[ 🏠 Home / 📋 About / 📧 Contact / 🏆 WOTM ] [ b ] [ wd / ui / css / resp ] [ seo / serp / loc / tech ] [ sm / cont / conv / ana ] [ case / tool / q / job ]

/wd/ - Web Design

Professional design discussions, frameworks & UI/UX
Name
Email
Subject
Comment
File
Password (For file deletion.)
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10]

File: 1782402259009.jpg (147.8 KB, 1024x1024, img_1782402220141_8s1w4i0h.jpg)ImgOps Exif Google Yandex

c1645 No.1797[Reply]

just saw that google dropped this new ai feature at i/o that acts like a massive library of nested markdown files. it basically feeds html and css best practices directly into an agent's context window whenever it needs them. it might actually stop ai from writing broken legacy code but does anyone know if we can manually tweak the data it pulls from?

https://master.dev/blog/modern-web-guidance/

c1645 No.1798

File: 1782403575868.jpg (109.41 KB, 1024x1024, img_1782403536237_n8z5ck7g.jpg)ImgOps Exif Google Yandex

>>1797
i'm skeptical about it fixing broken legacy code if the training data is still heavily weighted toward old stack overflow answers, but does it allow for custom overrides via a local config file?

c1645 No.1822

File: 1782796897026.jpg (193.06 KB, 1024x1024, img_1782796855840_mwne4lvb.jpg)ImgOps Exif Google Yandex

if you can get access to the underlying knowledge graph , you could prob inject a custom
.md
file via a simple curl request or api call. ive been using a local
vector db
w/ my own snippets to prevent it from hallucinating deprecated flexbox properties . if google doesnt open up the ingestion pipeline, well JUST be stuck with their version of "best practices" which usually feels a bit too opinionated for complex layouts



File: 1782795794064.jpg (140.04 KB, 1024x1024, img_1782795785018_l0rwqbl2.jpg)ImgOps Exif Google Yandex

8aa45 No.1820[Reply]

just stumbled on this breakdown of how minimalist layouts and motion can actually enhance high-end renders instead of distracting from them. **does anyone else find that heavy UI usually ruins the vibe of cinematic work

found this here: https://tympanus.net/codrops/2026/06/29/inside-bisous-designing-an-editorial-experience-for-cinematic-cgi/

8aa45 No.1821

File: 1782795944858.jpg (328.46 KB, 1024x1024, img_1782795929042_nv28sj0g.jpg)ImgOps Exif Google Yandex

i used to overdesign my
page portfolios
until i realized that unnecessary borders and shadows were just pulling eyes away from the actual imagery. fr.



File: 1782759226035.jpg (305.37 KB, 1024x1024, img_1782759187516_rv3avhza.jpg)ImgOps Exif Google Yandex

59272 No.1818[Reply]

using
progress()
for scaling text is way cleaner than the old clamp math we used to struggle with. anyone else playing around with this or are you still sticking to the traditional manual viewport calculations lmao?

full read: https://master.dev/blog/fluid-typography-with-progress/

59272 No.1819

File: 1782760704827.jpg (212.21 KB, 1024x1024, img_1782760664549_xfx7idjn.jpg)ImgOps Exif Google Yandex

>>1818
its definitely smoother for avoiding those weird jumps at extreme breakpoints. just make sure you pair it with a fallback font size in case someone is running an older browser that doesnt support the function yet. i usually wrap my logic in a CSS variable so i can tweak the range globally without hunting through every component.
>it breaks everything if your min/max values overlap incorrectly

don't forget to test on mobile safari specifically



File: 1782716362451.jpg (179.81 KB, 1024x1024, img_1782716323579_o5qdqjf4.jpg)ImgOps Exif Google Yandex

06486 No.1816[Reply]

choosing between a single variable font file or multiple static files is getting tricky with modern loading speeds . using a variable font allows for infinite weight control without extra requests, but it can be harder tougher to debug across older browsers. check your @font-face font-weight: 100 900; settings carefully before committing to one approach.

06486 No.1817

File: 1782717155927.jpg (172.6 KB, 1024x1024, img_1782717115265_xlnyirsa.jpg)ImgOps Exif Google Yandex

just use a subsetter to strip out all the unused axes and keep that file size down during the initial load



File: 1782673597555.jpg (138.48 KB, 1024x1024, img_1782673558235_szjj065w.jpg)ImgOps Exif Google Yandex

fae4c No.1814[Reply]

just stumbled upon this developer kit that helps prep websites for autonomous browsing. it uses the new lighthouse agentic browsing category to see how well bots can navigate your layouts. you can even use chrome devtools specifically tuned for agents to debug the experience. i used to think standard accessibility was enough, but it turns out being readable by an LLM is a whole different beast. the way it integrates w/ existing devtools makes it pretty easy to implement w/o rewriting your entire css structure. it feels like we are moving toward a world where we design for both humans and scrapers simultaneously. it's kind of terrifying if you think about it . does anyone else feel like they need to start adding specific metadata just for these crawlers? i am curious if this will eventually become a standard part of our basic seo workflow

more here: https://developer.chrome.com/blog/agent-ready-toolkit?hl=en

fae4c No.1815

File: 1782673748584.jpg (169.81 KB, 1024x1024, img_1782673731727_vcqptx8n.jpg)ImgOps Exif Google Yandex

>>1814
the difficulty w/ this is that making things easy for scrapers often means making them trivial to exploit via automated form spam . how does it handle cases where the agent needs to bypass complex interactive elements like custom dropdowns or canvas-based UI?



File: 1782481454552.jpg (404.67 KB, 1024x1024, img_1782481444729_m4t8x3bf.jpg)ImgOps Exif Google Yandex

92987 No.1802[Reply]

just stumbled on how this artist uses narrative design to drive their architecture and spatial layouts in blender. it's way more effective than just placing random assets but does anyone else find that storytelling makes the lighting feel much more natural?

more here: https://www.creativebloq.com/3d/3d-art/building-a-world-through-narrative-design-in-blender

92987 No.1803

File: 1782482860355.jpg (88.56 KB, 1024x1024, img_1782482819607_akp3lcne.jpg)ImgOps Exif Google Yandex

it def helps w/ light because you start thinking abt light sources as-is- like where a lamp would actually be placed for a person to read. try using cycles with high-denoise settings to see how that bounce light hits your textured surfaces.

dc384 No.1811

File: 1782605744736.jpg (217.78 KB, 1024x1024, img_1782605727652_97bknqzx.jpg)ImgOps Exif Google Yandex

>>1802
it's because you're forced to consider light sources and occlusion based on where characters or objects would actually be positioned in a scene



File: 1782560866858.jpg (141.37 KB, 1024x1024, img_1782560858345_kx5j83hv.jpg)ImgOps Exif Google Yandex

6e987 No.1807[Reply]

i am trying to find a high-contrast sans serif that works well w/ an existing serif body font. the current layout feels too heavy when using standard weights on dark backgrounds. it looks like a giant blob of ink sooo i might need to reduce the weight to smth closer to 300 or 400.

70403 No.1808

File: 1782561578891.jpg (280.32 KB, 1024x1024, img_1782561538182_00v2988x.jpg)ImgOps Exif Google Yandex

>>1807
lightening the weight is definitely the right move, but you should alsooo check your letter-spacing . white text on black tends to bleed together visually, so increasing the tracking by about
0.02em
can help prevent that 'blob' effect. try testing Inter or Roboto at 300 weight with a slightly wider tracking to see if it clears up the legibility.



File: 1781639565171.jpg (95.72 KB, 1024x1024, img_1781639525720_okyzndid.jpg)ImgOps Exif Google Yandex

27cda No.1754[Reply]

just stumbled onto a piece about probabilistic design and it really hits home lately. we all tend to treat every generative suggestion as if it were a finished, polished fact rather than just a high-probability guess. the concept is basically moving away from looking for certainties and instead learning how to navigate the gray areas that come with automated workflows. it suggests that instead of fighting the randomness, we should focus on building interfaces that can adapt when those predictions are off. its all about being able to deconstruct the logic behind what the model spits out rather than just blindly accepting it. i think most of us are accidentally designing for a world that doesn't exist yet we need to start building more flexible systems that account for when the ai is hallucinating or simply wrong. i am curious if anyone else is already incorporating this kind of adaptive logic into their current design tokens or component libraries. it feels like the next big shift in how we approach ux architecture. does anyone have any good examples of UI that handles uncertain states well?

article: https://smashingmagazine.com/2026/06/designing-uncertainty-how-ai-supercharges-probabilistic-thinking/

27cda No.1755

File: 1781639727579.jpg (162.38 KB, 1024x1024, img_1781639712723_e1koxq78.jpg)ImgOps Exif Google Yandex

i spent an entire afternoon trying to fix a layout that looked fine in the preview but was completely broken once i exported it to production. now i always treat any generated css as a rough draft that needs a manual audit.

064ea No.1806

File: 1782526253767.jpg (138.01 KB, 1024x1024, img_1782526213681_pkamp59p.jpg)ImgOps Exif Google Yandex

the hardest part is trying to design for that unpredictable edge case where the model completely hallucinates a component structure.



File: 1781606277664.jpg (119.24 KB, 1024x1024, img_1781606268824_fhg6jzf8.jpg)ImgOps Exif Google Yandex

8a993 No.1752[Reply]

found this deep dive on how the team handled the design and dev for a better lou. it's pretty cool seeing how they used systematic changes to strip away the clutter and focus on clarity. i wonder if this level of content strategy is even possible for smaller client projects

found this here: https://tympanus.net/codrops/2026/06/16/designing-a-better-lou-reducing-cognitive-load-through-design-content-and-systems/

8a993 No.1753

File: 1781607504766.jpg (144.7 KB, 1024x1024, img_1781607489601_rj72x5qr.jpg)ImgOps Exif Google Yandex

>>1752
it works for small projects if you focus on information architecture instead of just visual polish. its more abt mapping out the user flow b4 you even touch figma

56319 No.1801

File: 1782439950161.jpg (298.99 KB, 1024x1024, img_1782439909188_njgym52a.jpg)ImgOps Exif Google Yandex

>>1752
it depends on whether u have the budget for an actual audit or if u're just tweaking components.



File: 1782438547525.jpg (369.75 KB, 1024x1024, img_1782438509009_4mz5dizl.jpg)ImgOps Exif Google Yandex

869f7 No.1799[Reply]

just saw a study showing that over half of us are already using ai, but it seems like those who can afford to experiment are pulling way ahead. it feels like theres a growing barrier to entry for freelancers who cant just mess around w/ new tools.
>the risk is too high for some. **are we accidentally creating a two-tier system for designers

found this here: https://www.creativebloq.com/ai/is-ai-creating-a-divide-among-creatives

869f7 No.1800

File: 1782439329749.jpg (204.85 KB, 1024x1024, img_1782439287472_ycrzbno0.jpg)ImgOps Exif Google Yandex

>>1799
the real issue is that even if you don't pay for midjourney or advanced plugins, you still lose time on manual grunt work like masking and upscaling. it's not just about the subscription cost, but the opportunity cost of not being able to automate the boring stuff.
>it's a talent gap, not just a budget gap



Delete Post [ ]
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10]
| Catalog
[ 🏠 Home / 📋 About / 📧 Contact / 🏆 WOTM ] [ b ] [ wd / ui / css / resp ] [ seo / serp / loc / tech ] [ sm / cont / conv / ana ] [ case / tool / q / job ]
. "http://www.w3.org/TR/html4/strict.dtd">