Recent content by mirrorbender

Achillies, My suggestion for the widescreen labeling is as follows: The current "TV" option should be labeled "Standard Widescreen", perhaps with an "if unsure, use this option" message. In most cases, this is the best option, and even when used on a monitor, it's only a tiny bit (~1%) off...

A note about the different widescreen modes: Since I came up with the numbers for the three different versions of 16:9, I authored a change to how dolphin handles aspect ratios. By default, dolphin will now emulate the aspect ratios found on TV's, meaning that by default, the "TV" version will...

Good question. Dolphin's "Force 16:9" option now lines up with what a TV displays, so the "TV" version of the code works best with that. If you are stretching the aspect ratio to fill a 16:9 monitor, then the old "dolphin" version is still best, but I'd actually recommend using the "TV" code...

Update: Nintendont recently added support for manually setting the width of the output (i.e. doing exactly what this code does), so if you use nintendont to load melee, you don't need to use this code anymore, you can just set the width to 720, use the true 16:9 code and it should be beautiful.

A correction about S-Video cables: the Wii (not sure about the gamecube) simultaneously outputs both composite and s-video, so as long as the cable is made correctly, there is no problem with using an S-Video/Composite combo cable. Here is documentation for the pinout of the Wii AV multi-out...

Yeah, wierd bug. My first guess is that there is some kind of conflict with the registers being used to load the FOV multipliers. I didn't write the original code, @Dan Salvato did, so maybe he has more insight as to what might be causing it. It definitely could just be a 20XX problem.

If the two values set to be the same, the code should do nothing. It is multiplying the FOV by the XXXX and then dividing it by YYYY (which are each hexadecimal floating point numbers), so if they are the same XXXX/YYYY = 1, and nothing should change.

I think I've worked out how the scaling register works, so I'm going to mark this thread as completed. The best value is 227 (0xE3, second to last line), which scales the width by 256/227. (256/227)*640 = 721.76, which is as close to 720 as you can get using fractions with a numerator of 256...

Hey guys little OCD update here. I have decided that the best benchmark for a perfect circle is the shield, rather than the CSS token, because a) it is a 2D sprite, so there are no weird perspective transformations going on, b) I know it is supposed to be a perfect circle (the shield sprite is...

This code doesn't do anything in dolphin, as dolphin grabs the image directly from one of the frame buffers, before the video interface creates the analog signal. Here is a simulated image of what the difference looks like on a console. With code vs Without code: [spoiler] With code: Without...

I spent this past weekend working on this. Here's the code: C234FD8C 00000002 7D484378 3908EC00 60000000 00000000 C234FDB8 00000004 7D000378 91010000 39000000 39080050 7C080214 81010000 60000000 00000000 42000000 CC000000 0200204A 000010E3 42000000 80000000 For best results, combine with...

Hey Dan, thanks for posting this so that more people can find it (and for giving me my credit). RazzleDazzle, the difference has to do with the area of the screen that is actually game image in all 3 scenarios. For dolphin, it's the entire screen. On 16:9 monitors with a console, it's less...