Killer Instinct (Arcade Mega Patch)
Zeldix :: MSU-1 Hacking :: MSU-1 Hacks Database :: Fighting :: Other
Page 10 of 12
Page 10 of 12 • 1, 2, 3 ... 9, 10, 11, 12
20191028
Killer Instinct (Arcade Mega Patch)
Patch* and sound/data files:
https://1drv.ms/f/s!Ah8ksmUOX0QQiR-hlWgB-wczn08C
*Apply on headerless US 1.1 ROM
Light patch (only music for SD2SNES)
https://drive.google.com/open?id=1sOpDIBma5u_Khazl2hR7HULS5mfZaDWV
MSU-1 IS NOT ONLY FOR PLAYING PCM MUSIC.
You can stream graphics, fmv, sound, sprites. Ignore all previous patches. Released 28/10, as the arcade Killer Instinct release date.
gizaha- Since : 2015-05-13
Killer Instinct (Arcade Mega Patch) :: Comments
Re: Killer Instinct (Arcade Mega Patch)
You have a point there. It's totally cool, it's great as it is, anyway.gizaha wrote:Not at any cost due to respect. It's not at the arcade, it makes the game looks like bootleg or Mugen (no difference between these two) and it's against the original direction. I'm against of adding personal touches (i think the only exeption is the fade in the heartbeat).@gizaha, if you would be interested in adding Eyedol to the character selection
I noticed the heartbeat today, thought it was a very nice touch.
I tried again, erasing the previous .srm just in case, but nope, still not saving. Tested in Snes9x 1.53 and 1.60.gizaha wrote:I put them because the arcade had them so. It save every time at the winner screen, if a "xxx qualify" is shown. It might have bug. Any other user confirm this?
I know I'm late to the game but I hope this gets seen. I know little about MSU1. For all of my MSU1 stuff I'd had to use prepatched roms on my SD2SNES. Now I'm trying to use MSU1 on my Polymega and have been unsuccessful. I may be having issues on the Polymega as this MSU1 patch seems to be only useable on a pre-patched ROM (1.1 No Header... not sure what that means). But, with Polymega it has to be able to apply the patch on the fly to the game you have installed. Since I'm using a retail version of the game that might be my issue. So, this hack was not created for use with a retail release of Killer Instinct, is that correct?
There are 2 revisions for the Killer Instinct SNES cartridge. Since you own 1.0 and the hack was made for 1.1, it seems you won't be able to use the copy which the Polymega is ripping from your game. Unless there is a way to load external ROMs like the Retron5, you are out of options, i think.DarakuTenshi wrote:I know I'm late to the game but I hope this gets seen. I know little about MSU1. For all of my MSU1 stuff I'd had to use prepatched roms on my SD2SNES. Now I'm trying to use MSU1 on my Polymega and have been unsuccessful. I may be having issues on the Polymega as this MSU1 patch seems to be only useable on a pre-patched ROM (1.1 No Header... not sure what that means). But, with Polymega it has to be able to apply the patch on the fly to the game you have installed. Since I'm using a retail version of the game that might be my issue. So, this hack was not created for use with a retail release of Killer Instinct, is that correct?
You are correct on the 1.0. After I was able to figure out which version of the game I had I was able to create a working patch for the 1.0 version of the game. It works great on my PC emulator, but the patch is not working properly on the Polymega.
Amazing Patch! Eyedol's music is the same as Cinder's what did i do wrong?
gizaha wrote:In the bridge yes because the arcade does it that way.
Yes! i get it now. My mind is blown with this job!!!! So awesome!!!!!
Hi, i've been trying to step into the Snes hacking stuff, been into many Snes programming sites, i admit i didn't understand much, and many people recommend to have "some" programming knowledge in order to know how things work. I feel like every assembly examples, teachings or lessons are orientated to people that already have some basic knowledge. So, what programming language is the best one to achieve that basic knowledge?, help will be appreciated, thanks. By the way, what are you up to lately?, any KI update in store for us?. thanks again.
1. Please open a new thread if it goes off-topic.
2. ASM is a so-called second level language, every higher (3rd level) language like c gets transcribed into asm... Or better 1st level language that are bits and no human can read.
So if you script in c, python or whatever, the compiler translates it into a hex binary, which consists of opcode instructions and addresses, e.g. LDA #$04 is a9 04 in bytes. This would be int x=4 in a higher language.
Means of you know ASM you have a deeper understanding what your machine does and may code more efficiently. Many compilers (those things that translate your code into ASM) are a bit inefficient (gizaha can probably tell you more), c allows you for example to directly inject ASM (which is intel ASM I think, a bit different from SNES but basically very similar).
If you want to start try maybe this. I made it decades ago:
https://www.zeldix.net/t134-coding-for-absolute-beginners
2. ASM is a so-called second level language, every higher (3rd level) language like c gets transcribed into asm... Or better 1st level language that are bits and no human can read.
So if you script in c, python or whatever, the compiler translates it into a hex binary, which consists of opcode instructions and addresses, e.g. LDA #$04 is a9 04 in bytes. This would be int x=4 in a higher language.
Means of you know ASM you have a deeper understanding what your machine does and may code more efficiently. Many compilers (those things that translate your code into ASM) are a bit inefficient (gizaha can probably tell you more), c allows you for example to directly inject ASM (which is intel ASM I think, a bit different from SNES but basically very similar).
If you want to start try maybe this. I made it decades ago:
https://www.zeldix.net/t134-coding-for-absolute-beginners
Thanks for answering!, should've opened a new thread, i got carried away, and i will definetely try it, thanks again!.
https://sd2snes.de/blog/
That sounds too good to be true... Can anybody test if gizaha's patch works now... Please
[All] Fixed data caching bug in MSU1 data extension which could cause wrong data to be streamed
That sounds too good to be true... Can anybody test if gizaha's patch works now... Please
Relikk said it boots but has hard flickerings. But we forgot whether the old firmware had this problem too (or even booted), so we can't say whether it's an improvement. Does it boot on your side?
I can only get into the character select screen, with a lot of flickering and graphical issues. It hangs before the vs screen.
It mostly runs for me (on an old 64GB Platinum 0.313/0.799ms and a 200GB A1 card 0.484/3.842ms(!)) but there's a lot of prolonged vertical blanking going on. Sometimes I'm getting nonsensical data requests like this one:
If there are any look up tables stored inside the MSU file it would be best to cache them beforehand in WRAM or store them inside ROM in the first place to save seek time just to look up a pointer or similar. Just guessing at large though...
EDIT: It appears that redundant lookups are more rarely the case. The patch seems to do a lot of extraneous data requests but most of them are cached away by the FXPAK's MSU1 implementation. Still in some cases it seems that 3-4 different data offsets are accessed in "parallel" during VBlank, e.g. to stream two sprite animations and a background video such as seen in the city rooftop stage. Very cool, I might add
However, VBlank is ~2.3ms in 60Hz when overscan is off; random access latency should be assumed to be 1-2ms. Therefore multiple random accesses will not complete inside VBlank and extend forced blanking into the next frame, hence the flashing black bar on top of the screen.
It might be possible to alleviate this a bit by requesting the first data offset some time before VBlank to save at least a bit of access time, or stream all the required data into WRAM/SRAM during active frame time and only DMA it into VRAM during VBlank. I guess this might add a single frame of delay to the character animations but I doubt there'd be another way to achieve this on real hardware.
- Code:
Data requested! Offset=180c869a page1=002e2a00 page2=002e4a00
offset 180c869a out of range (002e2a00-002e49ff, 002e4a00-002e69ff), reload
seek to 180c8600, res = 0
read res = 0
read 0 bytes
If there are any look up tables stored inside the MSU file it would be best to cache them beforehand in WRAM or store them inside ROM in the first place to save seek time just to look up a pointer or similar. Just guessing at large though...
EDIT: It appears that redundant lookups are more rarely the case. The patch seems to do a lot of extraneous data requests but most of them are cached away by the FXPAK's MSU1 implementation. Still in some cases it seems that 3-4 different data offsets are accessed in "parallel" during VBlank, e.g. to stream two sprite animations and a background video such as seen in the city rooftop stage. Very cool, I might add
However, VBlank is ~2.3ms in 60Hz when overscan is off; random access latency should be assumed to be 1-2ms. Therefore multiple random accesses will not complete inside VBlank and extend forced blanking into the next frame, hence the flashing black bar on top of the screen.
It might be possible to alleviate this a bit by requesting the first data offset some time before VBlank to save at least a bit of access time, or stream all the required data into WRAM/SRAM during active frame time and only DMA it into VRAM during VBlank. I guess this might add a single frame of delay to the character animations but I doubt there'd be another way to achieve this on real hardware.
Thought: I might be able to reduce random access time from my side a bit; currently when it encounters a cache miss it will refill the entire MSU data buffer (16384 bytes) before clearing the busy flag. It might be possible to get away with filling only half the buffer, clearing the busy flag, then filling up the rest.
Filling the full buffer currently takes about 1.6ms (16384 bytes at ~9MB/s) so slicing that in half could help quite a bit. Maybe even less data could be pre-buffered.
I must ensure that the CPU won't catch up though.
Filling the full buffer currently takes about 1.6ms (16384 bytes at ~9MB/s) so slicing that in half could help quite a bit. Maybe even less data could be pre-buffered.
I must ensure that the CPU won't catch up though.
What is the correct track order for the Light Patch? All the tracks seem to play out of order
If we Snes fans don't have a program like SGDK for Megadrive, the Snes will become obsolete, just by looking at the projects that can be done with a program like that (Street Fighter 2 Perfect, Street Fighter 3), and a Street Fighter 5 tech demo, shows how much easier things can be when progamming. This a must for Snes.
First time poster here: @gizaha this hack is absolutely brilliant work, your achievement is truly amazing. Since MSU-1 was implemented into the SNES MiSTer core, it is playable also on this system. It has been tested and confirmed fully functional, with just a few glitches - could be in the MiSTer implemention of course.
As this thread has become somewhat more quiet recently: is the project complete? Do you have more things you would like to implement in KI or more ideas for other games, now that you have acquired such skill set? I would be happy to contribute to a Patreon account for instance or crowdfund for a MiSTer, if helpful at all.
Best wishes!
As this thread has become somewhat more quiet recently: is the project complete? Do you have more things you would like to implement in KI or more ideas for other games, now that you have acquired such skill set? I would be happy to contribute to a Patreon account for instance or crowdfund for a MiSTer, if helpful at all.
Best wishes!
I have a list but not have the corresponding courage to complete them. The time they require would not return equally good result.
KI:
Breaker sprite taken from arcade.
Rest knockdown graphics (abandoned)*.
Transparent projectiles (abandoned).
Edit stage data (Orchid sky, Sabrewulf's stairs etc)*.
Edit sd2snes firmware for faster sd load (abandoned).
If there was anything i missed and users wanted, be sure they would already say it.
SFA2:
Finish the Z full frames animation*
Lose sounds (they exist in stock rom).
SA1 conversion for full speed (not need, existing speed tweaking is good enough, only Rose would need it)
Bad apple 60fps (without msu).
SF2 CE intro.
SSF2T snes (very hard).
KI2 demo (hard).
Mario paint patches (already did the instant save/load).
Mk3/umk3 arcade additions and especially the choose your destiny tower zoom with mode7.
NBA hangtime screen transition with the purple bars.
*superfamiconv have bugs. I mailed the programmers, they didn't fixed them.
KI:
Breaker sprite taken from arcade.
Rest knockdown graphics (abandoned)*.
Transparent projectiles (abandoned).
Edit stage data (Orchid sky, Sabrewulf's stairs etc)*.
Edit sd2snes firmware for faster sd load (abandoned).
If there was anything i missed and users wanted, be sure they would already say it.
SFA2:
Finish the Z full frames animation*
Lose sounds (they exist in stock rom).
SA1 conversion for full speed (not need, existing speed tweaking is good enough, only Rose would need it)
Bad apple 60fps (without msu).
SF2 CE intro.
SSF2T snes (very hard).
KI2 demo (hard).
Mario paint patches (already did the instant save/load).
Mk3/umk3 arcade additions and especially the choose your destiny tower zoom with mode7.
NBA hangtime screen transition with the purple bars.
*superfamiconv have bugs. I mailed the programmers, they didn't fixed them.
What's the idea behind KI2 demo? Implement some art assets from KI2 into the KI1 ROM? Sounds very interesting indeed.
Something like that. Jago vs Jago in a stage that can be exactly replicated in snes. Without the moveset, only walking and jumping, but with the target to be visually replica of the arcade.
Loose sounds, what is that for SFA2? Why would cleaning those up not give a good result? SA-1 for SFA2, that I can see what you mean. All the headache for only one fighter to truly benefit from it, not worth it.
Lose sounds is the sound that the player does when he's KO.
They will not have equally good results because they would take months or years.
They will not have equally good results because they would take months or years.
gizaha wrote:Lose sounds is the sound that the player does when he's KO.
They will not have equally good results because they would take months or years.
Oh, now I understand. If it could take months or years to get fully done, it's not worth it. I think mk3/umk3 might be something to look at for fixing issues then sfa2. But it's your choice. I am grateful and thankful for all that you have done.
Similar topics
» Final Fight 1 MSU-1 + FastROM + 2-Player Patch + Arcade Remix
» help to patch Mega Man X2 with MSU1
» Mega Man Mesen HD Audio Patch
» Final Fantasy IV - Ultima
» Mega Man - Rockman to Mega Man (EU) / Roll-chan + No Skid + Restore Health with Score Points and Instant Glitch + MSU-1
» help to patch Mega Man X2 with MSU1
» Mega Man Mesen HD Audio Patch
» Final Fantasy IV - Ultima
» Mega Man - Rockman to Mega Man (EU) / Roll-chan + No Skid + Restore Health with Score Points and Instant Glitch + MSU-1
Permissions in this forum:
You cannot reply to topics in this forum