack of self awareness while you preach at others who are very aware. comical fun you cunt and you assume I dont consider myself coocoo for cocoa puff in ways. how cliche of you.

pic related to way to many idiots here on 4chan
>>
>>
>>109149681
Yeah, totally aware and totally not triggered "because their spergtism got triggered over a word" (*preach*) and completely not making up "fake "super villains" in their heads to stalk, even posting the parody in boards where I don't regularly post, showing a level of obsession that goes beyond the mere trolling. I guess you did answer my hypothetical question, indirectly. Thanks, I guess.
>>
File: Candy_BodyModelSheet.jpg (2.04 MB, 2048x2048)
2.04 MB JPG
>>
>>
>>109150621
Man, those are sad, narrow hips
>>
File: 1756035399218622.webm (2.88 MB, 1280x720)
2.88 MB
2.88 MB WEBM
>vgh... what we've lost...
>>
>>109149797
lol, lmao stay big trig kid. I was posting on /tg/ in 2023 so piss off with your big larp you're stalked and not showing up where Ive posted for years. try again with your pretend.
>>
>>
>>
>>
>>109140139
>https://labs.google/fx/tools/flow
Are there any semi-automation scripts for flow? For example: make 1 gen with NBP and 3 gens with NB2 or make two horizontal gens and two vertical gens?
>>
>>
>>109149681
>>109149797
>>109150935
>>109150942
Is it too much to ask that you pissbabies stop shitting up these threads with your pussyfighting?
>>
File: downgrade.jpg (229 KB, 1178x620)
229 KB JPG
I miss Dall-E 3, bros
>>
>>
>>109152204
You're proving my point that you're a pissbaby.
>>
>>109152204
>I'm not the problem, everyone else is.
>>
>>
>>
>>
>>
>>
>>
>>109152072
The irony in the whole thing felt too strong to ignore, even if I knew that I was being largely rhetorical. My bad.
>>
>>
>>109153111
>>109153364
model and prompts?
>>
>>109153598
Nano Banana Pro via Flow:
https://pastebin.com/DHL87DZa
https://pastebin.com/5RgMUNJ5
>>
BIC has a completely redone interface. So BICEDIT is 100% incompatible with this new version. It's not just a tweaked version of the older one. Oh well. It wasn't very useful for a long time now, anyway.
>>
>>
>>109154011
I can confirm it on my end as well, both on Edge and Firefox. I suppose we could remove that section completely, together with the mention of Imagen4, in the next OP.
>>
>>109154011
I've been seeing it go back and forth over the past few days
>>
File: wwg1wga.jpg (1.04 MB, 1376x768)
1.04 MB JPG
>>
>>
>>
>>
>>109143221
Cat!
>>
>>
>oh no, an invasive plant causing local damage
>>
File: fish tank.jpg (818 KB, 768x1376)
818 KB JPG
>>
>>
>>
>>
File: powerful...jpg (2.46 MB, 2400x1792)
2.46 MB JPG
>>
>>
>>
Test post.
>>
File: download-1 (9).jpg (206 KB, 1600x2848)
206 KB JPG
What are the best online tools / models to use for lewd headswaps? I want to feed the model a headshot and prompt (pose, outfit, scene) and for it to generate a realistic looking pic that keeps the likeness of the character?

Grok seems to be the best by far for faces, but its sensitivity filters are prohibitive.
>>
>>109157357
You can just check it out here:
https://4chan.org/banned
>>
>>
Morlock maintaining old infrastructure, programming in Ada/SPARK counts as technolo/g/y, right?
---
Reality is not anthropocentric. Genes, species, ecosystems, and their interactions exist independently of human thought. Biodiversity is a human concept for describing a real, non-human phenomenon.
>>
>>109152275
>actual schizo chimes in
back to your own containment
>>
>>
>"schizo chimes in"
>no image posted either a known troll move to not gen things while shit posting outside of >>>/b/
>ftfy
>>
File: 1752598984950497.png (3.88 MB, 1920x1920)
3.88 MB PNG
>>
File: 1775934465309104.jpg (3.15 MB, 1920x1920)
3.15 MB JPG
>>
>>
>>109159151
What did you prompt for Frankie? I've tried to gen her in the past but Gemini doesn't seem to know who she is
>>
>>109144191
Nice.
>>
File: 1782306024102337.jpg (203 KB, 1280x720)
203 KB JPG
>>109159209
I give Gemini this image some other anon made, and then ask for the actual prompt.
Gemini is trolling you, it knows who she is.
>>
File: test.jpg (1.48 MB, 1024x2048)
1.48 MB JPG
>>109159209
>>109159301
This is what I get if I prompt Frankie Foster without reference image. Only thing it "knows" is she has red hair.
>>
I found this on Sonic Stadium just now.

https://docs.google.com/document/d/1zmcEuocbSMAPH0Cdu0neR7qOYrOad6dB_PXEnPRM24Y/edit?tab=t.0

Which makes me wonder what are some long-term projects done with AI ? are comic books made by AI and animations made by AI finally feasible? that's some very detailed text and a lot of separate detail for each character. >>109147762
>>
File: infographic.png (2.5 MB, 1672x941)
2.5 MB PNG
>>
>>109159387
It managed to make the mansion and Mac, it might be avoiding the focus character due to copyright filters, or it is trolling you.
When you give Gemini reference images it works way better.
>>
Can't use Dall-e anymore, but Gemini is pretty good, and at least for now it is able to make hot characters
>>
File: imaginary friends at park.jpg (1.38 MB, 1792x1200)
1.38 MB JPG
>>109161123
I know it's part of the data, but it's not necessarily perfect with every franchise it's trained on
right: Drawn in cartoon style, no text. Mac from Foster's Home for Imaginary Friends, and friends, playing in an old eastern european playground under a dull cloudy sky. A red-haired woman with high ponytail, named Frankie, is smoking from the left side of the background, facing toward the right side of the image.
left: is watching from the left side
Similar results without mentioning Frankie. Missing hairclip, and wrong one when prompted. Anyway, was there a need to "troll" and single her out specifically? Whether it's doing it on purpose at inference level remains speculation and unlikely since it sure doesn't give a shit (on Flow, idk about Gemini the app) in general, except the external "prominent people" filter (god knows who it's trying to match),
>When you give Gemini reference images it works way better.
Yes, because you gave it an image. This part fulfills anon's original question of "how".
>>
File: cheerleaders.jpg (890 KB, 1200x896)
890 KB JPG
Yesterday I had a little funny where it added their names unprompted, given an input image.
>>
>>
>>
File: 1763117058106819.png (3.87 MB, 1920x1920)
3.87 MB PNG
>>
File: 1780368838652373.jpg (3.15 MB, 1920x1920)
3.15 MB JPG
>>
>>
File: 1774347579907314.jpg (1.2 MB, 1440x2560)
1.2 MB JPG
>>
>>
>>
>>
>>
The requisite beach episode.
>>
>>109142133
Made me chuckle, but then I imagined actually trying to sleep in the middle and it filled me with dread.
>>
File: cheebo.jpg (2.8 MB, 2400x1792)
2.8 MB JPG
>>
>>
>>
>>109157612
>Reality is not anthropocentric. Genes, species, ecosystems, and their interactions exist independently of human thought. Biodiversity is a human concept for describing a real, non-human phenomenon.
I would disgagree on the species point, in particular. Species are a human construct that doesn't actually exist indipendently in nature, if for instance you use the definition " largest group of organisms in which any two individuals of the appropriate sexes or mating types can produce fertile offspring, typically by sexual reproduction" all the so called species in the "Canis" genus become one single species (Eastern Coyote says "hi"), and they are hardly the only example (birds in particular have several similar cases where fertile interbreeding between different "species" is possible). And while descriptively "biodiversity" can be a neutral statement, prescriptvely it's far from neutral, it turns into a human-centered project focused on reproducing a particular ecological balance that existed in a particular historical moment, indipendently from how actually viable it is.

Yet, your post was the *only*, non-image-only, one that was left standing by the janitors/mods; also I was banned from *all* of 4chan for 24 hours, which is funny because I post almost exclusively here, while you apparently remained able to post on /tg/ and /trash/, at least; so I guess that your logic/point is approved by the right wing elites owning this place, at least. Rejoice.
>>
>>109164730
WTF 24 hours is possible? I thought the minimum is 3 days.
>>
>>109164805
Yep. It's at least the second time actually, but I can't remember whether the other 24 hours ban was also a general one or a board specific one.
>>
>>109164805
I was banned from /co/ for 24 hours for posting a screenshot from The Wind Waker, even though I was making a point about people complaining about art styles. I think the mod assumed I was some /v/ troll.
>>
>>109164892
I'm not sure about how it was on /co/ before, but... at least since early 2024, this board used to give only 3 days bans. The 24 hours one is effectively a relatively new thing here.
>>
Last few days in my country.
>109164730
>Yet, your post was the *only*, non-image-only, one that was left standing by the janitors/mods;
In case.you haven't noticed one of my posts was also removed (I got a warning) and the one you replied to was posted hours after your was removed.
>also I was banned from *all* of 4chan for 24 hours
Have you ever considered the possibility that if you keep getting banned then there's something wrong with you?
>which is funny because I post almost exclusively here, while you apparently remained able to post on /tg/ and /trash/, at least; so I guess that your logic/point is approved by the right wing elites owning this place, at least. Rejoice.
What is the medical term when someone sees connections where there are none?
>>
>>109164579
Dude was not supposed to be in there, but cool idea for an anime
>>
>>
>>109164962
>In case.you haven't noticed one of my posts was also removed (I got a warning) and the one you replied to was posted hours after your was removed.
I noticed that, but notice how there's a difference between a warning a ban for taking part in similar ways in the same "oh so terribly off-topic" discussion.
>Have you ever considered the possibility that if you keep getting banned then there's something wrong with you?
Funnily enough, until recently I wasn't really getting banned in *ages*. So I guess I'm going to blame an external factor, instead. New janitors, people multi-reporting me out of spite, who knows. Not me, as I've not really changed what I post.
>What is the medical term when someone sees connections where there are none?
The one for those who don't see connections is blindness.
>>
>>109165041 (Me)
>difference between a warning a ban
difference between a warning *and* a ban
>>
>>
>>109157362
None I know of. Sorry bro.
>>
>>
>>109153649
Why the fuck are you gunning words at Nano Banana anon? It works best with strong central concepts and related images as references. Using lots of small words just confuses it.
>>
>>
>>
>>109165479
He doesn't use reference images, which is mildly cool but infeasible for things you'd really want a reference for if not trained on.
The second prompt is fine, it's written in prose that is concise as is.
First prompt is intuitively "less legible" to the average human reader simply due to being one block of text, not terrible, more loaded than the other. Suppose "beautiful" doesn't mean anything to most models as a natural state, as opposed to "ugly". I don't think "perfect straight fringe" did anything.
>>
Antimatter bump.
>>
>>
>>109165479
I like words and it can take them. More seriously, like the anon above said, I rarely use references, when I do it's for specific logos, rarely objects, which NBP doesn't know, what I appreciated in the model is its capability to approximate several visual styles with decent/good results and its capability to follow quite complex prompts. It's *heavily* non-deterministic, it's not helped by the fact that it likely rewrites even long prompts that were already written by Gemini (AI Studio, that's what I use to write my prompts), but it completely fails only occasionally and curiously those failures seem to interest only NBP, NB2 tends to remain more faithful to the actual prompt (but stylistically it's inferior, in my opinion).

Prompt for picrelated: https://pastebin.com/UaEXbQvH

Also, my workflow:
>Please, try to write me a detailed and structured, narrative-based, image generation prompt for a masterful 2 pages (color) manga illustration by *x artist (optional medium description)*, western reading order, featuring:
>*character(s) description*
>It should be about *whatever.* (Optional: Keep a medium distance shot for panels where the face is shown). Include balloon dialogues in the description of each panel, don't worry about technical limitations.
The last part is important, Gemini actually believes that you're using some old (de3/mj) model that can't handle very complex prompts and it would try to give you separate prompts for each panel, so you have to instruct it to ignore those limitations. The handy thing is that now you have something that you can edit/tweak, if you don't like something in specific panels/dialogues. If you just instructed NBP to directly generate a manga about something it would likely take a different approach and write different dialogues each time.
>>
>>
Kinda recycled bump.
>>
>>
>>
>>109140139
Is this the right general to ask for cloud video generation alternatives to Google Flow?
I'm genuinely in love but managed to empty two Pro accounts with 1k credits this past week and don't have any more accs.
>>
Nano Banana 2 Lite has been released, for the record:
https://x.com/OfficialLoganK/status/2071988351083921690
Besides API, it can be accessed through Gemini by selecting "Flash-Lite" as the model. It's very fast and... that's all that one can say about it, I think. Used, for comparison sake, the same prompt that I tried for one of my "instant-ramen" generations, it's quite similar, I guess the model didn't change much if at all since then: https://desuarchive.org/g/thread/109057464/#q109076920
>>
>>109170347
>Nano Banana 2
>Here is the worse faster version of the thing that was only okay
Google has completely lost the plot
>>
>>109171043
Yeah, things look bad with Google lately, they were supposed to release 3.5 Pro this month, instead they released more cheap/fast models... One can only dream about NBP2 at this rate. I wonder if we can just blame agents burning tokens/computation and energy as if there was no tomorrow...

FWIW, they've also made NB2 Lite available on AI Studio (for free, without API key or subscription, unsurprisingly, given the low cost even on API). Interestingly, both Gemini and AIStudio output PNG, whereas instant-ramen on Arena was giving subsampled JPEGs. Also, resolution seems to be fixed at 1k.
>>
It looks like NB2 Lite *replaced* the original, Gemini Flash 2.5 based, Nano Banana on AIStudio, actually. I didn't notice that initially.
>>
>>
>>
>>
>>
Buffalo gals won't you come out tonight
Come out tonight come out tonight
Buffalo gals won't you come out tonight
And we'll dance by the light of the moon
>>
>>
>>109172403



[Advertise on 4chan]

Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.