We actually removed it because it's not really relevant to how most content is created these days. Almost all materials have some tiling component, detail texture mapped, or multi-frequency textures mapped to it which means density display doesn't give you much information.
There still is a lightmap density mode though.
I've been a Polycount lurker for a bazillion jillion years, but I finally registered (or maybe I registered years ago and just forgot) to address this post.
I guess I'm mostly confused by this reasoning; nothing has really changed in the year or so between UDK's last meaningful release and UE4. Well okay, tons of things have changed, but not really in the way content is created. Things have changed drastically since UE2 - we've gone from BSP dependency to mesh dependency - but not so much in this area.
Texture density is an incredibly useful diagnostic tool. The things mentioned here as reasons for why this diagnostic tool is no longer available are themselves part of the reason it was useful. So either I'm having an exceptionally bad brain fart moment, or this was the equivalent of "because cars now have so many options for on board computers, we no longer feel it necessary to have a device that tests the functionality of on board computers."
Maybe we're just using texture density for different things
Going through that hallway now and trying to decipher how he set up his materials...I'm so lost. I guess it would be easier to understand if I saw the process as apposed to just the finished product, but I'm trying my best haha. I'm glad I have it to slowly, but surely learn from it.
Going through that hallway now and trying to decipher how he set up his materials...I'm so lost. I guess it would be easier to understand if I saw the process as apposed to just the finished product, but I'm trying my best haha. I'm glad I have it to slowly, but surely learn from it.
I haven't looked at it yet, but the article said he mostly uses master materials.
I'm wondering if you guys had an idea about a problem I'm currently facing. I'm trying to do some triplanar projection mapping in worldspace. I just can't figure out the proper way to compute my normals properly. Since I'm in worldspace, I can't rely on the data provided by the mesh using the shader.
As far I'm aware, there is no way to recompute tangent/binormals manually in a shader in the Unreal Engine, anybody know an alternative ? I found the math here : http://www.terathon.com/voxels/
I'm wondering if you guys had an idea about a problem I'm currently facing. I'm trying to do some triplanar projection mapping in worldspace. I just can't figure out the proper way to compute my normals properly. Since I'm in worldspace, I can't rely on the data provided by the mesh using the shader.
As far I'm aware, there is no way to recompute tangent/binormals manually in a shader in the Unreal Engine, anybody know an alternative ? I found the math here : http://www.terathon.com/voxels/
However I don't see how to apply it.
if you find a way to get this into unreal4 please tell it!
I'm wondering if you guys had an idea about a problem I'm currently facing. I'm trying to do some triplanar projection mapping in worldspace. I just can't figure out the proper way to compute my normals properly. Since I'm in worldspace, I can't rely on the data provided by the mesh using the shader.
As far I'm aware, there is no way to recompute tangent/binormals manually in a shader in the Unreal Engine, anybody know an alternative ? I found the math here : http://www.terathon.com/voxels/
However I don't see how to apply it.
You're looking for this material functions (or just right click in the mat editor and type worldaligned in the search bar)
Be warned, doing normals 'properly' in tri-planar can get expensive quick!
Remember you have to plug a TextureObject into the texture slot on these.
Exactly what I was looking for, thank you !
Perfect normal mapping is far from being my goal, especially since it's only to texture some parts of my environments (like certain organic walls).
I'm currently stuck on importing an animated mesh. It's just a simple group of a few objects in Maya with a rotation on the group itself, but no matter how I try to import it I just can't get it to work. Has anyone an idea what I could be doing wrong?
I'm currently stuck on importing an animated mesh. It's just a simple group of a few objects in Maya with a rotation on the group itself, but no matter how I try to import it I just can't get it to work. Has anyone an idea what I could be doing wrong?
You are trying to import them as a static mesh with animation? You either need a skeletal mesh and animation sets... or, it shouldn't be that hard to create a blueprint if it's just a simple rotation. The only way you are going to move a static mesh is with a blueprint.
Thanks to everybody who responded to the item socket problem I was having a week ago, turns out the issue was just related to the fact that my character was set up incorrectly (had a duplicate object in the Components tab named "Mesh" so the Blueprint I made was pointing to the wrong thing.)
I've come with another noob question, though. Does anybody have any idea how one could go about fixing this "floaty" movement problem? I have this problem with my own character so I figured I would dig through the animation content examples but it seems like the issue is there as well. If you try walking on a ramp the same thing happens.
I have minimal experience with animation but I thought an inverse kinematic set up was supposed to fix this, as it seems that way from the documentation screenshot, but again I have very little idea about what I'm doing here lol.
are you animating the object/group? or do you have geometry weighted to bones?
Since there are a hand full of objects I tried to animate the group itself with simple keyframes, but I also tried to rig the group and multiple objects at once with bones. Nothing worked out.
You are trying to import them as a static mesh with animation? You either need a skeletal mesh and animation sets... or, it shouldn't be that hard to create a blueprint if it's just a simple rotation. The only way you are going to move a static mesh is with a blueprint.
See above, I tried to import them as static mesh and skeletal mesh. Neither worked for me.
But I'll try if I can figure out how to rotate them with blueprints, if that works out I have to find a way to add the rotation to multiple objects at once.
I convert my project from 4.1 to 4.2 but I have not so loooow FPS and I dont know why. Befor on 4.1 I've got great FPS. I use alienware laptop and I dont have any problems with low FPS in UE4.
Can you help me with this problem please? I try also to but details for project to low details but nothing happens. Just maby +4 fps
I also try sample projects but it is the same on 4.2 is low fps what ever I try. Maby I make something wrong.
I convert my project from 4.1 to 4.2 but I have not so loooow FPS and I dont know why. Befor on 4.1 I've got great FPS. I use alienware laptop and I dont have any problems with low FPS in UE4.
Can you help me with this problem please? I try also to but details for project to low details but nothing happens. Just maby +4 fps
I also try sample projects but it is the same on 4.2 is low fps what ever I try. Maby I make something wrong.
I convert my project from 4.1 to 4.2 but I have not so loooow FPS and I dont know why. Befor on 4.1 I've got great FPS. I use alienware laptop and I dont have any problems with low FPS in UE4.
Can you help me with this problem please? I try also to but details for project to low details but nothing happens. Just maby +4 fps
I also try sample projects but it is the same on 4.2 is low fps what ever I try. Maby I make something wrong.
Ever since I upgraded my project o 4.2, some of my textures have such a low resolution. But when I open the static mesh editor of any mesh, the resolution goes back up to normal, however, I have to keep the static mesh editor open. As soon as I close it, the resolution goes back down. Not sure what's going on.
EDIIT: Weird, in order to fix this, I just have to open and close any random associated material in the material editor, and it fixes everything... *shrug*
I'd love to have more realtime lighting solutions for UE4. Is that something that works with non-baked lighting?
Yes DFAO, works with dynamic lighting and for now, static meshes only (albeit you can move them in game or remove or whatever).
It doesn't work with Landscape (which is serious limitation for now), and BSP.
Skylight GI is only for Stationary Skylight.
I recommend to watch stream where Daniel presents it in real time. The difference it makes for scene is just day and night.
If there would be solution for Large Scale AO to complement the Medium (DFAO) and Small( SSAO), it would probably melt brain completely.
This AO technique (it's GI actually, but can be used just for AO), seems more efficient. But it's probably locked only to NVIDIA GPUs and have problems like every other Volume based rendering technique.
Leaking trough thin objects.
This AO technique (it's GI actually, but can be used just for AO), seems more efficient. But it's probably locked only to NVIDIA GPUs and have problems like every other Volume based rendering technique.
This is remarkably similar to SVOGI, and suffers from all the same problems. It's performance is WAY worse than what Epic are currently using; the stats are in the document.
AO only gives 14.3ms on a GTX650, and with GI on top a whopping 28.1ms.
If you're running at 60fps, your AO is taking up 87% of your total time to render a frame. That's batshit insane. With GI, you can't do 60fps, and at 30fps you're still using 85% of time to render a frame on it.
NVidia's conclusion:
[quote="PRACTICAL REAL-TIME VOXEL-BASED GLOBAL
LLUMINATION FOR CURRENT GPUS"]Fully dynamic GI becomes practical on mainstream GPUs
Low end GPUs can benefit from much higher-quality AO[/quote]
This is remarkably similar to SVOGI, and suffers from all the same problems. It's performance is WAY worse than what Epic are currently using; the stats are in the document.
AO only gives 14.3ms on a GTX650, and with GI on top a whopping 28.1ms.
If you're running at 60fps, your AO is taking up 87% of your total time to render a frame. That's batshit insane. With GI, you can't do 60fps, and at 30fps you're still using 85% of time to render a frame on it.
NVidia's conclusion:
My conclusion:
...on the upside it isn't locked to Nvidia GPUs.
:thumbup:
The karl pinkerton picture cemented that post as legendary.
This is remarkably similar to SVOGI, and suffers from all the same problems. It's performance is WAY worse than what Epic are currently using; the stats are in the document.
AO only gives 14.3ms on a GTX650, and with GI on top a whopping 28.1ms.
If you're running at 60fps, your AO is taking up 87% of your total time to render a frame. That's batshit insane. With GI, you can't do 60fps, and at 30fps you're still using 85% of time to render a frame on it.
Wait. You sayin that there are people who are using GTX 650 for gaming ? Insanity!
According to these slides for GTX 770 AO is taking about 3 ms.
And it seems that data strcutres they are using to hold information, about scene is more efficient than sparse voxels.
To further my original post - the GTX650 isn't an unreasonable benchmark, as it's performance lies somewhere between the Xbox One and Ps4's graphics sets - as most games are multiplatform, in order to maintain graphical coherency, you'd want that feature set to be as similar as possible across all platforms - especially where things as critical as lighting may be concerned since that can be both time consuming to iterate on and have a high impact on the look and feel of a game.
The GTX770 on the other hand, the performance is reasonable, but still pretty expensive given that it's still 20% of your available rendering time being thrown at ambient occlusion (45% including basic GI)!
It could work on a PC only title, targeting top-tier hardware, in a few years time.
More to the point though, if these guys are calling a GTX650 'low-end' and a GTX770 / Titan 'mainstream' - what the effing fuck is a top-tier card ?
It'll probably be something like 5 years before the average gaming pc has something as powerful as a gtx 770, really far away, and since this feature isn't as easy to implement as ssao and it's variations, you probably won't see it it in games for a long time.
DFAO is also moronically expensive. We clocked it as taking a full 5ms of execution time on fairly modern hardware, which is huge.
The rendering programmer on the stream said that it was pretty unoptimized for the time being. Do you think they can get it down to reasonable low end perf later on?
The rendering programmer on the stream said that it was pretty unoptimized for the time being. Do you think they can get it down to reasonable low end perf later on?
I think there is a good chance they will, considering they REALLY need something like this to make their own game, fortnight, pop. As it stands, fortnight looks very flat.
It'll probably be something like 5 years before the average gaming pc has something as powerful as a gtx 770, really far away, and since this feature isn't as easy to implement as ssao and it's variations, you probably won't see it it in games for a long time.
People who have medium-low end hardware are not the ones who are interested in playing A+ (in terms of visual quality) games on PC, they do so on consoles.
The rendering programmer on the stream said that it was pretty unoptimized for the time being. Do you think they can get it down to reasonable low end perf later on?
I genuinely have no idea. Sometimes stuff can be optimised, sometimes optimisation just fails to give the results you need. It'd be nice if they can get it down to something more workeable.
because they've used (assumingly) an in-house engine for all their games until now..
I'd rather see them continue using an in-house engine rather than going for a third party one..
I prefer companys having their own engines and what not rather than everyone relying on the same one in the end..
global federation of unreal engine?
also, the pathetic performance of UE3 on consoles last gen with texture popins is also a reason why I am not hyped for UE4 or whatevs.. also, the pc-console comparisons of UE4 just showed that consoles are not powerful enough to be all dat..
well, the choice to use a 3rd party engine comes down to cost.
if it costs them less in terms of both dev time and financial investment to use a third party, of course they're going to.
and since now they can just pay $20 per person in their studio ONCE, and build everything they need/want with that version (bearing in mind they have access to the source code so they can modify anything and everything they need to), that's going to be WAY cheaper than developing an engine in-house. and a 5% royalty is pennies to them.
doubt that they'd develop an in-house engine from scratch..
granted, harada mentioned that they needed to make huge overhauls/rework from scratch to the engine from tekken 6 to tag 2....
Also UE4 makes it more likely that PC could get a Tekken release.
Well they stated already they picked it because they can ship it at pretty much all platforms they desire, and i really hope Tekken 7 will land on PC , shut up and take my money.
UE 4.3 is out, and the Substance Plugin is now free for everyone and does not require a recompile of the engine, simply unzip it in your project and you're good to go!
Replies
https://www.unrealengine.com/blog/new-release-on-marketplace-scifi-hallway
Download ahoy!
Article on PC gamer about some UE4 levels including Wiktor Ohman's sci-fi hallway
I've been a Polycount lurker for a bazillion jillion years, but I finally registered (or maybe I registered years ago and just forgot) to address this post.
I guess I'm mostly confused by this reasoning; nothing has really changed in the year or so between UDK's last meaningful release and UE4. Well okay, tons of things have changed, but not really in the way content is created. Things have changed drastically since UE2 - we've gone from BSP dependency to mesh dependency - but not so much in this area.
Texture density is an incredibly useful diagnostic tool. The things mentioned here as reasons for why this diagnostic tool is no longer available are themselves part of the reason it was useful. So either I'm having an exceptionally bad brain fart moment, or this was the equivalent of "because cars now have so many options for on board computers, we no longer feel it necessary to have a device that tests the functionality of on board computers."
Maybe we're just using texture density for different things
I haven't looked at it yet, but the article said he mostly uses master materials.
As far I'm aware, there is no way to recompute tangent/binormals manually in a shader in the Unreal Engine, anybody know an alternative ? I found the math here : http://www.terathon.com/voxels/
However I don't see how to apply it.
if you find a way to get this into unreal4 please tell it!
You're looking for this material functions (or just right click in the mat editor and type worldaligned in the search bar)
MaterialFunction'/Engine/Functions/Engine_MaterialFunctions01/Texturing/WorldAlignedNormal.WorldAlignedNormal'
here's the non-normal version:
MaterialFunction'/Engine/Functions/Engine_MaterialFunctions01/Texturing/WorldAlignedTexture.WorldAlignedTexture'
Be warned, doing normals 'properly' in tri-planar can get expensive quick!
Remember you have to plug a TextureObject into the texture slot on these.
Exactly what I was looking for, thank you !
Perfect normal mapping is far from being my goal, especially since it's only to texture some parts of my environments (like certain organic walls).
I'm currently stuck on importing an animated mesh. It's just a simple group of a few objects in Maya with a rotation on the group itself, but no matter how I try to import it I just can't get it to work. Has anyone an idea what I could be doing wrong?
You are trying to import them as a static mesh with animation? You either need a skeletal mesh and animation sets... or, it shouldn't be that hard to create a blueprint if it's just a simple rotation. The only way you are going to move a static mesh is with a blueprint.
I've come with another noob question, though. Does anybody have any idea how one could go about fixing this "floaty" movement problem? I have this problem with my own character so I figured I would dig through the animation content examples but it seems like the issue is there as well. If you try walking on a ramp the same thing happens.
I have minimal experience with animation but I thought an inverse kinematic set up was supposed to fix this, as it seems that way from the documentation screenshot, but again I have very little idea about what I'm doing here lol.
Thanks!
Since there are a hand full of objects I tried to animate the group itself with simple keyframes, but I also tried to rig the group and multiple objects at once with bones. Nothing worked out.
See above, I tried to import them as static mesh and skeletal mesh. Neither worked for me.
But I'll try if I can figure out how to rotate them with blueprints, if that works out I have to find a way to add the rotation to multiple objects at once.
Is there a way to overlay wireframe over texture in UE4 ?
update : found it
Show >>>> advanced >>>> mesh edges
I didn't know that it's necessary to seperately import the bones as an animation. anyway it works now. Thanks guys!
I convert my project from 4.1 to 4.2 but I have not so loooow FPS and I dont know why. Befor on 4.1 I've got great FPS. I use alienware laptop and I dont have any problems with low FPS in UE4.
Can you help me with this problem please? I try also to but details for project to low details but nothing happens. Just maby +4 fps
I also try sample projects but it is the same on 4.2 is low fps what ever I try. Maby I make something wrong.
Thank you for all who will help me
I convert my project from 4.1 to 4.2 but I have not so loooow FPS and I dont know why. Befor on 4.1 I've got great FPS. I use alienware laptop and I dont have any problems with low FPS in UE4.
Can you help me with this problem please? I try also to but details for project to low details but nothing happens. Just maby +4 fps
I also try sample projects but it is the same on 4.2 is low fps what ever I try. Maby I make something wrong.
Thank you for all who will help me
Hi,
I convert my project from 4.1 to 4.2 but I have not so loooow FPS and I dont know why. Befor on 4.1 I've got great FPS. I use alienware laptop and I dont have any problems with low FPS in UE4.
Can you help me with this problem please? I try also to but details for project to low details but nothing happens. Just maby +4 fps
I also try sample projects but it is the same on 4.2 is low fps what ever I try. Maby I make something wrong.
Thank you for all who will help me
In other update news, Mikktspace is finally in 4.2!
EDIIT: Weird, in order to fix this, I just have to open and close any random associated material in the material editor, and it fixes everything... *shrug*
https://www.unrealengine.com/blog/unreal-engine-43-update-preview
Skylights also have 1 bounce of GI in lightmass.
Yes DFAO, works with dynamic lighting and for now, static meshes only (albeit you can move them in game or remove or whatever).
It doesn't work with Landscape (which is serious limitation for now), and BSP.
Skylight GI is only for Stationary Skylight.
I recommend to watch stream where Daniel presents it in real time. The difference it makes for scene is just day and night.
If there would be solution for Large Scale AO to complement the Medium (DFAO) and Small( SSAO), it would probably melt brain completely.
http://on-demand.gputechconf.com/gtc/2014/presentations/S4552-rt-voxel-based-global-illumination-gpus.pdf
This AO technique (it's GI actually, but can be used just for AO), seems more efficient. But it's probably locked only to NVIDIA GPUs and have problems like every other Volume based rendering technique.
Leaking trough thin objects.
This is remarkably similar to SVOGI, and suffers from all the same problems. It's performance is WAY worse than what Epic are currently using; the stats are in the document.
AO only gives 14.3ms on a GTX650, and with GI on top a whopping 28.1ms.
If you're running at 60fps, your AO is taking up 87% of your total time to render a frame. That's batshit insane. With GI, you can't do 60fps, and at 30fps you're still using 85% of time to render a frame on it.
NVidia's conclusion:
[quote="PRACTICAL REAL-TIME VOXEL-BASED GLOBAL
LLUMINATION FOR CURRENT GPUS"]Fully dynamic GI becomes practical on mainstream GPUs
Low end GPUs can benefit from much higher-quality AO[/quote]
My conclusion:
...on the upside it isn't locked to Nvidia GPUs.
:thumbup:
The karl pinkerton picture cemented that post as legendary.
Wait. You sayin that there are people who are using GTX 650 for gaming ? Insanity!
According to these slides for GTX 770 AO is taking about 3 ms.
And it seems that data strcutres they are using to hold information, about scene is more efficient than sparse voxels.
The GTX770 on the other hand, the performance is reasonable, but still pretty expensive given that it's still 20% of your available rendering time being thrown at ambient occlusion (45% including basic GI)!
It could work on a PC only title, targeting top-tier hardware, in a few years time.
More to the point though, if these guys are calling a GTX650 'low-end' and a GTX770 / Titan 'mainstream' - what the effing fuck is a top-tier card
The rendering programmer on the stream said that it was pretty unoptimized for the time being. Do you think they can get it down to reasonable low end perf later on?
I think there is a good chance they will, considering they REALLY need something like this to make their own game, fortnight, pop. As it stands, fortnight looks very flat.
People who have medium-low end hardware are not the ones who are interested in playing A+ (in terms of visual quality) games on PC, they do so on consoles.
I genuinely have no idea. Sometimes stuff can be optimised, sometimes optimisation just fails to give the results you need. It'd be nice if they can get it down to something more workeable.
I'd rather see them continue using an in-house engine rather than going for a third party one..
I prefer companys having their own engines and what not rather than everyone relying on the same one in the end..
global federation of unreal engine?
also, the pathetic performance of UE3 on consoles last gen with texture popins is also a reason why I am not hyped for UE4 or whatevs.. also, the pc-console comparisons of UE4 just showed that consoles are not powerful enough to be all dat..
if it costs them less in terms of both dev time and financial investment to use a third party, of course they're going to.
and since now they can just pay $20 per person in their studio ONCE, and build everything they need/want with that version (bearing in mind they have access to the source code so they can modify anything and everything they need to), that's going to be WAY cheaper than developing an engine in-house. and a 5% royalty is pennies to them.
granted, harada mentioned that they needed to make huge overhauls/rework from scratch to the engine from tekken 6 to tag 2....
Also UE4 makes it more likely that PC could get a Tekken release.
Well they stated already they picked it because they can ship it at pretty much all platforms they desire, and i really hope Tekken 7 will land on PC , shut up and take my money.
RTT blueprint!
Paper2d!
Distance Field AO!
Integrated Twitch streaming! (WIP)!
https://www.unrealengine.com/blog/unreal-engine-43-release
looooottsss