I made a Canvas class that takes in a 2-dimensional array with Color3 values and creates frames accordingly, with a greedy meshing algorithm to reduce the frame count. This will be used in tandem with another project I’m doing, but I’ll make another post on that soon
It takes ~5s to draw the below 1920x1080 image with only 2 colors using 19282 frames, compared to the theoretical 2073600 if it used a frame per pixel.
However, this is a very niche dataset, and performs terribly on random pixels because it can’t group them together. I chose perlin noise because it comprises of contiguous spaces and is therefore easier to group. I will need to do more to make it faster and more efficient with frames. I plan on using the UIGradient technique shown in this post to further reduce frame counts.
If you have any feedback on how I could further reduce frame counts, do let me know
No, it reads Color3 values from a 2-dimensional array and displays them. Here is it generating a random color image (192x108)
The total pixels in the array are the same as the frames created, so this is the worst-case scenario. It takes 6 seconds and 20736 frames, which is already worse than the above example.
Very neat indeed!
Thanks for the link, I’ve never heard of greedy meshing, do you think it could increase performance in my application as well?
Also, I have to ask how in the world are you creating so many frames? In my testing, and on the account of others tests, it’s found that at a certain point gui objects will stop rendering to save on memory.
Any idea’s on why your scene specifically defies those previous tests?
I thought it might’ve been because I was using a SurfaceGui, so I painstakingly rendered 62500 (250x250) pixels to a ScreenGui and found that no, it still works.
