ACCEPTED ENBSeries (by Boris Vorontsov)

[rpsgc]

  'techangel85' pid='20911' dateline='1358679283 said:

I don't agree with these performance marks' date=' though I've been using Cinematic. I've left him a message and am going to be doing some extensive testing once I find out his test method and tools because [b']I didn't see that kind of performance drop from v.119 to v.139[/b].

Because you have an NVIDIA card.

 

Most people with an AMD card see a big performance drop.

https://forum.step-project.com/showthread.php?tid=744&pid=16062#pid16062

[TechAngel85]

  Quote

  'techangel85' pid='20911' dateline='1358679283 said:

I don't agree with these performance marks' date=' though I've been using Cinematic. I've left him a message and am going to be doing some extensive testing once I find out his test method and tools because [b']I didn't see that kind of performance drop from v.119 to v.139[/b].

Because you have an NVIDIA card.

 

Most people with an AMD card see a big performance drop.

https://forum.step-project.com/showthread.php?tid=744&pid=16062#pid16062

If that's the case I wish he'd go ahead and do his magic and update the ENB and simply provide two versions of it: One for NVIDIA/v.139/whoever doesn't have performance issues... and one for AMD/v.119/whoever does have performance issues.

[rpsgc]

  Quote

If that's the case I wish he'd go ahead and do his magic and update the ENB and simply provide two versions of it: One for NVIDIA/v.139/whoever doesn't have performance issues... and one for AMD/v.119/whoever does have performance issues.

I don't think Boris cares enough about his 'fans' to do be arsed to do something like that.

[Besidilo]

Skyrim particle patch for ENB. Could be an important addition to STEP since ENB is the core part of the guide now.

 

Like I predicted, Boris has brought back some previous ENB versions, namely v132 and v139. v121212 is the same ENB as v126, but with no GUI. Not sure what the difference between v139 and v132 is, but there are a few nice updated configs supporting the later ENBs.

 

K ENB (v132)

SES FX ENB (v141 preview)

Beautiful Skyrim by UnrealWarfare (v139)

Tiamat Fantasy ENB (v141)

 

On another note, a lot of ENB modders (if you can call them that) suggest using Anisotropic Filtering of 8x that comes with ENB instead of the one set in drivers or in-game.

 

Has anyone got experience with the later ENB configs and can comment on the quality of that AF? I think it could save a few fps in the long run with little to no quality decrease.

[TechAngel85]

  Quote

Skyrim particle patch for ENB. Could be an important addition to STEP since ENB is the core part of the guide now.

 

Like I predicted, Boris has brought back some previous ENB versions, namely v132 and v139. v121212 is the same ENB as v126, but with no GUI. Not sure what the difference between v139 and v132 is, but there are a few nice updated configs supporting the later ENBs.

 

K ENB (v132)

SES FX ENB (v141 preview)

Beautiful Skyrim by UnrealWarfare (v139)

Tiamat Fantasy ENB (v141)

 

On another note, a lot of ENB modders (if you can call them that) suggest using Anisotropic Filtering of 8x that comes with ENB instead of the one set in drivers or in-game.

 

Has anyone got experience with the later ENB configs and can comment on the quality of that AF? I think it could save a few fps in the long run with little to no quality decrease.

I'll try to remember to test the ENB AF whenever I do my other testing in a bit. On NIVIDA drivers, you see around a 4% decrease in FPS with AFx8 and AFx16.

[TechAngel85]

I have completed my extensive ENB version testing. Below are the results. It's lengthy!

 

 

ENB Version Tests

 

  Reveal hidden contents

 

I did two tests... The first test (test one) replicated MTichenor's test. Standing outside the cave coming from Helgen. This is one of the most graphically intense areas in Skyrim. I used a stopwatch and stood motionless for one minute. The second test (test two) is running from this location to the other side of the stone bridge on the far side of Riverwood. I timed this perfectly so a time change to dusk would occur during the run. This provided a more realistic test for actual gameplay.

 

Test Conditions:

• SkyRealism with Cinematic preset
• Only mod installed is SFO (to replicate MTichenor's conditions)
• Transparency AA - Supersampling 2x for SFO (no other AA used)
• No AF
• FRAPS used to log stats
• v.141 beta was not tested as it only changed SSAO settings and SkyRealism already uses custom settings.
• Each version is a complete version step up. DLL and all enbseries.ini changes were included from each version (this is a step further than MTichenor instructions to update the enbseries file which can be see above in my previous post)

Test One - No SSAO

Vanilla  // Frames: 3065 - Time: 61937ms - Avg: 49.486 - Min: 48 - Max: 50

v.119  //  Frames: 2146 - Time: 61750ms - Avg: 34.753 - Min: 33 - Max: 36

v.132  //  Frames: 2119 - Time: 61579ms - Avg: 34.411 - Min: 33 - Max: 36

v.139  //  Frames: 2147 - Time: 62188ms - Avg: 34.524 - Min: 33 - Max: 36

 

Test One - with SSAO

Vanilla  // Frames: 3065 - Time: 61937ms - Avg: 49.486 - Min: 48 - Max: 50 (no SSAO - reference)

v.119  //  Frames: 1661 - Time: 61328ms - Avg: 27.084 - Min: 26 - Max: 29

v.132  //  Frames: 1751 - Time: 61922ms - Avg: 28.278 - Min: 27 - Max: 30

v.139  //  Frames: 1689 - Time: 61829ms - Avg: 27.317 - Min: 26 - Max: 29

 

Test One - with SSAO and Reflections

Vanilla  // Frames: 3065 - Time: 61937ms - Avg: 49.486 - Min: 48 - Max: 50 (no SSAO/Reflections - reference)

v.119  //  Reflections not available in .119

v.132  //  Frames: 1655 - Time: 61922ms - Avg: 26.727 - Min: 25 - Max: 28

v.139  //  Frames: 1594 - Time: 61688ms - Avg: 25.840 - Min: 25 - Max: 27

 

Test Two - No SSAO

Vanilla  // Frames: 8807 - Time: 155984ms - Avg: 56.461 - Min: 41 - Max: 62

v.119  //  Frames: 6018 - Time: 151797ms - Avg: 39.645 - Min: 29 - Max: 46

v.132  //  Frames: 6252 - Time: 156328ms - Avg: 39.993 - Min: 28 - Max: 48

v.139  //  Frames: 6013 - Time: 152437ms - Avg: 39.446 - Min: 28 - Max: 46

 

Test Two - with SSAO

Vanilla  // Frames: 8807 - Time: 155984ms - Avg: 56.461 - Min: 41 - Max: 62 (no SSAO - reference)

v.119  //  Frames: 4499 - Time: 152750ms - Avg: 29.453 - Min: 22 - Max: 34

v.132  //  Frames: 4783 - Time: 152625ms - Avg: 31.338 - Min: 22 - Max: 37

v.139  //  Frames: 4597 - Time: 152422ms - Avg: 30.160 - Min: 21 - Max: 35

 

Test Two - with SSAO and Reflections

Vanilla  // Frames: 8807 - Time: 155984ms - Avg: 56.461 - Min: 41 - Max: 62 (no SSAO/Reflections - reference)

v.119  //  Reflections not available in .119

v.132  //  Frames: 4427 - Time: 152172ms - Avg: 29.092 - Min: 21 - Max: 34

v.139  //  Frames: 4271 - Time: 153000ms - Avg: 27.915 - Min: 20 - Max: 32

 

Final Verdict:

As far as going from no ENB (vanilla) to using an ENB, the performance hit is as expected. That is pretty much where the performance loss stops. In the tests with no SSAO, the performance is pretty much the same for all versions of ENB. In the tests with SSAO turned on, v.132 performs just slightly better and the other two version are evenly matched. Turning on reflections, in the third test, is where you'll see a performance loss when using the latest version as v.132 out performs v.139 here.

 

During this testing and using SFO, I would have to recommend SSAO be turned off and same goes for reflections. Although SSAO adds some realism to shadows around buildings, it does horrible things for the foliage. It adds unrealistic dark shadowing to the bottom of green foliage and it makes Vurt's dead foliage look as though it's been burnt. So turn it off and reap the benefit of the performance boost. Same goes for reflections because simply turning it on comes with a performance hit. Interestingly enough, reflections will effect performance even where there are no reflections...say in the middle of the forest. That doesn't make since to me since no reflections are being rendered.

 

As far as MTichenor's report of performance loss, I'm clueless. I'm seeing no performance loss between v.119 and v.139. In fact, the later versions perform slightly better with SSAO turned on. The only thing I can think of is a difference in video card manufacturers, as rpsgc pointed out. I'm running NVIDIA and he's running AMD.

 

 

 

Anisotropic Filtering (AF) Testing

 

  Reveal hidden contents

 

These tests were ran similar to Test One. I stood still (in the water shot) for one minute. NVIDIA Inspector was used for force driver AF.

 

Anisotropic Filtering x8

Vanilla     //  Frames: 2726 - Time: 61750ms - Avg: 44.146 - Min: 43 - Max: 46 (no AF - reference)

ENB AF    //  Frames: 2493 - Time: 62468ms - Avg: 39.908 - Min: 38 - Max: 42

Driver AF //  Frames: 2416 - Time: 60422ms - Avg: 39.985 - Min: 39 - Max: 42

 

Anisotropic Filtering x16

Vanilla     //  Frames: 2726 - Time: 61750ms - Avg: 44.146 - Min: 43 - Max: 46 (no AF - reference)

ENB AF    //  Frames: 2401 - Time: 60594ms - Avg: 39.624 - Min: 39 - Max: 41

Driver AF //  Frames: 2484 - Time: 62391ms - Avg: 39.813 - Min: 38 - Max: 42

 

Final Verdict:

As you can see, there is no performance difference from using either the ENB's included AF or forcing it through the drivers. So lets check out the visuals. The forum has limitations here so go to this wiki page for these visual compares: ENB Compares.

 

From the visual standpoint there is still very little difference between the driver AF and the ENB's included AF solution. Here, the choice is yours. ;)

 

 

 

Antialiasing (AA)

 

  Reveal hidden contents

 

I have first confirmed that hardware AA no longer works on versions past v.119. MTichenor explains why:

  Quote

When you make changes to anti-aliasing levels in your graphics drivers, those drivers are making changes to the d3d9.dll file in your Windows directory (not exactly, but close enough). When you use ENB, you're placing a new d3d9.dll file in the application directory which over-rides the one from your Windows directory. [therefore, hardware AA no longer works]

So, sad day? Not really. The ENB's included AA solution does a fairly good job at making up for this.

 

Vanilla   //  Frames: 5628 - Time: 153187ms - Avg: 36.739 - Min: 26 - Max: 45

ENB AA  //  Frames: 5361 - Time: 153078ms - Avg: 35.021 - Min: 24 - Max: 41

(This test was doing the "Test Two" run from above with AFx16 enabled)

 

Vanilla (link)  >>  ENB AA (link)

*included links to full size images for better compare.

 

The ENB does a pretty good job with AA and at very little to no performance loss which will be a Godsend for modders with lower or mid-range systems.

 

 

[Daagar]

Nice comparison. What I really hope to see more of in the future though is a greater number of ENB presets that make use of SweetFX (SMAA) instead of FXAA. I've been toying with a few ENB presets on top of a Neovalen-based 'Skyrim Revisited' setup, and the FXAA just makes things a horrible blurred mess. I need to do some more tinkering to try a few of these without FXAA enabled to judge things better.

 

I just found that TheCompiler has RCRN Plus, which _is_ SweetFX-based. Hoping to give that a shot too, despite the RCRN authors saying that it doesn't go quite far enough with fixing/preserving the colors of the original RCRN.

[Korentin]

Most ENB presets made in the last few months make use of SweetFX or standalone SMAA actually. Never seen one recommending FXAA.

[Daagar]

Eep... my mistake then. I'll go back and review my setup then, as most have looked terrible. I might be having conflicts somewhere with incompatible settings. (I use SweetFX in borderlands2 and settlers7, for example, and it works wonderfully there).

[TechAngel85]

SMAA is a standalone product that can be used in combination with any ENB. If an ENB is using an FXAA injector just simply swap it out with SMAA and done! Instructions are on the ENB Guide.

[Daagar]

Ah, well that solves all issues then! However, for myself personally it all may be moot at this point. I had a chance to try TheCompiler's RCRN Plus, and unless something else blows me away I think I've found my preference :)

[TechAngel85]

WOOT! MTichenor said he's going to update to v.139!!! Testing paid off! AMD also released new drivers that increases performance for ENB users.

[Besidilo]

  Quote

WOOT! MTichenor said he's going to update to v.139!!! Testing paid off! AMD also released new drivers that increases performance for ENB users.

That's great news! Even better when you consider that jjc71 is working on a new Climates of Tamriel release shortly after Dragonborn is out!

[Deathwing514]

VERSION .139 Has Been Released For SkyRealism

 

Go Check It Out

[TechAngel85]

Anyone know how to get rid of the hazy effect in ENB? I've been playing with a lot of settings with no success. I know it's not bloom because it's present regardless if bloom is enabled or disabled.

 

 

Full Size Image