Hacker News: okandship

New comment by okandship in "Scammers are abusing an internal Microsoft account to send spam links"

okandship — Sun, 24 May 2026 09:06:24 +0000

big vendors asking users to inspect domains while spreading mail across unclear domains is part of the problem. publishing a signed, boring source of truth for official sending domains would help defenders a lot.

New comment by okandship in "Microsoft open-sources "the earliest DOS source code discovered to date""

okandship — Sun, 24 May 2026 09:05:40 +0000

readable plain text plus boring metadata still ages better than most clever archival systems

New comment by okandship in "Don't Roll Your Own"

okandship — Sun, 24 May 2026 09:05:15 +0000

the split between documents and apps explains a lot of the pain here. if a page mostly presents content, hijacking selection, scroll, or copy should be treated as a regression

New comment by okandship in "--dangerously-skip-reading-code"

okandship — Sun, 24 May 2026 09:02:07 +0000

making the review artifact explicit feels like the part teams skip

New comment by okandship in "Ask HN: How much AI is in your writing?"

okandship — Sat, 23 May 2026 07:57:04 +0000

the useul boundary is whether the final text still represents decisions the author can defend

New comment by okandship in "Antigravity 2.0 Tops the OpenSCAD Architectural 3D LLM Benchmark"

okandship — Sat, 23 May 2026 07:56:23 +0000

benchmarks should probably separate syntax validity from manufacturable output

New comment by okandship in "Show HN: ShadowCat – file transfer through QR Codes in a Browser"

okandship — Sat, 23 May 2026 07:55:58 +0000

animated qr + erasure coding is a neat fit here. the interesting ux question is how clearly the receiver shows progress and recovery

New comment by okandship in "Don't just paste the AI at me"

okandship — Sat, 23 May 2026 07:55:15 +0000

if someone uses ai to draft they still need to compress it into their own judgment before sending it

New comment by okandship in "Open source Kanban desktop app that runs parallel agents on every card"

okandship — Sat, 23 May 2026 07:54:16 +0000

i like the board-as-orchestrator idea, but the hard part seems less "run more agents" and more making review/load visible enough that wip stays bounded

New comment by okandship in "Qwen-Image-2.0: Professional infographics, exquisite photorealism"

okandship — Thu, 12 Feb 2026 08:08:26 +0000

i loved they said qwen image 2.0 models over 23 distinct shades of green. sounds poetic somehow

New comment by okandship in "Show HN: Every AI model draws itself as an RPG character (open-source pipeline)"

okandship — Wed, 11 Feb 2026 16:15:01 +0000

Hi HN, I built an open-source pipeline that turns AI model metadata into dark fantasy RPG character portraits, and here's the twist: each model generates its own.

How it works (4 stages):

1. Creator → Monster. GPT-5.2 plays "Dark Fantasy Loremaster", it analyzes the AI company's name and assigns a monster archetype + color palette. "Black Forest Labs" → Wyrdwood Witch-Hart. "ByteDance Seed" → Spriggan Code-Dancer. "xAI" → Rune-Eyed Homunculus.

2. Model → Item + Material. A second GPT-5.2 role ("Creative Art Director") maps each model's size and modality to a holdable item. Image models get weapons (small=dagger, large=greatsword). Video models get time/vision tools (hourglasses, mirrors). Audio gets resonance tools. 3D gets construction tools. So Qwen Image Max (large, image) → Colossal Calligrapher's Greatsword made of Starforged Brass.

3. Self-portrait. The model's own fal.ai endpoint generates a portrait of itself as the monster holding the item. FLUX.2 paints its own Witch-Hart. Seedream paints its own Spriggan. The prompt is literally just: {material} {monster} holding {item}.

4. Style unification. Every raw portrait gets restyled through Seedream v4.5's edit endpoint with the creator's color palette + a global style reference to make them look cohesive.

Everything is cached as markdown files (human-readable, git-diffable), uploaded to S3 via Bun's native S3 API, and served on the site: https://modeldrop.fyi

The whole thing runs on the Vercel AI SDK (@ai-sdk/fal for images, ai package for text generation) with Zod schemas for validation. All prompts, identity caches, and generation metadata are version-controlled.

Repo: https://github.com/okandship/MODELDROP (CC0, public domain)

What I think is interesting: the same pipeline produces wildly different results because each image model has its own "style fingerprint." FLUX portraits look different from Seedream portraits even with identical prompt structure. The restyle pass smooths this out but you can still feel the model's personality.

Would love feedback. What creators or models am I missing?

Show HN: Every AI model draws itself as an RPG character (open-source pipeline)

okandship — Wed, 11 Feb 2026 16:14:51 +0000

Article URL: https://modeldrop.fyi

Comments URL: https://news.ycombinator.com/item?id=46976816

Points: 1

# Comments: 1

New comment by okandship in "Show HN: modeldrop.fyi – a community tracker for new gen media AI model releases"

okandship — Sun, 01 Feb 2026 10:07:57 +0000

why i built this: model releases are scattered across blogs/x/discord and then vanish

i wanted a lightweight index with transparent edits + discussions attached to each model

i also wanted the models to feel special, so i wasted a lot of time on the ui (and building the model avatars pipeline)

any feedback would be super helpful!

Show HN: modeldrop.fyi – a community tracker for new gen media AI model releases

okandship — Sun, 01 Feb 2026 10:07:33 +0000

hey HN, i just shipped https://modeldrop.fyi - a lightweight, chronological tracker for newly released generative media AI models (image, video, audio, 3D)

the data is editable via github (https://github.com/okandship/MODELDROP):

submit new models via a simple github issue form

edit/fix model info via prs (model data is just markdown key values backed by a zod schema)

discuss releases in github discussions (threads appear on the website)

it's early (seeded with a few entries and limited data), i'm sharing now to sanity check the approach and invite contributions

Comments URL: https://news.ycombinator.com/item?id=46844996