VOGONS


dgVoodoo 2 for DirectX 11

Topic actions

  • This topic is locked. You cannot reply or edit posts.

Reply 2160 of 3949, by Expack3

User metadata
Rank Member
Rank
Member
Dege wrote:
Thanks for the newest reports!! […]
Show full quote

Thanks for the newest reports!!

Expack3 wrote:

What, exactly, does the new DX resolution override functionality do? From my observations, it seems to scale everything, from textures to polygons to everything else, to the chosen resolution, stretching the image if necessary. I ask because, despite the announcement that resolution override is enabled in dgVoodoo2 2.52, section 7 of ReadmeDirectX.txt has only this to say:

7. DirectX setup options wrote:

- Resolution: It is disabled in this version, sorry.

Uhmm, I forgot to update that part of the readme...
I thought nobody reads it, not even me. 😀 But the setup-app tip is correct.

I can only speak for myself, but I read it because 1) it's a very clear explanation of what dgVoodoo2 is doing behind-the-scenes, and 2) I can point my technology-minded friends and co-workers to it when they ask "how does dgVoodoo2 do [x]?" or ask for a reference regarding my statements on its functionality.

Dege wrote:
dgVoodoo renders every GPU-drawn parts in the selected resolution. This includes GPU-drawn 2D HUDs which simply get scaled up. S […]
Show full quote

dgVoodoo renders every GPU-drawn parts in the selected resolution. This includes GPU-drawn 2D HUDs which simply get scaled up.
Software (CPU)-rendered parts are simply scaled up as 2D images, too.
So, for example if you run a DDraw-only app (scene demo) in large resolution then only a 2D-scaling is what you'll see.
One of the best examples is Resident Evil: if I force it to large resolution then the backgrounds are very pixelated (320x200 or so) while the characters are rendered nicely in high-res onto that.

I'm assuming it doesn't actually touch the viewport size. Otherwise, I'd expect it to look more like this than this (both in 16:10 4K resolution).

Reply 2161 of 3949, by Dege

User metadata
Rank l33t
Rank
l33t
duchU wrote:

I have tried Might and Magic 7 & 8 and they look much better with higher resolution setting combined with Lumasharpen filter from SweetFX but there is a problem with mouse cursor during Arcomage game. Cursor lags very badly. Is there any chance to fix this? Besides this mouse seems to work fine... If you set resolution to Unforced then this bug is gone but of course the games don't look that great anymore.

If resolution is forced then the cursor is unfortunately a sw emulated one, so if the game goes at low fps at that point then the cursor does, too.
I encountered this problem in Settlers3 menu, but there is a chance that I could fix it.

Expack3 wrote:

I can only speak for myself, but I read it because 1) it's a very clear explanation of what dgVoodoo2 is doing behind-the-scenes, and 2) I can point my technology-minded friends and co-workers to it when they ask "how does dgVoodoo2 do [x]?" or ask for a reference regarding my statements on its functionality.

Good to hear that. 😎
I was just joking about the readme, it's important to maintain that but I wasn't careful enough last time.

Expack3 wrote:

I'm assuming it doesn't actually touch the viewport size. Otherwise, I'd expect it to look more like this than this (both in 16:10 4K resolution).

It does. If the rendering buffer size is changed under the game then dgVoodoo has to scale up the phyisical viewport to match it. So, when coordinates within the logical viewport (what the game see) are transformed into the normalized coord space and then further transformed to the physical viewport then you'll get the 2nd case.

Reply 2162 of 3949, by BuckoA51

User metadata
Rank Member
Rank
Member

Fighting Force working properly. Checking the options i guess this is another crappy Playstation emulation.

I can get Fighting Force to start, which is an improvement, but only by using the FFORCE_NOVID.EXE that was floating around this site that disables all the videos, and then music won't work either. I guess given that the port really has no enhancements compared to the PS1 version it's better to emulate it anyway.

Couple more games I tried:-

Crazy Taxi (DirectX 8 I believe) - Works very well, looks nice in 1080p 1440x1080. Also fixes an incompatibility with the game and Nvidias recording/streaming tools.

Future Cop LAPD - Works but videos don't play, all you hear is sound.

This is in Windows 10 x64.

play-old-pc-games.com

Reply 2165 of 3949, by MaLDo

User metadata
Rank Newbie
Rank
Newbie
stranno wrote:

Fighting Force working properly. Checking the options i guess this is another crappy Playstation emulation. There are options for Z-Buffer and Texture Perspective Correction. Features that PSX lacked of and quite stupid options for a PC native game.

Just tested and some textures have problems. I don't know if it's fixable but your screenshots show the same problem.

This is my game today, no metal gate texture

fforce2016-06-0922-04izs2a.jpg

and this is how it was

oldg5sqk.jpg

Reply 2167 of 3949, by Bakhtosh

User metadata
Rank Newbie
Rank
Newbie
Dege wrote:
Bakhtosh wrote:

Mystery of the Druids:
But it should be possible (that doesn't mean that it has to be one of your goals ) to make the special edition (updated with the Vista patch) work. The special edition (updated with the Vista patch) already kind of works with Windows 10 64 Bit.

The patched special edition should have compared with the unpatched one to see what the difference is. If it's only a little hack then the same could be applied to the original version. But, if the patch contains rewritten and recompiled complete parts of the executable then it's impossible.

Something I can do without having decompiling/deassembling skills? I can only imagine a checksum- or Bit-comparison of the unpatched/patched exe. But that won't really tell how far the differences are. Only thing I know is that the game uses an engine named cougar engine and that it was updated from the original version to the special edition to be able to play the new voice over sound files. But I don't know if the engine update also included some changes in the graphics part. And I have no idea what was updated with the Vista patch for the special edition.

Dege wrote:
Bakhtosh wrote:

With The longest Journey the problem with the partially rendered character models and the flickering menue buttons is gone with 2.52. The "cursor leaves some white pixel behind if you move the mouse" with "Antialiasing (MSAA)" set to "2x", "4x" or "8x" in dgVoodoo (see here) is still present.

It's a problem that cannot be generally "fixed". The game renders the cursor at pixel centers but if MSAA is forced then one pixel contains more than one samples, and some of them are outside of the rendered triangles (at the edge) so only half-pixels are affected there, that's why garbage is left behind.

Ok - I thought it would be possible since if I don't use dgVoodoo and force AA using the AMD Crimson driver settings (Game / global settings / AA-mode: override application settings, AA-method: Multi-Spampling, AA-filter: Edge-detect, AA-level: 8x) Antialiasing works without having the cursor leaving white pixel behind. But if I understand it correctly dgvoodoo uses a different AA-method/filter and with this method/filter the pixels can't be avoided due to the way the game renders the cursor, right?
So with an AMD card I can either play without dgvoodoo and forced AA by the AMD drivers (forcing doesn't work with NVidia cards) but with some black message boxes. Or with dgvoodo without any black message boxes but also without AA. Well - could be worse. 😁

Reply 2168 of 3949, by ZellSF

User metadata
Rank l33t
Rank
l33t
Bakhtosh wrote:

Ok - I thought it would be possible since if I don't use dgVoodoo and force AA using the AMD Crimson driver settings (Game / global settings / AA-mode: override application settings, AA-method: Multi-Spampling, AA-filter: Edge-detect, AA-level: 8x) Antialiasing works without having the cursor leaving white pixel behind. But if I understand it correctly dgvoodoo uses a different AA-method/filter and with this method/filter the pixels can't be avoided due to the way the game renders the cursor, right?
So with an AMD card I can either play without dgvoodoo and forced AA by the AMD drivers (forcing doesn't work with NVidia cards) but with some black message boxes. Or with dgvoodo without any black message boxes but also without AA. Well - could be worse. 😁

Forcing AA is possible with Nvidia cards, see attached screenshot from my 670. You can supposedly (I haven't tried it yet) fix the message boxes with DxPrimaryEmulation shim with -DisableMaxWindowedMode argument. Obviously, that does not apply to dgVoodoo2.

If you want to run it with dgVoodoo2, at the moment there's no reason to other than allowing you to use post-processing injectors like ReShade... Which you can also use for post-process antaliasing. That will obviously impact text readability, but it does have advantages like smoothing out some 2D art that's aliased for some reason. See the jaggies under the rock April is standing on on the antialiased screenshot? Those would be fixed with post-processing AA.

Attachments

Reply 2169 of 3949, by stranno

User metadata
Rank Member
Rank
Member
Dege wrote:

I've started a new page on my site, listing my testing results and general notes on individual games (only DX games and only rendering aspects):

http://dege.fw.hu/Gallery/DXGamesGallery.html

It's a short list now, but I'm continuously updating it.

I would like to see a better background and font but this kind of list, like the nGlide one, is pretty useful with patches and tricks 😀

Colin McRae Rally looks great, its a shame there is no widescreen hack available, while the PSX version supported widescreen.. (anamorphic of course)

Reply 2170 of 3949, by Bakhtosh

User metadata
Rank Newbie
Rank
Newbie
ZellSF wrote:
Bakhtosh wrote:

Ok - I thought it would be possible since if I don't use dgVoodoo and force AA using the AMD Crimson driver settings (Game / global settings / AA-mode: override application settings, AA-method: Multi-Spampling, AA-filter: Edge-detect, AA-level: 8x) Antialiasing works without having the cursor leaving white pixel behind. But if I understand it correctly dgvoodoo uses a different AA-method/filter and with this method/filter the pixels can't be avoided due to the way the game renders the cursor, right?
So with an AMD card I can either play without dgvoodoo and forced AA by the AMD drivers (forcing doesn't work with NVidia cards) but with some black message boxes. Or with dgvoodo without any black message boxes but also without AA. Well - could be worse. 😁

Forcing AA is possible with Nvidia cards, see attached screenshot from my 670. You can supposedly (I haven't tried it yet) fix the message boxes with DxPrimaryEmulation shim with -DisableMaxWindowedMode argument.

First I want to thank you for the -DisableMaxWindowedMode tip (never heard about the ACT before). It works great (no black message boxes anymore). And it also solved the problem that with AMD cards AA forced by driver settings only worked in the window mode (you had to fake a fullscreen mode using a desktop resolution of 640x480 and a borderless game window). Now it works in fullscreen mode. I've read quite a few threads about TLJ (black boxes problem, AA, etc.) and this was never mentioned.

With NVidia I should have been more precise. Yes forcing AA with the driver settings does work. But with many NVidia card/driver combinations the result are problems with characters are shown in front of instead of behind certain objects and some objects are missing completely. I've read that some people don't have these problems, but most people have these problems (including me using a GTX 260) even if they use exactly the same driver settings like the people where it does work. I've attached two screenshots with forced AA via driver settings. One AMD Screenshot and one NVidia Screenshot where April is displayed in front of a sack instead behind of it and where an axe and a lamp are missing.

Edit:
I noticed that this is getting off topic - since it's about get the game running better without dgVoodoo instead of with it. But if you have another fine tip about how to get rid of the NVidia AA problem (like the one about the black message boxes) I would be glad if you send me a private message. 😀

Attachments

  • Example1_NVidia.png
    Filename
    Example1_NVidia.png
    File size
    308.49 KiB
    Views
    3752 views
    File comment
    NVidia AA Example
    File license
    Fair use/fair dealing exception
  • Example1_AMD.png
    Filename
    Example1_AMD.png
    File size
    321.59 KiB
    Views
    3752 views
    File comment
    AMD AA Example
    File license
    Fair use/fair dealing exception

Reply 2174 of 3949, by Bakhtosh

User metadata
Rank Newbie
Rank
Newbie

Can't be (at least not with all cards). I already had 341.95 installed. The NVidia AA Example screenshot 3 posts ago (with the missing objects and April displayed in front of instead of behind the sack) was made with 341.95. It's the most current one for the GTX 260. But from time to time drivers for legacy cards are still released. So it's interesting from which driver version number AA works correctly with The longest Journey.

Edit
I have to correct myself. I can get forced FXAA to work without the described problems if I set bool_IsDoubleBuffer=1 in the preferences.ini (turn of double buffer). But as soon as I force MSAA or SSAA I get the described problems (no matter if bool_IsDoubleBuffer=1 or bool_IsDoubleBuffer=1 or if FXAA is on or off additionally).

Since with FXAA the text is also antialiased it looks a bit blurred (whole graphics looks slightly blurred). Also FXAA doesn't look as good on 3D models as MSAA 8x f.e. So I would prefer MSAA 8x. But even with FXAA the game looks better than with no AA at all.

So thanks you ZellSF & lowenz for your tips and having me tinkering with the game/driver settings again. 😀

Reply 2175 of 3949, by ZellSF

User metadata
Rank l33t
Rank
l33t

If you're using dgVoodoo2, you can also use ReShade to force different kinds of post-processing AA, I made a comparison here. If you really want to waste a lot of time getting a game running well rather than just playing it, ReShade allows you to only apply post-processing to certain parts of the screen (so you could make the bottom text bar not use antialiasing)... And of course you can always just increase the text size in the game's files to minimize the damage post-process AA does. So nice to have a new enough Nvidia card not to have to bother with any of that 😜

No clue which Nvidia driver version you need, but yeah, newer than available for a 260. I'm at 362.00, because Nvidia keeps breaking stuff in newer driver versions.

Starship Troopers works nice with dgVoodoo2. Resolution forcing and antialiasing no problem. Aspect ratio forcing results in mouse issues though.

Starship Trooper does present a problem with resolution forcing that Populous 3 also has: the game switches between 4:3 menus and FMVs and 16:9 gameplay.

If I force resolution to 1920x1440, menus and FMVs will look ok, but unless the game is set to 4:3, gameplay graphics will stretch and have heavily distorted aspect ratio.
If I force resolution to 2560x1440, gameplay works great in fantastic widescreen, but menus and FMVs will stretch heavily distorting aspect ratio.

Not sure I see an easy solution, but have you though about that problem?

Reply 2176 of 3949, by ZellSF

User metadata
Rank l33t
Rank
l33t

Conquest: Frontier Wars works. Resolution forcing too (but no point).

The Fifth Element sort of works. Focus issue when playing Indeo FMVs, might lead to crash. Not sure if resolution forcing works, but with dgVoodoo2 you can select any resolution ingame. With Windows ddraw it just shows 0x0 and is locked to 640x480 rendering resolution. Seems like a really bad PSX to PC port at any rate.

Glide version of The Fifth Element seems fine. The watermark and splash screen are not showing though and only reason I know it's using dgVoodoo2 is because it's affected by the Glide forced resolution setting.

Reply 2177 of 3949, by UCyborg

User metadata
Rank Member
Rank
Member

Few pages back, someone mentioned Drakan Demo crashing when entering graphics menu. I can confirm this occurs when the game tries to populate resolution list to display in that menu and it apparently uses fixed size buffer on the stack, overwriting return pointer and making it crash when there's too much resolutions available. I fixed this in full game (AiO Patch) by changing the code there to write to a new empty section I made in the Dragon.rfl (which is a DLL file) instead, which has plenty of space.

Arthur Schopenhauer wrote:

A man can be himself only so long as he is alone; and if he does not love solitude, he will not love freedom; for it is only when he is alone that he is really free.

Reply 2178 of 3949, by Dege

User metadata
Rank l33t
Rank
l33t
ZellSF wrote:
Starship Trooper does present a problem with resolution forcing that Populous 3 also has: the game switches between 4:3 menus an […]
Show full quote

Starship Trooper does present a problem with resolution forcing that Populous 3 also has: the game switches between 4:3 menus and FMVs and 16:9 gameplay.

If I force resolution to 1920x1440, menus and FMVs will look ok, but unless the game is set to 4:3, gameplay graphics will stretch and have heavily distorted aspect ratio.
If I force resolution to 2560x1440, gameplay works great in fantastic widescreen, but menus and FMVs will stretch heavily distorting aspect ratio.

Not sure I see an easy solution, but have you though about that problem?

Yes. Recently I tried some stuffs on a 4K monitor and encountered a little annoyance: what resolution should I force to 1) max it out and 2) keep the 4:3 aspect ratio. I had to calculate and enter it manually. Not to mention varying in-game aspect ratios...

My first approach is introducing a new type for resolution forcing. We now have 'unforced' and a bunch of other automatically offered ones. I'd like to add a new "max" one or sg like that (didn't yet find a cool name) that would automatically choose a maximum resolution for the given scaling mode/app resolution pair. So, its effect would be the same as of 'Unforced' for scaling modes 'unspecified', 'centered' and 'stretched' (res coming from the app is used) but it'd calculate the max available resolution based on the native one for the others ('Stretch, keep AR', 'Stretch, 4:3 AR').