Sabitlenmiş Tweet
SK
1.2K posts
SK
@hey2SK
My linktr: https://t.co/NW7QpsNOUr Our webapp: https://t.co/R40oAWtBEo, New product: One-Click Explainer Video Creation
Katılım Ağustos 2023
288 Takip Edilen142 Takipçiler
SK retweetledi
大家好，我來自日本 👋🏻
做了一款 app，叫「hibi」
一天那麼長，
最後濃縮成兩個字。
也許是「疲憊」，也許是「微光」，
也許只是「還好」——
兩個字，卻什麼都說到了。
不用寫完整的句子，
不用交代來龍去脈，
漢字自己會呼吸。
原是為日文使用者而做，
沒想到華語的朋友們也悄悄找來了。
繁體中文圈的你，
今天，是哪兩個字？
中文
SK retweetledi
Meet Z-Anime by @seesee. Built on the powerful Z-Image Base architecture, this model brings flagship-level diversity and precise prompt control to anime generation.
It inherits full support for complex negative prompts and extreme customizability. Ready to push your anime workflows to the next level?
Grab the weights on Hugging Face:👇
huggingface.co/SeeSee21/Z-Ani…
English
When I want to learn something new, or dig into a paper, I have Claude generate a HTML for me.
This works surprisingly well (especially in Claude, since Codex generated HTML is still kinda ugly...)
It's better than Google NotebookLM. Podcasts are nice, but reading is much higher-bandwidth than listening to a podcast.
HTML has a key advantage: they can show things.
Diagrams. Charts. Interactive bits. You can actually poke at the idea, not just passively consume it.
Then I iterate. Ask questions. Refine sections. Add missing pieces. The HTML evolves with my understanding.
Over time, this compounds into a personal knowledge base.
"The input/output mind meld between humans and AIs is ongoing and there is a lot of work to do and significant progress to be made, way before jumping all the way into neuralink-esque BCIs and all that." 💯
Andrej Karpathy@karpathy
This works really well btw, at the end of your query ask your LLM to "structure your response as HTML", then view the generated file in your browser. I've also had some success asking the LLM to present its output as slideshows, etc. More generally, imo audio is the human-preferred input to AIs but vision (images/animations/video) is the preferred output from them. Around a ~third of our brains are a massively parallel processor dedicated to vision, it is the 10-lane superhighway of information into brain. As AI improves, I think we'll see a progression that takes advantage: 1) raw text (hard/effortful to read) 2) markdown (bold, italic, headings, tables, a bit easier on the eyes) <-- current default 3) HTML (still procedural with underlying code, but a lot more flexibility on the graphics, layout, even interactivity) <-- early but forming new good default ...4,5,6,... n) interactive neural videos/simulations Imo the extrapolation (though the technology doesn't exist just yet) ends in some kind of interactive videos generated directly by a diffusion neural net. Many open questions as to how exact/procedural "Software 1.0" artifacts (e.g. interactive simulations) may be woven together with neural artifacts (diffusion grids), but generally something in the direction of the recently viral x.com/zan2434/status… There are also improvements necessary and pending at the input. Audio nor text nor video alone are not enough, e.g. I feel a need to point/gesture to things on the screen, similar to all the things you would do with a person physically next to you and your computer screen. TLDR The input/output mind meld between humans and AIs is ongoing and there is a lot of work to do and significant progress to be made, way before jumping all the way into neuralink-esque BCIs and all that. For what's worth exploring at the current stage, hot tip try ask for HTML.
English
I agree with it, and now my agent has accepted this purpose, and its medium text assets all used HTML.
Why not, because markdown only is compatibe with some people who is willing to make friends with computer rather than human.
Andrej Karpathy@karpathy
This works really well btw, at the end of your query ask your LLM to "structure your response as HTML", then view the generated file in your browser. I've also had some success asking the LLM to present its output as slideshows, etc. More generally, imo audio is the human-preferred input to AIs but vision (images/animations/video) is the preferred output from them. Around a ~third of our brains are a massively parallel processor dedicated to vision, it is the 10-lane superhighway of information into brain. As AI improves, I think we'll see a progression that takes advantage: 1) raw text (hard/effortful to read) 2) markdown (bold, italic, headings, tables, a bit easier on the eyes) <-- current default 3) HTML (still procedural with underlying code, but a lot more flexibility on the graphics, layout, even interactivity) <-- early but forming new good default ...4,5,6,... n) interactive neural videos/simulations Imo the extrapolation (though the technology doesn't exist just yet) ends in some kind of interactive videos generated directly by a diffusion neural net. Many open questions as to how exact/procedural "Software 1.0" artifacts (e.g. interactive simulations) may be woven together with neural artifacts (diffusion grids), but generally something in the direction of the recently viral x.com/zan2434/status… There are also improvements necessary and pending at the input. Audio nor text nor video alone are not enough, e.g. I feel a need to point/gesture to things on the screen, similar to all the things you would do with a person physically next to you and your computer screen. TLDR The input/output mind meld between humans and AIs is ongoing and there is a lot of work to do and significant progress to be made, way before jumping all the way into neuralink-esque BCIs and all that. For what's worth exploring at the current stage, hot tip try ask for HTML.
English
SK retweetledi
ChatGPT Images 2.0
プロンプト：
まるで小さな子供がクレヨンで描き直したような、やさしくて少し不器用な絵本風のイラスト。
人物の顔や手、服の細かい部分は描き込みすぎず、丸く柔らかい形で単純化する。
表情はあたたかく、少しいたずらっぽく、友達との楽しい記憶がにじむように描く。
色は実際の色に縛られず、明るいパステルカラーを自由に使う。
ピンク、ミント、レモンイエロー、ラベンダー、水色などをランダムに混ぜ、子供らしい想像力で再解釈する。
線はかなり揺れていて、塗りはクレヨン特有のざらざらした質感、色むら、にじみ、紙にこすれたような跡を感じさせる。
子供が大切な思い出を楽しく飾ったような自然な配置にする。
全体は子供のクレヨン画の夢のような世界がやさしく重なったイメージ。
懐かしく、可愛らしく、少しセンチメンタルで、絵本の中にしまっておきたくなるような一枚。
写真に書かれていない文字は入れない。
日本語
@ChillaiKalan__ Why the prompt is so long even if you've already used the most advanced model.😂
English
GPT IMAGE 2 on ChatGPT
Prompt:
Create a bold Y2K pop-street editorial collage featuring a young person as the central subject, combining playful graffiti typography, orange poster background, sticker-like graphic elements, and fashion magazine layout energy.
MAIN SUBJECT AND FRAMING:
Place the young person on the right side of the composition in a close-up to upper-body portrait. The face should be large and dominant, cropped from the chest upward, with the subject slightly leaning forward into the frame. The composition should feel asymmetrical, with graphic typography filling the left side and the subject anchoring the right side.
Do not control or describe the hairstyle.
OUTFIT DETAIL:
Style the subject in a minimal black streetwear outfit. They wear a black turtleneck or high-neck top with a soft fitted silhouette, creating a clean dark shape against the bright orange background. Add a black beret or rounded cap with a small embroidered logo detail on the front, giving the look a stylish retro street-fashion feeling.
Add a small colorful cheek sticker or decorative bandage on one cheek, with cute tiny graphic details, making the styling feel playful and idol-like. Keep the outfit mostly black so the blue graffiti typography and orange background stand out strongly.
BEHAVIOUR AND BODY LANGUAGE:
The subject’s body language should feel calm, closed-off, and slightly moody. The shoulders are relaxed but slightly rounded forward, creating a soft guarded posture. The head tilts subtly downward and to the side, giving the portrait a cool detached attitude.
The subject does not need a big hand gesture in the main portrait; instead, the behaviour comes from the face angle, lowered posture, and intense sideways eye movement. The pose should feel like someone caught in a quiet bratty moment, not fully engaging with the viewer.
FACIAL EXPRESSION:
The expression is pouty, serious, and mildly irritated in a fashionable way. The lips are softly pushed forward into a small pout, brows slightly tense, and the eyes look sideways with a suspicious or unimpressed gaze. The mood should feel bratty, stylish, aloof, and quietly rebellious.
GRAPHIC BACKGROUND:
Use a saturated orange background with a large white starburst shape behind the subject’s head and shoulders. The starburst should create strong comic-book energy and make the face pop.
Add hand-drawn black scribble text across the orange background, like messy diary writing or lyric notes. Include phrases such as:
“EENIE MEENY MINEY MO”
“4:03”
“LOVE DAY”
“HIDE YOUR LOVE”
“BOYZ”
The handwriting should look playful, imperfect, thin, and chaotic, but still visually controlled.
MAIN TYPOGRAPHY:
Place a large blue graffiti-style word “Your Name” across the left-center area. The letters should be chunky, rounded, irregular, and outlined in thick black, like a playful bubble graffiti tag. The typography should overlap slightly with the subject’s face area without blocking the eyes.
Add pink star sparkle stickers around the face and cap area, giving a cute pop-idol accent.
BOTTOM FILMSTRIP:
At the bottom of the composition, add a horizontal filmstrip with four rectangular photo panels. These panels show smaller full-body and close-up poses of the same young person wearing the same black outfit.
Panel pose ideas:
1. Side close-up portrait with head tilted back slightly, cheek sticker visible, eyes looking away.
2. Full-body stance with legs apart, hands near pockets or belt area, posture confident and relaxed.
3. Repeated close-up side portrait for rhythmic collage effect.
4. Repeated full-body stance, slightly different angle or crop.
The filmstrip should feel like a fashion contact sheet or idol editorial preview.
MOOD AND STYLE:
The overall mood should be playful, bratty, nostalgic, and street-fashion inspired. It should feel like a Y2K teen magazine layout mixed with graffiti poster design, idol concept styling, and scrapbook-like pop graphics.
English
@venturetwins As a aigc engineer, I am very interested into testing a new image generation model.😘
English
I've been testing a new image model dropping soon 👀
It's a breath of fresh air - the visuals are insane, and it makes image generation feel like art again.
I have a couple invites, comment or DM if you want to try it!
English