OASIS (working title)

[johnparker007]

Had another play with this new simple process for enhancing low-res pixelated layouts... the idea is that one day, as part of the layout editor, it will be a easy to use built-in tool, with some tweakable settings... 

This time was Super Hyper Viper at a whopping 16x upscale for an 8K layout, here's some before/after, using the simple fixed process (upscale, rebalance color, unsharp mask):

Before:

After:



Before:

After:



Before:

After:


Before:

After:

It's not all there yet by any means though:

Before (notice the text is actually much more legible):

After: (text is an unreadable mess)


I believe in future multimodal LLMs would be able to guess the correct text based on knowledge of the game though the MAME internal classic layouts, and fonts used on these machines, and then redo the text as fonts with layer effects... probably a long way off yet, purely due to these not being particularly in-demand generative AI skills, but it does seem feasible... will be a handy stopgap for those machines where no high res images are available, and no highres redraw has been done  

Edited August 3, 2023 by johnparker007

[SomeRandomGuy]

On 03/08/2023 at 18:36, johnparker007 said:

Had another play with this new simple process for enhancing low-res pixelated layouts... the idea is that one day, as part of the layout editor, it will be a easy to use built-in tool, with some tweakable settings... 

This time was Super Hyper Viper at a whopping 16x upscale for an 8K layout, here's some before/after, using the simple fixed process (upscale, rebalance color, unsharp mask):

Before:

After:



Before:

After:



Before:

After:


Before:

After:

It's not all there yet by any means though:

Before (notice the text is actually much more legible):

After: (text is an unreadable mess)


I believe in future multimodal LLMs would be able to guess the correct text based on knowledge of the game though the MAME internal classic layouts, and fonts used on these machines, and then redo the text as fonts with layer effects... probably a long way off yet, purely due to these not being particularly in-demand generative AI skills, but it does seem feasible... will be a handy stopgap for those machines where no high res images are available, and no highres redraw has been done  

That's quite interesting... on that first example the text looks somewhat funny and the numbers on the squares are screwed, but considering the source material this thing is working with, the parts that work look pretty good. Interesting stuff!

Got an MPU2 update as well - two VFS games went in, Triple Chance and Fruitopoly, the latter actually got dumped only slightly more than a month ago, so thank you EddieC on the Mecca! You don't see new dumps of machines that old very often, so that's always nice to see. These two and some Barcrest test ROM that I added are the last of the ROMs that I found, so now it's on to cleaning up and testing. Most of the way there for the next pull request now

[johnparker007]

1 hour ago, SomeRandomGuy said:

That's quite interesting... on that first example the text looks somewhat funny and the numbers on the squares are screwed, but considering the source material this thing is working with, the parts that work look pretty good. Interesting stuff!

Got an MPU2 update as well - two VFS games went in, Triple Chance and Fruitopoly, the latter actually got dumped only slightly more than a month ago, so thank you EddieC on the Mecca! You don't see new dumps of machines that old very often, so that's always nice to see. These two and some Barcrest test ROM that I added are the last of the ROMs that I found, so now it's on to cleaning up and testing. Most of the way there for the next pull request now

Yeah man, these AI tools are getting pretty good, at least for some parts of the layout from the low res sources  

Great news on the MPU2 update, you're doing great work, thanks dude!  

[Mrbeanstreak]

Will we see a demo release of this new arcade sim ?

[johnparker007]

On 17/08/2023 at 17:47, Mrbeanstreak said:

Will we see a demo release of this new arcade sim ?

It's a very long way off, I'm not up to doing much coding at the mo, projects this complex take a lot of time and effort... you do have the existing Arcade Sim to keep you going at least!

[johnparker007]

Ok a small update on the Layout Editor component of Oasis, here is a demo of it importing an MFME Extraction (of Vectra's Popeye layout).  It now places the lamps, and flickers them just to show they are there as separate editor components within the project:

 

[johnparker007]

Another little update to Oasis Layout Editor - a bunch of hardcoded bits, but this now shows proof of concept of the new way of interfacing to read component states from MAME:

There is no 'screen-scraping' taking place here to flash the lamps in the editor, and so also no need for 'data layouts' (previous MAME/MFME work for Arcade Sim has all relied on the hideousness that is screen scraping data layouts). 

I'm instead pulling a component data feed direct from MAME.  Need to suss out how to get the MAME.exe running as a totally hidden child process - good to see the idea works anyway, this new technique of reading the lamps etc has not been done before  

Edited September 10, 2023 by johnparker007

[johnparker007]

Done a little more hacky testing, this renders basic 2d reels, synced with the emulation.  Kept MAME window visible so I can press keys in MAME for the moment to drive the emulation (play the machine).  Ultimately MAME will be a hidden child process and Oasis will be passing inputs through:

 

[johnparker007]

I am keeping a very close eye on these new 'Unity Install Fee' developments, as they relate to Oasis.  Currently the plan is to build Oasis on Unity, however they have recently announced an insanely greedy plan that will potentially kill off lots of small studios (especially ones that create 'free to play' games, like mobile games with IAPs).  In a lot of cases (i.e: commercial free-to-play games), the model one tries to achieve is many millions of users, but a very small 'ARPU' (average revenue per user).  If Unity want say $0.15 per user, and you only actually earn $0.10 per user, you've now gone from a profitable game, to one that in reality costs you more than you receive!

While this new 'Unity Install Fee' would not affect Oasis (since Oasis is free), I suspect this may signify the beginning of the end for the Unity game engine.

Viable yet unpreferable (for a bunch of technical reasons I don't want to get into right now) alternatives are Unreal Engine, and Godot Engine.

Unity is 'going public' soon (so people can buy shares in it on the stock market), and is also now being run by the a-hole who used to run EA, with all the lootboxes and other toxic cashgrab IAPs, so it's clear why they are driving this excellent game engine into the ground; ca$$$$h.

Ugh, the enshitification of yet another great product, thanks modern capitalism.

On the plus side, we already did the very 'game engine specific' proof of concept with the initial Arcade Simulator project (to prove we could run a massive arcade with 100s of machines on a modest spec PC)... so a lot of the upcoming near-term work in Oasis has to do with stuff that is fairly easy to port to another engine.  Things like finishing up the MAME emulation IO, the Oasis Layout Editor project format, the 'safe screen scraper MFME extractor' system, the MAME layout export system, the machine+layout formats (for use in the new 2D and 3D Oasis modules like Arcade Sim v2, standalone machine player, etc) - these things are not so closely tied to Unity in the new Oasis project (they were very entwined with Unity in Arcade Sim, but the plan was already to decouple these formats with Oasis, so got a little lucky there).

So as and when I can do bits and bobs of work on Oasis, it will be in those areas... and we'll see how this whole thing plays out!

Here's a quick rough video a dev has made about this, I'm sure more will follow as this news (about Unity, not Oasis!  ) goes mainstream:

 

Edited September 13, 2023 by johnparker007

[Altharic]

Not going down well for them they have done a Bud Lite

[johnparker007]

1 hour ago, Altharic said:

Not going down well for them they have done a Bud Lite

It will be interesting to see if that decline continues.  Platforms like Unity exist because of developers, and game studios.  The whole community is quite rightly completely up in arms at this barely legal bait-and-switch.  While the effects won't be immediate, as studios large enough to be shafted by this, are also often complex enough that the Unity development pipeline cannot simply be changed (so we're kinda locked in for a good while).  We're all eyeing Unreal and Godot (and there are a couple of other potential options) at this point for long term game engine/tech stack transition plans.

So this will make Unity's profits go up in the short term... but long term they have put the final nail in the coffin (Unity has been rotting for the past 3-4 years).  Unity will almost certainly wheel this back to something less shitty, but the damage has been irrevocably done.  No new studio in their right mind is going to choose Unity as their tech stack, and existing studios are now considering the long term strategy to move away.  It's the end of an era - Unity was and is a great engine, but the only thing constant is change I guess  

The guy who is now CEO of Unity, was previously CEO of EA - he was planning to have an in-app purchase when you reload your gun in Battlefield.  He publicly called all Unity devs who don't want to heavily monetise their games via in-app purchases as 'fucking idiots' (his exact words)... and he's been selling off chunks of his Unity shares pretty much every month for the past year while buying none back.  The whole thing is screwed!

Edited September 13, 2023 by johnparker007

[johnparker007]

For those that are interested in behind the scenes info on Arcade Sim/Oasis, providing a small update on this unfolding major crisis in the game engine industry (Unity Engine Install Fees).  In lock-step with some other factors relating to my career, we will hopefully have the finalisation on which game engine Oasis/Arcade Sim will be built on around Jan-Feb 2024.

The current shortlist to replace Unity Engine for all FME-related projects from me and any future collaborators (on Oasis and its related sub-projects such as the new Arcade Simulator) is Godot Engine, Unreal Engine or Flax Engine.  This is subject to change before Q1 2024, but that's where we are currently.

Of course this will slow things down compared to sticking with Unity, and yes, a project like Oasis/Arcade Sim wouldn't be actually affected by these Unity license/fee changes as they currently stand since it is free, but; Unity is no longer viable for me, as I will be continuing to use the same engine for my hobby projects as I do in my career, so that I'm continuing to develop complementary skills to literally pay my bills.  Such a shame, but this is where the industry is at right now.

This is also the sort of action that needs to happen (migrating even free projects with potential to become quite popular in the coming years), as a cautionary tale for any future execs who think they can pull something like this, and not destroy the game engine company they are working for.  It's so frustratingly myopic and short-term share price focused, sigh...

Here's another video by a prominent youtuber, continuing to cover the events as they have unfolded so far:

 

[SomeRandomGuy]

On 11/09/2023 at 23:13, johnparker007 said:

Done a little more hacky testing, this renders basic 2d reels, synced with the emulation.  Kept MAME window visible so I can press keys in MAME for the moment to drive the emulation (play the machine).  Ultimately MAME will be a hidden child process and Oasis will be passing inputs through:

 

Ooh, anything to do with MAME integration is always nice to see . I was about to mention -video none for hiding the window but it looks like you've already got that one figured out . Worth talking to MAMEdevs about this stuff at some point anyway, I don't think MAME used as a background component is a use case which has been explored that much, and Vas & others know a lot more about this stuff than I do.

On 13/09/2023 at 10:41, johnparker007 said:

I am keeping a very close eye on these new 'Unity Install Fee' developments, as they relate to Oasis.  Currently the plan is to build Oasis on Unity, however they have recently announced an insanely greedy plan that will potentially kill off lots of small studios (especially ones that create 'free to play' games, like mobile games with IAPs).  In a lot of cases (i.e: commercial free-to-play games), the model one tries to achieve is many millions of users, but a very small 'ARPU' (average revenue per user).  If Unity want say $0.15 per user, and you only actually earn $0.10 per user, you've now gone from a profitable game, to one that in reality costs you more than you receive!

While this new 'Unity Install Fee' would not affect Oasis (since Oasis is free), I suspect this may signify the beginning of the end for the Unity game engine.

Viable yet unpreferable (for a bunch of technical reasons I don't want to get into right now) alternatives are Unreal Engine, and Godot Engine.

Unity is 'going public' soon (so people can buy shares in it on the stock market), and is also now being run by the a-hole who used to run EA, with all the lootboxes and other toxic cashgrab IAPs, so it's clear why they are driving this excellent game engine into the ground; ca$$$$h.

Ugh, the enshitification of yet another great product, thanks modern capitalism.

On the plus side, we already did the very 'game engine specific' proof of concept with the initial Arcade Simulator project (to prove we could run a massive arcade with 100s of machines on a modest spec PC)... so a lot of the upcoming near-term work in Oasis has to do with stuff that is fairly easy to port to another engine.  Things like finishing up the MAME emulation IO, the Oasis Layout Editor project format, the 'safe screen scraper MFME extractor' system, the MAME layout export system, the machine+layout formats (for use in the new 2D and 3D Oasis modules like Arcade Sim v2, standalone machine player, etc) - these things are not so closely tied to Unity in the new Oasis project (they were very entwined with Unity in Arcade Sim, but the plan was already to decouple these formats with Oasis, so got a little lucky there).

So as and when I can do bits and bobs of work on Oasis, it will be in those areas... and we'll see how this whole thing plays out!

Here's a quick rough video a dev has made about this, I'm sure more will follow as this news (about Unity, not Oasis!  ) goes mainstream:

 

Quite the clown show is all I'm going to say about that situation. @johnparker007 has already said everything important that you need to know about Mr. Unity CEO here anyway

 

Some emulation updates too - in the MAME thread I planned to look at Summit Coin 8035 mechanical hardware and Cleartone's Bonus Talker hardware. So I've done some looking now, took Bonus Talker's code apart and started gradually doing the same for Summit games too. No work on the MAME side yet. Bonus Talker should definitely be doable, as long as I can get the Digitalker sound chip that's doing the speech hooked up right, which is really the main draw of the machine anyway. There are some things that need guessing though, such as some additional sound hardware that's there in addition to the speech. There's a 555 timer on the board with some oddly-valued resistors close by that reminds me a lot of a tone generator with similar hardware used on Black Box. Problem here is there are no videos of this thing, so how it sounds has to be a complete guess. Some clocks need to be guessed as well, which means it'll run a bit too fast or too slow. No schematics to help here unfortunately, there's a manual that's very helpful but it only has a boardview in it. Anyway, it should definitely be possible to make it playable, but some parts won't be quite right.  Summit games should also be doable, one thing that kinda sucks though is that there's ROMs for a Dutch machine called Supershot, but compared to the regular UK games it also has an extra sound card on it. That card unfortunately has a ROM on it that's not dumped . The guy who has/had the machine actually dumped the game ROMs himself and is on the Mecca, only problem though is that his last post was made more than 10 years ago

[johnparker007]

5 hours ago, SomeRandomGuy said:

Ooh, anything to do with MAME integration is always nice to see . I was about to mention -video none for hiding the window but it looks like you've already got that one figured out . Worth talking to MAMEdevs about this stuff at some point anyway, I don't think MAME used as a background component is a use case which has been explored that much, and Vas & others know a lot more about this stuff than I do.

Hiya mate   Yeah I got the -video none sussed out, though when it comes to games with actual video output (like all the video games, and some video fruits), I'll have to figure out/implement a bit more so I can get some kind of texture pointer or something.

Also had it sending inputs, though was running into thread blocking issues with doing that whilst also pulling out the data for the lamps/reels etc.  I could get around it via having MAME send it's output over the local network port, but it means end users needing to make a local firewall exception for mame process which is grotty (and still wouldn't solve a good way of getting video, which I think will need work in Mame source).

So the next iteration is doing the lua plugin approach like BletchMAME project (worker_ui), working - it mentions various stuff about blocking in the plugin and the Qt c++ desktop app code, so he ran into the same issue, and managed to fix it with his plugin approach.  I don't think I can use his approach for video though (he's pointing it at a different Hwnd/widget handle, it's all a bit windows specific and hacky, it works for the BletchMAME use case, but not really for cross platform Arcade Sim use case), we'll need something better for more control... 

Great news that you are looking at more fruit machine emulation, good work man!     And once I have the basic non-video IO working without blocking thru Lua plugin, I'll get a couple of the older Blackbox / Mpu1/2 going for a demo that'll be running your drivers on the Mame backend (as this new approach can use the latest mame.exe)

Edited September 16, 2023 by johnparker007

[johnparker007]

A little more work on the MAME/IO stuff - now using a plugin to process input commands, while simply reading state change output from console.  MAME runs as a hidden child process of Oasis Layout Editor.  Still some minor issue with 'debouncing' where a button can get stuck down in the emulation, I managed to get Hold 3 to stick later in the video... will need more work, but this technique seems good overall.

 

[johnparker007]

22 hours ago, johnparker007 said:

Still some minor issue with 'debouncing' where a button can get stuck down in the emulation, I managed to get Hold 3 to stick later in the video... will need more work

I realised what this was whilst watching TV in bed   I'd made the output sends (that transmit the button inputs from Oasis framework into MAME process) asynchronous, so they could arrive out of order, from back when I was having thread blocking issues before the Lua plugin strategy and was trying various stuff.  I simply changed the output sends back to the standard synchronous behaviour, and now the inputs are 100% reliable, no stuck buttons issue  

[johnparker007]

Just thought I'd record a very quick play with some of those new emulation features we get 'for free' (since this is using MAME for emulation backend), such as the save states...  

 

[Ginge]

On 18/09/2023 at 19:25, johnparker007 said:

stuff.  I simply changed the output sends back to the standard synchronous behaviour, and now the inputs are 100% reliable, no stuck buttons issue  

I normally have to take button apart and clean all the beer residue to stop my buttons from sticking . 

[johnparker007]

Quick test of an MPU4.  Seems the fix we put in (as Adders and Ladders synth sound tunes were running too slow), has also sped up the sound chip, so everything sounds too fast/high pitched!  Will be fixed in a different way at some point (2x xtals, or different values per machine or something...).

 

[spa]

Sounds like a Liverpool rom version lol

[Ginge]

Flo sounds like janet krankie .Fandabidozi

[johnparker007]

Started feeling a bit better over Xmas, so I've been chipping away at 7 segment displays for the layout editor over the past couple of nights  

Next job is the 14/16 segment displays - unlike the 7 segments, I can't port them over from Arcade Sim, as I want to make the new ones more versatile (at the expense of a tiny bit of speed).  This is needed, as any 3d machine will be able to be built that MAME runs (like Korg synthesisers for instance), and these may use different 16 segment displays to the 3 main types found in fruit machines...

[Ginge]

Glad you’re feeling better .

[spa]

Good to hear you are feeling better

[thealteredemu]

Yes, I concur, great to hear you’re starting to feel a little better JP   take care bud.

I’ve been getting into rom hacking, spending a lot of time in the debugger, I’m currently getting my head into the 20p version of Project Coin Bullseye (v29) it’s the final £6 token version and once offering skill shots you can just go for hold after nudges but it must have a hold after wins pot unlike earlier version so a watchdog to monitor reel wins.  I’m slowly working bits out but I hope to put this part of the emptier back in, also going to try to put the bullseye trick back in, it isn’t fixed like the fixed 10p version was, that has some extra code and jumps around that portion of the game, I’m hoping it was more a hacky fix  I’m certain it is just clearing the 20 bill mystery ? So it’s still in memory when the machine exits that feature with no credit.

Progress is slow but I’m learning a load of stuff along the way, keeps me out of mischief anyway

I also slowly working on some artwork for the 20p Bullseye and will likely continue if I successfully put the emptiers back in

J