pervokpetr

I'm sending new update, some members will receive update tomorrow

- RGB sliders and displacement check fixed - have no clue why it was broken

- IBL Mask with MSAA fixed

- added new algorithm for anisotropic surfaces

- attempt to fix VR

Anisotropic

I found new way to implement anisotropic lighting (some kind). To keep previous version I created param in custom_shading.fx - ANISOTROPY_MODE. If this parameter will be set to zero, then current technique will be used, if you will set this param to 1, then will be used new algorithm - it produce more interesting result. Once again amount of anisotropy can be controled by texture in stage4/blue channel. This is amount so it should be constant for material and not depend from normals, albedo and so on.

VR

My headset not supported anymore, so I cant test it completelly right now, but I tried correct whole rendering for both eyes surfaces and I'm think that result are correct, but then both surfaces should be passed in correspond services (oculus/openvr) and final result depends from it - and I cant test it, my headset do not show final result.

- trick for fullscreen

According to koyot1987 this trick seems works on win10 and for me it works well on win7, to run game in like fullscreen mode:
1. set game to fullscreen resolution but windowed mode
2. set iDelayedInitialization to 1
3. set iFroceBorderlessWindow to 1
with this game will start in window, but will hide border and title, it will looks like fullscreen but still will be in window mode, so there is will not be any problem with alt+tab. But, according to koyot1987 screen goes a bit larger than desktop resolution on win10, so some clippin occure, on win7 it works flawlessly. In this case I can advice try create custom resolution in videocard settings and try run with it.

Another problem.
One of members told me about weird hairworks bug. Hairworks model appear without intermediate hairs, looks like only hairguides, and without intermediate points on hairline, like strong line between two points:
picture:  www79.zippyshare.com/v/jCI32L0M/file.html
It's first time when I see something like this, model placed correctly but hair simulation completelly broken, so I'm ask tell me if you faced with such bug

 I'm going to send new update. I set delay between each sending to try avoid gmail blocking (no more than 300 per hour) - but have now clue will it help or not.

   According to previous post I tried reduce game starting time. In last days no matter what I tried, no matter what I did I could not pass trough 29~30 sec for shader compilation. I reviewed all shaders that I wrote since begining of hook5 development, almost all effects/post effects was separated from each other and reorganized, now main pack contain only object rendering, lighting and shadow calculations, but as I said I cant reduce compilation for main pack without serious changes in hook code - I'm not ready for this right now, I even have no idea for this right now, maybe someday in future.
   Also I tried optimize every shader that was moved to another dll and fixed several errors - so you can notice some performance gain.

   There is some changes in archive, so be sure that you will overwrite all hook files when will place it in binaries. I added function to settings menu that will allow you load settings, so you can rename your "main11.fx" to "anything.fx", place new one and then load your settings from your "anything.fx" - I know I had to do this long time ago  http://modsgarden.cc/Smileys/clabicons/nah.gif.
   
   I do this update because significant changes in shaders and effects rendering sequence was made - I did not faced with some bugs, but maybe you will find such, so will wait your reports.


EDIT (next day):
1. Trick with gmail did not worked - I again was blocked on 24 hours
2. For those who had "black" screen variations or crash with iDelayedInitialization = 0 -> try install files again and be sure that you OVERWRITED ALL FILES, NOT ONLY d3d8.dll
3. For those who had light shadows problem - try disable iUseSeparateShadows
4. For those who had crash with iDelayedInitialization = 1 send me mainfx_errors.txt
5. Hook selectors already fixed
6. @Kraegar - show me comparsion for refractions - I not sure I understood problem.

I'm start sending new update, some members will recieve update tomorrow - gmail will block me on 24 hours:
Fix from 26.05.2020 - sending delayed by one day, due to couple fixes - I hope that 27.05.2020, rest of users will receive files

- fixed parenting with GUI and saving for parenting
- fixed script attaching with GUI and saving for attached scripts, parenting do not affect attached scripts
- fixed glowing edges, was caused by experiments with bent normals (another failed attempt)
- added "as_backround" property to sky section of level

As background
With this option enabled, hook will expect ordinal 2D texture in skybox texture property instead of cubemap texture, texture will be always stretched to whole screen and not depend from camera orientation - this will allow you load any texture as background (possible using some chromakey texture) will ignore any lighting, but will allow you adjust it with skybox parameters - brightness, curve, saturation. It was request - I decided to did it because it can be usefull.

    I tried....
    Since new year I start spent more time, than usually, to play game, and less to code something, I tried understand what's wrong and what can be improved. Due to it I did several request from various members, because I found that it really required. But also I found some flaws and some questions appear in my head. Some of this questions moved me back to PBR rendering, I found that I have some doubts about reflections and hireflective surfaces - Im think that reflections to faster fades with decreasing glossiness, I found that AO techinique produce not pleasant results with hireflective surfaces, I got some questions about gamma conversion (especially after Humus article where he says "developers should stop use power of two - ? ! ? have no clue why he says so, because it's used to convert from gamma space to linear space). I got some questions about ToneMapping - H5MConverter and hook with using the same PBR and lighting technology produce TOO different rendering results and difference almost only in ToneMapping (tone mapping take from Uncharted4 article).
    In my opinion hook close to it ending and should be finalized - i.e polished and optimized, I really want to improve redering quality and performance, I tried some things and found that I'm literally trying avoid serious changes, because of...
    Because of two things. First one is a your reaction - when I do any significant change in hook - I got some members who start "O, my levels looks different - why I should redo my rooms", "Skins brighter", "Skins darker", "Shadows blured", "Shadows not blured", "IT"S COMPLICATED", "WHO NEED THIS" and so on, so on, so on. Now I'm declare -
                                                                               

I DON'T CARE
    I will redo it again and again - until I WILL BE SURE that every f...g equation are correct and I WILL BE SATISFIED with rendering results - quality and performance. If you don't like it, if it anoynig you - you know what to do.
    Second reason more serious - shader recompilation time - any changes in shader code require to shaders recompilation, for me it take from 40 to 45 seconds. So I change something, do recompilation and look at monitor almost minute - after couple hours of such work I'm ready jum to the window with scream "I done with this shit" - it's literally hell.
    To resolve this problem I decided to separate shaders to several independed modules with reducing shaders complexity, I want move all post effect to one module and all debug shaders to another one. I hope that it help me reduce loading speed for game launch (not sure about this, because amount of shaders will remain same, only complexity wil be changed), but mostly I hope reduce recompilation time, because I will recompile only some part of shders, not all of  it like right now - this should free my hand for more hard experiments and research.
   
    I played with some hook objects and got some doubts about reflections and AO, I tried use hairworks for pubic hairs and found that something wrong with skinning trasnforms (and this is weird because in physx hairs it works correct), I tried optimize some shaders and got bugs like glowing edges and such of kind, I tried bent_normals (again) and again failed (I understand theory of bent normals very well, but always fail with technical solution) - I'm close to finish project, but in the same time it looks like I'm far away from it - it start anoying even me. So my plan for futher steps (declare it to better understanding by myself):
   1. Separate shaders for several independingly recompiled modules.
   2. Check all rendering shaders especially PBR algorithms and match results to UnrealEngine results (ligthing and IBL).
   3. Check all related to gamma conversion and tonemapping.
   4. Implelent Bent Normals (highly required for IBL shading)
   5. Resolve weird things with hairworks to make it possible use for pubic hairs and beards
   6. Resolve rendering for hairworks - I'm still think that there is something wrong
   7. Finally made "Optimization step 2" declared who knows how many times ago
   I should do it before continue or finish project, I will not do any requests untill finish this steps. This steps can change hook lighing, reflections, settings and so on - I'm telling it to let you know it, to you be prepared. Also it will take some time, so don't expect "two weeks" results (but who knows?).

I'm sending new update (some members will receive update tomorrow - gmail blocked me on 24 hours):
- Renaming for hook5 objects in property window
- buttons for assign parent and script in property window
- Binding groups for Customizable skin

Property window:


BindingGroups
Allow to change parameters for several layers of CustomizableSkin - very helpfull for brows, nipples and such things. To declare group/groups you need create section below others with such style:
   

[binding_groups]
   parameters = layers

Example from my skin:

   [binding_groups]
   pos, scale, angle, hsb, ds, ns, disp, gs, lvis, dblend, nblend = SkinHead\BrowsL, SkinHead\BrowsR
   hsb, lvis, gs, dblend = SkinBody\BodyTan, SkinBody\BodyTan2, SkinGenit\GenitTan

As you can see first group will affect more parameters then second one - when some changes will be used on one of listed layers, all listed layers will be matched by each listed parameter. Mirroring parameters and textures not affected.

Possible binding groups parameters and their meanings:
pos - position of layer
scale - scale of layer
angle - angle of layer
lvis - layer visibility
hsb - diffuse color hsb modificator
dblend - layer diffuse map blend mode
ds - diffuse strength
nblend - layer normal map blend mode
ns - layer normals strength
disp - layer displacement strength
gs - layer glossiness scale
 

WARNING - DISPLACEMENT CHANGES

It should be ready in next couple days and I'm think that unlike prevs updates I should warning you before it comes, in general due to changes in displacement calculations. Due to last tests and infos (see previous post) I changed zero level of displacement to what it should be i.e. 0.5, so after new update you will have to change/adjust your displacement maps, in other case you will get ugly results with your models. Long time I can't relax because this question bother me a lot, but now when all programs show me correct middle level - I know thats everything correct, so I removed all adjusting coefficients from displacement code and BASE/ZERO LEVEL now equal to 0.5, I also added ability to check displacement maps by hook, in overlay category you will find option to load and enable displacement map, with this, hook will fill screen with this texture and will fill with red zones that equal 0.5 in respect to some treshold.

Also update will contain:
- SSAO, MXAO, HBAO fixes for MSAA mode (still not catched edge glowing, reported by some member) with/without iUseHalfSSAOSurfaces
- SSLR fix for MSAA mode
- Loading for CustomizableSkin
- Renaming for CustomizableSkin layers and correction for clone operation
- Visibility groups (in thoughts BindingGroups, but probably not in this update)
- Changes in paths definition for skin textures, since lot of member report about problem (for me both versions of this algorithm works just fine)
- Cloth displacement fix - (can't beleive that I forgot include middle level in calculations )
...

Loading CustomizableSkin
Will load skin from current skin folder and set it as active in _skin_definition.txt. You will not be able to load skin from anywhere, only from current skin folder

Visibility Groups
There is no gui to create it (it will require coding up to 3~4 Imgui windows). To create group you need create section in skin file below other sections like this:

  [visibility_groups]
    MyVisGroupName = visible state(true or false), BodyPartName\LayerName, AnotherBodyPartName\AnotherLayerName and so on
Example with stockings in my skin:
   [visibility_groups]
   Stockings = false, SkinBody\Stockings, SkinFeet\StockingsFeet
   HighHilsFix = true, SkinBody\LegsHHFix, SkinFeet\FeetHighHeelsFix

Skin textures
Have no clue how it can not works. But once again: if you use _hook5data inside ActiveMod you don't need provide any path in main11.fx parameter cSharedHookFolder, if you use _hook5data outside game folder you have to provide path to folder that contain _hook5data or (from now) to _hook5data, in my case I'm use _hook5data that not in any game installation and I use cSharedHookFolder parameter to provide path to folder that contain _hook5data:
   C:\Games\SharedHookFiles, from now path like C:\Games\SharedHookFiles\_hook5data also will be correct

Cloth Displacement fix
From now middle level of displacement for cloth will be expected as 0.5(50%, 128 and so on). For photoshop users I advice change gray settings (previous post, hfg2 comment) - bring up color settings window (shift + ctrl + K ) and change next parameter from "dot gain" to "gray gamma" or "sgray" probably depends from PS version:

with this phototshop will not change gray color of your maps. If you use GIMP, then as according to hootie it display correct gray value. Onve again zero(middle) level of your displacement maps should be equal to 50% of gray (128 from 256), so you need adjust brightness of your displacement maps alpha channels to bring it to this value.

mouth_eye_fix below contain two displacement maps that I shared earlier, one for eyelids fix fro my head mod, second for open mouth fix

mouth_eyes_fix.zip

New update was sent (some of members will recieve update tomorrow, gmail blocked me on 24 hours
- implemented Delayed initialization
- added setting for shadow pool
- hook5 gui open dialog now work with one click
- environment lighting applied from open dialog
- fixed tessellation vs mirrors
- fixed normals computations for Customizable skin layers with rotation
- atempt to implement saving for level_definition in packed addon for VX

Delayed initialization
According to previous post another variation of initialization of GAPI resources cause faster game start for win10 users, but crash with fullscreen mode. With option "DelayedInitialization" in main11.fx you can use this opportunity if you not use Fullscreen mode - game should start faster but need test

Shadow pool size/count
Until now amount of shadow maps was hardcoded in hook, it was equal to 16, i.e 16 shadow maps could be calculated and stored due to frame rendering. Sun light took 4 maps, every spot took 1 map and every omni took 6 maps. Exiding this amount cause shadow glitch and flicering. Now you can set desired amount of possible shadow maps (lucky users of strong PC) - "iShadowPoolCount", but note that shadow map/target can use only video memory so large amount of it can cause "out of video memory" issue.

hook5 gui - Open dialog
Tried force imgui recognize double click - without success, I got only triple click working(LOL), then I realized that we had no any other functionality with folder in OPD only open it, so decided make things simple - now it can be opened with one click - tested and liked it more than previous variant with button "open"

Environment ligting
Mostly by request - now environment cubemap will be applied by clicking on it in Open Dialog - so you can view and compare the result, there is will be some delay due to cubemap prefiltering

Tessellation and mirrors
Have no clue when this bug appeared, but as I found it really was - now tessellated body normals calculated correct in mirrored scene

Normals and rotated layer
Accidentally found that normals of layer of CustomizableSkin with some rotation angle are incorrect, the image was rotated correct, but since content of image is not just color, but 3d vector it should be rotated also to preserve similar space for blending between layers - but it was not done, now fixed

Saving for level_definition in packed addon VX
Functionality implemented, but seems to require some correction from avarteas side - right now I passed data to avarteas functions but saving not happens, even datetime of file was not changed - anyway I'm think we will finish this functionality in future

New update was sent:
- Fixed export for layer_masks of CustomizableSkin
- Fixed saving for tracking parameters
- Added BAKE function for CustomizableSkin

Regarding bake
First - diffuse maps will be saved as it come to shader stage i.e. flipped by verticaly and secod - I used unpacked dds files, so result files will be huge in size, but you can resave it in any format you want.

Regarding long game start
All modern games use precompiled shaders that only loaded when you start game, I use compilation on fly, because always mess with shaders so all this ~40sec wasted for shader compilation. It becomes a problem for win10 users where by unknow reason, win10 or game reject initialized GAPI objects and start initialization again, so it can waste 3 and more times longer. I tried change initialization, first version did not works (tested by hunter), second variant reduced loading time to normal 1 step initialization (40sec), but as bBird said this variant had crash on start in fullscreen mode. I tried figureout whats can be wrong, but could not find reason (I'm win7 user and did not have such problem) so I had to revert back old initialization code. I have no clue what's wrong on win10, I had one small idea to try some thing - and I will in next days - BUT I PRETTY SURE that this will not help with problem.

Regarding Customizable Skin not save correct paths - tested and for me it works as intended all paths saved correctly, so I need more info about this problem. One idea - maybe you have external _hook5folder, but did not provided correct path to it in main11.fx?

New update was sent (some members will recieve update tomorrow - gmail blocked me on 24 hours)
- fixed swap speed for diffuse override
- added experimental function to dump skin textures

I added two buttons on CustomizableSkin window - "BAKE" and "EXPORT". Bake not working for now, I'm faced with some problems, it should save combined textures into skin folder. Export works - it create "Export" folder inside skin folder and copy all parts and layer textures with preserving folder structure - this should help a lot with skin sharing. This functionality require testing, I have only couple skins, so can't test all cases.

I'm started sending of new update - it contain fixes for issues reported earlier, mostly related to update from 18.12.2019 - Fixed loading problems for meshes and textures, fixed sequencer room (or whatever) I'm still not realized what is this.

Some members will recieve update tomorrow, gmial will block me on 24 hours as always.

KNOWN ISSUE FOR VX VERSION - HEAD TEXTURE NOT PROPERLY UPDATED FOR CUSTOMIZABLE SKIN - will be researched in future, right now use ALT+R to get proper result, in VX this operation performed faster, in 7.5 working as it should

New update was sent (some members will receive update tomorow, gmail blocked me on 24 hours):
- added possibility load hook files (passes, levdefs, h5m's, textures) from packed addons for VX version of game
- added separate parameter for level_definition tracking iLevelDefinitionTrackingEnable in main11.fx
- added separate parameter for render settings tracking iSettingsTrackingEnable in main11.fx
- fixed mirrors vs resolution scale bug
- fixed UI transparent element bug
- fixed parenting for hook fire object
- redesigned a bit hook5 objects creation algorithm - will automatically switch selection mode to object type, to user can see what he created
- redesigned loading procedure for h5m model file - should be faster

KNOWN ISSUE FOR VX VERSION - HEAD TEXTURE NOT PROPERLY UPDATED FOR CUSTOMIZABLE SKIN - will be researched in future, right now use ALT+R to get proper result, in VX this operation performed faster, in 7.5 working as it should

Packed addons/archives of VX version
This functionality require updates for both sides - hook5 and some VX dll's. I'm not sure that required update for VX was done already. Perhaps I tested not published special test update (0.93d), so it will work only after such update. For me it works fine, except issue with head texture, I loaded successfuly every object of hook5 - textures, pass files, level_definition, models in h5m and so on. Tried it for rooms and clothes, both works fine. So at least in VX I will not have anymore "rooms without textures and hook5 effects" 

New update was sent, some of members will recieve update tomorow (gmail blocked me on 24 hours):
- fixed problem with lighting/shadowing reported earlier
- found problem with crash on switching hairworks
- possible fix for crash when changing "cull_mode"
- could not catch crash on alphablended clothes
- added saving for parameter "select mode" in main11.fx - it will store last used mode and load it on game launch

Hairworks crash
As Siberia suggested the problem was in other place, not in hook code. I found problem in my replacers...several months I checked the hook code trying understand where bug and only after I disable all related to "pass" files I understood that hook code has no relation to problem. I redid hairworks replacers and could not reproduce crash anymore, so here new replacers (for both 7.5 and vx):

Cull mode
Perhaps I fixed it, at least it did not crashed yet for me, I found one possible place that could lead to wrong memory access. Need test it.

Alt+Tab
Firstly - hook do not allow game use "exclusive full screen mode", hook force game use borderless windowed mode on fullscreen. Second alt+tab do not call any resource recreation (at least on win7), it just swithc between app like with normal app "windows" of Windows. And last - even window size changing that lead to resource recreation performs without errors (at least for me on win7). So I assume that problem related to win10 or VX version, as for code I could not find any place of it that can lead to this problem. Will check this issue with VX later.
 

New update was sent(some of members will receive update tomorrow, gmail blocked me on 24 hours):
- Added hairworks smaa
- Fixed displacement shadowing bug
- Fixed parenting for lights, mirrors, fires
- Fixed parent transform bug when used hook gui

Hairworks SMAA.
SMAA applied only for hairworks - iHairworksMSAA , can be switched dynamicaly in runtime - use rendering settings window. iUseSMAA shouls be 0 to use this option. This will affect performance in depends from distance to hair and it density, but not so much as full MSAA rendering. Good compromise between non-MSAA rendering and MSAA rendering.

Displacement/shadowing bug.
Accidentally found that some of parameters for displacement was passed incorrect at shadow rendering stage, this caused to some lighting flaws at general rendering stage, now after fixing it works as intended - got correct shadows even for huge displacement.

Still not found reason for crash when Alt+Tab used, I only removed/collapsed/reduced various parts of hook without adding something when did optimization, so can't understand/catch problem - will search more further. The problem that for me alt+tab works without problem - so maybe bug related to hook5 and windows10, a bit hard to catch problem.

New update was sent, some of members will receive update tomorrow. I did a bit more optimizations, removed another couple render targets and change some shader code - this is for non msaa rendering. For msaa rendering I removed one msaa render target, succeded with avoiding several resolve operations, so it should work a bit faster, also resolved some problem with msaa rendering, but I'm not happy with results anyway. As many in net told MSAA not designed to work with Deferred Rendering, it should be used with forward rendering where all msaa operations performed on hardware level. Now I agree with this people, msaa with deferred works bad - it create very good results on G-Buffer filling (resolved normals, albedo or PBR textures looks very good), but further operations - lighting calculations and post effect became too expensive (performance and amount of code) and produce artifacts that broke antialiasing effect - sucks, another one fail for hook5. With this I will try implement only hairworks MSAA - hope this will improve hairworks rendering, but will keep performnce on acceptable level.
 

New update was sent, some of members will receive update tomotrrow. Optimized main rendering pipeline, removed several 64bit per pixel render targerts and several 32bit per pixel targets, removed some of shaders (old or experimental versions). Optimized shaders: SunLight(Environment lights), Spot lights, Subsurface, AverageHDR, ApplyHDR, shaders for depth calculations. You can see difference with profiler window.

I need to you test it, I need to know about glitches, crashes and so on, because there is one render target that can be removed in my opinion, but after I did it I got game crashes in random moments, so I turn it back, hovever I did not found any reason for crash related to this render target.

MSAA not working for now, after optimization I will adjust MSAA rendering.

New update was sent(some members will receive update tomorrow, gmail blocked me on 24 hours). Update contain MSAA implementation, I assume this update as EXPERIMENTAL. There is some flaws, I still had some questions, with using MSAA I got significant fps loss. In main11.fx serach new parameter iUseMSAA section [global_constants] - 0 mean no msaa, 1 mean MSAAx2, 2 mean MSAAx4, 3 mean MSAAx8, you can switch MSAA in runtime change value in main11.fx and hit recompile shaders.



original: https://pichost.org/images/2019/08/25/MSAAvsSMAA.png

Posted: July 29, 2019

http://modsgarden.cc/index.php?action=dlattach;topic=4184.0;attach=384030;image

I thought that I'll finish it today, but caould not. Anyway - I want ask you to test this not finished version, maybe tomorrow I'll finish it, so will be good to know about possible issues.

As you know by my plan I should work with antialiasing functionality in hook, but after I tried enable it in hook I got just black screen...Find out where problem too hard with current state of hook - too complicated rendering process, too many post effects, too many questions. So I decided shoot two targets - rework converter and try antialiasing in more simple app. I successefully enabled antialiasing in converter, and, after reading some nuances and posts from various people in net, I can say that it's not very complicated in such app like converter - only one (base) render target, no any post processing and plus - ForwardRendering. So it just works. In opposite - hook has lot of render targets different size and format, lot of post processing effects and Deferred Shading, which can create a lot of problems. Directx11 antialiasing completely differs from directx9 antialiasing, the things I see (I'm thing that I understood it correct, but maybe not) tell me that it will not be easy in hook.......Had no clue for now how I should proceed all this MSAA surfaces with DeferredLighting and PostProcess, for now I understood only how to enable it in simple forward rendering app....After I finish converter - I'll try antialiasing in hook, I will disable most of rendering code, trying from first steps trace how it will work... if honestly...I'm scared...or better say nervous....

As for converter - the new one more comfotable, has simple IBL implementation( maybe I'll do full like in hook later), support antialiasing (I know, not so important for converter, but read above) and it do TBN calculation FASTER, so once again - test it and tell me how it's going. Converter do not have saving and batching for now - but it's just because I could not finish it yet....
 

UPDATE:

Converter did not work at all, for all - EXCEPT me. I tried find out reasons, but without success. So I just started new project and moved all code to it. I did it by portions, every time testing how it's works. I want say great thanks to panjabi, norm and pes1972 for helping me with testing. Several hours agao I added last portion of code - and.........everything works fine. If you ask me what was it :

https://www.youtube.com/watch?v=VJfK-QLv27o

This is first time when I faced with such situation. It's the same code, whole code the same...but:
v2_1. did not works for anyone except me, but for me it works just fine, has weight 470 kb
v2_2. this one seems works fine, has weight 910 Kb
 http://modsgarden.cc/Smileys/clabicons/huh.gif
nothing to say more...
I spent last two days for nothing, just copy/past with some not important changes...
I'm still don't understand what caused problem and if problem was so critical, why converter worked on my PC...

Before you will try new converter check readme in archive.

New update was sent (several members will receive update tomorrow - gmail blocked me on 24h)

Added "iUseHalfSSAOTargets" to main11.fx as global constant parameter, 1 or 0. 1 mean enabled, allow calculate SSAO effects on surface twice lower than base game resolution. Such method used in all modern engines by default, because SSAO effect are very performance consuming effect, I did not implemented this before because with my old resolution it took 1.2ms and I decided that this was not a problem, but since I got 2K monitor this effect took 3.5ms with significant fps hit. With this implementation I got 15~20% fps boost, in depends from scene, almost without significant changes in rendering qulity. Note: this parameter should be adjusted before game start.

Added "fResolutionScale" to main11.fx as global constant parameter, allow to perform rendering with different resolution than game resolution. You can set it higher or lower in depends from your tasks. For example to run game in fullsized window I should setup resolution 2560x1440, of course such resolution lead to low performance (in comparison with my old monitor), but now I can set fResolutionscale to 0.8, for example, and hook will render everything with resolution 2048x1152, at final stage result will be scaled to game resolution. Even with parameter 0.9 I could get fps 70~72 on my pc. But of cource this parameter will affect picture quality. Note: this parameter should be adjusted before game start.

Fixed problem with manipulators and resolution - this was caused by unfinished experiments with resolution scaling, I'll update basic version later. About glow in basic version - you need add "glow_intensity" to your pass files. Onve again - lighting calculations was synhronized with extended version of hook, in my case rooms looks identical in both versions.