6 ms·
I often need to find logos and was not impressed by the existing logo search engines, so, in classic bikeshed mode, I built my own. Because it makes total sens
by Fileformat 6y ago
I often need to find logos and was not impressed by the existing logo search engines, so, in classic bikeshed mode, I built my own. Because it makes total sense to try to compete with Google. And Wikipedia. And countless others (some of which I list on the alternatives page).
Why is mine better?
- just logos
- instant results
- only SVGs
- hyperlinked directly to the source
- no right-click hijacking (or other dark patterns)
- totally awesome domain name
It is open source (AGPLv3), so you can see the tools I use and how it works (and criticize my code). The sources have their own licenses, so be sure to check there before using.
It is still a work in progress and there are some glitches (I mean, who would use a percent sign in a file name? And I'd love to try to fix the Wikipedia redirected images if I could only find the source...), but I want to see if people find it useful.
So let me know what you think! And I'm always looking for more sources!
- alessioalex 6y agoYou were not kidding about the 'instant results' part. I like that, as well as the domain name.
- nathanganser 6y ago.ch is great!
- the_duke 6y agoLooks useful. Feedback: the result rendering is very "jumpy" due to logos loading in one at a time. You could either render sized wrapper containers for each logo, or load them in in the background and only show them once ready (with a loading indicator somewhere).
- Fileformat 6y agoI have tried various solutions, but nothing that works better than what's there now. I only force the height for the logos, not the width, so wrapping doesn't really improve things. Waiting to display until they are all loaded kills the "instant" tagline. I'm hotlinking the images, so some images can take awhile.
- whoisjuan 6y agoIn theory you should be able to get the natural height and width of the image before loading it. Not exactly sure if it works with SVGs but it does work with PNGs
- EmilStenstrom 6y agoYeah, you could get the size of the images at indexing time, and return that in the search result.
- Fileformat 6y agoYou are assuming that I actually load the image at index time: I'm really only indexing the file name. And getting the width/height of a minimized SVG requires rendering it (or at least more SVG mojo than I possess).
- setr 6y agoHe's suggesting your index store the filename, height & width, and retrieve all 3 on search
- AHASIC 6y agoIt's a non-issue, don't sweat it.
- Ayesh 6y agoYou can use preconneft tags to make the browser preconneft to the likely source of the images. This usually gives a pretty big boost in speed because browsers will have already connected (DNS + TLS handshake) to the icon source's CDN.
- dfee 6y agoDid not find a logo for Logosear.ch though!
- Fileformat 6y agoHeh. You got me! That'll be an easy one to fix though
- airstrike 6y agoThanks for building this! I actually obsess about having high quality logos for companies in the pitchbooks I write, so have built a personal collection over time... it's not big, but it has the companies I care about. Some things I thought I'd share from that experience: * The hardest thing is knowing which logo is actually the most recent and therefore correct. Searching logosear.ch for, say, IBM, yields countless logos that are virtually identical, as well as some of their products (which is fine) but also many old logos like this[0], which I don't think is what your user would want to find 9 out of 10 times (i.e. as a user, if I'm looking for a niche / ancient logo, I don't mind doing the research on my own and understand it will take a while. But when I want "the right logo", I'll consider visiting a specialized logo search engine to get the right result and get it fast) * In my experience, The best sources for finding the company logo in SVG format are: - company website header (hit or miss as SVG adoption at first was slow, but it's increasingly good for S&P 500 type companies and tech-savvy firms... anything outside of that venn diagram is pretty bad) - Wikipedia page either in English or in the language matching the company's headquarter (i.e. better luck with German companies in de.wikipedia.org) - For large companies where both of the above fail, I search for "company name filetype:pdf" and look for vector logos embedded into those files, then use Adobe Acrobat to extract them in a very manual and ugly process by deleting unnecessary pages / content and cropping, and then visiting cloudconvert.com to generate a SVG (hey, it works...) * One final comment is PowerPoint doesn't take SVG files, so I always have to convert them to .emf or .wmf before dropping them in, so maybe you should consider exporting your SVG file in other formats too, though probably not limited to just those two –––––––––– 0. https://commons.wikimedia.org/wiki/File:IBM_Logo_1956_1972.svg https://commons.wikimedia.org/wiki/File:IBM_Logo_1956_1972.s...
- Fileformat 6y agoWould you be willing to share your collection of high-quality logos with the community? I can index any github or gitlab repo with logos. A lot of the current sources are people's personal collections that they have shared. This is actually how I started: I made my personal collection into a website (https://www.vectorlogo.zone/ https://www.vectorlogo.zone/). My collection is more curated, definitely in terms of size/layout, and hopefully more current, but curation is a lot of work, and not what I'm trying to do on LogoSear.ch. I do have an SVG to PNG converting working on VLZ. I think there are open SVG to EMF/WMF libraries, so I could add that as well, but I would need some help testing it since I'm not a big PowerPoint user. Would you be willing to help?
- tracker1 6y agoVery cool... I think I found a bug, in that results from a prior entry would come up on a new entry... For example, I typed mercedes, then selected all and amg over that... in the end, some mercedes results showed on the amg screen... should probably tag your search request with the value in the input, and abandon the request/results if they no longer match.
- Fileformat 6y agoSigh. I hate debugging timing related issues that I cannot duplicate. But I already include the input in the search results API response, so your suggestion should work. I'll try it shortly.