Thread #58892256 | Image & Video Expansion | Click to Play
Alice Anon
/napt/ AI Pokémon Thread #247: Bea's Takigyo Edition 02/02/26(Mon)11:12:02 No.58892256
/napt/ AI Pokémon Thread #247: Bea's Takigyo Edition 02/02/26(Mon)11:12:02 No.58892256
/napt/ AI Pokémon Thread #247: Bea's Takigyo Edition Alice Anon 02/02/26(Mon)11:12:02 No.58892256 [Reply]▶
File: Bea Waterfall.jpg (398.8 KB)
398.8 KB JPG
Previously on /napt/:
>>58868479
This is a general thread for people to post their AI art and/or make requests from others who may want to attempt making their own Pokémon-related gens.
Local Generation using Stable Diffusion requires a decent video card. Guides for local genning can be found in the /g/ threads:
>>>/g/sdg
>>>/g/ldg
Other Stable Diffusion threads are on: /h/, /e/, /d/, /b/, /vt/, /aco/, /trash/
NovelAI v4.5 is a paid AI generator. Opus tier users have "unlimited" Small/Normal generations, though editing tools require credits.
Midjourney is a paid AI generator, also capable of video. Usage requires a monthly subscription, with basically infinite gens.
Microsoft's Bing AI is a free image generator that is powered by DALL·E 3.
Civitai is free using its ‘buzz’ currency. You get a certain amount every day, plus more for liking models and gens.
>civitai.com
Otaiku is a (mostly) free mobile app. An ad watch is required between each image generation unless you pay.
>ailla.biz/otaiku
free video generation
>videomaker.me/app
Tensor is another generator that gives free daily credits, as well as a vast library of LoRAs to train your prompts on. Subscriptions are available for those who wish to pursue more than this.
>https://tensor.art/
Pastebin of Pokémon artists:
>https://pastebin.com/U0mmK3Tm
Characters with under 50 images will require more tags to have them look correct. This is a list of all Pokémon characters with over 300 images:
>https://danbooru.donmai.us/posts?tags=*_%28pokemon%29+&z=5
[Multiple images to share? conserve space. collage them]
A Message from ThatOneAIAnon:
>Heya, fellow /napts, I've updated the Novelai artist tool to 4.5, and added an ELO ranking system so you can rank your favourite artists more easily.
>https://github.com/theapexxx/novelai_artist_showcase
120 RepliesView Thread
>>
>>
File: 00013-148103672.png (732.4 KB)
732.4 KB PNG
Good morning everyone!
>>
>>
File: 636de881-385a-42ba-8328-6e628f944847.png (829.4 KB)
829.4 KB PNG
>>
>>58892256
>>58892260
This but it's Serena, please.
>>
File: RealLillie.jpg (318.7 KB)
318.7 KB JPG
>>
>>58888543
Is it possible to do orla with anima model or not yet? Could someone try it with this prompt please?
>best quality, masterpiece, highres, 1girl, solo, orla (pokemon), pokemon, pokemon horizons, @sugimori ken, brown hair, orange hair, green eyes, multicolored hair, parted bangs, sidelocks, large breasts, yellow jacket, short sleeves, jacket around waist, gloves, black shirt, t-shirt, orange pants, smile, posing, abstract background, dynamic pose
>>
>>
File: ComfyUI_00369_.png (933.5 KB)
933.5 KB PNG
>>58892621
>>
>>
>>
>>58892647
Is it too early to ask for Professor Mable? You could test it with, mable (pokemon), pokemon legends: z-a, sugimori ken (style), mature female, aqua hair, aqua eyes, hair rings, lipstick, large breasts, safety glasses, long sleeves labcoat, open white labcoat, aqua necktie, black dress, black belt, black pantyhose, aqua nails, id card
>>
File: ComfyUI_00371_.png (669.6 KB)
669.6 KB PNG
>>58892714
>>
>>
>>
File: ComfyUI_00419_.png (794 KB)
794 KB PNG
kinda weird that ralts doesnt work here but pikachu does. cant imagine the model doesnt know what a ralts is
>>
File: trainer card, bikini bottom only, s-933461122.png (1.5 MB)
1.5 MB PNG
NovelAI just put up a new feature. More tools for taking info from an already existing image.
https://blog.novelai.net/precise-reference-release-character-style-ref erences-reference-inpainting-175d69 e44cfa
>>
>>
File: 1753574010848125.png (1.7 MB)
1.7 MB PNG
For Valentine's Day, would it be fine to see something wholesome between Wally and Aqua Grunt-chan? Hearts, chocolates, dressing him up like a female Aqua grunt for funsies, etc.
>>
File: IMG_3674.jpg (359.2 KB)
359.2 KB JPG
https://files.catbox.moe/6b4zg8.webp
>>
>>58890662
Can I ask something like this but with Rosa and Kris dressed in the Pasio uniform?
>year 2025, newest, best quality, masterpiece, highres, 2girls, rosa (academy) (pokemon), kris (pokemon), pokemon, pokemon masters ex, 3d, official alternate costume, rosa (pokemon) (cosplay), oppai loli. Both girls have huge breasts wearing the same school uniform, pressing their breasts together while looking at viewer. The school uniform has open grey jacket, long sleeves cropped jacket, red vest, yellow neckerchief, white collared shirt, white pleated skirt. Rosa have brown hair, double bun, hairband and blue eyes. Kris have green hair, twintails, red beret hat and green eyes. outdoors, day, sky, clouds, sun, school, garden, park, pink petals, tree, spring (season), building, breast contest, smile,
>>
>>
>>58893474
BREAK is a tag I've seen in some works that's supposed to separate the prompt into parts to make sure certain things get applied to specific subject
>Change position
do you just tell Novel the position in the prompt or is it somewhere else?
>>
File: ComfyUI_00450_.png (1.1 MB)
1.1 MB PNG
>>58893426
>>
>>
>>
>>58893614
Hehe, thanks. You do a great job, but the 3D style makes it look really weird vibes. I was expecting something like in-game screencap. Looks like this model doesn't have that effect, unlike others.
Would you be so kind as to remove just the 3d tag and add official art, official style? Also, please add yellow buttons to the uniform description.
>>
>>
>>
>>58893614
That looks horrible, but I'm genuinely fascinated by how the AI almost created a 2010s Deviantart OC look to it with the shadingless flat colors and the background that looks copied from a video game in paint.
It's nostalgic almost lol, and I did not expect these feels
>>
>>
>>
File: ComfyUI_00453_.png (615.8 KB)
615.8 KB PNG
>>58893636
an attempt was made
>>
>>
File: ComfyUI_00456_.png (1.2 MB)
1.2 MB PNG
>>58893647
this was with realistic added, I'll try sugimori next
>>
File: ComfyUI_00460_.png (1 MB)
1 MB PNG
>>58893647
>>
>>58893678
>>58893700
Hey, it's not bad now, thanks. The 3D now looks like the GameCube or PS1 in HD, lol. Sugimori looks better, but it seems like they just have different sized breasts, a bit weird.
>>
File: ComfyUI_00472_.png (652.2 KB)
652.2 KB PNG
>>58893721
ah yeah just noticed myself. guess some things are still a bit wonky
>>
>>58893700
>>58893721
Kinda like watching the Thing merge with someone. Freaky.
>>
File: emma vs mable.png (2.4 MB)
2.4 MB PNG
>>58892690
>>
>>
File: ComfyUI_00500_.png (943.7 KB)
943.7 KB PNG
>>
File: comfyfurnishing.png (2.6 MB)
2.6 MB PNG
>>58893808
I use Sora for this with macro-prompting (style, mood, character description, camera effects and etc). The eyes and hairstyle where the hardest to nail, I wanted to make them real-life like the Lillie pic but retain their on-model designs.
After multiple tries i just gave up and went with the one you got here.
>>
>>
File: rhyhornrampage.png (2.5 MB)
2.5 MB PNG
>>58893872
I used the character renders as image reference but the text did most of its job.
Yes anon, you are welcome!
>>
File: margocity23o2.jpg (160.8 KB)
160.8 KB JPG
Dates in Fula City.
>>
File: ComfyUI_00522_.png (777.1 KB)
777.1 KB PNG
>>
>>
File: Selene Sundress Set.jpg (683.5 KB)
683.5 KB JPG
The full Selene sundress collection went up today, for those who want to see them all: https://www.pixiv.net/en/artworks/140674824
I only have five characters left to put up for this collection at the moment, and with my current posting schedule, they'll all be up by the end of next week. I'm going back and forth on which collection to share next. Either it'll be the maids, or the kimonos. If I go with the latter, I may end up redoing some of the characters that were done through Otaiku and have it done in Tensor instead, to try and keep things relatively uniform. What do you guys think?
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
File: ComfyUI_00545_.png (944.7 KB)
944.7 KB PNG
>>
File: ComfyUI_00547_.png (838 KB)
838 KB PNG
the contrast between the innocent face and huge muscular body is cracking me up for real over here
>>
>>
>>
>>58894154
>>58894160
Can you try make the official young cynthia from anime?
Maybe some tags can be cynthia (pokemon), pokemon journeys, pokemon (anime), loli, blonde hair, grey eyes, blue bow, hair bow, cleavage, medium breasts, long sleeves, blue dress, white socks
>>
>>
File: ComfyUI_00557_.png (670.2 KB)
670.2 KB PNG
>>58894195
this is what it gave me
>>
>>
File: ComfyUI_00567_.png (655.1 KB)
655.1 KB PNG
>>58894240
>>
>>
>>
>>58894283
Then maybe is the problem, if you can try it with masterpiece, best quality, cynthia (pokemon), pokemon, pokemon (anime), pokemon journeys, anime coloring, anime screenshot, and then all the previous tags of her appaerance pls
>>
File: ComfyUI_00581_.png (313.8 KB)
313.8 KB PNG
>>58894293
>>
>>
>>58894178
>>58894327
There are random black kids in Z-A, aren't there? Just use one as a character reference, then turn it off and throw in the other character's reference using the new character reference inpainting. Removing the background on the reference pics seems to help, also.
>>
>>
>>
File: 43a38aab-af66-4c7e-b9d4-daf41bef1764.png (1003.8 KB)
1003.8 KB PNG
>>58892640
big day for large orlas
>>
>>
>>
>>
>>
>>58894593
They changed it from character reference to precise reference, allowing you to use multiple reference images (at 5 anlas per gen each, so still no point in using more than one if you don't want to pay up every 20 seconds) and choose whether to copy character, style or both. Problem is the generations are of much lower quality and consistency than before, frequently getting details wrong and making mistakes where the previous model would get them right. It's kind of like the intitial version of character reference before they made it better except worse.
>>
>>
>>
File: ComfyUI_00613_.png (1 MB)
1 MB PNG
>>58894597
I wonder if the release was rushed in response to Anima.
>>
>>
>>
>>58892647
Anima can made hgss karen wearing a bunnysuit? I made this prompt for her:
>Year 2025, newest, best quality, masterpiece, highres, 1girl, solo, karen (pokemon), pokemon, pokemon hgss, sugimori ken (style), blue hair, long hair, blue eyes, cleavage, large breasts, bursting breasts, white bunnysuit, rabbit ear hairband, yellow bowtie, white leotard, playboy bunny, fishnet pantyhose, blush, standing, holding cards, casino, indoors,
>>
File: ComfyUI_00631_.png (1.1 MB)
1.1 MB PNG
>>58895356
for some reason it came out looking fried or something. I'll try with the artist tag instead of the style one
>>
>>
File: ComfyUI_00633_.png (989 KB)
989 KB PNG
>>58895377
the saturation seems a bit high. this is with the artist tag
>>
>>58895386
It's not very noticeable, but the style is definitely there. I think using only Sugimori Ken gives you a very basic retro look, sometimes too much so, while adding (style) makes it look more modern, with softer and more detailed shapes. I think this varies depending on the base model.
Thanks for both versions.
>>
File: Torchic chilling.jpg (248.5 KB)
248.5 KB JPG
>>
>>58893614
>>58893678
>>58893700
Can I get this but with the style of @m-nyuugyou and doing peace gesture, winking, from above?
>>
>>58895212
I'm sure making Nai less precise will stop people flocking to the competition. Supposedly they're already scrambling to try and put the previous character reference model back alongside the new one in response to a multitude of complaints.
>>
File: 89952734_p11_master1200.jpg (381 KB)
381 KB JPG
Alice-Anon I need more sun and lillie!
>>
>>
>>58896346
Thanks anon, no problem. That model is only for audio, right? How do you create music and voices for chatbots? Wait, did you say it takes 20 minutes?! Do you mean the song or an image?
Another thing I wanted to ask is if it's possible to do other resolutions with Anima, like portrait and landscape, and not just square. Have you tried using Euler or Euler_a in Sampler? I read that it gives better results in certain cases.
I'm sorry for so many questions.
>>
>>58896383
20 min or so for a 2min song. This model is only for music. For like text to speech youd need something else. I haven't really looked into TTS much but I think VibeVoice is supposed to be good and the Qwen TTS that was released recently. For Anima it can handle different orientations like portrait and landscape, just can't really go higher than 1024x1024 since its not done training. I can do like 832x1216 though since thats still 1 MP (megapixel). For sampler I usually just stick with euler A since its hard to really tell if its better or just a better seed.
>>
>>58896395
Thanks for that info, it takes almost as long as a WAN video for 2 seconds then. That resolution is fine, 800x1100 is what I usually use online in XL, it takes up little space and is cheaper.
I only know that Euler ancestral works better for anime, but it's a hassle to configure the correct steps and cfg.
>>
>>
File: arcaninepainting.png (3.4 MB)
3.4 MB PNG
>>
>>
>>
File: Elio and Lillie Beach Set.jpg (688.5 KB)
688.5 KB JPG
>>58896268
There a specific scenario you'd like to see them in? Keep it wholesome, if possible.
Have them at the beach in the meantime.
>>
>>
File: ComfyUI_00727_.png (1011.5 KB)
1011.5 KB PNG
>>58895680
its not really wanting to do a "from above" view even when i add they're looking up etc
>>
>>
File: ComfyUI_00728_.png (949.5 KB)
949.5 KB PNG
>>58896768
the background isnt really great in any of them. might just be the model but i can double-check the prompt
>>
>>
>>
>>
File: 5572660.jpg (337 KB)
337 KB JPG
>>
>>
File: GYJ7HBVRFM8BFEMA459JSB29B0.jpg (454.2 KB)
454.2 KB JPG
>>58896824
Sure, why not?
>>
File: Iono Flamenco Set.jpg (736.9 KB)
736.9 KB JPG
>>58896824
Here's some Iono flamenco dancing that I did a while back.
>>58896740
I'm going to pretend that this is a different Anon attempting to troll us...
>>
>>
>>
>>58892256
Animooted, worksafe:
https://files.catbox.moe/poyp2e.mp4
>>58892456
I'll see what I can do later.
>>
>>58892420
I didn't see the request anchor until it was too late, so... >>58893302