bigmuscle

Once again - the blur effect is there, those "trails" are actually blur effect rendered on wrong X coordinate. But what I don't understand is that you say that the test build didn't contain it but your screenshots from the test build contains it: http://abload.de/img/unbenannt-1cud8k.jpg http://abload.de/img/unbenannt-1bxiqe.jpg (it's on both of your screenshots) What I need to find is why X coordinate is wrong now (the change from test build is that it took -X before) but Y seems correct.

The trails are actually wrongly positioned blur effect. It was also present in the test build (it was what I commented on your last screen, but your background was too solid so it was not noticeable). The only difference was that it was mirrored - I fixed the bug but it seems that X coordinate is still invalid. Is it always attached to the edge of screen or to the window edge? Could you hold F11, move with window and then send me debug.log + printscreen?

It's bug in your theme. There is no reason to do this. You can have custom atlas even for other themes. Actually, I never said that it changed somehow, I only said that it's not fully possible to respect theme settings. And it really still works as in previous beta except the bitmap is read from theme (custom atlas) so you can see something different only if your theme/atlas contain image different from glow.bmp. This is how it looks here: There is no such option in Win8.1.

I reuploaded fixed version.

yes, if it crashes then minidump is important for me so I can find a cause. edit: it's ok, I was able to reproduce the bug

AERO GLASS for Win8.1 Beta 4 Here is Beta 4 which contains a lot of internal changes, so I hope I didn't break anything. * HKCU atlas loading fix * multi-displays and rotated displays fixes * optimization: do not inflate dirty regions and do not use glass safety zone when rendered region does not contain glass * glow bitmap is read from system theme (or custom atlas if used), however, only horizontal sizing margins are respected due to the fact how Win8.1 deals with window caption (it seems to be a bug in DWM itself); vertical margins are always zero and glow bitmap is stretch over whole caption height * because there is no appropriate theme part to be used for reflection bitmap (aero peek background is not suitable), the reflection bitmap is read from PNG file specified in registry HKCU(or HKLM)\Software\Microsoft\Windows\DWM:CustomThemeReflection (SZ_STRING) * renamed registry settings CustomThemeResource to CustomThemeAtlas You still need to manually download symbols for udwm.dll, dwmcore.dll and dwm.exe else it won't work.

http://glass8.berlios.de/guide.html

hehe, I just have one 128 GB SSD + one HDD and it seems enough to make me mad when I have to work on HDD only at work

I must admit that I don't know how it will work in raid configuration because I have never had such system. My algorithm just takes serial number of HDD/SSD where OS is installed to and if you license key does not match, it takes the 1st HDD/SDD found in the system. If machine ID is different for Win8 and Win8.1 and the rest of configuration is 100% then it's either a bug which should be fixed, or some drive blocks serial number detection on one of the system (this situation should be inspected too to minimize it) Machine ID for Win8.1 Beta is displayed in console window. However, I do not recommend to generate license key unless you are a serious tester because you don't gain nothing (donation+license key does not really turn unstable beta version into full final version). And one thing which I should finally change is renaming license.key to something like donation.key to more reflect the reality - you don't buy any license by donation.

Windows Style Builder says you rectangle and margins for each object.

paul70: not now. we will see after another beta release mirecek1965: when the development phase is finished

it's ok, I've already found a bug in affine transformation.

But look at left on the console - at your desktop background. You will see that the background is blurred in the shape of the whole console. So the position is either mirrored (because the blur touches the console window - the blue rectangle of the MSFN forum is blurred in Firefox) or shifted. Could you make another log with holding F11 (from this fixed build) + also make screenshot when you release F11 (so last log item matches the screenshot)?

I see the blur is here but on invalid position - look at left side of console, but it is hardly noticeable due to your background.

btw, does blur appear on your rotated display?

Very good! And yes, it could be good when screen preview is attached with the topics. I will then link to this topic from my website

It is buffer size which is still 1600x1200 but transformation matrix is: which ensures the rotation + translation of rendered rectangle from [1124.0, 218.0, 2223.0, 1225.0] to [218.0, 897.0, 1225.0, 1200.0]. This is correct (and psst, but I copied this thing directly from Win8.1 function), but I probably forgot something. Does your problem appear always or only under certain condition? (especially order and number of displays)... what if there is only one (but rotated displays)? what if the 1st display is rotated and 2d is not and vice-versa? Does anybody know if there is some possibility to simulate screen rotation in VMWare? EDIT: so I have found where the problem is - rendered region is clipped against wrong coordinates.

If more people are working on custom atlas files, it would be good to create separate topic for it so they don't get lost here and they can be found all together at one place

I finally got time to analyze the log but I could not find any problem. Rendering does not fail and everything seems to be rendered correctly, so I don't understand what could be wrong.

It is StartIsBack.

I don't know what you mean by "CMD came back with everything failed" but my Aero Glass for Win8 contains uninstaller (from Add/Remove programs control panels) which uninstalls everything and deletes the remaining files after restart.

Actually, there is one thing in AeroGlass which can increase CPU usage a bit when playing videos or in fullscreen - in Win7/8, glass safety zone is not used when rendered primitive does not contain glass, I use it always because I did not find a way to detect whether there is glass or not (internal DWM flag does not exist anymore). However, the difference would be hardly noticeable (it can be found in the ETL file above but not under the branch with highest CPU usage - try to find CHwndRenderTarget::RenderDirtyRegion). Btw, I don't know why but loading this ETL says me that my PDB file is unmatched although it matches beta 3 DLL. In VMWare, I noticed a strange thing which is, I guess, the problem of VMWare display driver - when multiple displays are used, the stuff rendered in the second display is very jerky. But it happens even without Aero Glass.

it's ok, it logged what I needed and I think the problem is what I thought Someone here already suggested adding timestamps to logs so I should really do it. But now it looks that settings and theme were loaded at the same time so it is just race condition when it asks for CustomThemeResource. I guess putting the settings into HKLM will work.

This is the verbose debug build: http://glass8.berlios.de/stuff/DWMGlassVerbose.7z NoelC: could you use this build a provide debug log when custom resource is not applied? Rotated display problem: could you use this build a hold F11 while moving some window into non-functional area? In both cases, it starts logging needed data into debug.log so it may be slow as hell during it. But then upload debug.log. It should not be used in other cases!

I never said when it will happen. However, it is weird that you get popup if: