SF2 demo game engine enhancements?

[rocketman]

This from the "Read Me":

*    Faster graphics, including FPS improvements, especially for infantry-heavy maps and faster video cards.
*    Possible speed improvements depending on video card hardware and drivers.

I have played the "Day at the Beach" USMC scenario and it runs so smoothly. Don't think I have ever had a better performance. This on a Nvidia GTX 970M with latest driver. Am I imagining something or are we in for a treat with the patch for the rest of the games?

[JoMac]

Yeah, in general, all Demo's tend to run better & smoother then the Full Versions...

Edited October 17, 2018 by JoMc67

[The_MonkeyKing]

22 minutes ago, JoMc67 said:

Yeah, in general, all Demo's tend to run better & smoother then the Full Versions...

I am not convinced

[MikeyD]

I'm notorious for hesitating to play huge maps and dense urban areas. But I've pretty much forgotten my fears playing CMSF2. Admittedly, a huge city map with a battalion of infantry won't run the same framerate as a small open desert map with a few tanks scattered about, but it certainly runs good enough, and MUCH better than it used to back-in-the-day. If you insist on pushing the limits of the game's capabilities you might experience some push-back from the game. But most scenarios and most maps don't approach that limit.

[rocketman]

So there is some merit to the claims in the "Read Me" - that some improvements have been made to ensure better performance?

[MikeyD]

I guess you could ask 'Better compared to what?' It runs VASTLY better than the old CMSF1, that's for sure. It shares the same game engine as the other titles but CMSF2 maps generally lacks the heavy forested maps of other theaters, which helps framerate a lot. Plus limiting the building types to 'modular' allows large city maps to be built without over-straining the graphics card. Charles has very likely further enhanced general v4 performance, he's always tinkering. If he did it'll show up in the eventual across-the-titles v4 game engine patch.

Edited October 17, 2018 by MikeyD

[domfluff]

To be fair, that's in the section "WHAT IS NEW SINCE SHOCK FORCE 1".

The CM Engine *definitely* runs better than the engine that powered CMSF 1, so that line is entirely accurate. Whether there's been much improvement going from, say, CMFB to CMSF? That's less clear. Battlefront do try to optimise and improve where they can between releases.

For a number of reasons, including terrain and the weather, the Shock Force maps are generally less complex than some of the CMFB, CMBS or CMRT ones (not always, especially with some of the heavier urban ones). Less complex maps will run better in general.

[domfluff]

Or, yeah, what MikeyD said

[rocketman]

I didn’t read the header properly ...

but good to know that independent buildings where a bigger strain on performance than modular ones. 

[AttorneyAtWar]

The German scenario gives me a hard time when trees are set to on but everything else runs very smoothly on full settings.

[Buzz]

1 hour ago, MikeyD said:

I guess you could ask 'Better compared to what?' It runs VASTLY better than the old CMSF1, that's for sure. It shares the same game engine as the other titles but CMSF2 maps generally lacks the heavy forested maps of other theaters, which helps framerate a lot. Plus limiting the building types to 'modular' allows large city maps to be built without over-straining the graphics card. Charles has very likely further enhanced general v4 performance, he's always tinkering. If he did it'll show up in the eventual across-the-titles v4 game engine patch.

In my experience ^ MikeyD has it spot on correct. 

On Mac side... Running H2H comparison = CM Shock Force 2 Demo (2018)  is a much improved performance product vs. Mac CM Shock Force v1.32 (2012). Charles is always tinkering and whatever magic he has done shows on the Mac side... for me. 

Buzz

[BlackAlpha]

I've compared the training level to the same training level in BS and the performance is exactly the same. If you don't put the model detail on the lowest or near lowest setting, then your FPS will tank. I find it disappointing how in Black Sea I played with a GTX 680 and the performance was crap if you didn't put the model detail to the lowest. And now I'm playing BS and SF2 with a GTX 1080 and the performance is exactly the same. From what I remember, SF1 had similar issues. There just seems to be something in the engine that destroys the performance, no matter what hardware you run on it.

I'm still hoping that one day they'll do real engine updates (not just these minor patches).

Edited October 18, 2018 by BlackAlpha

[37mm]

7 minutes ago, BlackAlpha said:

I've compared the training level to the same training level in BS and the performance is exactly the same. If you don't put the model detail on the lowest or near lowest setting, then your FPS will tank. I find it disappointing how in Black Sea I played with a GTX 680 and the performance was crap if you didn't put the model detail to the lowest. And now I'm playing BS and SF2 with a GTX 1080 and the performance is exactly the same. From what I remember, SF1 had similar issues. There just seems to be something in the engine that destroys the performance, no matter what hardware you run on it.

My investigation of the Demo tree issues indicates that graphics settings & resolution have relatively little effect on frame rates... the "tree trunks only" setting makes the game run smoothly under almost any circumstances.

Currently I've also noticed that, unlike other CM titles, it is best to NOT to have a graphics card profile for CMSF2... just allow the base program to do its thing unhindered.

[BlackAlpha]

15 minutes ago, 37mm said:

My investigation of the Demo tree issues indicates that graphics settings & resolution have relatively little effect on frame rates... the "tree trunks only" setting makes the game run smoothly under almost any circumstances.

Currently I've also noticed that, unlike other CM titles, it is best to NOT to have a graphics card profile for CMSF2... just allow the base program to do its thing unhindered.

For me it's the same in BS. BS with trees runs the same as SF2 with trees. BS without trees runs the same as SF2 without trees. But anyway, it's why I always play with the tree trunk mode to improve the performance. If you want to play with trees, then you'll need to drop the model detail. Just compare the lowest model detail setting to the highest, the difference is night and day.

[MikeyD]

In 'options' 'Tree Detail' turned off causes the trees to stop swaying in the wind which gives the graphics card much less to do. Swaying trees, as an indication of wind strength, comes in handy if you're using smoke screens. Scenario order text will often tell you up front wind strength & direction (for use of smoke). If you don't need swaying trees turn off 'tree detail'. That swaying tree feature killed me in large heavily forested (pre-v4) CMBN maps.

Oh, and indi buildings aren't 'more' of a strain on performance compared to modular, they're just additional stuff on the map to render. Another 25-ish building models with 2-3 texture options each.

[Hister]

I'll never understand how you guys can say the game plays smooth. Looks like the meaning of smooth can vary a lot from person to person. I could call it a smooth slideshow if I would have used the smooth word describing my experience with the game and it has been like that on all the different hardware I had in these years up until the beastly rig I sport now. The only thing I haven't yet tried is using gsync monitor that purpotedly makes the gameplay "very smooth" but I am not falling for that any more, heh.     

Edited October 18, 2018 by Hister

[A Canadian Cat]

6 minutes ago, Hister said:

I'll never understand how you guys can say the game plays smooth. Looks like the meaning of smooth can vary a lot from person to person.

Yep - clearly. I am happy if the wasd keys move the screen with out stalling and the replay does not stutter. For me I am frequently happy as a clam when the FPS counter shows 20. When it gets up to 30 to me it's not different that when its down at 20. Frequently I don't' even notice when it drifts down to 17 even. I am happy to play and I call it smooth. I have no doubt that if someone who is being more picky came and sat next to me pointed out the issues I would notice them. I am also sure that after they left I would get on with playing and never notice them again

 

1 hour ago, BlackAlpha said:

I'm still hoping that one day they'll do real engine updates (not just these minor patches). 

Totally - a full rewrite is the kind of thing it would take. The limits of the engine performance we have now is largely a result of choices made based on the state of the art circa 2004 or so. A lot has chanced since then but BFC aren't able to be rewriting the engine ever year or two. Granted they have made a lot of improvements over the years but not the kind of change that a rewrite would allow.

[37mm]

10 hours ago, MikeyD said:

In 'options' 'Tree Detail' turned off causes the trees to stop swaying in the wind which gives the graphics card much less to do. Swaying trees, as an indication of wind strength, comes in handy if you're using smoke screens. Scenario order text will often tell you up front wind strength & direction (for use of smoke). If you don't need swaying trees turn off 'tree detail'. That swaying tree feature killed me in large heavily forested (pre-v4) CMBN maps.

Oh, and indi buildings aren't 'more' of a strain on performance compared to modular, they're just additional stuff on the map to render. Another 25-ish building models with 2-3 texture options each.

If you check out my investigation thread you will see I have compared both "high detail" & "low detail" tree settings... also neither the CMRT or CMSF scenes I was comparing had any wind.

CMSF2 foilage triples (sometimes quadruples) the "draw call" counts of a running game. This does not occur for other CM titles like Red Thunder.

Edited October 19, 2018 by 37mm

[37mm]

9 hours ago, IanL said:

For me I am frequently happy as a clam when the FPS counter shows 20. When it gets up to 30 to me it's not different that when its down at 20.

One of the things I noticed during my investigation is that the "frame rate stat" is less important for CM than you would think, I think Steve has mentioned something similiar in the past.

However the time it takes your system to render one frame does seem to be pretty important. For me I found that if it's less than 100ms your gameplay will be smooth, if it's over 200ms then you will notice stuttery play & poor animations.

[Bulletpoint]

19 hours ago, Hister said:

The only thing I haven't yet tried is using gsync monitor that purpotedly makes the gameplay "very smooth"

I have a Gsync monitor and it only makes other games smooth, not Combat Mission. It does, however, sometimes make the screen flicker like mad, so I generally turn off Gsync for CM.

[db_zero]

21 hours ago, BlackAlpha said:

I've compared the training level to the same training level in BS and the performance is exactly the same. If you don't put the model detail on the lowest or near lowest setting, then your FPS will tank. I find it disappointing how in Black Sea I played with a GTX 680 and the performance was crap if you didn't put the model detail to the lowest. And now I'm playing BS and SF2 with a GTX 1080 and the performance is exactly the same. From what I remember, SF1 had similar issues. There just seems to be something in the engine that destroys the performance, no matter what hardware you run on it.

I'm still hoping that one day they'll do real engine updates (not just these minor patches).

What are the rest of the specs on your computer? GTX1080 is a great graphics card.

I have an older GTX 970 and I run the game maxed out on all CM games including Black Sea. I’m also running at 2560x1440 on a widescreen monitor. Only on the very largest of maps with tons of units do I get a very very slight pause when scrolling the map. It’s barley noticeable.

My settings also allows for the trees to sway in the wind which is a nice effect.

My system is intel I7, 16GB ram, GTX 970 and 2 SSD drives.

 

Edited October 19, 2018 by db_zero

[Hister]

3 hours ago, Bulletpoint said:

I have a Gsync monitor and it only makes other games smooth, not Combat Mission. It does, however, sometimes make the screen flicker like mad, so I generally turn off Gsync for CM.

Ah, no real surprise there. Thank you for letting us know. 

[Lethaface]

31 minutes ago, Hister said:

Ah, no real surprise there. Thank you for letting us know. 

I don't have the flickering problem with my GSYNC monitor, luckily. And yes CM doesn't run on 165 fps, but I consider this quite 'smooth' and IIRC I have the AA tuned up for CMBS:

 

 

Edited October 19, 2018 by Lethaface

[Bulletpoint]

18 minutes ago, Lethaface said:

I don't have the flickering problem with my GSYNC monitor, luckily.

The flicker only appears when the game frame rate fluctuates a lot. It doesn't appear all the time. Sometimes, it's not there at all. Sometimes, it's just barely perceptible. But in other situations, when the camera is pointing in a certain direction, it will flicker like crazy.

19 minutes ago, Lethaface said:

And yes CM doesn't run on 165 fps, but I consider this quite 'smooth'

I think you rotate the view mainly using the keyboard? I find the most sluggish performance comes from holding down right mouse button and looking around that way.

[Lethaface]

12 minutes ago, Bulletpoint said:

The flicker only appears when the game frame rate fluctuates a lot. It doesn't appear all the time. Sometimes, it's not there at all. Sometimes, it's just barely perceptible. But in other situations, when the camera is pointing in a certain direction, it will flicker like crazy.

I think you rotate the view mainly using the keyboard? I find the most sluggish performance comes from holding down right mouse button and looking around that way.

Regarding the flickering: strange. My brother build the exact same pc as me, only difference is he got a TN gsync panel while I got an IPS. He did get some flickering in certain situations, for example in the TW warhammer menu screen. IIRC it had to do something with certain semi static graphic objects being rendered a certain way. The strange thing was that I didn't have the flickering at all. And yes I also checked his screen to see whether I could see the flickering. Not sure if he still has it and he doesn't play CM unfortunately. 
I haven't noticed flickering in CM at all.

In the video and during normal play I do use the keyboard mainly for rotating. I just tried rotating with the mouse in the CMSF2 demo beach scenario and I don't see a real difference compared to the keyboard. There is a bit of difference between 'wide' and '1.0x' screen adjustment in CM (using the 'c' key), the 'wide' option seems smoother.

I have a 165hz Gsync monitor and while CM only runs at 20-30 fps on average use, it seems quite smooth. A lot smoother compared to CM on my previous PC's! 

Edited October 19, 2018 by Lethaface