DeviantArt Muro

I completely agree with you WRT downvoting... I'd much prefer a discussion like this to a buried comment, regardless of the merit of the original comment. Perhaps it was due to the perceived tone of your comment: "I don't understand why anyone would think it would be impossible" ~= "Who would be dumb enough to assume this isn't possible?" As for why anyone would claim the task is impossible - not enough imagination! :) Although it's more likely that they meant they thought it was impossible to achieve at a decent speed/framerate, which is a pretty reasonable assumption at first glance.

Technically, I think you're right - you could implement a pseudolayering system with only one canvas element. The limiting factor, however, is redrawing speed. You have essentially two options - every time a layer moves, you could redraw the entire canvas, which is conceptually easy but slow. Or you could do some calculation to determine exactly which pixels need to be redrawn - as you say, a matrix operation to determine a delta matrix. This may be faster than redrawing the entire canvas, but it gets much slower as you increase the number of layers. Speed may not seem like a huge issue here, but imagine clicking a layer and dragging it around - this would require constant (un-hardware-accelerated) canvas redrawing at ~10-30fps, which is probably pretty difficult to achieve with any significantly-sized canvas.

The brilliant canvas-on-top-of-canvas implementation avoids this problem altogether. While the layer being dragged still has to be redrawn, the process of doing so is made a lot easier because you don't have to shuffle around the data in the other layers in the process, you just have to worry about one layer at a time.

Impossible might have been hyperbole on my part. I recall a discussion about a different, simpler HTML5 drawing app (no layers) here on HN in which someone claimed that HTML5 is no alternative to Flash based solutions since layers would be – and that’s how I remember it – impossible with Canvas. It is possible that he just said hard, not impossible.

They're a company so their goal is (among other things) to make money which they do (among other things) by charging for premium membership. Assuming there's a business case for this (very cool) tool, the question is: will they make more money by allowing people to try the tool before signing up?

Seeing how the tool requires certain amounts 'points' for some of the brushes to be available, this would make the benefit of a premium account even more clear after toying with the tool.

So, to sum it up: for me, not being able to try out the tool immediately is a small frustration. Presumably, for other people without Deviant accounts this is also annoying. I believe, as argued above that they might be able to make more money if they let people play with the tool first. Regardless, that might not be their goal at all.