Show HN: Repaint – a WebGL based website builder (repaint.com)

236 points by benshumaker ↗ HN
Hey HN! We're Ben and Izak, founders of Repaint (YC S23). Repaint is like Figma, but for creating real websites.

It has panning, zooming, and dragging elements around. The settings closely follow html/css. We think an open canvas is a big improvement over other website builders. Everything is easier: styling, consistency, responsiveness…

But making it work was hard! We thought HN would appreciate the tech behind it: - A custom WebGL rendering engine (w/text, shadows, blurs, gradients, & opacity groups) - A partial implementation of the css spec - A custom text editor & text shaper - A transformer to turn designs into publishable html/css

Repaint is a design tool for html/css. But internally, it doesn’t actually use html/css itself. All your designs live in a single <canvas> element.

We wanted users to be able to see all their content on one screen. We target +60fps, so frames only have 16ms to render. The browser’s layout engine couldn’t handle simple actions, like zooming, with many thousands of nodes on the screen. To fix that, we wrote a rendering engine in WebGL.

Since we use custom rendering, we had to create a lot of built-in browser behavior ourselves.

Users modify a large dom-like data-structure in our editor. Each node in the document has a set of css-like styles. We created a layout engine that turns this document into a list of positions and sizes. We feed these values into the rendering engine. Our layout engine lets us display proper html layouts without using the browser's layout engine. We support both flex and block layouts. We already support multiple layout units and properties (px, %, mins, maxes, etc.).

We also can’t use the browser’s built-in text editor, so we made one ourselves. We implemented all the common text editor features regarding selection and content editing (clicking, selection, hotkeys, inline styling, etc.). The state consists of content and selection. The inputs are keystrokes, clicks, and style changes. The updated state is used to render text, selection boxes, and the cursor.

When you publish a website, we turn our internal document into an html document. We've intentionally structured our document to feel similar to a dom tree. Each node has a 1:1 mapping with an html element. Nodes have a list of breakpoints which represent media-queries. We apply the styles by turning them into selectors. All of the html pages are saved and stored on our fileserver for hosting.

We made a playground for HN, so you can try it yourself. Now that the tech works, we’d love feedback and ideas for improving the product experience.

And we’d love to meet more builders interested in the space. If that’s you, feel free to say hello in the comments! Or you can reach Ben from his website.

Playground: https://app.repaint.com/playground

Demo Vid: https://www.loom.com/share/03ee81317c224189bfa202d3eacfa3c3?...

Main Website: https://repaint.com/

Contact: https://benshumaker.xyz/

166 comments

[ 5.8 ms ] story [ 413 ms ] thread
This is very cool! How do you guys compare to the incumbents like Webflow and Framer?
Thanks!

We think our ux foundation is stronger. Many high-frequency actions are faster: moving between pages, copying content between them, multi-select editing. Especially if you're used to Figma. A lot of designers feel like Webflow is notoriously cumbersome to use.

But we also want to keep it very close to html/css. We think it's the only way too push power really far. Webflow did that. Framer's diverged quite a bit. We imagine it might go beyond a website builder to be a generally good "design tool for code"

I'm obviously biased because we've been in the same YC batch, but I just want to say that one thing that always impressed me about you guys is how obsessed you were with building the best product you could.

When you explained how you had just reimplemented the CSS spec I was mindblown.

You guys rock. Repaint is amazing.

Congrats on launching. WebGL is really hard, I built a similar engine and it's tougher than it seems.

Your budget is only 8.33ms if you target 120hz too btw :) And realistically less because you don't get all of it.

I would argue a web designer might not even need 30fps.
Whenever my frame rate drops on larger projects I can feel the difference. We probably don't need higher frames, but it feels better.
640K should be enough for anybody
Will it ever be possible to export the code, or will it follow the Framer playbook?
We don't have any plans either way, but we've been thinking about this a lot. We've realized Repaint might be useful as a Figma that makes real code, so we're open to alternative business models.

Frankly, we've been very focused on making Repaint a great design tool to build websites and haven't thought about hosting that much.

Would you want to self host? Or maybe edit the code output?

As a Dreamweaver alternative would be interesting.
I wouldn't even consider Repaint if i don't have access to the source code. I see it as a good starting point (like Figma), but i'll never rely on a WYSIWYG editor alone.

As presented only makes sense for competing with Wix for power users.

Same. If there were a good WYSIWYG editor that would output well-formed Javascript (especially React) that was easy to edit/maintain by hand, I'd pay in a heartbeat. Everything I've tried in the past had too much friction. Something like Tailwind UI in an editor that spat out the same type of code as Tailwind UI gives would be killer.

But to the point, if my output is locked to the tool, I'm not going to use it. If I can get at least an uglified version of the source that I can host on my own Nginx, then it's a possibility, but not having at least that is a total deal breaker for me.

Why is it a deal breaker? For marketing websites and landing pages, I would prefer not to have to write any code because the frequency of changes is so high and I'd rather have my engineering team focus on building product.

If you're talking about wanting to use this tool as a starting point for actual applications, then I would see why you'd need to export JS/React code.

Well maybe you found a very specific niche, but for mkt landing page most ppl would prefer one of the very mature competitors that collect analytics, offer insights, capture email, have templates, etc.
Congrats on launch, incredible work!

Btw, your pricing is broken :)

I wonder if there is ever a world where the websites themselves are just WebGL, and no one has to worry about CSS ever again. Ignoring a11y and CPU/GPU usage, that is...

Edit/addition: Congrats on your product! Always great to see a high-quality, pure software project get funded and realized. If I ever do YC (non-ivy so probably not in the cards) I'd probably do something like this.

We dream about this all the time. Unfortunately, it breaks too much default behavior like browser extensions, password managers, SEO...

Maybe one day :)

I guess there'd need to be an API for extensions to treat WebGL rendered text as synthetic DOM node. Maybe WebGL apps expose and update a "virtual DOM" that represents how extensions/Google should interpret what is rendered...

Repaint might be in some sense the bridge, though!

Web.GPT.GL_AgI bot proxy to handle such pedestrian digital quibbles, indeed ;-)
I went to a non-ivy school and dropped out. There's always hope :)
Things that annoy me about Figma:

- No way to set defaults for new frames etc ( that I know of )

- Some things rely on mouse input (setting width modes etc) which breaks flow

- Generally not for power users

- No decent code export

- No stylesheets for retheming components /showing variation (I know there's a Variant workflow for it but it's not simple to restyle an entire hierarchy)

Would love to see you tackle those if you're not already. Great to see you following CSS more closely, makes so much sense.

But if you aren't allowing us to extract the code / self host then I'll be passing on it. Digital feudalism

Congratulations. The product looks very interesting.

Also congratulations on building a canvas engine library! Nothing is easy when it comes to canvas. Creating one for the WebGL canvas is magnitudes of more difficulty compared to 2D (which is what my library maps to).

I'd love to hear your horror stories when it came to coding up the text layout functionality and text editor - that is, in my view, the hardest part of the job: I may have invented a few new swear words while tackling that work.

I'd also be interested in your plans for accessibility. Not just making the tool accessible, but how such a tool may push designers to consider accessibility as a first class requirement for the end user in the projects they build with the tool.

Best wishes and great fortune!

Man you're right. The text layout system was brutal. I had a particularly memorable day where I couldn't figure out why our line height didn't match the browser (and each browser was different). I fixed it by adding a constant that was buried in the font files. It took 12 hours. It was a one liner.

Haven't planned accessibility help yet. But I think simple warnings could go far. When I break guidelines, it's usually by accident.

I’m betting the value you were looking was in the OS/2 table in the TrueType/OpenType files.

Every time I have to interact with font metrics data directly, I pour one out for IBM’s greatest and most tragic PC desktop OS.

This looks cool. Here's my feature request list (can't tell if these are there yet):

- import existing html/css. This would allow me to use it for non-greenfield projects, and allow for back-and-forth workflows.

- mark nodes with my own semantic ids that are included in the html export. I would use these to post process exported html to create dynamic templates/components/views. I don't want to be tied to a specific framework.

Looks insane! Since you mentioned Figma, a Figma project import feature would be very useful for a project like this.
I’m curious what you guys used for generating text. Canvas 2D to a texture?

And congrats. I’ve dabbled a little in trying to use WebGL in places where DOM would the typical choice, and quickly ran away. :)

We use msdf textures. Generate them dynamically when people upload fonts. We tried 3 methods, and we're sticking with msdf for now because its very fast.
Don't zoom in too close on the text :)
I just wish it would let me get the feel without any data collection first and as soon as I look like I'm interested .. I want to share, save, do something more advanced .. only then go fish.
We made the playground so that you can try Repaint without signing up.
(comment deleted)
what playground? the linked website has only a "Get started" button which leads to a signup form.
At the top of this post there is a link to the playground.

Here's the link: https://app.repaint.com/playground

So anyone discovering repaint organically has no chance to find it? Essentially gatekeeping it to HN insiders? interesting approach
how do you deal with Information Architecture, SEO, Interlinking, etc etc..?
Honestly we haven't focused on this yet. It's important that our websites are accessible and score well for SEO. Users will have settings to improve SEO on their site.
About pricing, it's strange that neither specify access to the generated source code?

Not sure what audience you have in mind.

This is very impressive!

The rendered pages have noticeable aliasing when zooming out in the playground. Is there a reason why the WebGL context isn't configured with anti-aliasing enabled?

Yeah, I really want to improve the AA. The built-in AA only supports 4 samples. In testing, it didn't seem to do much. My research suggested most apps like this handle it at the shader level, and we just haven't gotten to perfecting them all.
Yes, WebGL's built in antialiasing is MSAA, not supersampling. MSAA only affects the edges of triangles. It doesn't help aliasing introduced by shaders at all.

Supersampling is easy to implement, just set the canvas size to a multiple of the viewport size and scale it down with CSS. It will help with all forms of aliasing, however it gets very slow very quickly. Shader aliasing is better solved analytically whenever possible.

One trick that you can use when aliasing is difficult to solve is to use supersampling but only when the scene is static. Google Maps WebGL does this. While you are dragging the map it is aliased. As soon as you stop dragging and the map becomes stationary, it renders 16 frames with subpixel offsets and blends them together to produce one supersampled frame (this is better than resizing the canvas because it can be interrupted in the middle if the user starts dragging again). This doesn't work for dynamic games but it might be suitable for your application with a lot of static content.

Whoa, that's brilliant. I think that just works for out app. Is there a public reference I could look at?
Try searching for accumulation buffer antialiasing. OpenGL ES (WebGL) doesn't directly support an "accumulation buffer" the way desktop OpenGL does, but you can emulate it with render to texture and appropriate blending settings.

Now that I think about it, it may not work as well for your app as Google Maps. The thing about Google Maps is that when you zoom or pan the map the whole screen moves. In your app when you drag something, everything else remains static. The static items on screen will flicker between aliased and smooth whenever you drag something else. That will be distracting.

This looks great!

Ill definately be building something with this. Il give more detailed feedback after use - but from vid:

Have you used SPLINE at all and looked at their events menu system for anims in Spline (which BTW you should take a look at how to integrate and work with the Spline folks as their app would be an indeal way to make interactive 3d UI/X on this as well...

But also - (il test this,) but UI scaling for the windows to get them out of way given canvas, the ability to s=zoom a single element/page on the canvas?>

And the ability to float menu panels to a second screen? (I have a USb monitor on laptop, move all the menu panels to that usb monitor and have the canvas full screen laptop)

-Excited to try this out later today.

https://old.reddit.com/r/Spline3D/top/

Thanks for the notes! I'll look into spline.

I've never of a web app that lets you pull panels to a second screen. Sounds neat though. Is there a good reference?

You can use window.open to create a child window whose DOM is writable by the main thread of the parent window. You can then mount a React root in the child and go off to the races.

This issue with this strategy is usually over reliance on 3rd party code that assumes that globalThis.window is the only window, and event handlers don’t end up working or something.

If you are using good libraries that really understand the DOM (React itself handles this well) or you write things yourself, it’s a bit of bother to add “target window” as a parameter to various functions that were previously using the global but quite doable.

I built an electron app that has one “root” window that is hidden that I use to spawn child windows without needing to load an entire new web page & instance of the JS bundle into each window. I used Tldraw and had to do some patch-package to fix global window assumptions in a few places.

Here’s my code: https://github.com/justjake/tlshot/blob/main/src/renderer/ed...

Looks great!

The video felt a little sped up. Not sure whether that's your natural talking cadence, or the video was 1.5x, but slightly slower would be great.

Also, at first the video focusses on the nerdy interesting things like fast pan & zoom, vs the things less-technically-inclined users would care about.

I noticed the same thing. For some reason, the video (player?) defaults to 1.2x speed. You can change it in the video controls at the bottom.
That's the default speed that Loom plays, unfortunately we can't update it.
My slack status at work is “please don’t use loom” and the emoji is the loom logo crossed out

    How does Repaint handle responsiveness?
    
    First you design your website for desktop. Then you make adjustments on different breakpoints. In design terms, Repaint uses autolayout and breakpoints.
I'm curious, why did you choose to go with desktop-first rather than mobile-first here? I'm not saying it's wrong. I'm mostly curious what the thought process was, because my very limited experience has been that mobile-first is a lot easier.

It's easy to go all out with a desktop design and discover it's not practical for a mobile view, nor is it easy to find ways to scale it down and have it make sense both aesthetically and technically for a smaller screen. On the other hand, it's not so difficult to start with limitations that can then scale out into something else as screen real estate allows.

Are these potential issues mitigated by how the tools allow you to layout and design things in Repaint?

Another thing, what are your plans around accessibility and semantics? How is the SEO for these sites? I noticed "Welcome to the Playground" in the playground is a p tag with an id. Almost everything is divs and p tags. Not to detract from what is clearly incredible software so far; you're all doing an awesome job. I'd be extremely happy with myself if I accomplished this.

edit: I ran a lighthouse test on the preview and I see the SEO and accessibility are higher than I expected. The performance score is excellent. The others aren't great, but they likely beat a lot stuff out there. I mean, visit any recipe website and weep for the modern internet.

edit edit: I'm realizing you could probably generate some decent document structure/semantics by having the user indicate the intent of some root items, like headers, subtitles, and sections. That's plenty to get some great meta and intent, right? You've got the hierarchy of the page, you can infer some info from properties of the containers and text, etc. There's a ton of potential to get excellent results here without too much knowledge or input from the users.

My totally unasked for suggestions:

    - a section in the editor for adding these details
    - components
Honestly, we did desktop first because that seemed to be the default for most websites we looked at.

We plan on giving the users control over what tag each element uses for SEO purposes (h1, section, etc.).

We have some SEO related settings that can be set (descriptions, title, etc.) and plan to add more.

You're right that we could make good defaults for SEO, it would probably improve our scores a lot. It would be cool if we could add some sort of checker directly in the editor. It could also give tips on how to remediate.
Looking at the results from lighthouse scores, you can totally translate that into a helpful guide for your users. You can even highlight the objects on the page with missing/poorly defined attributes.

This actually sounds like fun to implement. I suppose it should be opt-in, because some users genuinely won't care. And I imagine doing as much as possible by default would be ideal too, because this can turn into a huge chore which might make your product seem less efficient and convenient by no fault of its own.

Framer does the same thing regarding responsive layouts: desktop > tablet > mobile.
What is an open canvas? What problem does it solve?
An open canvas means you can see all the content in your website on one screen, like Figma. This is different from other website builders.

This makes editing your content and navigating your project easier.

This is incredibly impressive engineering and a wonderful UI! As a dev, I would have to agree with some of the other comments here about access to generated code. I'd definitely suggest this to non-tech friends but I wouldn't see much of a use for it personally without code output.

I was curious what your plans are for code output in the future, have you considered some kind of React integration? If this tool had Storybook-style codebase integration I would 100% pay for a subscription. For example, being able to export the website design as a React project with organized components, allow me to modify the component code to customize behavior, and continue to iterate on the design and apply changes to my codebase?

I think an important thing for me would be to have the ability to continue to modify the design after exporting and be able to apply the design changes without having to lose my code changes.

Obviously just my personal preference so take that with a grain of salt, but I know with that feature this would completely replace Figma in my workflow and would definitely be worth paying for.

Congratulations on your launch!

Code export wasn't our original plan, but we've been thinking about it more and more. It seems like Repaint is only a couple steps away from a pretty sweet "design tool for real code".

You're right about the exporting though. We're not sure exactly how to synchronize a code editor & Repaint editor for two way updates.

What if the Repaint Project required me to link it to a git repo and make my repaint account have git access and my files live and read out of my git repo?
I'm confused - why can't this export as static HTML / CSS? Would make it useful for github pages. If you make it free for exported code as GPL or some super-restrictive license that only works for true free software, where nothing commercial can ever be used for what is exported, seems quite useful
I don't get it. If you can't export HTML and it can only be hosted on servers you control, why did you output HTML internally? Why not just stick to your own proprietary layout format and rendering engine?
Since almost all websites use HTML / CSS, its the output format we wanted to target. If you don't use HTML it breaks things like SEO, accessibility, etc.
this doesn't seem to support non-english language that requires input methods.
IMHO, your strategy has always been the Correct Answer™. "Web browsers" should be based on URI/URL, HTTP, a canvas, and a stock virtual machine (eg WASM).

HTML/CSS support should just be another MIME type among many. Java Applets, Flash/Shockwave, PDF, Markdown, VRML-97, Quake .map, or whatever. Most sites would use one of the many stock renderer & runtimes. Some would implement their own, like your CSS reimplementation, as needed.

In other words, just skip the need for HTML shims.

Lastly, your efforts bring us that much closer to bringing us full circle. Next step: implement Wayland (and/or X11 if you're nostalgic).

You're right, we feel trapped in the current model for serving content on the internet. We've talked internally about a browser API that would allow you to override certain CSS layout behavior. If I could extend / modify the browser's layout engine, I would gladly do so.
Login to try a website? No, thanks.
There's a playground in the post if you want to try it without making an account :)
Dude. PLEASE MAKE IT EMBEDDED!!!

I need to integrate a "site builder" for a landing page service. I cant use things like the other options out there because they require actually going to the site. It makes it useless where i want a fully self contained "builder" for a number (bunch) of different landing page domains.

What do you mean by embedded? Like the editor can be embedded on other sites? Or the websites we publish are embeddable?
Make the editor embedded in another app. so I can just parameterize the "builder." I want to have the HTML on my own though so i can actually host the page myself. really I just want some fancy editor that outputs a JSON schema.

I literally almost built this myself twice, and just never got around to it.