Not a real recomp sorta kinda

[D0gg0Man]

Title, this is not really a recomp more then it is a semi nativifier.
Not hating but dev wise its not human readable, I cant do anything with this as its all byte extraction.
My suggestions:

MPHRead already does a fantastic job at rendering, please look there to make things look good as your supersampling does not work as intended from what I can tell - https://github.com/NoneGiven/MphRead

Extract the data assets and allow loose file loading for devs to play around with

KB/mouse control settings, need to be able to adjust things like sens. Maybe actually make the keyboard and mouse function native instead of emulating it.

Oh and allow us to lock aspect ratio for heavens sake lol, make it a setting

Bugs :
Arch Lin, 6950xt and AMD 7800x3d here, black screen constantly in the 3d render window when 3rd person and in the morph ball
Extremely resource intensive for what the game is as well.

[mstan]

Hi. This issue feels a bit more like a conversation than entirely an issue report. If that's what this is, I'm happy to oblige.

Definition wise, the tooling lends itself more to a recomp than a decomp. These aren't strictly defined terms, but it is often the case that recompilations are, in fact, less legible when transformed into code from binary.
If a decompilation or annotated disassembly exists, you can certainly overlay it onto the decoder to make it more digestible. For example, I did this with the annotations for Super Mario World for SNES. Another example could be the decompilation of Mario Kart: Super Circuit for GBA using MKSC as a reference.

It is also important to note that yes, there is interpreter fallback in the project as the ecosystem is just starting out (more on that below). Those are stop gap measures that will disappear as ndsrecomp is able to mature. I know other projects, like ModernGekko for Gamecube and Wii, has a similar fallback strategy. But as a project matures, that should become increasingly inert.

MPHRead already does a fantastic job at rendering, please look there to make things look good as your supersampling does not work as intended from what I can tell - https://github.com/NoneGiven/MphRead
Happy to take a look at this and import what I am able, as licenses permit. I also very much welcome community contributions.

KB/mouse control settings, need to be able to adjust things like sens. Maybe actually make the keyboard and mouse function native instead of emulating it.
These can be modified in the MODS section before launch. They can't be done in game, but that will be an eventual future target. Likely something that a decompilation will help with in function replacement.

Extract the data assets and allow loose file loading for devs to play around with
For this, a decompilation is far better suited as these regions can be more readily defined. With regards to a recompilation, this is less approachable, given that the asset types cannot be readily identified nor what types of files they are.

I do not think this project will be forever mutually exclusive from such tasks. However, I think this conversation is teetering into the semantics of R&D I've done with decomp-annotated-recomps. I wrote an article about it here that I encourage you give a quick read. (Being pedantic, two of the three demonstrations are technically annotated disasm over retranslated asm, but the idea is the same)

Bugs :
Arch Lin, 6950xt and AMD 7800x3d here, black screen constantly in the 3d render window when 3rd person and in the morph ball

Extremely resource intensive for what the game is as well.
That is expected at this point. The way in which I build my ecosystems is iteration occurs on each game, and that where abstractable, I abstract it into the runtime. Each game's finalized and polished success benefits the recompiler which rolls forward to all other games that build against it as well (and the inverse is true. As other games are improved, so does future iterations of games like Metroid Prime Hunters). ndsrecomp is early yet, and has Metroid Prime Hunters as its first "released" title.

Optimizations will come, but I often rely on the community's crowdsourcing efforts as I'm still only one person at the end of the day, doing this as a volunteer effort. By comparision, psxrecomp is another project I run that has had a lot of community contribution, which has led to it receiving netplay, rollback, save states, rewind, extensibility, and tons of optimizations.

Metroid Prime Hunters nor ndsrecomp come close by comparision, but ongoing community efforts are what make this possible. The community providing feedback, as well as the community contributing. For example, I am very grateful to @MysteriousAeon who has been providing PRs to ndsrecomp and engaging with the community here to improve the project. It has already been materially improved through efforts put forth, and I seek to have more of that, as it will pay dividends for this project and any future project that anyone chooses to pursue using the ndsrecomp ecosystem.

I hope this was insightful and gives you an idea of where this project is and where it is heading.

[mstan]

https://github.com/mstan/MetroidPrimeHuntersRecomp/releases/tag/v0.6.1-alpha

Following up, the adaptive widescreen mod has been modified to offer adaptive or one of several fixed resolutions. Let me know if this fits your expectations. Happy to take feedback if not.