Isle of marooned

Share with us how are you using the powerrrr of the force

Moderator: Alain

Re: Isle of marooned

Postby Sury » Sat Sep 07, 2013 2:08 pm

Wow guys, thank you. :D
At first I bougth some packages with models from different sites, and I still have them, but that approach was very restrictive and I had to use what they were offering, my story and vision should be made according to that art. Now I'm able to materialize my ideas and imagnation by modeling what I need and rendering it how I want it to look. That gives me freedom and I feel like only the sky is the limit.
Of course there are lots of very powerful and affordable engines, like UDK, Unity etc.
Because of that and other factors, there is fast growing competition on the feild. Many small indie developers can grab Unity ot UDK are now able to put a game that looks much better than mine, and they will put much less effort and time, and will ship to many more platforms with a click of a mouse.
User avatar
Sury
 
Posts: 193
Joined: Sat Aug 14, 2004 5:32 am
Location: Bulgaria

Re: Isle of marooned

Postby Julio Jerez » Sat Sep 07, 2013 2:40 pm

Sury wrote:At first I bougth some packages with models from different sites, and I still have them, but that approach was very restrictive and I had to use what they were offering


that's my problem, I am a complete moron when it comes to modeling anything.
I can do simple things. but that's about it.
Julio Jerez
Moderator
Moderator
 
Posts: 12249
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Isle of marooned

Postby JoeJ » Sat Sep 07, 2013 2:51 pm

I don't think those engines look much better than yours. Just fix the postprocessing glitches, do smoother blending on fog planes, so they disappear before you can recognize the planes, etc... small issues.
Your lighting looks very good. State of the art as long as Enlighten or similar is not part of those 'free' Engine versions.
Also, you could not use Newton with those. Reason enough for me to do an own engine.
Many games tried to do something like Amnesia, but no one had those intense physic enviroment - too much instability in other physics engines?
And if i'd be AAA company and hiring, i'd choose you over any UDK/Unity wizard, even if i'd use those engines.

What are your plans about characters?
For me that's the point where i have most interrest but also most worrys.
Worrys because i can do modeling, but i think animation is too much work for the one person game idea (i try to do that too).
That's why i try to make living ragdolls to avoid any animation at all. I found Newton good enough to really handle that, but the ai is very hard...
I could have done a lot animation since i've started that :)
User avatar
JoeJ
 
Posts: 1451
Joined: Tue Dec 21, 2010 6:18 pm

Re: Isle of marooned

Postby JoeJ » Sat Sep 07, 2013 2:54 pm

Julio Jerez wrote:that's my problem, I am a complete moron when it comes to modeling anything.


... you have enough other skills :wink:
User avatar
JoeJ
 
Posts: 1451
Joined: Tue Dec 21, 2010 6:18 pm

Re: Isle of marooned

Postby Sury » Sun Sep 08, 2013 3:39 pm

JoeJ wrote:What are your plans about characters?
For me that's the point where i have most interrest but also most worrys.
Worrys because i can do modeling, but i think animation is too much work for the one person game idea (i try to do that too).
That's why i try to make living ragdolls to avoid any animation at all. I found Newton good enough to really handle that, but the ai is very hard...
I could have done a lot animation since i've started that :)


Ahhm, I've tried to animate characters ( mostly biped ) in 3ds max, but I wasn't satisfied by the results. Then I tried to search for some free .bvh files and had success with them to animate some characters I was modeled and rigged before. Files from sites like this : http://gfx-motion-capture.blogspot.com/
I see people are getting results with Kinect( and other hardware ) to motion capture their movement :
http://www.youtube.com/watch?v=omm7Hya1uGY
http://www.vali.de/archives/938
User avatar
Sury
 
Posts: 193
Joined: Sat Aug 14, 2004 5:32 am
Location: Bulgaria

Re: Isle of marooned

Postby JoeJ » Mon Sep 09, 2013 4:37 pm

Motion capturing with Kinect? That's a solution. I wouldn't have thought this works so well.

Take a look those cute war machines, to get an impression how well this could do in games:
http://www.youtube.com/watch?v=cNZPRsrwumQ
http://www.youtube.com/watch?v=tFrjrgBV8K0
User avatar
JoeJ
 
Posts: 1451
Joined: Tue Dec 21, 2010 6:18 pm

Re: Isle of marooned

Postby Sury » Fri Sep 13, 2013 8:07 am

Yeah, those machines always look frightening to me, because they confuse the brain of the observer and the brain fears the unknown + it knows that something is wrong, but cannot figure what exactly and that forces it switch to "caution mode".

JoeJ wrote:Your lighting looks very good. State of the art as long as Enlighten or similar is not part of those 'free' Engine versions.
Also, you could not use Newton with those. Reason enough for me to do an own engine.
Many games tried to do something like Amnesia, but no one had those intense physic enviroment - too much instability in other physics engines?
And if i'd be AAA company and hiring, i'd choose you over any UDK/Unity wizard, even if i'd use those engines.


Thanks, but I'm actually not very pleased about my lighting model at the moment, because it does not support dynamic real time global illumination. I've tried different solutions some time ago in that direction and one of these was an attempt to implement Instant Radiosity with virtual point lights. I used Newton to help me with those, as it is very capable of shooting rays very fast through the scene geometry.
Here is a short video showing dynamic generation of 200( if I remember correctly) point lights in real time.



I have pretty much crapped it, and now I'm trying to follow the steps of the big boys (Unreal, Crytek) and implement something similar to their version of global illumination solvers. I need fully dynamic solution, since I intend to make fully dynamic scenes with respect of geometry, lights, and of course camera.
If you have any resources, samples, papers about that subjects, I'll be glad if you share them with me.
User avatar
Sury
 
Posts: 193
Joined: Sat Aug 14, 2004 5:32 am
Location: Bulgaria

Re: Isle of marooned

Postby JoeJ » Fri Sep 13, 2013 12:13 pm

Sury wrote:Yeah, those machines always look frightening to me, because they confuse the brain of the observer and the brain fears the unknown + it knows that something is wrong, but cannot figure what exactly and that forces it switch to "caution mode".


Yep, the dilemma if something is very close to look real, we tend to accept it less likely than we would accept a cartoon.
May become a problem - but not more than other things related to caharacters. For now my ragdoll looks pretty natural, sometimes it makes a few steps by accident, but always looks either silly, funny or natural - never as creepy as those real robots.

For the GI, i've done a lot work here and i think i'm done with 90% of research, but i do on CPU and need to learn GPU somewhen soon to proof it's fast enough.
The main problem i see with current algos is the question: "Is it really worth to spend so mutch cycles for single bounce reflection?".
I'd say no, even if you get two bounces. Unlimited bounces is a must for me, and you can get those for free if you use a technique that stores results and reuses them in the next step.
I think Enlighten is the only current game approach that may use this(?), but there's no reflection from dynamic objects - they're still only receivers?
Crassins method is promising - mainly because it supports speccular reflections - but i heared that Unreal engine gave up on it - too slow.
Cryteks method is fast, but as long as they sample the world only from screenspace it's the same fake as SSAO.

The VPL method you use misses shadows - fixing that led to very interesting ideas:
"Imperfect Shadow Maps" + "Micro Rendering" (Tobias Ritchel, Carsten Dachsbacher...)
"ManyLods" (Mathias Hollaender, T.Ritchel)
Very inspiring... be sure to read this stuff.

Georgios Papaioannou did something similar than Crytek, but used world space voxelization like Crassin, and tried some other ideas.

Cyril Crassin for sure.

Also the 10 years old stuff is still valid.



After reading this and a lot more, i came up with a mix influenced by most of those techniques with those properties:
* does not use rasterization or raytracing to compute diffuse interreflection
* supports unlimited bounces at the cost of temporal lag
* needs to store spherical harmonics samples either over surface using LOD tree structure (fast), or in a cascaded volume grid (GPU friendly)
* can support speccular reflections from samples and volumetric light for particles / fog.
* dynamic world and lights (for sure)
* may allow "force < X ms" at the cost of more temporal lag (the missing 10 %)

Sounds good on paper, but let's see first... Maybe real GI still have to wait for PS5?
Looking at your screenshots, if architecture is non destructable, SH radiance grid would suffice?
User avatar
JoeJ
 
Posts: 1451
Joined: Tue Dec 21, 2010 6:18 pm

Re: Isle of marooned

Postby Sury » Sat Sep 14, 2013 4:16 am

JoeJ wrote:
Sury wrote:Yeah, those machines always look frightening to me, because they confuse the brain of the observer and the brain fears the unknown + it knows that something is wrong, but cannot figure what exactly and that forces it switch to "caution mode".


Yep, the dilemma if something is very close to look real, we tend to accept it less likely than we would accept a cartoon.
May become a problem - but not more than other things related to caharacters. For now my ragdoll looks pretty natural, sometimes it makes a few steps by accident, but always looks either silly, funny or natural - never as creepy as those real robots.

For the GI, i've done a lot work here and i think i'm done with 90% of research, but i do on CPU and need to learn GPU somewhen soon to proof it's fast enough.
The main problem i see with current algos is the question: "Is it really worth to spend so mutch cycles for single bounce reflection?".
I'd say no, even if you get two bounces. Unlimited bounces is a must for me, and you can get those for free if you use a technique that stores results and reuses them in the next step.
I think Enlighten is the only current game approach that may use this(?), but there's no reflection from dynamic objects - they're still only receivers?
Crassins method is promising - mainly because it supports speccular reflections - but i heared that Unreal engine gave up on it - too slow.
Cryteks method is fast, but as long as they sample the world only from screenspace it's the same fake as SSAO.

The VPL method you use misses shadows - fixing that led to very interesting ideas:
"Imperfect Shadow Maps" + "Micro Rendering" (Tobias Ritchel, Carsten Dachsbacher...)
"ManyLods" (Mathias Hollaender, T.Ritchel)
Very inspiring... be sure to read this stuff.

Georgios Papaioannou did something similar than Crytek, but used world space voxelization like Crassin, and tried some other ideas.

Cyril Crassin for sure.

Also the 10 years old stuff is still valid.



After reading this and a lot more, i came up with a mix influenced by most of those techniques with those properties:
* does not use rasterization or raytracing to compute diffuse interreflection
* supports unlimited bounces at the cost of temporal lag
* needs to store spherical harmonics samples either over surface using LOD tree structure (fast), or in a cascaded volume grid (GPU friendly)
* can support speccular reflections from samples and volumetric light for particles / fog.
* dynamic world and lights (for sure)
* may allow "force < X ms" at the cost of more temporal lag (the missing 10 %)

Sounds good on paper, but let's see first... Maybe real GI still have to wait for PS5?
Looking at your screenshots, if architecture is non destructable, SH radiance grid would suffice?


Thanks, I'll keep the right to comment on different points later.
As for the SH world grid - yes, but I guess I need to recalculate the probes each time the lighting enviorment changes. That means, the geometry that bounces light moves or changes color material, or the light that bounces off geometry moves or changes propetries(color, intensity, etc). I'm not sure that rerendering a dense grid of cubemaps and converting them to spherical harmonics coefficients in real time would be fast enough, if I needed to do that.
User avatar
Sury
 
Posts: 193
Joined: Sat Aug 14, 2004 5:32 am
Location: Bulgaria

Re: Isle of marooned

Postby JoeJ » Sat Sep 14, 2013 3:51 pm

What i meant was a precomputed SH grid from static scene without dynamic objects. Dynamic objects then would be only receivers at runtime.
Nothing new about this, but it would suffice if dynamic objects are not too large.
For stuff like doors you could precompute the difference and update the grid area of influence, same for static on/off lights.

If a door opens slowly, blending precomputed grid data would create the illusion of dynamic lighting,
and because grid contains only indirect light, this trick should not become visible.
Same for a swinging lamp - direct light and shadows are accurate, and the static indirect still looks correct,
as long as the light stays somehow close to the initial position.

... no cool algorithm, but good looking and the limitations might be acceptable.
Full dynamic GI will not look as good as precomputed for mainly static scenes soon.
Quake 3 Lightmaps are much more detailed than Battlefield 3 - just turn off SSAO and textures - Q3 is awesome, BF3 is *.
Even with the approach i suggest you will not come close to Q3. SH grid means low detail and sampling artifacts, but you have SSAO & textures too to hide that.
Some guys snap the SH lobes close to near surface to fight sampling artifacts.

EDIT:
To add indirect light from flashlight, a reflective schadow map might work.
The missing shadows should be acceptable if you use spotlights at hit point normal direction instead point lights?
User avatar
JoeJ
 
Posts: 1451
Joined: Tue Dec 21, 2010 6:18 pm

Re: Isle of marooned

Postby Sury » Sun Sep 22, 2013 5:33 am

Thanks, but I prefer to jump into deep waters and implement fuly dynamic approach with VPL's generated using RSM and injected into a LPV - something like Crysis solution.
The problem is, I need to do this with DirectX 9.0 and that means no way to render into volume texture directly, but with unwrapping it should work and trilinear filtering done "manually" instead of hardware as with volume textures.
User avatar
Sury
 
Posts: 193
Joined: Sat Aug 14, 2004 5:32 am
Location: Bulgaria

Re: Isle of marooned

Postby Sury » Wed Oct 02, 2013 2:02 pm

Here is a small test scene of my implementation of something like irradiance volumes. Lights are static for the moment, since rerendering the probes in my implementation is slow for now. Uses spherical harmonics for the irradiance storage.

GI off
Image

Gi on
Image
User avatar
Sury
 
Posts: 193
Joined: Sat Aug 14, 2004 5:32 am
Location: Bulgaria

Re: Isle of marooned

Postby Julio Jerez » Wed Oct 02, 2013 2:58 pm

wow, that start to look real good. are you doing that with shaders?
Julio Jerez
Moderator
Moderator
 
Posts: 12249
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Isle of marooned

Postby Sury » Thu Oct 03, 2013 12:46 am

Yeah, I basically spread light probes automatically across an area (or only on key areas) calculate the incoming light at those points, send this data and shade affected pixels. @JoeJ described similar approaches above.
This only one bounce of indirect light. I believe I can extend it several bounces, which would look much better on complex scenes, and the real-time cost (performance) should be the same.
User avatar
Sury
 
Posts: 193
Joined: Sat Aug 14, 2004 5:32 am
Location: Bulgaria

Re: Isle of marooned

Postby JoeJ » Thu Oct 03, 2013 4:35 am

Yes, looks good.
The volume resolution seems pretty lowres, typically people work against that with SSAO.
I remember two good ideas for replacemaent of SSAO:

Screen Space GI - calculate reflected light from surrounding pixels instead of occlusion (brighten instead darken)
Screen Space Bent Normals - modify normal direction and length based on surrounding occlusion (improoved AO look)

Personally i hate SSAO, but both those techniques looked very good, performance might be practical.
Can't remember authors or axact titles from the papers.
User avatar
JoeJ
 
Posts: 1451
Joined: Tue Dec 21, 2010 6:18 pm

PreviousNext

Return to User Gallery

Who is online

Users browsing this forum: No registered users and 1 guest

cron