Blender 2.32 (BF) on Unix (Mac OS X 10.3.3)

log0n · Post by **log0n** » Fri Apr 16, 2004 10:05 pm

How much ram do you have in your PB? I've got the same Powerbook 17 (1GB) and I just created a quick scene w/ about 120K faces (ico&uvspheres). Viewport interaction is slower, but still quite usable. Also, double check your Power Management settings in the Energy System Preference. Make sure that CPU Performance is set to Highest on the Power Adapter selection (battery, leave at automatic, or better yet, reduced).

One major problem with OpenGL on OSX (10.3.3) is zooming in and out with Option and Control keys. Mouse tracking somehow gets lost and the zoom will extend indefinitely. It's like a weird acceleration issue or something...

JoOngle · Post by **JoOngle** » Sun Apr 18, 2004 3:03 pm

Hi there,

The thing is....we all perceive "acceptable-speed" differently,
I work with 3d on a "Pro-level" meaning that I must deliver
stuff that brings food on the table, for that...I use both
Blender and 3dstudio max (4).

The Powerbook is technically a very STRONG computer
and it's comparable to my previous MyBook (vega pc)
that was a whopping 2.8 Ghz P4 with a Radeon 9000 card
onboard. There I could easily reach 200-300 K polys and still
have about 30 FPS in my viewport...making it workable
in blender...but that was under Windows...

Here...I have a 17" Powerbook G4 with a RADEON 9600 CARD
and theoretically it should burn 9000's top off like a breeze,
but under Mac OS X somehow...blender doesn't

Unreal-tournament & Quake 3 runs WAY faster....using
a LOT more polys in realtime than I'm able to squeeze out
of Blender 3d under mac os x

I just tried the 140.000 vertice/faces test...and it slows
down in blender to 3-5 fps pr. second...that's useable FOR SOME
but not me...I've been doing 3d for ages...and with a
computer like this, I'd expect it to work WAY faster than
my previous Pentium-4 Portable....So something MUST be
wrong.

Take care
/S
http://www.hamsterking.com

cekuhnen · Post by **cekuhnen** » Sun Apr 18, 2004 3:35 pm

i wouldnt blame it on the hardware.

most times it is the software!!!

claas

cekuhnen · Post by **cekuhnen** » Mon Apr 19, 2004 12:26 am

i have the impression that ini this case here we see that drawback of plattform independet solftware!

log0n · Post by **log0n** » Mon Apr 19, 2004 7:54 pm

JoOngle wrote:Hi there,

The thing is....we all perceive "acceptable-speed" differently,
I work with 3d on a "Pro-level" meaning that I must deliver
stuff that brings food on the table, for that...I use both
Blender and 3dstudio max (4).

As do I, but I use primarily Maya (work, a little bit of LW now and then) and C4D (personal). I stopped using Blender around the 2.0 betas because I wasn't interested in the game development and that was also around the time I switched full time to Mac.

The Powerbook is technically a very STRONG computer
and it's comparable to my previous MyBook (vega pc)
that was a whopping 2.8 Ghz P4 with a Radeon 9000 card
onboard. There I could easily reach 200-300 K polys and still
have about 30 FPS in my viewport...making it workable
in blender...but that was under Windows...

Here...I have a 17" Powerbook G4 with a RADEON 9600 CARD
and theoretically it should burn 9000's top off like a breeze,
but under Mac OS X somehow...blender doesn't

Unreal-tournament & Quake 3 runs WAY faster....using
a LOT more polys in realtime than I'm able to squeeze out
of Blender 3d under mac os x

I just tried the 140.000 vertice/faces test...and it slows
down in blender to 3-5 fps pr. second...that's useable FOR SOME
but not me...I've been doing 3d for ages...and with a
computer like this, I'd expect it to work WAY faster than
my previous Pentium-4 Portable....So something MUST be
wrong.

Take care
/S
http://www.hamsterking.com

It's been my experience that Q3/UT (though newer games are changing) demand more out of a 3d cards memory/bandwidth than it's triangle crunching power and aren't good benchmarks to judge 3dapps by. I'm not a gamer, but I was when Q3 first came out so I don't think that's changed all that much (low poly modeling).

I can work with huge scenes in Maya (1.5+M faces, etc) and it's quite usable* (both desktop g5 & pb17) so it's not a limitation of the hardware, more on Blender's part. For it's part though, Blender is still highly usable (IMO) in spite of this slowdown - esp. as it's free.

*we have a different interpretation of useable (my hardware expectations are oldschool - if it gets 3-5FPS for me I'm pleased; hell, it wasn't long ago that 3-5 FPS on a 150K vert scene was something to be ecstatic over (Wildcat 4000))

I guess my only real advice is get more ram

and change your expectations (or learn to code and optimize the heck out of Blender for OSX

- not sure which would be easier

)

wavk · Post by **wavk** » Mon Apr 19, 2004 8:11 pm

I just did a test in LightWave on OSX, I can't say it's much faster. And as for games, Quake III runs smooth on my powerbook (1.25 GHz), but Unreal Tournament doesn't. It's very shocky even with all settings set to low.

I guess Blender can be optimized more, but until then, use layers. That's always been the best way to speed things up.

cekuhnen · Post by **cekuhnen** » Mon Apr 19, 2004 8:44 pm

years ago they used proxy models and in render time the high res objecst. this works in blender as well special with layers and the screnn redraw enhancement and the s-mesh optin for preview and render.

but i think a bit of os x optim. might help as well!

claas

log0n · Post by **log0n** » Mon Apr 19, 2004 9:43 pm

If anyone's interested, a 100x100x100 cube in C4D w/ 100x100x100 segments gets about 5-8FPS guesstimate (wireframe or shaded wireframe) on my PB17. Pretty good IMO, but I haven't used a PC in years so I can't compare it to current PC performance.

Are Radeon 9600 AA lines done in hardware (ie: AA lines on a Quadro)? I don't often use AA but there was no frame loss due to AA lines being on or off. Very kewl!

Does blender support viewport AA?

cekuhnen · Post by **cekuhnen** » Mon Apr 19, 2004 10:28 pm

interesting to see that.

dont think that this AA feature is enabled here but it should normaly work. it did not slow down c4d at all???

claas

log0n · Post by **log0n** » Mon Apr 19, 2004 10:46 pm

nope! I know in the past that AA lines on Geforce (non Quadro) cards were done essentially in software (unless you had the SoftQuadro hacks) and there was a performance pentalty.

1600x1200@32, pretty smooth considering how many vertices are moving aboot..

I'm going to need to start blendering again

wavk · Post by **wavk** » Tue Apr 20, 2004 8:42 am

Well, log0n, a 100x100x100 segment cube only has 100x100x6 polygons if you think about it. So that's ony 60000 polys and that's even worse if your powerbook is already staggering at that amount of vertices.

Nice new word though, guesstimate

log0n · Post by **log0n** » Tue Apr 20, 2004 2:23 pm

That's a good point.. but I'm not sure how C4D handles primitives. They aren't really meshes until you make them editable so I'm not sure if it's saying 6sided cube/100, or cube/100 in X, Y and Z (if that makes sense). I'll look into it later today.

Staggering/5-8fps isn't accurate. It's smooth updating, it's just not fluid updating. But if it is slowing on only 60k, that's interesting. Hrm..

(non gaming) How many polys can a 'modern' PC/3d card get on screen and keep up a 10-15fps framerate?