1
Newcomer's Board / Re: NES Mapper Conversion
« on: March 03, 2020, 02:50:14 pm »
Gotcha, that makes more sense now. Thanks for the clarification of that (and of course for being very patient with me throughout this process)! Should the bad news be the end of this project, I feel like I learned quite a bit already thanks to you.
I do have the games as individual ROMs and confirmed they are MMC3 (mapper 4) and NROM (mapper 0), at least based on their iNES headers. I have also reached out to the source of where I got them from for any potential verification on the matter. However, the games seem to have no issues in emulators. Using Super Mario Bros. as an example of NROM and Dragon Power as an example of GNROM, I changed the mappers of each game to 0, 3, 4, 5, 6, 9, and 66 (numbers selected for no reason in particular) in the iNES header and ran the modified games to see the results. Dragon Power only would display a blank screen on the mappers aside from 66, whereas Super Mario Bros. would still be playable on all of the above, with some minor corruption in a couple of the mappers and fairly major corruption on mapper 9. The games listed as NROM behave exactly the same as Super Mario Bros, and not as Dragon Power. Not saying you or the wiki is wrong by any means, just my observations in the spare time I've had today.
Edit:
Yeah this project sounds dead, unfortunately. I'm assuming this is less about the games individually and more about how they were stored/swapped. From my source:
I do have the games as individual ROMs and confirmed they are MMC3 (mapper 4) and NROM (mapper 0), at least based on their iNES headers. I have also reached out to the source of where I got them from for any potential verification on the matter. However, the games seem to have no issues in emulators. Using Super Mario Bros. as an example of NROM and Dragon Power as an example of GNROM, I changed the mappers of each game to 0, 3, 4, 5, 6, 9, and 66 (numbers selected for no reason in particular) in the iNES header and ran the modified games to see the results. Dragon Power only would display a blank screen on the mappers aside from 66, whereas Super Mario Bros. would still be playable on all of the above, with some minor corruption in a couple of the mappers and fairly major corruption on mapper 9. The games listed as NROM behave exactly the same as Super Mario Bros, and not as Dragon Power. Not saying you or the wiki is wrong by any means, just my observations in the spare time I've had today.
Edit:
Yeah this project sounds dead, unfortunately. I'm assuming this is less about the games individually and more about how they were stored/swapped. From my source:
Quote
That is probably the worst ROM for starting to learn mapper conversion, because mapper 534 has quite a few obfuscatory aspects to it. The "GNROM" term used on the mapper 534 wiki page should be more understood to be "GNROM-like" rather than actual GNROM. Mapper 534 has GNROM-like functionality, meaning that the fixed bank at $E000-$FFFF is overcome by making MMC regs 6 and 7 apply both to $8000-$BFFF and $C000-$FFFF, as described before the "Note" on that wiki page. (In addition, there is also the outer bank register that can switch 128 KiB or 256 KiB PRG, which *also* affects the fixed bank.) The best way to understand this is to make some kind of truth table that tells you which address bit will come from which register (MMC3, mapper outer bank, CPU) depending on the CPU and mapper outer bank mode bits.Being realistic, this is sounding beyond my capabilities at this point, even with my hand held. Bit bummed about that, but at least I did learn quite a bit, and perhaps others can as well. At some point, I shall revisit this when I have more knowledge and experience; maybe it would be possible to recreate the menu with a more appropriate mapper. In any case, thank you very much, Disch, for all the knowledge you've shared with myself and everyone here! I hope to still make use of it, just perhaps with a different project.
The problem with converting this thing to mapper 52 is that mapper 52 has no such GNROM-like functionality, so it's an unsuitable choice for a conversion's target mapper number.