I started with http://hello.jpg.to/ and was delighted by the results, thinking that the web app translated my phrase into a variety of languages and then made a stylish motif. Eagerly, I typed in http://goodbye.jpg.to/ to see, again, what looked like a totally custom image based on random text that I put in. I have to admit, I found this to be http://totallyawesome.jpg.to/ at this point, pressing F5 and hoping for other stylized generations.
After not seeing any, I decided to just try my name ... and found a football player.
Good observation.
But if you embed it on webpage (in which refer header is sent)
it would turn into real image file, which means it WORKS when you include it in forum, blogs, etc. :)
Refer header is not a good way to distinguish between embedded images and images accessed directly by user. AFAIK that header will have the same value both when you load an embedded image and when you click on a direct link to the image.
I did a little bit of research about it few years ago; the summary can be seen here:
Basically, I was able to distinguish between embedded and directly accessed images by analyzing the content of Accept header. It wasn't perfect (it didn't work for Opera), but I think it could be good enough to do something useful with it.
No, please don't change it that way! The best use case for this is sharing images quickly. If I can't trust that my http://homer%20simpson.jpg.to/ is going to be the same one loaded by my friend, then I won't share it with my friend. This could be pretty handy in it's current state.
Note that that particular word is a katakana phonetic import from English; in romaji that reads "hakkaa". So while that works as an example of multilingual character support, it may not be the best example for a foreign word per se. :-)
It is the first image result on google image search. The previous comments on different keywords verified this. To amaze yourself, try "sex" and "male"
behind-the-scene technical aspect are not difficult to realize (either google image API, which is deprecated, or parse the result of http://www.google.com/search?tbm=isch&hl=en&source=h... and get the first image url after a predefined string anchor, say <span class=rg_ctlv>), but the idea to simplify input and output is brilliant.
To deal with not-so-good image search result:
since google image search is presenting the result in a thumbnail group, it might be worthwhile to look into their ranking scheme for the result. It might be that the first one (on the top left) is not the most relevant result. It won't shock me if google ranked the relevance of result from center to peripheral. In the end that's how we look at a pile of images--we tend to start from the middle. Try a few examples, from the ones I tried the middle row middle column image is much more relevant than the top left result.
I take that since your website didn't state any call limit, you must be using the second method (parsing google image result page).
Give user option to get more photo (in ways like: /1, /2) as the index of result page. Or even randomize the result (provide an option to random deliver the result, so same call doesn't result in the same image) might serve the user good. lots of ideas.
Thanks for your ideas. In fact it's using the image search api and I think I would have implemented social voting to replace search before the limit runs out (if any).
Quizlet is awesome, but the use-case for this new tool could be:
a) mentioning new vocab off-the-cuff (or in response to a question) in class and immediately showing a picture of it
b) students trying to remember a word and me showing them pictures of words they are guessing until they see the right thing
c) Us making up sample sentences on-the-fly and using a background picture as an impromptu slide: "Darth Vader is hungry. He could make a sandwich. He may toast some broad. He might eat it later": http://darth.vader.bread.jpg.to/
128 comments
[ 3.7 ms ] story [ 144 ms ] threadhn.jpg.to, on the other hand, is probably not related to hacker news :)
http://gigawatt.jpg.to/
http://bing.jpg.to/ (what you'd expect)
http://google.jpg.to/ (not what you'd expect)
Top result in Google Images? Looks like it anyway.
Thanks for giving me something to pay with while waiting on the slowest server in the world at work :)
After not seeing any, I decided to just try my name ... and found a football player.
I did a little bit of research about it few years ago; the summary can be seen here:
http://temporal.pr0.pl/devblog/2009/09/14/eksperyment-podsum... (pl_PL)
http://translate.google.com/translate?sl=pl&tl=en&js... (en_US)
Basically, I was able to distinguish between embedded and directly accessed images by analyzing the content of Accept header. It wasn't perfect (it didn't work for Opera), but I think it could be good enough to do something useful with it.
Awesome! THanks.
For example: http://kittens.jpg.to/301.jpg
Another feature could be to return a random result for that image search.
For example: http://ass.jpg.to/random.jpg
Being able to choose what I want for my placeholder would be extremely useful.
http://food.jpg.to/random300x500.jpg
http://placekitten.com/
And it's not even limited to jpg, cf. http://drumroll.jpg.to/
But then if I needed a 500x500 picture of Bruce Willis, because who doesn't, I could go to http://bruce_wilis.jpg.to/500x500
Fun little app. Nice work.
http://large.gorilla.eating.a.taco.jpg.to/
http://baby%2belephant.jpg.to/
behind-the-scene technical aspect are not difficult to realize (either google image API, which is deprecated, or parse the result of http://www.google.com/search?tbm=isch&hl=en&source=h... and get the first image url after a predefined string anchor, say <span class=rg_ctlv>), but the idea to simplify input and output is brilliant.
To deal with not-so-good image search result: since google image search is presenting the result in a thumbnail group, it might be worthwhile to look into their ranking scheme for the result. It might be that the first one (on the top left) is not the most relevant result. It won't shock me if google ranked the relevance of result from center to peripheral. In the end that's how we look at a pile of images--we tend to start from the middle. Try a few examples, from the ones I tried the middle row middle column image is much more relevant than the top left result.
just my two cents.
Give user option to get more photo (in ways like: /1, /2) as the index of result page. Or even randomize the result (provide an option to random deliver the result, so same call doesn't result in the same image) might serve the user good. lots of ideas.
The only issue is: I need a much larger rez image to put up on the big projector. Since this is doing Google Image Search... any chance for a:
http://keyword.jpg.to/large
http://keyword.jpg.to/medium
[plus]
http://keyword.jpg.to/photo
http://keyword.jpg.to/clipart
[plus]
http://keyword.jpg.to/red
http://keyword.jpg.to/green
http://keyword.jpg.to/white
a) mentioning new vocab off-the-cuff (or in response to a question) in class and immediately showing a picture of it
b) students trying to remember a word and me showing them pictures of words they are guessing until they see the right thing
c) Us making up sample sentences on-the-fly and using a background picture as an impromptu slide: "Darth Vader is hungry. He could make a sandwich. He may toast some broad. He might eat it later": http://darth.vader.bread.jpg.to/
Works like expected: http://fffffffuuuuuuuuuuuu.jpg.to
Reddit will have a field day with this.
There doesn't appear to be any size checking at all. I got a doozy:
http://senator.jpg.to/