rikschennink

I imagine that would be the same as a black fill, just in a different color with noise.

I've used the original image data as a fill, but scaled down, so you get a mosaic effect. Then I randomise the tiles in that mosaic and then I blur. The result is that it seems like the original data was redacted but in reality the original data has been scrambled to such an extent that it can no longer be retrieved.

I'm the developer behind Redact.Photo [0], instead of blurring or pixelating the image, I've added an additional step to improve data security. See "how it works" section on the page.

1. Image is scaled down to create a mosaic effect

2. Pixels are moved around a random offset of x and y axis

3. Blurring is applied

Adding black bars over information will always be better but this does result in more smooth redaction that I think cannot be undone because of the randomisation step.

[0] https://redact.photo

Thanks! You’re right that this is targeted at people that are doing quick casual photo editing :)

Not op but I created https://edit.photo just for that :)

It’s 100% private, editing happens in the browser.

I've added a caption to try and make that more clear :)

It’s accompanied by new.photo

use https://new.photo/800/600/white to create a new blank canvas on edit.photo

Thanks!

That's a good point. In theory you could check network tab to see that I'm not sending any data over the line.

Suggestions on how I could create that trust are more then welcome :)

I wanted to build a more secure image redaction tool but not resort to black (which is I suspect the best way to redact)

So instead of blurring, this tool first scales down the image, then scrambles pixels randomly, and then blurs. This should (disclaimer: not a scientist) make it impossible to revert information back to its original state.

It should also prevent uncensored characters giving away the blurred shape of censored characters as those will be randomly scrambled.

Additionally the resolution of the blurred version of the image will be lowered when large parts of the image are redacted.

I thought I'd post it here on HN because I'm genuinely interested in if my assumptions are correct.

https://doka.photo/ an browser based image editor and https://pqina.nl/doka/ the editor component doka.photo is based on.

Disclaimer: I'm the developer behind these tools :)

Thanks :)

Every operation is just way faster in WebGL.

I currently use WebGL for display preview and 2D canvas for generating the actual image.

I suspect you'll run into performance issues if you plan to animate the image in a 2D canvas context (like I do when rotating / flipping / etc.)

I've been planning to use offscreencanvas but haven't gotten around to it yet.

Thanks, I'll look into this. If I understand correctly you do get the "WebGL isn't enabled" message?

There is a limit on canvas size in most browsers, unfortunately it's rather low on iOS, more than 12 megapixels (4000x3000) will result in empty canvas elements.

Indeed progressively loading high res textures or tiling is a solution to this.

Hey, thanks for reporting, that's very odd, if you're familiar with web development can you check if your browser is throwing any errors?

Nope, not for the online image editor, you can edit as many images as you like.

You're absolutely right, you either go for CSS 3D Transforms, or WebGL, WebGL being faster still.

Hi, Thanks for feedback!

In my experience loading huge images like those and animating them is way too slow so the preview is limited to 1500x1500 pixels (I'm testing on a MacBook from 2014). I have some ideas to work around that. With the recent switch from CSS transforms to WebGL (for the image preview) I've added some additional options.

Good luck with your project!

Awesome, let me know if you have any questions, just use that chat widget thingy on the product page.

I think you had to zoom out the image (scale down with scroll wheel or multi-touch), maybe this needs to be more clear.

Yes, it's the perfect fit, I have someone else to thank for that as I'm horrible at naming things.

It'll show the actual output size of the image, I'm not a 100% happy with this yet. The fact that you report it here (as a bug) is a good indicator that it needs to be changed.

Thanks! And excellent idea, added to the todo list.

Thank you :-)

Thanks!

Initially I rendered the image preview with CSS 3D transforms, while super fast on Safari, it turned out this wasn't as performant on Chrome when viewing on a retina display (weird right?). So I switched to WebGL, which sits closer (closest?) to the hardware and therefor yields better results. It's now super fast, it can animate filter transitions at 60FPS.

SVGs currently don't support hardware acceleration, so that's why I went with CSS transforms for other parts of the UI. The only things rendered with SVG at the moment are the markup, the UI icons, and the rotation control.

I specifically build it to work well across multiple different contexts :)

Thanks! And indeed, it's a full feature instance of Doka, I figured it might be useful for some people to do quick image editing and offering it for free feels good.

It isn't open source, I do use income from Doka to finance my other open source products (like FilePond).

I haven't been able to test this as my test device is too old. But, I'm betting my money on Firefox somehow rendering `background-attachment: fixed` differently from Webkit/Blink which is causing the slowdown.

I'm looking at switching to GPU accelerated transforms instead but that brings a couple of new challenges.

Nuts, that doesn't sound good, I haven't been able to test on Android because my Android device does not have a Gyroscope :\ Hope to fix this next week.

Thanks!

Apparently Apple has a patent pending.

Bob below has build what you describe above, so he might receive a letter from Apple in the near future ;)

https://twitter.com/bob_burrough/status/1080191577887141890