This is definately an odd one. If you're running NVidia's binary drivers for
XFree86, *and* you've got the antialiasing environment variable set to something
other than "no antialiasing" (in my case, I have __GL_FSAA_MODE="2"), then the
interface gets more and more blurred as you move your mouse.
The URL points to a screenshot on my home server that demonstrates this. If you
mouse over an interface element (such as a button or menu item) it'll redraw it,
leaving you with some clear items (the two buttons in the screenshot) and most
of the rest incredibly blurry. It does this as you move the mouse, and visually
what seems to be happening is that for each pixel you move the mouse, it shifts
the whole interface display up and left one pixel, and blurs it one step. So,
the more you move your mouse, the more blurry it becomes.
If you manually set the AA environment variable to 0, the problem goes away, so
I'm guessing this is a problem with NVidia's antialiasing routines. I'd be
interested to hear if this happens with ATI cards and AA turned on... wait, I
have one at home, if I remember I'll test it out tonight. :)
Steps to Reproduce:
1. Use NVidia opengl drivers
2. Set __GL_FSAA_MODE environment variable to a nonzero value
3. Run blender, and move mouse over interface
Same as what you'd see with __GL_FSAA_MODE=0.
Portage 2.0.48-r1 (default-x86-1.4, gcc-3.2.2, glibc-2.3.1-r4)
System uname: 2.4.20-openmosix-r4 i686 Intel(R) Pentium(R) 4 CPU 1.50GHz
CONFIG_PROTECT="/etc /var/qmail/control /usr/kde/2/share/config
/usr/kde/3/share/config /usr/X11R6/lib/X11/xkb /usr/kde/3.1/share/config
USE="x86 oss apm avi crypt cups encode gif jpeg kde libg++ libwww mikmod mmx
mpeg ncurses nls pdflib png quicktime spell truetype xml2 xmms xv zlib directfb
alsa gdbm berkdb slang readline arts aalib svga tcltk java guile postgres X sdl
gpm tcpd pam ssl perl python esd imlib oggvorbis gnome gtk qt motif opengl mozilla
gphoto2 cdr dga dvd fbcon flash gd gtk2 gtkhtml imap libgda mbox pda plotutils
samba sse tiff usb xinerama -3dnow"
CFLAGS="-O3 -mcpu=pentium4 -mmmx -msse -msse2 -mfpmath=sse -pipe"
CXXFLAGS="-O2 -mcpu=i686 -pipe"
FEATURES="sandbox ccache fixpackages"
Whoops, I just thought of an additional thing to test, and my guess was correct. Blender displays correctly for all __GL_FSAA_MODE values *except* 2. 2 is for NVidia's QUINQUNX (or whatever that silly name is) AA on my Geforce 3 card, so it's definately a problem with that kind of antialiasing.
I'd suggest pointing out the deficiency to NVidia, and closing the bug, since all other AA modes seem to be working fine.
I don't have a nvidia (it's a Centrino) and I have the same problem, so maybe it's only AA related
Is this still a problem with the new drivers?
This bug is 10 months old now so im just wondering if nVidia have fixed the issue, or do we need to mark it upstream and point it to them still?
I have no idea what you meant in comment #2 Andre, could you please clarify.
Since my problem isn't on a NVIDIA, I was wondering if the real issue was with Anti-Aliasing, as was commented on the Description.
So are you saying that the fix mentioned works for you aswell?
setting the env AA to 0 doesn't work for me, but I've read somewhere, don't remember where, that the problem was with the _animated cursors in X_, so I've moved .icons to icons just to test it out, and now with the default xorg cursor, blender works as expected.
Strange, because I *always* had the animated cursors enabled, but was able to make the problem disappear by changing the AA settings with an NVidia card. Mind you, now I don't have access to the machine that I initially reported the problem from, so I can't do any further NVidia testing. Perhaps it's a combined trigger, where you have to have (at least in some cases) antialiasing and animated curors enabled?
Please test this with 6111 which just went into cvs.
Report back on success or failures.
Can we close this ... no activity recently at all
Yeah, I no longer have any machines with NV cards to test this with, and since changing the AA settings seems to fix it, I wouldn't worry overly much.
Closing, i see no reason to leave on test.