Redflag RP5 V5.0 Preview
-
Well AS those guys have the tools to do the math and decide what is a killer or not, we don’t… we just see others and drull and want…
-
Well AS those guys have the tools to do the math and decide what is a killer or not, we don’t… we just see others and drull and want…
You were born as a baby with the same brain as them We (or i) dont drull and want …i try to find compromized solutions for all of us to make things better.
If there is a way…lets find it…if not…let´s move on, but don´t give up becuase your are “powerless”.“Humans communicate with language… animals by sniffing their butts” (an old saying).
-
Gents…before this goes south here again… …
IMHO this is already pointing there… Many interesting things and opinions have been exchanged in this thread and they need some time to be absorbed. Please do not drag it to the level of personalized ugly argument. If this is for me to reach too far to say such a things, forgive me, but my name and my pretty hard work also is involved here. I’ll leave then things to the power of moderation.
-
@A.S:
You were born as a baby with the same brain as them We (or i) dont drull and want …i try to find compromized solutions for all of us to make things better.
If there is a way…lets find it… but don´t give up becuase your are “powerless”. “Humans communicate with language… animals by sniffing their butts” (an old saying).I just mentioned that BMS guys have the code, tools and means to test and tell if your suggestions (which are not new, they where done long before u got in to Falcon…) are doable or not… Blueprint is on it from the very very beginning…
And u r doing nothing more then talk actually… If u want to find the solution for your suggestions get the RV code do what u say and prove that it works ok…
We all want the same thing and try to help eachother for the best you also where born as a baby with the same brain as them, didn’t see a we there and troubled me…
-
And u r doing nothing more then talk actually…
Then don´t use the redflag 5 theater….because its just “talk”
And don´t use Falcon Online, because it is only “talk” too. (oh, i forgot…you can´t).Arty, i did not offend you. You are interpreting too much into things before you reply.
Look. We all can find thousands of reasons in life why certain things don´t work. But its not about that. It is about finding the one reason why it can(might) work.
Demanding and complaining doesn´t help anyone - the least yourself. But participating and contributing does. -
@A.S:
hmmm… thanks for pointig me… i will try to have a chat with him…who knows.
For me it doesn´t makes sense to loose FPS just because of a more defined blank terrain “frame” or “grid” while the textures are kept in the same size as before (L2). That actually was the whole idea (trick). If we would use tiles for each new “plane” created instead (more smaller tiles)…yeah that would be silly …and kill alot of FPS. https://www.benchmarksims.org/forum/showthread.php?19146-Redflag-RP5-V5-0-Preview&p=276443&viewfull=1#post276443A.S I don’t understand what you mean by that example… if you think the FPS hit is from tiles then you are wrong. Tiles will cause FPS hit if the texture size is too big for example which may cause more swaps for the V-card and will eventually affect FPS. The terrain resolution is the “heavy duty” stuff in this regard. 10m resolution simply requires more triangles calculations and so will be heavier than a 1000m res, think the same as you have a 100K poly model and a 1K one, now multiply it with the size of the terrain that you see in a scene.
Terrain engine is exactly something that should answer all this questions, manage lowest resolution at low altitudes and smart enough to create relatively smooth transitions for far terrain when full res isn’t necessary.
-
@ I-Hawk.
The idea is simple:
- keep the textures size in L2 (1000m x 1000m) and don´t create smaller or bigger ones (no changes here)
- allow more than 4096 unique textures to be used, so we can consider covering a whole theater with indiviual textures instead of repeating patterns
- reduce elevation grid (or mesh) from 1000m (which is 1980 or older style tbh) to at least something more appropiate (like 50m or 100m)
Mind you (i dont know if you fly FSX)… THEIR elevation mesh can even go down to .30cm and with modern PCs (around 2 years old)… FPS is great !
If we take real textures from google map right now… and place them into BMS with only 1000m elv. mesh, the texture attributes don´t match with the low 1000m elv. mesh and we get again rivers flowing uphill and downhill or cities “hanging” in declined hills etc etc. It looks “squeezed” and “fake”.
Photoreal textures with wrong elevation data doesn´t match up. Photoreal textures only work nicely if appropiate elevation resolution is possible.
I am not proposing to change the size of TEXTURES … no, keep the same size you have now - Same texture size and tile-sets.
ONLY (if possible) re-code the ELEVATION DEFINTION to 50m or 100m for L0 or L2 …whatever you like and consider the AI understands it too.
Fartiles you still can kick out (as it is now) - no need and saves FPS.Once that is achieved…THEN we can litterally IMPORT … TEXTURES AND ELEVATION from the internet and create much more realistic terrain for BMS. We know how to do it, but we are limited by:
- only 4096 unique tiles, thus repetitions required
- only 1000m elevation resolution
Those 2 things cut us off in the creation of amazing looking terrain (forget what you see in redflag 5.0 preview pictures, because once that is possible… terrain can be much much better than the pictures shown in 1st post)
As you can see in the first post of this thread…Polak did an amazing job on the textures… i worked like an idiot on the WHOLE elevations for korea…making 250.000 tiles manually by hand (i had to, so it makes sense with textures used)BUT!!! we can DO EVEN BETTER THAN THIS, IF …those two limiations change.
My thoughts go towards “maybe we can utilize what we already have in better ways (with tweaks), instead of considering the creation of a whole new gfx-terrain engine”
Note: most DEVs (so i figured) haven´t even tried Redflag 4.0 as most don´t have 4.32 installed anymore for long. Once you see the textures+elevation is Redflag 5.0 … you will get the idea. At this point remember this what i wrote above… WE CAN DO MUCH BETTER, IF …those two limitations change (or look at that tileproxy video above).
-
always divide by 2 in Falcon world - so 1000m -> 500m -> 250m (L0) -> 125 m -> 62.5m -> 31.25 -> 16,625 m …etc…. are possible mesh subdivisions
BTW I like 1km patch, it is 90ths technology …80ths to early 90th is flat terrain with pyramid/hill objects (wireframe or solid) -
In the first step somebody pls. explain why had to give up the much more deteailed terrain of OF… I did not caused any FPS issue…
-
all 250m terrains “poped up” in front of player due to LOD switching…simply it was too heavy to see 250m resolution up to the horizont…
1000m terrain can be edited on triangle level
-
…and another problem - object placement, unit placement and movement…Z shifting is counted according L2
-
In the first step somebody pls. explain why had to give up the much more deteailed terrain of OF… I did not caused any FPS issue…
thats why
-
always divide by 2 in Falcon world - so 1000m -> 500m -> 250m (L0) -> 125 m -> 62.5m -> 31.25 -> 16,625 m …etc…. are possible mesh subdivisions
BTW I like 1km patch, it is 90ths technology …80ths to early 90th is flat terrain with pyramid/hill objects (wireframe or solid)31.25m for Le (extra) would be ideal. I still cant get my head around, that a better wireframe (and nothing else changed) would kill soo much FPS……espcially after having like 100FPS with my almost 2 year old system while running this:
-
@A.S:
31.25m for Le (extra) would be ideal. I still cant get my head around, that a better wireframe (and nothing else changed) would kill soo much FPS……espcially after having like 100FPS with my almost 2 year old system while running this:
imagine u are flying over 50 x 50km visible area
L2 res : 50 x 50 = 2500 polygons = 5000 triangles
L0 res: (50 x 4 ) x (50 x 4) = 40 000 polys = 80 000 tris if LOD switching off…if ON = pop ups… on the other hand 80k is not so much for most of u…so probably game logic part of it is rather show stopper here (but u also like 80k hires pit, F16 model, populated city with hi-res houses…)Edit: note to your video
FSX has much more inteligent and adaptive mesh tessalation compared to simple and regular Falcon grid/checker(F4 mesh is not adaptable)…see here:
http://www.microsoft.com/Products/Games/FSInsider/developers/Pages/GlobalTerrain.aspx
-
I get your point, what a pitty. So, bottom line… new terrain engine one way or the other is the only solution?
-
SFP1 terrain engine had small provision for local height map tile. It was small grayscale (256 values) 16x16 pixels so called nameofthetexture_HM.bmp, which superimposed locally heights with greater resolution on the mother tile. Terrain engine automatically was dividing that area into smaller squares (16x16 because of the size of _HM.bmp) and was taking into account height information from that small height map but only whenever there was nameofthetexture_HM.bmp present in the texture folder. Rest of the tiles were rendered with lower height resolution. I guess it was, what we call bump texture.
Tricky part of course was that higher resolution “meshed in” seamlessly with the rest of sorrounding tiles on the perimeter, but that was done by the smart algorithm within the engine. 16x16 did not have enough detail to sculpt say intricate terraform shapes, but I have experimented with much higher resolutions of bmp up to 256x256 with quite interesting results and no extra adverse effect on the performance. Again I repeat it was meant to be used only locally (say river tiles) and it was not meant to be widely used.
EDIT: only height information was taken from the _HM texture. Main also *.dds terrain texture still was covering area of the main larger grid.
-
-
Should I try the switch? Because I cannot see any real answer.
I havent exprimented with the “-g” switch in BMS yet. I know what it did in the past… so try it out.
-G31 is extreme high though. I remember screwing around with g-15 values in OF. -
It was said that cause of the new gfx engine this switch has no effect anymore in falcon…
-
@A.S:
I havent exprimented with the “-g” switch in BMS yet. I know what it did in the past… so try it out.
-G31 is extreme high though. I remember screwing around with g-15 values in OF.I set 99 terrain detail, and between 7-13. I did not see any change with -g31 or -g13 either.