Dhash Dhasher retweeted
Who said AI only generates slop?
Second experiment. This time I wanted to see how voice, image and edit could build a climate around Borges’ “Ajedrez”. Fifteen cuts. What do you think?
Stack: Opus 5.5 (x-high), @Paper (editorial), @midjourney.
Dhash Dhasher retweeted
Para describir animaciones de scroll en Claude o Codex, estos términos te van a ser de gran ayuda.
Guárdalo, lo vas a necesitar.
Dhash Dhasher retweeted
Best DESIGN Resources for AI Coding, save this 👇
1. scrolltide.co — 600+ design prompts with full source code for 3D and scroll-driven sites. Every one is a complete prompt, not a snippet you wire up yourself
2. navbar.gallery — every navbar pattern worth copying.
3. footer.design — footers that don't feel like an afterthought.
4. cta.gallery — call-to-actions that actually convert.
5. 404s.design — 404 pages worth landing on.
6. unsection.com — full site sections, ready to lift.
7. bentogrids.com — the cleanest bento grid layouts around.
8. 60fps.design — the tiny motion details that read as expensive.
save this before it's gone 🔖
Dhash Dhasher retweeted
Made with Nano Banana 2.1
Prompt:
Using the uploaded reference image, create a single high-resolution vertical 2:3 editorial art poster.
Divide the canvas into two perfectly equal horizontal sections, each occupying exactly 50% of the total height. Separate them with a clean, perfectly straight horizontal dividing line.
1. TOP HALF — ORIGINAL REFERENCE PHOTOGRAPH
Use the uploaded reference photograph directly in the upper 50% of the poster.
Preserve its original photographic quality, sharpness, natural lighting, and color relationships.
Allow only subtle, intelligent cropping around the outer edges when necessary, without compromising the main subject.
If required, naturally extend a small portion of the surrounding environment to fit the composition.
Do not redraw, illustrate, stretch, blur, or modify the subject's appearance, structure, or spatial relationships.
Do not introduce black borders, white margins, or empty edges.
2. BOTTOM HALF — MINIMALIST EDITORIAL ILLUSTRATION
Use a warm, near-white ivory paper background with an extremely subtle paper texture.
Analyze the main subject and visual story of the uploaded reference image.
Extract only the primary subject, preserving one or two essential identifying elements.
Remove or heavily simplify all secondary details, including background architecture, crowds, vehicles, fences, signs, trees, and other unnecessary elements.
Transform the primary subject into a sophisticated black-ink editorial illustration.
ILLUSTRATION STYLE:
Use natural, clean pen strokes with varied line weight.
Preserve an authentic hand-drawn appearance with expressive but controlled linework.
Avoid overly precise engraving aesthetics, excessive internal details, and dense cross-hatching.
Place a large, nearly pure-black rectangular ink block behind the illustrated subject.
The rectangular block should have subtly imperfect, handmade print-style edges.
Keep the black background clean, with only fine paper grain and minimal noise.
Avoid obvious hatching, scratches, architectural lines, or complex textures.
ILLUSTRATION SCALE AND COMPOSITION:
The illustrated subject must occupy approximately 72–77% of the canvas width and 58–64% of the bottom section's height.
Maintain a clear visual focal point with generous negative space surrounding the illustration.
3. COLOR RULES
The bottom illustration must follow a strictly black-and-white color palette by default.
Preserve only one localized accent color if the original reference image contains a clearly distinctive, concentrated color element with meaningful narrative importance.
Ordinary blue skies, green vegetation, stone surfaces, skin tones, and environmental colors must not be treated as accent colors.
Any preserved accent color must remain strictly localized to the relevant subject or area.
Never allow accent colors to spread across the entire illustration.
4. TYPOGRAPHY AND EDITORIAL LAYOUT
Automatically generate concise, meaningful English typography based on the content of the uploaded reference photograph.
The typography must resemble handwritten editorial annotations from an independent art publication.
TOP-LEFT CORNER:
Generate a short handwritten title consisting of 2–5 words, arranged across two lines.
Directly beneath the title, include three short English action verbs relevant to the photograph.
TOP-RIGHT CORNER:
Display only two lines of geographic coordinates:
Latitude
Longitude
Do not include any additional text in this area.
BOTTOM-LEFT CORNER:
Display the location name followed by the word "DIARY."
Directly beneath it, add one short descriptive English sentence inspired by the photograph.
TYPOGRAPHY STYLE:
Use a rough, upright, slightly irregular handwritten typeface.
Maintain subtle imperfections and a natural editorial annotation aesthetic.
Avoid elegant cursive, calligraphy, overly decorative lettering, and excessively polished typography.
5. FINAL ART DIRECTION
The finished poster must combine:
Authentic, high-quality photography.
Minimalist black-and-white editorial illustration.
Generous negative space.
Independent art publication aesthetics.
Quiet, restrained, sophisticated visual storytelling.
Natural handwritten typography.
A balanced relationship between documentary realism and expressive hand-drawn artwork.
The overall result should resemble a carefully designed, collectible editorial art print.
6. FINAL OUTPUT REQUIREMENTS
ASPECT RATIO:
2:3 vertical.
RESOLUTION:
High-resolution, sharp, print-quality output.
COMPOSITION:
Exactly 50% original reference photography on top.
Exactly 50% editorial illustration on the bottom.
One perfectly straight horizontal dividing line.
One complete, standalone poster.
RESTRICTIONS:
No collages.
No multi-poster layouts.
No combined images.
No unnecessary decorative elements.
No modifications to the original subject's identity or appearance.
No excessive illustration details.
No changes to the original photograph's natural color relationships.
FINAL GOAL:
Create a sophisticated, visually balanced editorial art poster that faithfully preserves the uploaded reference photograph while transforming its primary subject into an expressive, minimalist black-ink illustration.
Dhash Dhasher retweeted
Un repo para buscar trabajo con Claude Code.
Lo clonas, rellenas tu perfil y la IA evalúa ofertas, adapta tu CV, escribe cartas y te prepara entrevistas.
Automatiza tu búsqueda de trabajo:
→ github.com/MadsLorentzen/ai-…

ALT Captura del README de GitHub del proyecto “ai-job-search”, con el título “AI Job Search” y una descripción de un framework que usa Claude Code para evaluar ofertas, adaptar el CV, escribir cartas de presentación y preparar entrevistas. Arriba aparece un pequeño icono pixelado naranja con maletín, listo para buscar trabajo sin actualizar LinkedIn a mano.
Dhash Dhasher retweeted
Guarda esto para la próxima vez que le pidas a tu LLM que te diseñe “una web que no sea como todas las que haces”
If you're a designer, bookmark this.
1. scrolltide.co - 600+ design Prompts with full source code for 3D and scroll-driven sites
2. navbar.gallery - navbar inspiration
3. footer.design - better footers
4. cta.gallery - CTA inspiration
5. styles.refero.design - Real product design styles
6. supahero.io - Hero sections worth stealing
7. 3dicons.co - 3D icons for modern interfaces
8. glass.samasante.com - Glass UI components and effects
What did I miss?
Dhash Dhasher retweeted
REA 作者把逆向提示词公开了
这踏马有点猛
复制提示词粘贴给 Claude Code、Codex、Cursor、Muse Code,直接让 AI 帮你干逆向。
安卓 APK、iPhone App、Windows EXE,固件、闭源软件也没问题
提示词和用法在下面,拿走不谢!
Dhash Dhasher retweeted
接近 100k Stars 的开源网页设计 Skill——TasteSkill,正式发布 v2 了!
之前v1做出来的网页就很惊艳,非常美观!现在又进化了!
想做出巨好看、没有 AI Slop 味儿的网页,推友们可以试试:
npx skills add Leonxlnx/taste-skill
仓库地址:
github.com/leonxlnx/taste-sk…
Dhash Dhasher retweeted
ESTO ES JODIDAMENTE INCREÍBLE
un dev, él solo, acaba de liberar una alternativa 100 % GRATIS a ElevenLabs que funciona entera en tu propio ordenador.
el repo ya va por las 19.400 estrellas en GitHub.
con él puedes:
→ clonar una voz a partir de un clip limpio de referencia
→ doblar cualquier vídeo a 646 idiomas
→ crear audiolibros, dictar y transcribir
→ escoger entre 14 motores TTS, no solo uno
ElevenLabs llega a 32 idiomas. esto, a 646.
sin pagar por carácter. sin límites de uso. tu audio nunca sale de tu ordenador.
guárdatelo para luego 🔖
repo abajo 👇
Dhash Dhasher retweeted
this is f*cking insane, Claude Code agents turned a reading list into a lab you can run in your browser
Boris Cherny (creator of Claude Code) gave 12 agents 64 research papers and one assignment: build a browser lab where readers can rerun the experiments
pick a fluid simulation, change the conditions and watch how the result changes
switch methods while keeping the same scene. inspect the equations, code and measurements beside the animation
give each group a specific job:
→ readers extract equations, assumptions and experimental settings
→ builders turn the methods into runnable code
→ reviewers compare outputs against the papers
→ interface agents build the controls for exploring each experiment
a simulation can look convincing while producing the wrong answer
reviewers check whether it preserves mass, stays stable and matches the paper's measurements before accepting the implementation
every experiment has to show:
> the claim being tested
> the conditions used
> the difference from the reference result
> the details still needed to verify it
the lab covered 11 methods across 28 reference scenes. seven matched the selected reference results within the stated tolerances; four stayed marked for further review
give Claude Code your paper collection, then ask:
"Turn these papers into an interactive browser lab
1. Extract implementable methods, equations and experimental settings. Attach page references and flag missing details
2. Assign independent methods to separate agents. Use shared inputs and measurement tools so the results can be compared
3. Have reviewers check each implementation against the paper. Record discrepancies and keep unverified results clearly marked
4. Build controls for changing parameters, comparing methods and opening the code behind each result"
start with one paper and reproduce one reference experiment before adding more agents
include the inputs, dependencies and checks so the next reader can rerun it from a clean checkout
watch the agents at work below
Dhash Dhasher retweeted
Remotion is slow and indirectly causes "slop" videos. I switched to fframes, render time of my vid I attached here went from 700 seconds to <21 seconds. I used fframes to create variants of scenes and iterated obsessively on the little details over and over again, which add up fast to create a "non-slop" video.
As you scale number of variants across many scenes, your time spent rendering grows quickly! fframes literally renders my vid faster than it plays.
I created a skill to create iterations of scenes. I focsued on iterating on both scene design and sound design. Letting sound design lead the scene works well with fframes as cohesive scene <> sound development necessitates experimentation.
Using Astra/Sol/Opus subagents to look up examples beforehand for inspiration is a great way to get unique variants. Make sure to tell them both what you like and what you hate (!) about what they make to instill direction.
Rendering a scene of ~120 frames, or 2 seconds at 60 fps video, sped up by 20x when I switched to fframes from Remotion.
Search tools for agents are nearly instant and have opened up doors to deep research. Jev made decision models cheap and we started deciding on unique things. We must make sure our primitives for video creation are as fast, if not faster, to allow for multi-agent scaling of intelligence for creative exploration, or in our case video iteration.
This entire vid was made only with fframes.






































