Really appreciate this detailed feedback. These are the types of features that would make the platform more robust long-term.
Re: tagging/metadata - You're right that additional fields like sex, age, life stage would be valuable. Right now keeping it simple to reduce upload friction, but could definitely add optional metadata fields as the collection grows.
Re: user accounts - Good point about verification and editing rights. Currently everything is anonymous/CC0, but user accounts would enable better provenance tracking and let uploaders manage their own submissions. Thinking about how to add this without adding too much complexity to the upload flow.
Re: normalized taxonomy - Already using GBIF's taxonomy API for species names (auto-suggest on upload), so names are normalized to their database. Common names are free text for now since they vary regionally. Could potentially pull those from GBIF as well.
This is pretty helpful. Gives me a clear roadmap for v2 features. Trying to balance simplicity (to encourage contributions) with the kind of rich metadata that makes the collection more useful. Open to thoughts on how to strike that balance. Thanks!
Right, exactly. Good eye. Yeah well the tool that does the PNG transformation is far from perfect. Sometimes it cuts an image cleanly and other times not so much (open to improving this function).
Also a source/url would be a solid addition as well, even if the image is under a CC0 license. Appreciate your feedback.
Fair point. Right now the collection is small and somewhat random as I'm seeding it. The goal is to have multiple views/life stages for popular species over time, not just skulls or single angles.
Quality curation is definitely a challenge as it scales. The community verification system helps, but you're right that it requires active moderation. Thinking about how to balance growth with quality. Open to suggestions.
I kept running into the same problem: needing transparent PNGs of organisms (animals, plants, fungi) for design/education projects, but everything was either paywalled, had messy licensing, or required manual background removal.
Existing solutions:
- Stock sites: $$$, restrictive licenses
- Wikimedia Commons: mixed licensing, photos have backgrounds
- PhyloPic: silhouettes only, not full specimens
- iNaturalist: photos with backgrounds, not cutouts
So I built specimen.gallery – a searchable library of transparent specimen PNGs, all CC0 (public domain). Organized by scientific taxonomy. No accounts, no attribution required, just download and use.
Stack: Rails 8 + Postgres + Fly.io. Server-rendered, no React/Next.js, just clean HTML. Using Cloudinary for auto background removal.
MIT licensed.
Current status: ~90 specimens, growing daily. Looking for contributors and feedback.
Why PNGs specifically? They're the only widely-supported format with proper alpha transparency. SVGs don't work for photographic specimens. WebP support is still inconsistent.
Great point about the transparency grid for image search SEO. Hadn't thought about that... Might add checkerboard thumbnails for search engines while keeping the downloads as pure transparent PNGs. Thanks!
These look very nice thanks, hope it can be expanded. Interesting I wonder how does web devs handle transparency? There must be a better way than pngs.
Thanks! Yeah, expanding the collection is the priority right now.
Re: transparency - PNGs are still the most universally supported. WebP has better compression but inconsistent support in design tools (Photoshop, Figma, etc.). SVGs don't work for photographic specimens.
Could potentially offer WebP alongside PNG for modern browsers, but PNG is the safe default for now.
Re: tagging/metadata - You're right that additional fields like sex, age, life stage would be valuable. Right now keeping it simple to reduce upload friction, but could definitely add optional metadata fields as the collection grows.
Re: user accounts - Good point about verification and editing rights. Currently everything is anonymous/CC0, but user accounts would enable better provenance tracking and let uploaders manage their own submissions. Thinking about how to add this without adding too much complexity to the upload flow.
Re: normalized taxonomy - Already using GBIF's taxonomy API for species names (auto-suggest on upload), so names are normalized to their database. Common names are free text for now since they vary regionally. Could potentially pull those from GBIF as well.
This is pretty helpful. Gives me a clear roadmap for v2 features. Trying to balance simplicity (to encourage contributions) with the kind of rich metadata that makes the collection more useful. Open to thoughts on how to strike that balance. Thanks!