- Thread starter eerieone
- Start date
that´s what i was thinking
if an engy pulls back from cp2 to have a sentry up there early enough, he deserves a good sentry-spot
and its not that overpowered either, a demo shooting from the balcony backed up by a medic can take it down in a second. as spy i always had an easy time dealing with engis up there, they are in a stressful position where they try to manage staying on top of their gear and don´t spycheck everything
if an engy pulls back from cp2 to have a sentry up there early enough, he deserves a good sentry-spot
and its not that overpowered either, a demo shooting from the balcony backed up by a medic can take it down in a second. as spy i always had an easy time dealing with engis up there, they are in a stressful position where they try to manage staying on top of their gear and don´t spycheck everything
... amazingly well done man, when you keep on mapping
it could become an official map one day =)
it could become an official map one day =)
DontWannaName
L3: Member
- Jan 12, 2009
- 112
- 24
An Icy Mouse
aa
- May 23, 2009
- 205
- 34
cameraman should stick to his dayjob
but otherwise that looks really epic! love it. if only there was an explosion in hoodoo :/
If I must make one criticizm (and I must) I'd say your fire should go out the instant the door lands on it, it looks odd to see flames and smoke coming through solid metal. you can achieve this by simply sending a "stop" output to all the particles.
:thumbup:
but otherwise that looks really epic! love it. if only there was an explosion in hoodoo :/
If I must make one criticizm (and I must) I'd say your fire should go out the instant the door lands on it, it looks odd to see flames and smoke coming through solid metal. you can achieve this by simply sending a "stop" output to all the particles.
:thumbup:
kankle_king
L4: Comfortable Member
- Jan 31, 2009
- 161
- 86
the 'gimmick' as you called it is a great idea!!! I can check the map out w/o DLing etc. More ppl do this...
- Feb 26, 2008
- 1,626
- 1,325
Excellent explosion! Remember to disable the cap model after it happens. Also, when testing, you can just use console to make the cart keep moving: "ent_fire (name of tracktrain here) setspeeddir 1" 
- Apr 19, 2009
- 4,460
- 1,722
I am loving the new explosion!
Mr. Happy
L6: Sharp Member
- Jul 16, 2008
- 320
- 158
Please optimize Cashworks
OH NO SIR PLEASE DONT RELEASE YET. Cashworks is an excellent map and you have done a wonderful job, and yet I see you think you have optimized as much as possible! I am sorry to say this is not true That's ok though, optimizing is very difficult!
I want to help! I did a quick inspection of your map and at first glance see many possible optimizations! Below are many paintovers to show you some things that are possible.
For starters you really must get a better handle on the overdraw! There are many places where hidden or all but hidden faces are not nodrawed, where props are being rendered and where func_details could be func_lod's. I know what you are thinking, "func_lod is awesome new entity but it has alot of computational overhead!" this is true, however many places in your map could benefit from it's judicious use.
While your hinting is quite good it does leave alot to be desired You have forgotten the power of angled hints in many places, and are not splitting high action/view areas and vertical spaces enough! I have done some paintovers that might give you a few ideas.
Also in some places, vis could be controlled better simply by better "filling in" space (especialyl behind displacements) and further hint shaping from there.
Controlling the vertical could be very important to you. XY hint planes and large func_viscluster volumes will group the upper outdoor leaves into more efficient clusters.
Please keep working. You can do it!
Photobucket gallery of paintovers for optimization: http://s136.photobucket.com/albums/q192/MrHappy_photos/please optmiize cashworks/
OH NO SIR PLEASE DONT RELEASE YET. Cashworks is an excellent map and you have done a wonderful job, and yet I see you think you have optimized as much as possible! I am sorry to say this is not true
That's ok though, optimizing is very difficult!I want to help! I did a quick inspection of your map and at first glance see many possible optimizations! Below are many paintovers to show you some things that are possible.
For starters you really must get a better handle on the overdraw! There are many places where hidden or all but hidden faces are not nodrawed, where props are being rendered and where func_details could be func_lod's. I know what you are thinking, "func_lod is awesome new entity but it has alot of computational overhead!" this is true, however many places in your map could benefit from it's judicious use.
While your hinting is quite good it does leave alot to be desired
You have forgotten the power of angled hints in many places, and are not splitting high action/view areas and vertical spaces enough! I have done some paintovers that might give you a few ideas.Also in some places, vis could be controlled better simply by better "filling in" space (especialyl behind displacements) and further hint shaping from there.
Controlling the vertical could be very important to you. XY hint planes and large func_viscluster volumes will group the upper outdoor leaves into more efficient clusters.
Please keep working. You can do it!
Photobucket gallery of paintovers for optimization: http://s136.photobucket.com/albums/q192/MrHappy_photos/please optmiize cashworks/
kankle_king
L4: Comfortable Member
- Jan 31, 2009
- 161
- 86
- Apr 19, 2009
- 4,460
- 1,722
Wow! Mr. Happy that was one of the most informative and well made reviews I have ever seen for a map. For that I give you my thanks!
Mar
Banned
- Feb 12, 2009
- 607
- 63
OH NO SIR PLEASE DONT RELEASE YET. Cashworks is an excellent map and you have done a wonderful job, and yet I see you think you have optimized as much as possible! I am sorry to say this is not true
I want to help! I did a quick inspection of your map and at first glance see many possible optimizations! Below are many paintovers to show you some things that are possible.
For starters you really must get a better handle on the overdraw! There are many places where hidden or all but hidden faces are not nodrawed, where props are being rendered and where func_details could be func_lod's. I know what you are thinking, "func_lod is awesome new entity but it has alot of computational overhead!" this is true, however many places in your map could benefit from it's judicious use.
While your hinting is quite good it does leave alot to be desired
Also in some places, vis could be controlled better simply by better "filling in" space (especialyl behind displacements) and further hint shaping from there.
Controlling the vertical could be very important to you. XY hint planes and large func_viscluster volumes will group the upper outdoor leaves into more efficient clusters.
Please keep working. You can do it!
Photobucket gallery of paintovers for optimization: http://s136.photobucket.com/albums/q192/MrHappy_photos/please optmiize cashworks/
I'd also like to the say the same. There has been drastic improvement and by average FPS has doubled, but in many action intensive areas I tend to lag a bit. Any noticeable lag on a mid-high range PC will surely be noticeable on a low end and the low end base for TF2 is huge. So please optimize it.
For reference I got a FPS of 20 before RC, 40 on the RC, but it does drop to 30 some times. I get an FPS of 65 on all Valve and most fully optimized custom maps. (I set my FPS max limit to 65)
Last edited:
@Mr. Happy:
Don´t worry, not releasing yet there´s an rc2 coming
I´ve done much for that with working with occluders (thx andysss456), which now effectively kills the models past a corner which were drawn beforehand.
About your func_lod proposals
i know this func_lod bastard too well, and believe me, i func_lod whatever possible and get rid of props to be able to afford another lod-entity
tricky thing about lod is, stickies don´t stick to them, so making geometry the player uses func_lod makes a demos life unpleasant
putting a block-bullet brush over it won´t help either since they just don´t stick to brush-entities
i will surely try out the hint-setups you provided, since i´m no real expert on them
viewdistances... yes some might be shorter, it´s a try and error, i personally see every too short viewdistance in every map and they bother me much, i just don´t like things appearing out of thin air...
so, no worries, the optimization part is still underway
thx for the giant effort on putting together those screens, they will certainly come to good use, i promise
eerie out
Don´t worry, not releasing yet
there´s an rc2 comingI´ve done much for that with working with occluders (thx andysss456), which now effectively kills the models past a corner which were drawn beforehand.
About your func_lod proposals
i know this func_lod bastard too well, and believe me, i func_lod whatever possible and get rid of props to be able to afford another lod-entity
tricky thing about lod is, stickies don´t stick to them, so making geometry the player uses func_lod makes a demos life unpleasant
putting a block-bullet brush over it won´t help either since they just don´t stick to brush-entities
i will surely try out the hint-setups you provided, since i´m no real expert on them
viewdistances... yes some might be shorter, it´s a try and error, i personally see every too short viewdistance in every map and they bother me much, i just don´t like things appearing out of thin air...
so, no worries, the optimization part is still underway
thx for the giant effort on putting together those screens, they will certainly come to good use, i promise
eerie out
I'd have to say that occluders are fairly expensive, I've never used 1, and I've never honestly needed one. Heck Valve doesn't even use occluders anymore.
Mainly though I agree with everything Happy has mentioned, there is a load of visleafs in the area that can be simplified. Moreover you need to keep an eye out for your texture budgets.
I haven't checked on this within the map myself, but I feel that the world budget might be a little too high. That, alongside lightmap usage of the texture memory budget. Areas that hardly receive any light can be turned into scales of 32.
Mainly though I agree with everything Happy has mentioned, there is a load of visleafs in the area that can be simplified. Moreover you need to keep an eye out for your texture budgets.
I haven't checked on this within the map myself, but I feel that the world budget might be a little too high. That, alongside lightmap usage of the texture memory budget. Areas that hardly receive any light can be turned into scales of 32.
it´s not that i overuse them, i use them at the big turning points like after cp1, at the bottom of the hill, at cp3 and the big building at cp4
it surely can´t be more expensive to process if a player can see it than to actually having to render 20 props
i´m pretty sure valve is using occluders, i can see them in the visgroups, i think that they don´t get decompiled correctly, therefore we don´t see them in the vmf
it surely can´t be more expensive to process if a player can see it than to actually having to render 20 props
i´m pretty sure valve is using occluders, i can see them in the visgroups, i think that they don´t get decompiled correctly, therefore we don´t see them in the vmf
Last edited: