PDA

View Full Version : OGL Improvements in Modeler are Amazing! 11.6.3 -> 2018



tcoursey
01-02-2018, 07:52 AM
I come from 11.6.3 and in 2018 my modeler is flying fast compared to moving models around in 11.6.3 (I would say prob 5-10frames a second to over a hundred) I use two titan cards for Octane, looks like they might be used in OGL modeler too now *shrug*. Not sure how much of this came in previous versions but if your in the 11 series jump off to 2018 just for this alone! Wow...makes me wish modeler had even more progress in tools and such, looking forward to more progress LW3DG...great work!

tcoursey
01-02-2018, 08:03 AM
Yep looks like in 2015 I'm getting improvements as well, but not nearly as much. I would say in rough numbers 11.6.3 -> 2015 maybe 3-4X faster and now 2015 -> 2018 probably another 2-3X faster. Very nice welcome for a reasonable upgrade price. Wish they would have made more point of this, people think Modeler is not seeing any love (which on the surface is probably quite true) but behind the scenes there are some things being done, hopefully the flood gates open up soon!

Snosrap
01-02-2018, 08:10 AM
Well Modeler did need to be made to handle the new .lwo specification, so something could have changed. Modeler is still a clunker though. :) Just try to move a poly selection in a dense model.

Lewis
01-02-2018, 08:15 AM
I come from 11.6.3 and in 2018 my modeler is flying fast compared to moving models around in 11.6.3 (I would say prob 5-10frames a second to over a hundred) I use two titan cards for Octane, looks like they might be used in OGL modeler too now *shrug*. Not sure how much of this came in previous versions but if your in the 11 series jump off to 2018 just for this alone! Wow...makes me wish modeler had even more progress in tools and such, looking forward to more progress LW3DG...great work!

I'm happy that you find it such magnitude faster but i'm 100% sure you have "wrong" setting sin 11.6 so that is where this difference comes from 'coz mesh engine in modeler is no updated ;).
Last modeler speedup in OGL/performance (just for tumbling) was in LW 9.2 when they added GPU buffering (VBO) so just go to Display options in modeler/GL Tab and then set all same option in 11.6 as they are now in 2018 and you will see that there is no such or any difference in speed ;).

and BTW Fastest speed in modeler is with MultiTexture Shaders, VBO On and Transparency OFF.

tcoursey
01-02-2018, 08:15 AM
Well Modeler did need to be made to handle the new .lwo specification, so something could have changed. Modeler is still a clunker though. :) Just try to move a poly selection in a dense model.

For sure...looks like moving polys around is still slow but zooming and panning for me is amazing compared to 11.6.3 and faster than 2015. No doubt that modeler still needs lot's of love! But this should be a sign of some behind the scenes work...

tcoursey
01-02-2018, 08:20 AM
I'm happy that you find it such magnitude faster but i'm 100% sure you have "wrong" setting sin 11.6 so that is where this difference comes from 'coz mesh engine in modeler is no updated ;).
Last modeler speedup in OGL/performance (just for tumbling) was in LW 9.2 when they added GPU buffering (VBO) so just go to Display options in modeler/GL Tab and then set all same option in 11.6 as they are now in 2018 and you will see that there is no such or any difference in speed ;).

and BTW Fastest speed in modeler is with MultiTexture Shaders, VBO On and Transparency OFF.

Well I'm not sure what's going on then...I just changed to VBO in 11.6.3 which did speed things up but not at all the same as I see in 2018. I'd still say there is a much smoother response in '18 than I'm seeing in 11.6.3 *shrug* Probably 20fps to 60+ in '18.

I am disappointed some now that I see this, but still some difference is better than none for me. Even if it's some weird anomaly with my drivers or something. :)

Lewis
01-02-2018, 08:22 AM
It's simple, Older LWM versions had "wrong" settings as default so it was slower unless you adjusted settings later on, i think they changed it in 2015 too so defaults are faster. I'm heavily editing all my options/settings that i don't even remember what was the default mode ;).

tcoursey
01-02-2018, 08:25 AM
It's simple, Older LWM versions had "wrong" settings as default so it was slower unless you adjusted settings later on, i think they changed it in 2015 too so defaults are faster. I'm heavily editing all my options/settings that i don't even remember what was the default mode ;).

Gotcha, and maybe some of those settings are not public facing? Anyway, I've changed everything I can find to match and I still see a "difference" in '18 so for me it's going to be nice. Thanks for bringing me down from the clouds and being realistic about the modeler changes... :)

next_n00b
01-02-2018, 01:00 PM
I have compared LW2015 vs LW2018 on the same workstation with old Quadro FX 3800 and new Radeon Pro WX 7100 in MODELER. Both settings are with Buffered(VBO). LW2018 (tested with 1,5kk poly model) with Radeon is faster like 10 times and not much slower with Quadro. It is hard to say exactly how much, but things just fly! Flat Shade and Sketch aren't that fast.

tcoursey
01-02-2018, 01:04 PM
I have compared LW2015 vs LW2018 on the same workstation with old Quadro FX 3800 and new Radeon Pro WX 7100 in MODELER. Both settings are with Buffered(VBO). LW2018 (tested with 1,5kk poly model) with Radeon is faster like 10 times and not much slower with Quadro. It is hard to say exactly how much, but things just fly! Flat Shade and Sketch aren't that fast.

Glad to hear my thoughts are validated at least by one other. I def feel/see a difference, I was just hoping it wasn't a delusional effect! lol.

Again, this is mainly in spinning and panning around, the actual moving of geometry during commands is not any different that I can perceive.

next_n00b
01-02-2018, 01:10 PM
I have an old Xeon e5-2670 with not so good singlecore performance, I am pretty sure it would be 2x faster on modern i7 8700.