Rendering Improvements [WIP]

Announcements and news updates to your projects using Urho3D.
Forum rules
If you have images or videos, you can attach them in the showcase post. Please use small thumbnail as link to original media resources.
When you post in this forum, you have to agree that Urho3D main website as well as the Facebook page have the permission to republish your post with all the related images or videos. The original images must have width larger than 890px for best viewing result in all the supported viewports by main website.

Re: Rendering Improvements [WIP]

PostPosted by krstefan42 » 13 Aug 2016, 17:14

Nice work, always glad to see more modern features in Urho3D 's rendering.

dragonCASTjosh wrote:Pre-filtered IBL
This may be difficult to see the difference but take my word for it performance is massively increased. likely not viable for mobile due to file size limits.

Why can't the pre-filtering be done at scene load time? It doesn't need to be super accurate. Heck you could almost just generate mip maps from the environment map, put on trilinear filtering, and call it a day. You could add a 1 pixel pre-pass blur to each mip to hide the pixels, or do a quick 3-tap blur filter in the final shader. Then mix the lower mips into the higher ones to better match a wide specular lobe.

dragonCASTjosh wrote:the simplest way to solve the shader issue is just write some preprocessor change the syntax of one of the languages but this is likely not a good long term solution. Vulkans spirv may be the best solution

I've seen the preprocessor solution used in other code, and I think it works pretty well.
User avatar
krstefan42
Have some posts
Have some posts
 
Posts: 24
Joined: 28 Jun 2015, 06:16

Re: Rendering Improvements [WIP]

PostPosted by dragonCASTjosh » 15 Aug 2016, 16:34

krstefan42 wrote:Why can't the pre-filtering be done at scene load time? It doesn't need to be super accurate. Heck you could almost just generate mip maps from the environment map, put on trilinear filtering, and call it a day. You could add a 1 pixel pre-pass blur to each mip to hide the pixels, or do a quick 3-tap blur filter in the final shader. Then mix the lower mips into the higher ones to better match a wide specular lobe.


the current solution is in mixing pre-filtering witha mobile friendly solution. the performance is on par with traditional pre-filtering but saves 2 samplers on the gpu as prefiltered also needs the roughness vs ndl image and irridiance map.
User avatar
dragonCASTjosh
Moderator
Moderator
 
Posts: 205
Joined: 04 Aug 2015, 18:59

Re: Rendering Improvements [WIP]

PostPosted by sabotage3d » 24 Aug 2016, 16:29

Might be cool to test against some of these PBR examples: https://github.com/simongeilfus/Cinder-Experiments
User avatar
sabotage3d
Have many posts
Have many posts
 
Posts: 515
Joined: 25 Oct 2014, 13:26

Re: Rendering Improvements [WIP]

PostPosted by theak472009 » 09 Sep 2016, 09:53

Did the guy just give up? Seems reasonable since there is not much to improve XD
User avatar
theak472009
Have some posts
Have some posts
 
Posts: 24
Joined: 14 Aug 2015, 11:56

Re: Rendering Improvements [WIP]

PostPosted by dragonCASTjosh » 09 Sep 2016, 18:36

theak472009 wrote:Did the guy just give up? Seems reasonable since there is not much to improve XD


Im still working although a mix between lack of time and a lot of learning makes it hard. i pushed physically based light values not to long ago.

sabotage3d wrote:Might be cool to test against some of these PBR examples: https://github.com/simongeilfus/Cinder-Experiments


i actually looked through that whilst i was learning :)+
User avatar
dragonCASTjosh
Moderator
Moderator
 
Posts: 205
Joined: 04 Aug 2015, 18:59

Re: Rendering Improvements [WIP]

PostPosted by dragonCASTjosh » 14 Sep 2016, 13:25

Improvements to sphere lights using the approach detailed in epic paper, with this method it should be possible to have large amounts of area lights within the scene although its limited to sphere and tube lighting only. I may look into doing other lights shapes later down the line.

Images: http://imgur.com/a/69BFR
User avatar
dragonCASTjosh
Moderator
Moderator
 
Posts: 205
Joined: 04 Aug 2015, 18:59

Re: Rendering Improvements [WIP]

PostPosted by dwlcj » 22 Oct 2016, 23:46

DragonCASTjosh:
Rendering Improvements have new progress? Especially soft shadows.
User avatar
dwlcj
New user
New user
 
Posts: 12
Joined: 04 Jul 2016, 03:34

Re: Rendering Improvements [WIP]

PostPosted by dragonCASTjosh » 23 Oct 2016, 11:15

dwlcj wrote:DragonCASTjosh:
Rendering Improvements have new progress? Especially soft shadows.


Not much just yet, I have been very busy with University, i also managed to lose a lot of work in a system crash so i have to rebuild the tubelight stuff. I have some local work to improve the mip sampling and the attenuation of lights under PBR. Alongside that i have put some time aside to try and work out what is wrong with variance shadow maps as in my personal testbed renderer they work perfectly and the implementation so far looks identical
User avatar
dragonCASTjosh
Moderator
Moderator
 
Posts: 205
Joined: 04 Aug 2015, 18:59

Re: Rendering Improvements [WIP]

PostPosted by dwlcj » 23 Oct 2016, 23:34

All Right,Thank you very much,dragonCASTjosh.
I Use Physically based Rendered in Urho3d.
Image
Missing a few things:
SSR or Ogre2.1 Parrallax Cubemap
Soft Shadow
AO
GI
User avatar
dwlcj
New user
New user
 
Posts: 12
Joined: 04 Jul 2016, 03:34

Re: Rendering Improvements [WIP]

PostPosted by dragonCASTjosh » 24 Oct 2016, 07:30

dwlcj wrote:All Right,Thank you very much,dragonCASTjosh.
I Use Physically based Rendered in Urho3d.

Missing a few things:
SSR or Ogre2.1 Parrallax Cubemap
Soft Shadow
AO
GI


Awesome that someone is using the PBR stuff already :) let me know if there us anything i can do to make it easier to use.

As for the missing features they are all on my roadmap, parallax mapping should be simple i know JSandusky had a text file explaining how to implement it in his original PBR implementation but i no longer have access to that. As for SSR im fairly new to graphics and have never touched tracing so it will be difficult to learn, same goes with GI. AO should be a little more simple and same with soft shadows
User avatar
dragonCASTjosh
Moderator
Moderator
 
Posts: 205
Joined: 04 Aug 2015, 18:59

PreviousNext

Return to Showcase

Who is online

Users browsing this forum: No registered users and 0 guests

cron