seed
seed copied to clipboard
nodebox
Allow users to export animated GIFs
When creating an animation, it would be nice if users could export the animation as an animated GIF.
This is challenging for a number of reasons:
- The output are HTML / SVG elements, not a canvas tag, so we can't use something like GIF.js
- The element we want to convert to GIF can be any element, of any size. The size can even change between frames.
One idea is to create a server-side renderer that could take seed sketches and render them in a headless browser, like Headless Chrome or PhantomJS. This would work as a webservice: we request a URL and it returns a GIF image.
Are you planning to render the animation, then screencast it and finally convert it into a GIF using some other webservice?
I would suggest creating a special URL route that only shows the output (like the embed view, but without the source editor and navigation bar), then use Chrome Headless to screencapture the page, then use something like GIF.js to generate a GIF file out of that.
This requires setting up a custom webservice (preferably using Heroku or Docker) that can convert sketches to GIF on-demand.
Yes, a basic webservice that takes in a Seed URL or Sketch address and outputs a GIF image. This could be the API address:
https://seedgif.herokuapp.com/create?sketch=-L4uLB99JzFHMPcVoiTm
Here's some documentation on running Headless Chrome on Heroku: https://timleland.com/headless-chrome-on-heroku/
The official Heroku buildpack looks promising: https://elements.heroku.com/buildpacks/minted/heroku-buildpack-chrome-headless
I have had a look over the links that you provided. It seems like the first focus should be on being able to make and run the app locally.
I was trying headless chrome locally using Puppeteer
but I am facing this error
My code is this
Are you familiar with this kind of errors? I am not being able to figure a way out of this.
No idea. Could you try a different site? Maybe google.com does not support headless browsing?
Works for me:
Here's the code:
const puppeteer = require('puppeteer');
(async() => {
const browser = await puppeteer.launch()
const page = await browser.newPage();
await page.goto('https://seed.emrg.be/');
await page.screenshot({path: 'seed.png'});
await browser.close();
})();
The main issue now is that the actual sketch page loads the script async, so it shows the "loading..." indicator when screenshotting the page:
I've also linked issue #32 since we don't want to screenshot the entire UI, just the viewer portion.
Wait, how are you able to work with async - await syntax in Node.js without try - catch phrase.
Doesn't it report UnhandledPromiseRejectionWarning?
The ECONNRESET is an error that comes back from the network stack. It looks like the requests are blocked on the network level somehow — perhaps overly aggressive filtering by some router / firewall?
I think you should try with different sites.
I tried several sites with no hope.
Could you send me your package.json file please?
Here's mine:
{
"name": "headless",
"version": "1.0.0",
"description": "",
"main": "index.js",
"scripts": {
"test": "echo \"Error: no test specified\" && exit 1"
},
"author": "",
"license": "ISC",
"dependencies": {
"puppeteer": "1.1.0"
}
}
And here's the index.js:
const puppeteer = require('puppeteer');
(async() => {
const browser = await puppeteer.launch()
const page = await browser.newPage();
await page.goto('https://seed.emrg.be/sketch/-L4uLB99JzFHMPcVoiTm');
await page.screenshot({path: 'seed.png'});
await browser.close();
})();
Nope, I tried that too. Didn't work. I will have to try it on one of my peers' system.
@fdb I was finally able to get puppeteer running (though was not able to solve the above issue). Now I am able to take screenshots of the browser. However, have you decided upon the method to be used to record the screen, or is it yet to be figured out?
Since the built-in animations are entirely deterministic, you could calculate the duration * FPS to get the amount of frames. Then, loop through each frame and set the internal frame counter and render out each image in sequence. Then, make a GIF out of this image sequence.
Initially it would be good to have this working for just a static image import, which is useful by itself.
Here's a crappy mockup of how that UI might look:
@fdb Yes, the static image part can be achieved now. However, I doubt if puppeteer is so accurate to keep up with the FPS of animations. 🤔
For the animation I would just render out an image sequence with the frame counter set to a specific value. So we don't render animations realtime; instead we export individual images at the exact frame time we want. We can then stitch these image sequences together using GIF.js or ffmpeg.
Okay, I understand your method.
But how can we account for the delay that will exist between the script calling the screenshot method and Puppeteer actually taking the screenshot?
None of it needs to be realtime. We can just pass the frame we want to export as a URL parameter, e.g. https://seed.emrg.be/view/SKETCH_ID_HERE?seed=ABC&frame=5 Then do this repeatedly for every frame.