Most efficient manner to present graphics generated pixel-by-pixel performance wise?

[narkro555]

Hello,
I am designing an editor for an older video game as a side project. Part of it involves pulling graphics data from the game files, of which it is stored as interchangeable colors and pixels. I will need to generate this at runtime. Then, I will need to draw those 8x8 tiles to a scrollable grid.
I initially figured I'd see where it goes by experimenting, however, I soon realized that performance is definitely going to be an issue. The main grid is at least 255x255 tiles, which means the possibility of having 65k+ tiles being rendered. I have a couple of theories:

1. Render the 8x8 pixel arrays to a traditional image format like PNG or JPG, then place them in a 255x255 grid as backgrounds or images. Downside: massive number of image files in memory. Caching these beforehand would be even bigger since it would require saving a file for every color and pixel combination.
2. Make sub-grids inside the tiles that are 8x8 grids, each with a certain color, inside a 255x255 grid. Downside: Massive number of grid tiles in memory
3. Make the grid (255x8)x(255x8), being 1 tile per pixel, each with a colored background. Downside: Massive number of grid tiles in memory
4. Create a custom renderer. Downside: Will need to recreate all the functionality present from QT5

Which of these is the most pro-performance? Is there another option I'm not thinking of? This is for desktop, using C++.

Thank you

[SGaist]

Hi,

Did you consider OpenGL ?

[narkro555]

@SGaist That was my back-up plan. I would miss a lot of the functionality that QT provides out the gate, but if this framework is not right for this task, that's what I'll do.

[SGaist]

Qt integrates very well with OpenGL so I am unsure what you think you would be missing ?

That said, another possible way to do what you want could be the Graphics View Framework.