Page 5 of 6

Re: Help us to Optimize PR! (LOD Switching & Cull Distances)

Posted: 2011-06-16 14:19
by Psyrus
[R-DEV]Rudd wrote:thats the vBF2 one mate which we don't use because iirc it wasn't appropriate
Yeah I know but in Rhino's list it had ru_apc_btr90 [#27], yet I can't find that anywhere. Is that a mistake or is my PR missing something? In any case those values can be used if the BTR is ever reskinned and used in a faction :)
====================

For anyone thinking about doing some LOD switch analysis, it's really easy, but somewhat time consuming. I'm now what I would consider 'versed' in the art [having done 41 of them], and I just timed myself for the latest one (LAV25). It took 14:40 from opening the spreadsheet to see what the next one was to hitting "Save" for the edit. That's inclusive of photoshopping, uploading, checking etc. Easier models without issues would be 2/3 that time, and the worst case scenario would be about twice that long for a newcomer.

Get stuck into the Air Vehicles list if you really like PR as much as you say you do, especially if you're one of the people bitching about performance.

====================

Code: Select all

[b][u]ru_tnk_t90[/u][/b]

ObjectTemplate.cullRadiusScale 5.8

GeometryTemplate.setSubGeometryLodDistance 1 0 65
GeometryTemplate.setSubGeometryLodDistance 1 1 85
GeometryTemplate.setSubGeometryLodDistance 1 2 240
GeometryTemplate.setSubGeometryLodDistance 2 0 115
GeometryTemplate.setSubGeometryLodDistance 2 1 225

[b]Issues:[/b]
[list=1]
[*]
[*]
[*]
[*]
[/list]

[b]Coding:[/b] Psyrus
[b]Checked by:[/b] 
[b]Tweaked by:[/b] 
[b]Committed to REPO:[/b] 

Code: Select all

[b][u]suv_vip[/u][/b]

ObjectTemplate.cullRadiusScale [b][color=Red]N/A[/color][/b]

GeometryTemplate.setSubGeometryLodDistance [b][color=Red]N/A[/color][/b]

[b]Issues:[/b]
[list=1]
[*][b][color=red]Cannot find suv_vip in unzipped objects folder. [/color][/b]
[*]
[*]
[*]
[/list]

[b]Coding:[/b] Psyrus
[b]Checked by:[/b] 
[b]Tweaked by:[/b] 
[b]Committed to REPO:[/b] 

Code: Select all

[b][u]us_aav_avenger[/u][/b]

ObjectTemplate.cullRadiusScale 5.6

GeometryTemplate.setSubGeometryLodDistance 1 0 50
GeometryTemplate.setSubGeometryLodDistance 1 1 70
GeometryTemplate.setSubGeometryLodDistance 1 2 90
GeometryTemplate.setSubGeometryLodDistance 2 0 105
GeometryTemplate.setSubGeometryLodDistance 2 1 9001

[b]Issues:[/b]
[list=1]
[*]Geom_1 LOD 3: Final LOD poly count way above limit @ [b][color=Red]2410[/color][/b]
[*]Geom_1 LOD 3: Final LOD poly count way above limit @ [b][color=Red]2208[/color][/b]
[*]Geom_2 LOD 2: LOD 2 & 3 are mixed up, with the final LOD having [b]more[/b] polys than the one preceding it. Have set final LOD @ over 9000m so that it stays on the lower poly version. This geom_2 will have to be redone when it's corrected, if ever. 
[*]Geom_2 LOD 2&3: Wrong part of the texture is applied to the missile canopy compared to LOD0, should be fixed for consistency. 
[/list]

[b]Coding:[/b] Psyrus
[b]Checked by:[/b] 
[b]Tweaked by:[/b] 
[b]Committed to REPO:[/b] 

Code: Select all

[b][u]us_apc_aavp7a1[/u][/b]

ObjectTemplate.cullRadiusScale 8.2

GeometryTemplate.setSubGeometryLodDistance 1 0 20
GeometryTemplate.setSubGeometryLodDistance 1 1 55
GeometryTemplate.setSubGeometryLodDistance 1 2 130 [90 - See issue 1]
GeometryTemplate.setSubGeometryLodDistance 1 3 205
GeometryTemplate.setSubGeometryLodDistance 2 0 25
GeometryTemplate.setSubGeometryLodDistance 2 1 55
GeometryTemplate.setSubGeometryLodDistance 2 2 130
GeometryTemplate.setSubGeometryLodDistance 2 3 175

[b]Issues:[/b]
[list=1]
[*]Geom_1 LOD 4: Removed base of cannon results in floating geometry. Geometry can be extended down (just make polys bigger) to bridge the gap or the turret lowered a bit to close the floating bit. [[url=http://i235.photobucket.com/albums/ee94/psyrus_uraya/LODs/aav7_geom1_lod4_floating_turret.png]See Image[/url]]
[*]Geom_2 LOD 3: RHS of model's texture is a fair bit lighter on LOD 3 than the preceding LODs. This results in the lod switch being pushed out to 130 rather than ~90m (+/- 10m). If you don't think players will notice the texture shift, I suggest 90m for[i] setSubGeometryLodDistance 1 2 XX[/i]
[/list]

[b]Coding:[/b] Psyrus
[b]Checked by:[/b] 
[b]Tweaked by:[/b] 
[b]Committed to REPO:[/b] 

Code: Select all

[b][u]us_apc_lav25[/u][/b]

ObjectTemplate.cullRadiusScale 7.9

GeometryTemplate.setSubGeometryLodDistance 1 0 40
GeometryTemplate.setSubGeometryLodDistance 1 1 105
GeometryTemplate.setSubGeometryLodDistance 1 2 175
GeometryTemplate.setSubGeometryLodDistance 2 0 80
GeometryTemplate.setSubGeometryLodDistance 2 1 160

[b]Issues:[/b]
[list=1]
[*]Geom_2 LOD 3: Turret is floating slightly on LOD3. It makes no difference to the LOD switching, but it is disconnected enough to mention. [[url=http://i235.photobucket.com/albums/ee94/psyrus_uraya/LODs/lav25_geom2_floating_turret.png]See Image[/url]]
[*]Geom_2 LOD 0: Wheel colours are significantly lighter on the first LOD than the subsequent two, which makes the LOD switch rather stark at anything close. [[url=http://i235.photobucket.com/albums/ee94/psyrus_uraya/LODs/lav25_geom2_wheel_colors.png]See Image[/url]]
[*][*]Geom_2 LOD 1: Geometry has been added to the back of the turret in the second LOD, which makes LOD switching early on even harder. [[url=http://i235.photobucket.com/albums/ee94/psyrus_uraya/LODs/lav25_geom2_wheel_colors.png]See Image[/url]]
[*]
[/list]

[b]Coding:[/b] Psyrus
[b]Checked by:[/b] 
[b]Tweaked by:[/b] 
[b]Committed to REPO:[/b] 
Decent steps on average so far. I think attempts are doing quite well :)
Image

Re: Help us to Optimize PR! (LOD Switching & Cull Distances)

Posted: 2011-06-16 18:09
by Psyrus

Code: Select all

[b][u]us_apc_stryker[/u][/b]

ObjectTemplate.cullRadiusScale 8.8

GeometryTemplate.setSubGeometryLodDistance 1 0 50
GeometryTemplate.setSubGeometryLodDistance 1 1 75
GeometryTemplate.setSubGeometryLodDistance 1 2 115
GeometryTemplate.setSubGeometryLodDistance 1 3 230
GeometryTemplate.setSubGeometryLodDistance 2 0 35
GeometryTemplate.setSubGeometryLodDistance 2 1 75
GeometryTemplate.setSubGeometryLodDistance 2 2 145
GeometryTemplate.setSubGeometryLodDistance 2 3 190

[b]Issues:[/b]
[list=1]
[*]Geom_2 LOD 3: Back door of the stryker is removed which leaves a dark patch on the texture showing, resulting in a very stark LOD switch. LOD switching could occur earlier if not for this factor. [[url=http://i235.photobucket.com/albums/ee94/psyrus_uraya/LODs/stryker_geom2_back_darkness.png]See Image[/url]] 
[*]
[*]
[*]
[*]
[*]
[/list]

[b]Coding:[/b] Psyrus
[b]Checked by:[/b] 
[b]Tweaked by:[/b] 
[b]Committed to REPO:[/b] 

Code: Select all

[b][u]us_ifv_m2a2[/u][/b]

ObjectTemplate.cullRadiusScale 4.1

GeometryTemplate.setSubGeometryLodDistance 1 0 85 [40 - See Issue 1]
GeometryTemplate.setSubGeometryLodDistance 1 1 115
GeometryTemplate.setSubGeometryLodDistance 1 2 135
GeometryTemplate.setSubGeometryLodDistance 2 0 70
GeometryTemplate.setSubGeometryLodDistance 2 1 115

[b]Issues:[/b]
[list=1]
[*]Geom_1 LOD 1: Tracks are raised a bit in the first LOD switch which I assume is unintentional and pushes LOD switch out considerably (up from about ~40 +/-5). [[url=http://i235.photobucket.com/albums/ee94/psyrus_uraya/LODs/M2A2_geom1_tracks_raised.png]See Image[/url]]
[*]Geom_1 LOD 3: Final LOD is way above the poly limit @ [b][color=Red]2067 [/color][/b] polys
[*]Geom_2 LOD 2: Final LOD is above the poly limit @ [b][color=Red]1371[/color][/b] polys
[*]
[*]
[*]
[*]
[/list]

[b]Coding:[/b] Psyrus
[b]Checked by:[/b] 
[b]Tweaked by:[/b] 
[b]Committed to REPO:[/b] 

Code: Select all

[b][u]us_tnk_m1a1, us_tnk_m1a1_alt, us_tnk_m1a2, us_tnk_m1a2_alt[/u][/b]

ObjectTemplate.cullRadiusScale 3.9

GeometryTemplate.setSubGeometryLodDistance 1 0 45
GeometryTemplate.setSubGeometryLodDistance 1 1 85
GeometryTemplate.setSubGeometryLodDistance 1 2 100
GeometryTemplate.setSubGeometryLodDistance 2 0 90
GeometryTemplate.setSubGeometryLodDistance 2 1 120

[b]Issues:[/b]
[list=1]
[*]Geom_1 LOD 3: Final LOD is way above the poly limit @ [b][color=Red]2587[/color][/b] polys
[*]Geom_2 LOD 1: Removing the tracks so early in the LOD process forces the first LOD switch to occur far later than it would otherwise have to. Simple tracks being present would have allowed for about half the distance (45).
[*]Geom_2 LOD 2: Final LOD is way above the poly limit @ [b][color=Red]2384[/color][/b] polys
[*]The M1A1 and M1A2 were grouped as the minor cosmetic differences did not change the cull/LOD distances. 
[*]
[*]
[*]
[/list]

[b]Coding:[/b] Psyrus
[b]Checked by:[/b] 
[b]Tweaked by:[/b] 
[b]Committed to REPO:[/b] 

Re: Help us to Optimize PR! (LOD Switching & Cull Distances)

Posted: 2011-06-16 18:39
by Psyrus
Aaaaaaand that's me done.

I might return some day to dabble in some of the aircraft but upon my own little cost benefit analysis, the land and civ vehicles made a lot of sense, as they are prevalent in many maps and used quite often - near infantry.

Aside from transport choppers, the % usage of the respective air vehicles is tiny, and thus at this point in time I need to take a break and let some air-enthusiasts have their hand at optimizing their toys. I'll probably do any transport choppers that are left undone, but like SU25a.. used on all of what... Kashan? :)

36 air vehicles to go. 2 people doing 5 a week each will take less than a month, and constitute only just over an hour per week to each person. Something to think about ;)

Edit: It's truly a pity that someone has to go and check through each of the ones that I've (or others have done). Even if they only take half the time to check/tweak, it'll still be >5 hours of work just for the land vehicles. Good luck I guess to that poor dev :P

Re: Help us to Optimize PR! (LOD Switching & Cull Distances)

Posted: 2011-08-02 22:19
by lucky.BOY
Bumpy time :)

I tried to look at huey, since its the most used air asset i think.

Note: Values in brackets are ignoring the problem with glass (described below).

Code: Select all

[b][u]us_the_uh1n[/u][/b]

ObjectTemplate.cullRadiusScale 4

GeometryTemplate.setSubGeometryLodDistance 1 0 (35)   
GeometryTemplate.setSubGeometryLodDistance 1 1 (55)
GeometryTemplate.setSubGeometryLodDistance 1 2 230
GeometryTemplate.setSubGeometryLodDistance 2 0 50
GeometryTemplate.setSubGeometryLodDistance 2 1 100

[b]Issues:[/b]
[list=1]
[*]Geom_1, LOD_1, LOD_2: seems like these LODs have wrongly placed glass, [url=http://i.imgur.com/ESp4H.jpg]image1[/url], [url=http://i.imgur.com/uCctk.jpg]image2[/url]
[*]Geom_1, LOD_3: Final LOD above limit, at 1075 tris.
[*]Geom_1, LOD_3: LOD 3 is missing tail surfaces, thus making the transition very noticable when looking from above (or from below) and pushing it as far as 230m ([url=http://i.imgur.com/hBgQn.jpg]image1[/url], [url=http://i.imgur.com/vQnUR.jpg]image2[/url]).
[*]Geom_2, LOD_2: two floating bits near rotor, probably left-overs from deleted stabilizator. [url=http://i.imgur.com/CfbVN.jpg]image[/url]
[*]Geom_2, LOD_2: Final LOD above limit, @ 1000 tris.
[*]
[*]
[*]
[/list]

[b]Coding:[/b] lucky.BOY
[b]Checked by:[/b] 
[b]Tweaked by:[/b] 
[b]Committed to REPO:[/b] 


Questions:
How can i "unload" rotors when im working with geom_2 and dont need them?
What causes that red text going on in my Output window? It reads something along "Collmesh for mainrotor and tailrotor of us_ahe_uh1nrockets not found". I accuse it of slowing the editor really badly.

-lucky

Re: Help us to Optimize PR! (LOD Switching & Cull Distances)

Posted: 2011-08-07 21:02
by lucky.BOY
Really nobody? You make me feel lonely... :(

-lucky

Re: Help us to Optimize PR! (LOD Switching & Cull Distances)

Posted: 2011-08-08 09:42
by Rhino
lucky.BOY wrote:How can i "unload" rotors when im working with geom_2 and dont need them?
The only way you can do this is by going into the "Ammunition" mode of the objects editor (button found in to the right of where you select the objects editor, level editor etc, looks like a bullet going though a wall), and then scroll down on the right side in the editor bar to the bottom and you will find an "Explode" button, hit that and your vehicle will blow up as it would ingame and then the rotors etc will be removed.

http://img84.imageshack.us/img84/6178/0 ... 103922.jpg
http://img593.imageshack.us/img593/7502 ... 104014.jpg
http://img713.imageshack.us/img713/336/ ... 104020.jpg
http://img607.imageshack.us/img607/345/ ... 104036.jpg

lucky.BOY wrote:What causes that red text going on in my Output window? It reads something along "Collmesh for mainrotor and tailrotor of us_ahe_uh1nrockets not found". I accuse it of slowing the editor really badly.
Its due to the way we load the col meshes on some cloned choppers so we don't need to load duplicate col meshes etc. The problem is the editor can't seem to work out what we do (while the game can) and it gets all confused and crates a huge amount of lag in the editor...

No simple way to fix it I'm sorry to say but you can fix it by crating the "missing" col meshes its looking for in the correct dir, ie, crate a "us_the_uh1n_mainrotor.collisionMesh" file in the "/objects/vehicles/air/us_ahe_uh1nrockets/us_ahe_uh1nrockets_mainrotor/Meshes/" folder by copying the col mesh found in "\objects\vehicles\Air\us_the_uh1n\us_the_uh1n_mainrotor\Meshes" to that folder, just don't upload this with your work.

Re: Help us to Optimize PR! (LOD Switching & Cull Distances)

Posted: 2011-08-08 18:09
by lucky.BOY
Ok, thanks for info. Ill try to get some more done when i get home from vacation. Just to make sure, you dont need me to upload anything, my outcome are these

Code: Select all

 things i post here, arent they?

-lucky

Re: Help us to Optimize PR! (LOD Switching & Cull Distances)

Posted: 2011-08-11 03:37
by Valleyforge3946
Well, i was told to come here and help out with the chance that i could join the team someday so, i already have some coding expierence. i know basic python, have been editing tweak con files for over 2 years and have been working on my own minimod for over 2 years that started with a modded bf2demo server back when i was in high school. i have read the tutorial's and have practiced a bit so i am confident that i can help out. I have the time and with practice and trial and error... the knowledge. So i'll help do at least 2 models. Maybe 3. I'll post back here in a few hours with updates on my progress. ;D

Just want to know whats taken and whats not so if someone could give me a heads up.

Re: Help us to Optimize PR! (LOD Switching & Cull Distances)

Posted: 2011-08-11 04:26
by Psyrus
Valleyforge3946 wrote:Well, i was told to come here and help out with the chance that i could join the team someday so, i already have some coding expierence. i know basic python, have been editing tweak con files for over 2 years and have been working on my own minimod for over 2 years that started with a modded bf2demo server back when i was in high school. i have read the tutorial's and have practiced a bit so i am confident that i can help out. I have the time and with practice and trial and error... the knowledge. So i'll help do at least 2 models. Maybe 3. I'll post back here in a few hours with updates on my progress. ;D

Just want to know whats taken and whats not so if someone could give me a heads up.
The easiest way to track it is here IMO:

AIR - Incomplete

Summary Spreadsheet

I have this thread subscribed so within 24 hours of you posting a new completed one, I'll update the spreadsheet to reflect the new completion(s).

(lucky.Boy's added already)

Re: Help us to Optimize PR! (LOD Switching & Cull Distances)

Posted: 2011-08-11 04:43
by Valleyforge3946
:D Alrighty then. Sad to see all the land vehicles gone but thats alright.
I'll be attempting
ger_the_uh1d
idf_ahe_apache
Thanks.

Re: Help us to Optimize PR! (LOD Switching & Cull Distances)

Posted: 2011-08-11 06:31
by Psyrus
Valleyforge3946 wrote: :D Alrighty then. Sad to see all the land vehicles gone but thats alright.
I'll be attempting
ger_the_uh1d
idf_ahe_apache
Thanks.
Hey there, I just double checked and it seems like the apache was already committed by Leusch earlier, as when one faction's LOD distances are done, all vehicles that are shared between factions (Hueys, Chinook etc) are considered done. I have updated the spreadsheet to reflect the corrections. Apologies! I have put purple coloring on all air vehicles that are currently incomplete but used in multiple factions. If you have already finished the apache before you see this, at least it will give you a good idea if you compare the committed values (first post) to what you thought was good.

Use this slightly updated link to include coloring. [Link]

Can I suggest instead that you do one of the UH60 variants because currently the LOD switch distances are far too early, especially from 1 to 2.

See Video:


Re: Help us to Optimize PR! (LOD Switching & Cull Distances)

Posted: 2011-08-12 19:09
by Valleyforge3946
Okay. Thanks for the heads up. And thank you for the video. I appreciate it. I had already finished the apache but as u
You said, I will take that as good practice. I will do 2 uh60 variants later today as I am busy at the moment. I will post back progress later. If you have any questions to ask me simply reply back here since I am subscribed to this thread or send me a PM.
Odds are I will be askinh the questions. XD

Re: Help us to Optimize PR! (LOD Switching & Cull Distances)

Posted: 2011-08-12 19:31
by Psyrus
Valleyforge3946 wrote:Okay. Thanks for the heads up. And thank you for the video. I appreciate it. I had already finished the apache but as u
You said, I will take that as good practice. I will do 2 uh60 variants later today as I am busy at the moment. I will post back progress later. If you have any questions to ask me simply reply back here since I am subscribed to this thread or send me a PM.
Odds are I will be askinh the questions. XD
Cool well as I said in the post above, if the name is listed in purple, it means you only have to do one of the variants, as the model is exactly the same and generally the textures make SFA difference to the LOD switching (well for the 15 or so doubles that I looked at for the land vehicles). So you can do either the IDF or the US blackhawk, you needn't bother with both.

Re: Help us to Optimize PR! (LOD Switching & Cull Distances)

Posted: 2011-08-21 04:19
by Valleyforge3946
Psyrus wrote:Cool well as I said in the post above, if the name is listed in purple, it means you only have to do one of the variants, as the model is exactly the same and generally the textures make SFA difference to the LOD switching (well for the 15 or so doubles that I looked at for the land vehicles). So you can do either the IDF or the US blackhawk, you needn't bother with both.
You have proven most helpful and i have noticed. Why make 6 blackhawks when you can have 1 main model and just adjust it as needed. Makes sense. Unfortunately, this is not the post where i say DONE!! and all is well. I'm afraid real life caught up with me and i have been very busy over the last few days. However, tonight, finally, i will have the time to complete what i said i would complete. So, Thanks for the info. Going to be a very busy night. Lod's to adjust, servers to set up, a dinosaur in need of serious repair and then time allowing, sleep past the alarm and miss church. xD

Re: Help us to Optimize PR! (LOD Switching & Cull Distances)

Posted: 2011-09-19 03:31
by Valleyforge3946
Hello its me again. Guess who finished the LoD switches for the uh-60. I did. Anyways, personal life is crazier and sporadic than it ever has been. Appreciate the time and patience of everyone as no one message me. Would have posted the other night but seeing as my main rig is currently residing in a place with no internet and me never staying at one place for more than 24 hours, i was unable to post here. Promise next post will entail my work. (D'oh on me for not saving to a flashdrive. They were MIA when i finished and reappeared literally a minute before i left. FML) Will try to post Tuesday. :)

Re: Help us to Optimize PR! (LOD Switching & Cull Distances)

Posted: 2011-12-20 15:26
by Psyrus
Valleyforge3946 wrote:Hello its me again.
Any news?

Also, if the devs get around to this some day, there are things that were supposedly committed to the repo but have not had their values changed in the latest release, which leads me to believe the changes were lost when something was reverted. Just as an FYI.

Re: Help us to Optimize PR! (LOD Switching & Cull Distances)

Posted: 2011-12-28 11:52
by illidur
i started testing this with the " usthe_uh60" but i can't render it beyond 993.59 meters no matter what it seems.

is it normal that i can render a BH at 993.59 as a pixel with 3 cull?

Re: Help us to Optimize PR! (LOD Switching & Cull Distances)

Posted: 2011-12-29 16:42
by Psyrus
illidur wrote:i started testing this with the " usthe_uh60" but i can't render it beyond 993.59 meters no matter what it seems.

is it normal that i can render a BH at 993.59 as a pixel with 3 cull?
I got to 1311.74 with 3 pixels @ 3.7 cull distance (from the side)

Image

Always zoom out from the largest side. When you get to a point that it disappears... hit the "continuous rotation" on the right toolbar to make sure that it doesn't pop back into view/change size at your given level.

Re: Help us to Optimize PR! (LOD Switching & Cull Distances)

Posted: 2011-12-29 16:51
by lucky.BOY
Seems it depends alot on the resolution of your monitor, doesnt it? I did the huey on a HD monitor (1920x1080), but since im now left with only a 1366x768 monitor of my notebook, I can't make progress...

-lucky

Re: Help us to Optimize PR! (LOD Switching & Cull Distances)

Posted: 2011-12-29 17:05
by Psyrus
lucky.BOY wrote:Seems it depends alot on the resolution of your monitor, doesnt it? I did the huey on a HD monitor (1920x1080), but since im now left with only a 1366x768 monitor of my notebook, I can't make progress...

-lucky
It shouldn't make a huge amount of difference to be honest... although smaller pixels on some screens might be harder to see with the naked eye.