05-03-2017, 01:27 AM
(This post was last modified: 05-03-2017, 01:28 AM by Raccoon Sam.)
they are special effects drawn by HDMA but there's nothing particularly difficult or different about them.
Consider the Platform Ghost in 1-8 – it looks like it's smooth vector graphics on an SNES game. Technically, it is. You can find the coordinate arrays in the ROM.
So, the graphics aren't stored in the traditional sense of 8x8 tiles stored in VRAM, but as 2D polygonal frames where the coordinates of the points are loaded from the ROM and displayed on the 256x224 screen.
This has interesting implications, because with displaying the same data with modern engines, you are given an extremely rare chance to not only transform sprites (scale mostly, but sure, rotate and skew or whatever) losslessly but also actually access the in-between frames!
The fall walls in 1-4 are done with a similar technique, but afaik they're never been documented. What I'm saying however is that while there's absolutely nothing wrong with screenshot ripping the walls the traditional way, finding the coordinate arrays is much better because it gives you infinite scalability.
Your game probably won't run in insanely huge resolution or above 60fps though, but that's something to think about ^
Consider the Platform Ghost in 1-8 – it looks like it's smooth vector graphics on an SNES game. Technically, it is. You can find the coordinate arrays in the ROM.
tehaxor69 Wrote:there are 0x64 bytes per frame, they are in groups of 2 -X-axis and Y-axis, the 3rd byte group also controls the position of the platform.
The data at ROM 0x0375EF SNES 0x06F3EF points to where the data is located, add 0x060000 to each one
0xF40B
0xF46F
0xF4D3
0xF537
0xF59B
0xF5FF
0xF663
0xF6C7
0xF72B
0xF78F
0xF7F3
0xF857
0xF8BB
0xF40B
Here is a render of all the frames
![[Image: i85ZAYW.png]](/image.php?src=http%3A%2F%2Fi.imgur.com%2Fi85ZAYW.png&hash=cbc0a810eb6e059d96b837b20cf9bf77)
So, the graphics aren't stored in the traditional sense of 8x8 tiles stored in VRAM, but as 2D polygonal frames where the coordinates of the points are loaded from the ROM and displayed on the 256x224 screen.
This has interesting implications, because with displaying the same data with modern engines, you are given an extremely rare chance to not only transform sprites (scale mostly, but sure, rotate and skew or whatever) losslessly but also actually access the in-between frames!
The fall walls in 1-4 are done with a similar technique, but afaik they're never been documented. What I'm saying however is that while there's absolutely nothing wrong with screenshot ripping the walls the traditional way, finding the coordinate arrays is much better because it gives you infinite scalability.
Your game probably won't run in insanely huge resolution or above 60fps though, but that's something to think about ^