Mana Games

[manutoo]

More Porting Progress ?

I made some decent progress in porting TE4 to PlayStation for the past month.

First, I made the World Tour saved games work on PS ; it went smoothly enough, although I had to handle the PS user data system, which is a tad complicated due to cloud save.

After, I checked Mod.io & the Mods. So for the anims, they are working after the same kind of small patching I did for the Xbox version. For the Mod installation, I had 2 bad surprises :
- the Mod drive is case-sensitive for the file names ; this forced me to put everything in lowercase when installing the Mod and when looking for their files
- it cannot have more than 8 subfolders in total, which made the XKT Mod cannot install ; I had to tweak the Mod.io system to use fewer subfolders so I managed to give 6 subfolders for the Mods (including the Mod folder ; eg: "MySuperMod\Ini\Image.ini uses 2 subfolders)

All that was a bit cumbersome, but doable ; I still lost 1 week with all that, though.

Another good news : the PlayStation version will also have 2GB of space for the Mods, like the Xbox version.

Then came the Online mode : it's working as it is, so full Cross-Play for TE4 should likely come true (TE4 will have to pass Sony Certification 1st, though, to be 100% sure ).
The biggest issue here has been setting up the online identification from PSN, once again because the Sony docs aren't the greatest in the world...

2024-08 - TE4 - Mod Collection on PlayStation.jpg

When TE4 will be released on PlayStation ?

... is a question I hear very often lately...

So I still hope for October, as so far everything is ok-ish : most of the port is done, and out of the 30 Sony requirements I noted down, only 10 left are still not done, half of them being checking stuff is working when everything will be done.

Here's the (almost) complete list of my porting tasks, in addition to the 10 requirements :
• Invitations for online games
• Trophies
• Default Video settings for PS4, PS4 Pro, PS5
• Update ModIO before Sony Certification (the Mod.io plugin had a few issues preventing to pass the certification)
• Dynamic image loading for the Menus, to help lower the uses of GPU RAM for the HUD Mod(s)
• CPU vs GPU Memory Management without & with Mods

The last point is a delicate one : on Xbox, the memory is fully shared between the CPU & GPU so the share percentage can vary and I don't have to worry about it.

On PlayStation, I have however to decide once and for all what percentages of memory I give to the CPU & GPU. For an unmodded game, it can already be painful, but for a game very open to Modding, it's really complicated to anticipate all the different memory usages. There, the Dynamic image loading for the Menus should help as well, to keep the GPU memory use more stable.

Anyway, I hope all that will be done by the end of September. I'm not sure how long time I'll need to pass the Sony Cert. It'll likely have issues like for the Xbox version, and require some back & forth, so maybe the release will be in November instead of October. We'll see...


Ending on a personal note

Right now, I'm quite tired for the same reason as 1 year ago : during the summer holidays, I have my 2 sons at home most of the time, which means that most of the time they run, scream, hurt themselves or hurt each other, and ask me 1 thousand questions per hour.

So school time will be welcome, although I dread the millions of "hurry up or else we'll be late for school" I'll have to shoot every morning.

And I'm still in dire need of some holidays... I took 1 week off with my wife & the kids in the mountains, but I spent a too big part of it doing the police after the boys or doing activities for them and not for me, so it didn't feel like a holiday...

Anyway, that's all for today, cu next time for more game-dev adventures..!

A comment, cheer or question about this mini-blog ? => viewtopic.php?p=292596#p292596

[manutoo]

PlayStation Port is (almost) done !

I just submitted the game to Sony for certification. There's likely less than a 1% chance it'll pass it because I guess I didn't do correctly at least a couple of things, though...

For the Xbox version, it took me nearly 1 month to make it work as Microsoft wanted. I hope it'll be faster with Sony, but it's impossible to know in advance. From the little I've seen, it seems the process is noticeably slower with Sony, but hopefully, I won't have to submit too many versions ; the 2nd or the 3rd might be ok.

So in the past month, I did as planned the invitation for the online games, the trophies, default video settings depending on the PlayStation type, updated ModIO, and checked the memory usage.

2024-06 - TE4 - 1st pic on Playstation 4 !.jpg

And now for unusual technical notes !

I also spent a good bunch of time on 2 things.

The 1st is that right now on Xbox, it's quite slow to load big Mods ; it's actually due to an Xbox bug. Unfortunately, I won't get access to the fix because of how Unity & Xbox updates (don't) work together. So instead, I created a new approach and did some optimization to get back to a more normal loading time. That new approach concerns the animations : before I was loading all the Mod Animation Asset Bundles on game start which was taking up to ~70MB with the XKT Mod. Now, I create a small text file listing all the animations and load only the animations required by the players in the current match (or menu), saving a lot of loading time in addition to these ~70MB. (Side note : I'll likely update the Xbox version before the end of the week with these improvements)

The 2nd is that when checking memory usage I found out that one of the biggest memory consumption is from the crowd : I create big static meshes from all the people in the crowd, so only a few big meshes are rendered instead of many many small ones. It's much faster to do it like that, but it costs around up to 400MB of RAM. It's quite high on consoles where memory is sparse (only 5GB for the game on Xbox One S & PlayStation).

So I spent nearly 1 week checking the technic called GPU Instantiated Rendering : this allows rendering each person in the crowd separately, but using only the GPU (usually the slowest part in rendering is calculating on the CPU and then sending the data from the CPU to the GPU). The basics are not too complicated, but it actually was slower in my tests : it was because all the crowd was always rendered. So ~5000 persons was a bit slower than rendering only 200 on the screens. This means that I should have implemented the view frustum on the GPU, but that started to be quite complicated, plus I found an important limitation : GPU instantiating doesn't support static lighting with Unity. That was not acceptable, and working on a solution to bypass this limitation would likely take weeks of intense effort. So it felt like all-in-all the 400MB price wasn't that bad.

But I didn't stop there on the memory optimization. The PS4 DevKit has a very nice memory analyzer which helped me to pinpoint why the memory consumption was so high during the stadium init : it was because of the player texture init. I was compressing them, which was supposed to lower their footprint at the end of the frame, but it didn't (coz Unity is stupid ? ). After some fiddling, I found out I had to wait not 1, not 2 but 3 frames between each player to get the texture memory freed and avoid a big memory spike. This saved me more than 300MB in singles (and much more in doubles)..!

So between all this, the memory consumption became much more manageable. As a bonus, Stef updated his Mod following my optimization recommendations, so now all his big scoreboards do not take that much memory anymore, and so I didn't have to work on the Dynamic image loading for the Menus.

That's all for today, cya soon, hopefully for a release date on PlayStation !

A comment, cheer or question about this mini-blog ? => viewtopic.php?p=293028#p293028

[manutoo]

PlayStation Port is (almost) done ?

Yup, it's the same title as last time, but I changed the exclamation mark with a question mark.

The game failed 3 times the Sony certification so far. I had a bunch of issues to fix, some being silly mistakes on my part, and others being the consequences of the obscure Sony docs. And one crash possibly from Unity.

The process is very excruciating and I hope it'll be done soon. I already queued for the 4th check, but I think there's only a 20% chance it'll be ok this time.

So now, it's pretty clear the game won't be released in October. I hope it'll be ok for November 12th (ie: for the ATP finals). The game could even fail a 4th & 5th time and that date would still work.

Anyway, if you're very impatient to see the game released on PlayStation, you're still less than 1% impatient like I am. I'm really running on fumes here and I could use a significant break...

2024-10 - TE4 - PSN Logo.jpg

Xbox & Sales

Despite being focused on the PlayStation, I managed to do an update for the Xbox & PC versions, bringing some needed bug fixes and a few requested Modding features.

It's a good occasion to review the Xbox sales : from June to September, they brought about 25% less than the sales on PC. It's not bad, and it means my income this year is going to be pretty nice, but the sales already went noticeably down in September on Xbox compared to PC, and I'm afraid it means that the total will be much behind the PC sales, while I was hoping to earn as much on Xbox as on PC.

So now I hope the PlayStation sales will be noticeably better than the Xbox ones.

Thus said, I cannot complain too much, as my 1st income goal is almost reached for TE4+TEM2 : I count them together and compare them against TE3+TEM1 (reminder : TE3 includes TE2009, TE2011 & TE2013). It makes sense to do it like that coz a lot of work that went into TEM1 was useful for TE3 and the same with TEM2 for TE4. The 1st goal is to earn 66% of TE3+TEM1, and the 2nd is to reach 133%.

The 1st goal is under TE3+TEM1 because times changed a lot : TE3 sold pretty well from managames.com, while nowadays Steam has captured nearly the entire market (and they take 30% where my payment processor takes 9%). TE3 had nearly no competition on PC, while TE4 has many, many competitors : when TE2013 arrived on Steam in March 2015, there was only Virtua Tennis, which was then removed a few months later. There was no other tennis game on Steam till 2018.

And the 2nd goal is above TE3+TEM1, coz TE3 has never been released on consoles ; that goal won't be reached if the console sales aren't noticeably better, though. So we'll see in 1 or 2 years if it was a reasonable hope...

That's all for today. Hopefully, we'll get to see TE4 on PlayStation soon enough...
Cya next time !

A comment, cheer or question about this mini-blog ? => viewtopic.php?p=293693#p293693

[manutoo]

November 11th !



It has been intense, it has been nerve-wracking, it has been tiring, but after 5 attempts, the game has finally been certified by Sony !

I changed the release date to the 11th instead of the 12th, coz it's a holiday in France and a few other countries so hopefully many people will use their day off to buy the game and play it all day long..!

If you want to wishlist the game, it's still here => https://store.playstation.com/en-us/concept/10011758 .


Trailer incoming ?

These days, I'm trying to create a small trailer (as you recommended me to do it). I learned how to use Davinci Resolve (it's free and really good ), coz Windows Movie Maker couldn't let me do much (I used it for the TEM2 trailer).

Here's a little teaser :

2024-11 - TE4 - Trailer Teaser.jpg

I thought about showing the different bounce heights on the different surfaces, as a famous concurrent has the exact same bounce physics for all courts.

I'm not sure if it's good enough or not, but anyway, I need to do that to gather ideas and create footage. If it's really subpar, I'll see if I can hire a professional trailer creator.

Cu soon for more news.

A comment, cheer or question about this mini-blog ? => viewtopic.php?p=293693#p293693

[manutoo]

Pfffffffffffffffffffffffffffffff !

Ok, that's official, I have never been that tired for that long in my life without being seriously ill.

These 2 months around the PlayStation release really completely drained my energy.

I did the last significant PlayStation update almost 4 weeks ago and my mental energy is still not fully back : far from it actually, I just started to get a tad better a couple of days ago. Plus I have been a bit sick on and off these past weeks (as well as my kids & wife), so that didn't help either.

Also, I had to take care of small and bigger family, house & personal things that I had put on the side in the past 2 months, so anyway, I needed some time off.


What happened exactly ?

Back to dev news ; if you didn't notice the dedicated topic, I finished the trailer before the PS release :


After the release, some very frequent crashes have been reported when playing the game with Mods. I had tested, but not thoroughly enough, and seeking the cause led me to a rabbit hole. There are undocumented and completely unintuitive (read "stupid") memory handling in Unity. I never really noticed them so far : they came to light on PS because it's the system where memory is the most sparse. On Xbox, I had some issues, but just cleaning my own mess was enough as the memory is so much easier to handle, thanks to the Xbox system.

I asked for help from Unity support, but they didn't provide any ; on the contrary, they didn't understand what I explained and it made me even more annoyed.

So I had to study Unity's behaviors all by myself to understand the terrible mess it was. After 2 very intense weeks, I finally managed to find tricks to bypass that stupid Unity system. Basically, there were 2 main issues :
- when changing the camera screen size (en: when switching to split-screen), Unity doesn't release the camera render textures before assigning the new ones, leading to an extra 400MB of used memory in 4K ; if the system runs out of memory, it just crashes ; to avoid that, I added a wait time so the textures are released 1st, but I can't know if this wait time works in all cases or not, as it's not documented ; it's beyond me that such a system was made
- even worse, when uploading a texture from the CPU memory to the GPU memory (eg: when customizing the player outfit), instead of uploading directly from the texture CPU buffer, a copy is created ; this means if the texture is uploaded repetitively (eg: when the user is looking all the different T-shirts), then it creates several temporary upload buffers, which requires much, much more memory than expected ; in this case, it was up to 850MB !! Oh, for crying out loud, what a horrible thing ! The trick there has been again to add a waiting time between 2 texture uploads ; that's why you now can see the player's textures modified over a couple of seconds after a change

So all that plus the stress of the Sony submission process in October really took a toll on me. (and I guess the previous 8 years as well... )


Some good news !

At least, the PS sales have been good so far : almost 60% better in November than the Xbox ones in May, while the PS version was released on the 11th, and the Xbox on the 8th ; and still nearly 55% better for PS' December 1st-22nd against Xbox's June 1st-22nd.

So that's quite nice but it's important that the PS sales go on like that for at least all 2025 : the Xbox ones dropped just after 4 months... So it's still too early to conclude, but at least it's quite a decent start.


What now ?

In late October, I was hoping to release the v1.0 in late November, but the turn of events really pushed that away. I hope it'll be in January now, but I have a bunch of little and bigger things to do to reach that state.

I'll get more into details in the next post, this one has been long enough already.


Meanwhile, Happy holidays and merry Christmas (if you celebrate it).

A comment, cheer or question about this mini-blog ? => viewtopic.php?p=293806#p293806

[manutoo]

...

Oh damn, I was a bit sick when I wrote my previous post, but 2 days after (ie: Xmas) I got super tired and now I'm still quite tired... On the other hand, my mind is getting kinda fresh and I even tried to do some serious work 2 days ago, but then I got a headache in less than 20 minutes...

Hopefully, I'll get better soon, but it's really boring to be impaired like this for more than 1 month already.


Motion Matching Test #2

So I wanted to let you know what I did during December. It's not much but the result is encouraging. So during the 1st Motion Match test, I just took the example bundled with the Motion Matching plug-in I bought and put the TE4 3D player and tuned a bit the speed.

In that example, the animations control the movement of the player, but in TE, the physics control the movements and the animations have to follow it.

So I did a 2nd test, recreating the TE4 physics model but a bit simpler (it turned out to be quite difficult actually ), so it'd be close to the current TE4 movements, and after some fiddling, I got that :


I think overall the result is pretty neat and we can see a bit more diverse animations than in the 1st test because the TE4 physics system is more realistic than the weird system in the example.

However, the feet slide a lot more. This means I'll have to find a nearly professional tennis athlete who can move really fast, or else the resulting anim set I'll create by myself will have a lot of foot sliding as well...

Does anyone have Nadal's phone number around here ? I heard he has some free time now...

To create this 2nd test, I had to read the plug-in doc and understand the basics of using it, so now I think there's a good chance it'll work with the game. I still have to fully incorporate it to check if the blending with the strike animations will be ok or not. It might require a bunch of stitching so I'll need my body to have its energy back to do that final step before jumping into producing the new anims.

It'll be quite a challenge and at this point, I'm still not sure if I should use AI to create animation from video, or buy an inertia-based motion tracking set. The AI I know doesn't handle props and that might be quite annoying to position the racket correctly, but it has some ease of use and it might be cheaper than the motion trackers.

It might take me around 4 months, but I'm really not sure. It'll depend on how much hand-cleaning the animations require once they are captured. I'll likely hire an animator to do that.


Vesion 1.0 ?

I also would like to release the v1.0 within a few weeks.

The main things for it would be :
- updating player bases with rankings up to the end of 2024, so it'd be possible to start in 2025
- a few little but important gameplay tweakings
- a very basic in-game .ini editor so console players would have access to all the settings hidden in the Tennis.ini
- possibly a few other little things


Cu next time to see if I get better and get some serious work done...
Meanwhile, Happy New Year.

A comment, cheer or question about this mini-blog ? => viewtopic.php?p=294462#p294462

[manutoo]

Version 1.0 !

After nearly 6 years of development, we finally reached the version 1.0 of Tennis Elbow 4 !

So it has been 2 months since the previous mini-blog post, which means things didn't go as planned.

1st, I spent the 1st half of January checking a cheap solution to do motion capture : Ipi Software. It's not AI powered (except on a couple of side stuff), but the result seemed very nice. I tried to make a recording by myself to test the system, and it turned out to be quite a hassle. So finally, I just took the sample video on their website. And there, at my big surprise, it didn't give the result they were showing : it actually required to do a _lot_ of manual clean-up. So these 2 weeks went to the bin.

However, I also understood AI solutions weren't great, because they use video as well, so it requires to be able to set up the cameras far enough from the action to see the whole actor body, which prevents to capture animations in my living-room (coz I don't live in a castle with a big ballroom ), as well as a bunch of cameras if I want to capture a tennis player running fully left-right on a real tennis court.

So I was left with 2 options :
- a classic Motion capture studio, using visual captors (ie: what has been done for games since 25 years)
- an inertia-captor system, that I could use at home or anywhere

pnpro-2x@2x.png

Personal issue version #45897

After that, my son had to do a relatively small surgical operation at the hospital ; the operation itself was fine, but its consequences were quite unexpected and exceptional (less than 3% of the times, according to his surgeon) and I had to take care of him almost full time for 10 days, and after that he had to still stay at home for 10 more days !

So it took a big toll on my work for a long while, and left me quite stressed... Fortunately, he's better now, but still not 100% ok.


Back on track (again)

So I have been back to serious work a couple of weeks ago and went into integrating the Motion Matching plugin into the game. It was quite challenging as I discovered several issues, the main 2 ones being :
- the lack of sidestep, as expected, is annoying, especially as the plugin came with a strafe system that doesn't match the need for a tennis game
- the lack of small positioning footwork, which leads the player to walking when using the autopos

This means that, unlike what I hoped, I couldn't switch to using this system by default for the v1.0 of TE4. Instead, you can activate it manually in Settings -> 3D Match. It'll require to create a complete set of tennis animations to get all that to work correctly.

Additionally, some weird rotating animations were triggering more or less randomly. Fortunately, I found out they were from the anim set, so I had just to remove them and the issue disappeared.

I still used the plugin default strafe system, to bring animation variation, but after striking the ball, I had to revert to the legacy sidestep anims to avoid a pelvis orientation issue.

It was also missing backward running animations, so I put one similar to the legacy one.

So, overall, the result is pretty nice when running around, and really not that good when doing positioning footwork. After trying it for a couple hours, please let me know what you think of the current state here : topic7-36310.php .

I could have cleaned up more the issues, but it would have been lost time : it's much more logical to wait for the tennis anim set.


Back to the future

So I settled my choice for the Motion Capture. I asked a quote from a popular Motion Capture studio and its cost is high but still reasonable, but I think it's safer to go with the inertia-captor solution, for the flexibility and the autonomy ; ie: if I need to do more anims, it'll be much easier to proceed.

Also, now, I'm 100% sure I need to find a tennis player with near professional footwork to get a good result in-game. This is going to be a big quest. (if by chance, you have the footwork of Alcaraz or not far, and live in France, in Drôme or Isère districts, let me know )

Once I find such a player, I'll buy the inertia system, so it might take a while to get the new anims in the game. If we're lucky, it could be done for June...


Cu next time to see how all of this goes..!

A comment, cheer or question about this mini-blog ? => viewtopic.php?p=295470#p295470

[manutoo]

Actor found ! (*)

After a bit of research in the tennis clubs around my city, I found a quite good player : he wasn't that far from being a real pro, but fell short a bit. So his movement is probably not on par with Alcaraz, but he's really close to the average pro, as he has all the automatisms and the rigor. I spent a few hours checking videos on Youtube of amateur tennis, and it's flagrant how little effort most of them put into the legs when compared to the pros.

When I went to check on him, he was playing against an ex -4/6 (that would be 6.7 or 6.8 US rating) ; that other guy looked like he was on the beach and didn't put any real effort in his legs (and of course, at my best, I'd have never scored a point against him ), so it's pretty rare to find a non-pro having a pro movement.

This guy actually learned in the USA and tried to turn pro, and that's why he moves like that.

Here's the little test I asked him to do to see how fast and flexible he was :



I think it looks pretty good. Hopefully, you'll think too, coz Alcaraz still hasn't phoned me.

So I ordered the Inertia Motion Capture system. It's pretty expensive so I hope everything will be fine with it. Once I get it (within a couple of weeks if it's on time), I'll have to do a couple of sessions by myself (or maybe with my brother) to get the hang of its handling. Once I'm comfortable with it, we'll do the session with that player, hopefully within 3 or 4 weeks. After that, I'm not sure how long it'll take to clean all the animations and prepare them for the game.

(*) : in motion capture, the person whose motion is captured is called the "actor".


Stats & Records

Since the beginning of the year, the record for online matches per week has been broken almost every other week, and last week we reached 2'000 matches !

It's still far from the TE2013 record (about 10'000), but it's getting closer to the average, which was around 4'000. I hope it continues to grow and that people enjoy the online part of the game.

To go on with records, the number of TE4 concurrent players on Steam has just reached 140, while the TE2013 record was 144. This is very close, although when TE2013 got its record, there were likely more ManaGames.com players, so the real total on PC might be closer to 155 vs 245.
However, if we also include consoles, we might be near that record, but I don't have any tracking number, so I can't be sure...

Anyway, it's pretty nice to see TE4 being played & enjoyed that much !

20250318_094150_s.jpg

Cu next time with hopefully some report about my 1st motion capture tests..!

A comment, cheer or question about this mini-blog ? => viewtopic.php?p=295603#p295603