วีดีโอ

The most difficult zone to make is Jungle Zone because of the art (forests are always difficult to do with tile based art). I am definitely wanting most of those zones in the game!

@@doslogo There is Two Versions, 8-Bit Version (65C02S) And 16-Bit Version (65C816S)

@РусланИщук-щ8в Aiming for 65C02 8-bit, which has worked out greatly so far

It's fun though! Newer devices are easier to program for, and has more help and tools available

The Commander X16 even supports the same FM synthesizer sound as the original, so it will be really fun to get music and sound for this!

The audio is always difficult to dial in, but I rather have it like this to make it future proof against AI generated content, than to make videos that are too perfect that you can't tell it was made by a human or not

I will definitely give credits to anyone that helps out with game resources. At this current stage, everything will be placeholders, even so, if not made by myself, I will give visible credits in-game

There would be improvements with 3 hotspots for ground detection, but the original game only uses 2. I know Yoshi's Island (SNES) uses 3 where the middle actually improves the visuals. I might be limited to 1 just like Super Mario since the CPU isn't very fast. We'll see, it is hard to time these things until everything is implemented

@@doslogo on 8 bit I wouldn't use 3 feeler because I wouldn't use a heightmap lut, but a slope lut with less variations, basically branching logic on type of slope. And I would do tile collision on position hashed bitfield first (a nibble being 4 tile in square patten, organized in chunks) which bypass the need for feeler entirely (you get more context with less compute), the slope lut taking care of sub tile collision. There is probably an optimisation, if there is a clever way to differentiate full collision tile with partial collision tile or slope tile. Slope resolution being something like: grounded = if y < inTilePos.x >> slope //0 for 45, 1 for halving, permute x and y for rotation and flipping, I only have two type of slope, should be more than enough

@@timmygilbert4102 Good advice! Most of my luts are pointers to other luts in different banks, so I don't have to bitshifts. The 65C02 really likes to eat cycles on such basic operations as bitshifts, causing me headache

@doslogo yeah okay, it's more of a reference, the main idea was to show a quick slope computation, using the fact that y<x let's you have a quick diagonal comparison to avoid storing data, the bitshift is more to create a second slope, with half the height, but if you find another way to compute it. Maybe artificially doubling the axis by adding to itself might help, or maybe a single slope at 45 is enough lol, after all sprite of sonic only have 8 direction (two sets of sprite animation with permutations and flip) Anyway this discussion made me conceptualize doubling instead of halving as a cheap trick lmao 🤣 I'm trying to port some assassin's Creed 1 level to game boy, due to their boxy nature, and cheating like hell to get a hybrid of duke nukem sectors and Wolfenstein raycaster, I spend so many time trying to get random box intersections down to a few computes, which bogs down to arbitrary rectangle diagonal, square are easy in comparison all hail x<y lol

@@doslogo The CPU isn't very fast? It's an 8mhz 65c02 right? You're already faster than the SNES.

Thank you!

That's pretty cool, has you posted anything about it somewhere?

@retrodeveloper9156 Not yet, it's not available to the public at the moment. Rest assured, if the game get it's final update, I'll let you know!

@@InnovateSphereStudios Alright!

Wish you luck with the FM programming!

0:22 is the Commodore 64 port of the first Sonic game for Master System, and I have only good memories from this game. The port is crazy good! It is a good benchmark in direction, story and music. But it is not working the same as the Genesis games. Truth be told, Super Mario Bros. 3 (NES) is probably much better as a reference to get the basics right

There was that bootleg game...

@ a far cry from an accurate port made by passionate fans

The biggest problem with a Sonic game on the Super NES is that unless you want to keep everything except the Bonus/Special Stages in Mode 6, and put a Super FX and/or SA-1 chip on the cartridge PCB to handle extra graphical functions, Sonic, Tales. and the rest of the cast will look rather zoomed in. The Sega Mega Drive/Genesis featured a 320×224 as its main resolution mode. The Super NES had either 256×240 with hardware parallax scrolling, or 512×480 in 30Hz, with only one scrolling field and 64 total colors (four different CLUTs of sixteen colors. each), which looked terrible on.a broadcast television quality CRT if you didn't know what you were doing. At least the Commander X16 features a 320×240 resolution mode.

@@benbreeck3363 Discussing the technical aspects of video games is something I struggle with, I will say that someone already made a proof of concept demo that didn’t do it your way. IIRC the demo does not require enhancement chips, and it doesn’t sacrifice much except performance in certain parts and resolution.

512 kB is definitely the best option, since that's what the CX16's base model has. It is accessed though a narrow window of 8 kB, giving me 64 banks to read and write to. Going to be a lot of bank-switching, but it will work 👍

If you put it on a cartridge, or at least make a ROM for someone else to make a cartridge for it, the CPU could directly access 3.5 megabytes or 28 megabits, enough for Sonic 3D Blast.. If you put it on a CMD-4000 double sided, extended density floppy, which should plug right into the Commander X16's floppy drive port, You are looking at 3,200KB. If you do both at the same time, Sonic 3 and Knuckles should be quite possible. You could even add more characters like certain fan games. Just be prepared to do a LOT of garbage collection and background loading on Commander X16 with a stock RAM capacity.

Size won't be a problem right now. The Commander X16 can't process data that fast, so the more data won't change the visual outcome of the game. Like, how many animation frames can be used is limited by the speed the graphics data can be uploaded to VRAM, and that speed is just not fast compared to game consoles of the past (they had DMA). There are some design choices of the CX16 system that directly work against me (interrupt code is locked in ROM, wasting cycles, not that bad though), but I'm focusing on what can be done and not what can't be done

Once I fly into the sun, I'll make sure it gets entertaining for the TH-cam 👌🔥

Actually, you can use the 65C02's bit shift instructions for multiplication. It's even faster than trying to use MUL on a Motorola 68000, or any X86 processor before the Pentium outside of the NEC V series, which featured hardware multiplication and division. But it's still tedious to write out the operation in anywhere from seven to twenty-one instructions.

@@benbreeck3363 u sent the comment twice by accident

Other CPUs can multiply by 2 by adding the same value with itself, but the 65C02 must do such a thing with a memory location rather than a register, and the same goes for ROL and ROR to multiply 16 bit numbers (and moving that carry bit). It is very limiting, but at least it works

doslogo you can just bitshift left once to multiply by two

n bitshifts will multiply the number by 2^n

Absolutely true, all of the Sonic Retro guides are invaluable when working on this. Also I won't be afraid to ask for help if I run into problems. Thank you!

It has been a few crazy weeks, but it has been exciting the entire time! Thanks!

I don't know it it's AN exciting development, but it's exciting development.

Thank you!

Fantastic! A whole new level of nerdiness, I love it. Thanks for sharing 🤩

Microsoft was "forced" to make Windows ME because OEMs wanted a consumer version and Windows 2000 was still targeting business users. Microsoft never wanted to do it, and they didn't give enough development time so it could function correctly. They did however spend a huge amount of time to remove real mode MS-DOS, to impressive levels actually. Funny though, when OEMs found out about it, they got mad, so Microsoft had to recover real mode MS-DOS but only in the OEM release! So no patch is needed, you just need the OEM version and browse the CD

@@doslogo Yeah, at the time they were juggling a handful of projects pertaining to Windows like Neptune, which was intended to be a consumer NT release - which was also being developed in tandem with 2000 and Me at the time. Ultimately, that never fell through, though a lot of what it inherited was brought over to Me, and as a sort of nudge to prepare people for XP not having DOS, only NTVDM, they still kept being able to run DOS programs in a window, but real mode was heavily hidden. The emergency boot disk was a popular way to access it. Here's a really hot take... I don't think the public was necessarily ready for XP by 2001, at least for home consumers. Many were still using 98 well into 2004, so that does hold some weight. Perhaps they should have given ME more development time to be better polished and add features like fast user switching to hold people over, as releasing it alongside another desktop OS the same year isn't something you just do. Then, that can be followed up by a different XP in about 2003, right around the time hyperthreading would start to appear, requiring the NT kernel, to which it would have refined NTVDM to be more accurate. That would've helped transitioning to NT be smoother in the long run. There's a lot more I could talk about regarding their development, but I'll save it for another time.

@@bluehorizon3181 XP (the 2001 release) was definitely rushed, and quite buggy with existing programs. Dave Cutler was busy with the 2003 release and said in a TH-cam interview he was quite disappointed with the programmers. As for Windows ME, I have seen signs of working installations from online videos, but they are all based on virtualized hardware (such as VMWare and Virtual PC) where the hardware is all the same (Windows 9X really needs to be experienced on real hardware). From my own memory, at least my 4.90.3000.2 install always started crashing after a few weeks of use, with the updated simplified VWIN32.VXD blue screen, that new Protection Error blue screen at boot, and even a repeated booting loop into Safe Mode situation.

@@doslogo Virtual PC 2007 was pretty much my full introduction into the realm of virtualization. It may be dated now (especially since it doesn't work on current releases of Windows), but it has still managed to hold up in some oddball cases where no other hypervisor or emulator has surpassed fully. I've also never heard of the .2 decibel at the end of ME's build number, 3000. I knew that both releases of 98 had them internally and never showed them in system properties, but never heard one for ME. And yeah, there isn't a whole lot of videos showing it running on real hardware. I've seen way too many cases where they just install it on something like VirtualBox and don't install video drivers. I do look forward to seeing your full coverage on the innerworkings of MS-DOS 8, seeing what it's really made of.

I should have said improved nothing [over Windows 98 Second Edition and Windows 2000], which I believe was implied (they came out before Windows Me, in 1999). I am all about MS-DOS, and most of internet agrees with the assumption that Windows ME is the worst operating system that Microsoft produced (for some reason or another, crashes mostly). So, not an improvement. But, I secretly love Windows ME for other reasons and will do more videos in the future about it, hopefully being a little nicer with my wording

Game companies should have, then I don't think Microsoft Windows would have been so dominant as it became, since more games would "act" as operating systems, get more features, build communities, etc

I have fond memories of that time era, the PIII era. Hopefully your Athlon didn't overheat, they had issues back then

Thank you for testing! The system requirements are 225 MHz Pentium with 4 MB RAM (only ~160K is actually used, but placed below 4 MB) and VGA of course. Anything below will struggle. The barrier of rendering Sega Genesis layers and sprites at full speed with raw assembler code is a little bit above 225 MHz (but the title screen doesn't show too much at once), and PII is actually slower than Pentium MMX without setting write combining (which this does). Celeron requires 400 MHz to keep up. No floating point is used at all (I will always stay away from floating point for accuracy reasons). I optimized this to the point where only a few scanlines could be gained per 60 Hz

Thank you!

Just rip the graphics chip from your favorite video game console, slap it on a 8 or 16bit isa card Make your own "driver" Put right next to that sound blaster Now you have a nes or genesis vdp in a x86 machine Composite and svideo, analog rgb may be supported if you have a play choice 10 ppu

Just add some extra dma chip to do all the memory moves to and from the chip, ram, and isa bus and some fast floating point and calculus Trigonometry if you doing light effects like ray tracing

@@Clancydaenlightened No need for floating point. I am actually against floating point for many reasons. But for the rest, you are right that replacing or adding hardware will give features for the PC to act like a Video Game Console and I like that idea! I also like the idea of not doing any extra work, and just use the bare bone PC hardware to get there, which is demonstrated in the video

The NES's Japanese counterpart, the Famicom got a BASIC interpreter as well as a keyboard, making the Family Computer actually functions as a computer

Cool! I only know about one arcade using a PC, and that is Arika's Tetris Terror Instinct, but it is running under Windows XP where the game has zero control over the hardware or lag. Also, any emulation like Mame are under the mercy of the PC's operating system. That doesn't mean it can't run smoothly under perfect conditions, but I was aiming for a constant smooth frame rate until power down of the PC, conditions never seen running under an operating system

@@doslogo SEIBU SPI HARDWARE: CPU : Intel 386 DX-25 Graphics : Custom Seibu Sound CPU : Z80AP @ 8 MHz Sound Chip : YMF 271-F Fm towns computer also uses 386. But all these have custom gpus ofcourse, not standart vga. And yes all emulation are operating system dependable. Cant remeber if DOS based emulators with VESA3 support are any smooth or not (under pure DOS).

@@vasileios6301 I'll check that out! Thanks! My tests of emulators in DOS (mostly just VESA2) have given me both good and bad results. One thing that VESA is bad at is telling how video synchronization is actually done in hardware, meaning there is a lot of guessing and praying from us programmers. When it works, it works great, but compared to VGA that always (from 1990 onward) follow the rules (compatibility), VESA is just specifications that doesn't have to be followed. Some of the emulators do great VGA Mode X stuff, but DOS and the DPMI host eventually will interfere with the smooth scrolling

Thank you!

Thank you! Also, Abrash's book is awesome, I have used many techniques from that book in previous projects, but the issue is most of his optimizations are too slow after the hardware manufacturers moved on with their own accelerated graphics. Basically, his book became outdated quickly around the time of Quake's release. You already know DOOM runs at half speed, that is 35 frames per second instead of 70, but otherwise, both DOOM and the implementation used in this video accesses memory above 1 MB. We just do it differently, DOOM uses a DPMI host (more layers on top of DOS = potential added lag), where I temporarily jumped into Protected Mode to change the cached address space limits of a segment. Video memory is slow, and Mode X is always slow, no matter if it is DOOM using it or DirectDraw. There is no way around the plane switching, so everything else needs to be optimized to crazy levels, including memory access, and that needs to extend past the 64 kb boundary to work! Yes, sadly the audio used in this video was glossed over too quickly, else the video would have become too complex. I am still working on Windows Me stuff for this channel, and when that is done, the source code would become more relevant and potentially released

SNES can be converted as well, with titles such as Mario Kart and Yoshi's Island using background rotations together with the raster effects (two games I have worked on as well in the past). However, the VGA Mode X is too limited in the color depth so VESA VBE would be required (which is faster anyways, see the 16 million color video, but can't have a border color)

Thank you!

Windows 3.1x already has a very straightforward way to replace the boot screen. The only thing is the bitmap is required to be a 16 color RLE bitmap (which MS Paintbrush doesn't support saving). Search for win.cnf and splash screen on Google for more information

@@doslogo Yeah, I know how to do that, but the tricky part is exporting it as an RLE. Do you know of any programs that can both open and export RLEs?

@@ThatOneWindowsFan MSPaint (Win95) can open RLE, but won't save it. Adobe Photoshop can export BMP with different formats, so that's where you choose 4 bit RLE in the export options

Yep. Earlier Windows had those flying Windows logos as a screen saver (same as the star field screen saver), same design, same colors, same mindset from early 1990 design. The code for animating them in the boot screen is quite interesting with some pseudo-random number generation. The later boot screens had the palette cycling data stored in the bitmap header to give much more flexibility in the design (mostly explored in Microsoft Memphis, the codename for Windows 98)

Stay tuned for a tool that can edit the text "Starting Windows Millennium..." to any length and language in the future (on this channel). As well as all the juicy details about MS-DOS 8. I'm working on it

@@doslogo you had my curiosity ... but now you have my attention