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..
That was 1 year ago, back when PS4 had 2GB of ram instead of the 8GB.
UE4 also didn't get finished till March 2014 and even now it receives monthly updates.
We obviously have to wait for more UE4 games to come out before we can judge the differences.
Looking to get UE4 soon and start putting some scenes together in it, but I was wondering if someone can tell me how big it is to download? I'm on a monthly bandwidth limit is all. I remember UDK was only like 2 gigs or so. Is UE4 about the same?
I think it's more 3.5/4 Gb during the download because it's compressed, and 8/9 Gb once installed. The GitHub version is less than 2GB, but you have to compile everything by yourself (no binaries).
I've been having trouble getting good quality screenshots from unreal. I've been using the "HighResShot" console command, the problem is that the lighting values in the screenshot don't match how the scene is in the editor at all. For some reason the screenshot has extremely high contrast levels, so areas that have some decent GI turn out completely black in the screenshots, anyone mind sharing how the take their screenshots?
Newbie question please.. if I want to learn how to use the engine with UDK, how hard would it be to move to UE4 in the future? Or should I start with UE4 right away?
Hey guys,
anyone mind sharing how the take their screenshots?
+ 1
@DireWolf I'd recommend starting out with UE4 right off the bat. If you still want to experience Unreal and can't purchase UE4 at the moment, UDk is still viable. The transition is natural. I started out with UDK, switched to CryEngine for a bit, and when I picked up UE4, it felt like I was re-learning with a few changes rather than learning something entirely new.
I've been having trouble getting good quality screenshots from unreal. I've been using the "HighResShot" console command, the problem is that the lighting values in the screenshot don't match how the scene is in the editor at all. For some reason the screenshot has extremely high contrast levels, so areas that have some decent GI turn out completely black in the screenshots, anyone mind sharing how the take their screenshots?
Yeah I think HighResShot command is bugged right now. It seems it doesn't capture Anti-aliasing. What I do is I go to the Quick Settings and check "Hide Viewport UI" ckeck box (though it's bugged in 4.2.1 and doesn't hide anything) and then I press F11 to maximize the viewport and then I press Shift+F11 to go into full screen mode. Press G to enable game view mode and then I just use "Print Screen" button on the key board and paste the shots into Potatoshop
The HighResShot command was working for me under 4.0.2. With AA even if it's not really needed since you can do a good downsampling by cranking up the resolution.
Maybe the contrast problem came from the eye adaption ? Using a fixed value could resolve the problem (it would require fixed values in a post-process volume for that).
I use the HighResShot command. If you type a number after it, you get a higher res. I usually do HighResShot 3 and then downsample in Photoshop.
How far off are your lighting values? I have noticed that the screenshot tools and especially the movie capture in Matinee sometimes have a hard time with post processed effects like screenspace reflections, so maybe Froyok is right with the Eye Adaptation issue. Have you tried locking the exposure inside your post process volume?
@Froyok,@Harbinger I did remove the eye adaptation by setting the min and max values to 1. That's how i usually work as i find the eye adaptation annoying to when im lighting/ set dressing. I can post the screenshot when i get back back to my computer, everything looks extremely sharp and the exposure is really high
@Yuri , Thanks for the idea, im gonna give it a go and see if i get better results.
I updated a project of mine to 4.3. Not sure what it did to the reflections but I used to have a blurry but reflective floor surface.
And now it's basically a mirror. I didn't change any roughness values.
When I tried toning my roughness down, it made it less reflective, but it still keeps the clarity of the reflection? Is this how it's intended to work now?
I'm curious. I found many posts mentioning "subscribe UE4 and cancel it until you need update" or similar. What happen when you un-subscribe? You still get to develop your games in it forever?
I updated a project of mine to 4.3. Not sure what it did to the reflections but I used to have a blurry but reflective floor surface.
And now it's basically a mirror. I didn't change any roughness values.
When I tried toning my roughness down, it made it less reflective, but it still keeps the clarity of the reflection? Is this how it's intended to work now?
do u have sRGB turned off on everything except ur albedo texture?
I had a somewhat similar problem and this rectified the problem, correct me if im wrong but i think it has something to do with gamma correction on the textures?
They reduced the quality of reflections in the latest update. You can change this back to a softer look by going into your post volume and cranking up the quality of screen space reflections to 100. Just keep in mind that it's more expensive. Hope that helps!
Hi everyone.
Trying to build a simple room for a mobile platform (ipad) and on a mobile preview it has this banding. Almost like jpeg artifacts. It also appears on ipad.
Editor view:
Mobile preview:
Its weird as it mostly shows up on walls and not on worktops, floor or sofa.
Any ideas how to solve it?
They reduced the quality of reflections in the latest update. You can change this back to a softer look by going into your post volume and cranking up the quality of screen space reflections to 100. Just keep in mind that it's more expensive. Hope that helps!
O wow thanks! I just hate how at certain angles reflections or so damn grainy. But I guess it just has to do with lighting and such because I've seen plenty of environments that don't have that issue, or at least it's hidden well.
I have question. How can i change the scalar parameter of a post process blendable inside blueprints. I tried making it a MID, but apparently this doesnt work for postprocess blendables..anybody have any clues how to do this.
I also tried applying the blendable to the Player camera and accessing it that way, but i was unable to access the scalar parameter..Any insight would be greatly appreciated.
You need to make a MID indeed, but don't forget to assign your MID to the list of Blendable materials after (and remove the original, otherwise you would still get a MIC). I'm doing this and it works (but in C++). The process in blueprint is identical.
In fact I did it in Blueprint first for doing some tests. Then replicated the basics to affect all the volumes in my scene.
Just a general question though..For something like a scoped view for a rifle when you right click. Is it better to do that with Postprocess or use the HUD?
I didn't touch any camera, but I presume the behavior is identical.
by removing the original I mean : since you create a MID from your MIC, you get a new material, this one is dynamic and that's great. However as is it has no effect, to see the result you need to add it to your array of blendable materials. If you have it, you will end with two time the same shaders, one being dynamic and the other one static. So remove the first and keep only the dynamic one.
You can do this setup during a Begin Map event for example. In my case a did something a bit more advanced to do it during each tick, this way I can easily manage streamed map (and use only one global MID).
Just a general question though..For something like a scoped view for a rifle when you right click. Is it better to do that with Postprocess or use the HUD?
I think a post-process is better since they are shaders optimized for doing screen-space effects (if I remember correctly the blendables are merged together).
Note also that doing it with the HUD wouldn't allow you to distort the scene easily (like simulating a lens distortion) and you will probably get a large amount of overdraw which can be very expensive.
Im having an issue with the behaviour trees in U4. Whenever i try to set a blackboard decorator on my selector node and try to change the key type to ENUM, my editor crashes. Anybody know what the deal is? I even tried setting SelfActor as the first key in my blackboard.
Try grabbing the 4.4 preview which is now available through the launcher. That might work more reliably. I believe behavior trees come out of 'experimental' status in 4.4 ...
Try grabbing the 4.4 preview which is now available through the launcher. That might work more reliably. I believe behavior trees come out of 'experimental' status in 4.4 ...
Wow thats great news. I can definitely see how useful behaviour trees are gonna be.. but these crashes have been making me extremely frustrated.
EDIT: I got it to compile, the issue was because of my plugin folder, but the editor still crashes when i try to set an ENUM type key in the decorator.
How exactly does the distance field AO work? When I first saw the feature I assumed it was something like Ambient Occlusion Fields, but the UE4 documentation says its DX11 only, and people here have said its really slow. The method in that Ambient Occlusion Field PDF is super cheap to render. Its about equivalent to rendering a deferred light. Is UE4s implementation doing something crazy like ray marching?
I have mostly-implemented something similar to whats in the Ambient Occlusion Field pdf (in unity), and plan to use on mobile devices in very small controlled scenes.
I noticed that the updates dont happen immediately in the UE4 video which makes me think its not a direct deferred splat style application.
How exactly does the distance field AO work? When I first saw the feature I assumed it was something like Ambient Occlusion Fields, but the UE4 documentation says its DX11 only, and people here have said its really slow. The method in that Ambient Occlusion Field PDF is super cheap to render. Its about equivalent to rendering a deferred light. Is UE4s implementation doing something crazy like ray marching?
I have mostly-implemented something similar to whats in the Ambient Occlusion Field pdf (in unity), and plan to use on mobile devices in very small controlled scenes.
I noticed that the updates dont happen immediately in the UE4 video which makes me think its not a direct deferred splat style application.
I thought the distance field AO was just pre-calculated AO used with dynamic lighting.
Distance field AO is fully dynamic, although it doesn't work with characters yet. I'm not sure exactly how it works.
Not fully dynamic, if they are similar to the technique in the infamous 2 paper then they would be a offline bake process to generate the fields and then those would be splatted into the frame buffer dynamically, similar to deferred lights. I suppose the fields could be generated pretty quickly with dx11 using direct triangle voxelization and compute shaders. With my implementation I use a many shadow map approach to approximate voxel visibility and average geometry direction, which can be done is about 20ms with very high quality (400 shadow maps).
^Yeah, it takes samples over multiple frames, too, in order to clean up the result and make it less noisy. By dynamic, I meant it doesn't need any extra preparation or work from artists or content creators.
It can be. It is a much higher quality version of the subsurface model, so you should use it sparingly as it has serious performance implications.
It's actually the reverse, pre-integrated is cheaper.
The Preintegrated Skin Shading Model is very similar in nature to the Subsurface model, but geared toward low performance cost skin rendering on human characters. While not physically perfect, this Shading Model will often give nice results on your characters while being a bit cheaper to render than the Subsurface method.
Jerc, we used material layers on every asset in the Infiltrator demo, most environmental assets only had a couple layers and it still proved useful for them to have them. Also each layer itself is not complex, Most of them are a few tiling textures and some values, some are even solid colors for the base color.
One huge benefit of NOT baking these layers is you get really high resolution. Each layer has tiling textures so you're not limited to a baked resolution.
Oh I see. Yeah that makes sense. So you could have 256 layers in just 1 channel really.
So I'm running into this problem right now and I can't find anything in the documentation about it. Searched Google and it sent me to this thread.
How would you use more than 4 masks in a Color ID? I have a object with 6 materials (and I'm sure I will run into objects with more) and would like to figure out how to do the masking properly without having multiple Color ID maps for 1 model.
How would you use more than 4 masks in a Color ID? I have a object with 6 materials (and I'm sure I will run into objects with more) and would like to figure out how to do the masking properly without having multiple Color ID maps for 1 model.
I was playing around with that the other day and created a material function that gives you 8 channels (R,G,B,C,M,Y,K,W). There might be a better/more efficient way but here's what I cooked up:
Thank you for the reply. This should work for most of the assets in my scene. I posted a similar question on the UE4 forums so hopefully someone replies there as well.
So i've been playing around with the experimental UMG system that released with 4.4.0..does this essentially render the HUD Blueprint obsolete, as it seems easier and more practical to add widgets to the viewport from the player controller blueprint. Any insight would be helpful.
So I've been going back and forth a bit on the UE4 Answer Hub with no luck about an issue/idea, so maybe one of the Polycounters will be able to help. Here's the thread:
Basically there's a big visual gap between static and stationary lights, as static lights lose the high quality deferred lighting specular of stationary lights. I have a hack where I use a dummy sphere with an emissive shader with my light, so that the glowing sphere gets captured in a reflection capture actor and gives a nice specular to metallic surfaces, even with a static light. Basically, the light source itself gets captured, not just the light's interaction with the rest of the environment.
I'd like to set up a construction script blueprint to automate creating these emissive dummy spheres but I can't figure out how to hide the dummy geo during gameplay. Disabling the "Actor Hidden In Game" attribute also hides the geo from the reflection capture.
I would try to have them be deleted first thing that happens after the game starts. The game might prevent you from messing with meshes flagged as static though so not sure how easy this would be to achieve.
Another option might be to do some magic in the material such that it will no longer render when in game. Something like a masked material with a dynamic toggle so you can hide it in game.
Any particular reason why you can't use the stationary lights?
Replies
UE4 also didn't get finished till March 2014 and even now it receives monthly updates.
We obviously have to wait for more UE4 games to come out before we can judge the differences.
thanks.
I've been having trouble getting good quality screenshots from unreal. I've been using the "HighResShot" console command, the problem is that the lighting values in the screenshot don't match how the scene is in the editor at all. For some reason the screenshot has extremely high contrast levels, so areas that have some decent GI turn out completely black in the screenshots, anyone mind sharing how the take their screenshots?
@DireWolf I'd recommend starting out with UE4 right off the bat. If you still want to experience Unreal and can't purchase UE4 at the moment, UDk is still viable. The transition is natural. I started out with UDK, switched to CryEngine for a bit, and when I picked up UE4, it felt like I was re-learning with a few changes rather than learning something entirely new.
Yeah I think HighResShot command is bugged right now. It seems it doesn't capture Anti-aliasing. What I do is I go to the Quick Settings and check "Hide Viewport UI" ckeck box (though it's bugged in 4.2.1 and doesn't hide anything) and then I press F11 to maximize the viewport and then I press Shift+F11 to go into full screen mode. Press G to enable game view mode and then I just use "Print Screen" button on the key board and paste the shots into Potatoshop
Maybe the contrast problem came from the eye adaption ? Using a fixed value could resolve the problem (it would require fixed values in a post-process volume for that).
How far off are your lighting values? I have noticed that the screenshot tools and especially the movie capture in Matinee sometimes have a hard time with post processed effects like screenspace reflections, so maybe Froyok is right with the Eye Adaptation issue. Have you tried locking the exposure inside your post process volume?
@Yuri , Thanks for the idea, im gonna give it a go and see if i get better results.
btw im using 4.3
And now it's basically a mirror. I didn't change any roughness values.
When I tried toning my roughness down, it made it less reflective, but it still keeps the clarity of the reflection? Is this how it's intended to work now?
you can develop forever if you want to.
do u have sRGB turned off on everything except ur albedo texture?
I had a somewhat similar problem and this rectified the problem, correct me if im wrong but i think it has something to do with gamma correction on the textures?
Trying to build a simple room for a mobile platform (ipad) and on a mobile preview it has this banding. Almost like jpeg artifacts. It also appears on ipad.
Editor view:
Mobile preview:
Its weird as it mostly shows up on walls and not on worktops, floor or sofa.
Any ideas how to solve it?
I also tried applying the blendable to the Player camera and accessing it that way, but i was unable to access the scalar parameter..Any insight would be greatly appreciated.
In fact I did it in Blueprint first for doing some tests. Then replicated the basics to affect all the volumes in my scene.
What do you mean by remove the original? are using the player camera blendables or the volume in the level?
EDIT:@Froyok aah never mind i got it working. Thanks for the help mate.
by removing the original I mean : since you create a MID from your MIC, you get a new material, this one is dynamic and that's great. However as is it has no effect, to see the result you need to add it to your array of blendable materials. If you have it, you will end with two time the same shaders, one being dynamic and the other one static. So remove the first and keep only the dynamic one.
You can do this setup during a Begin Map event for example. In my case a did something a bit more advanced to do it during each tick, this way I can easily manage streamed map (and use only one global MID).
[EDIT]
I think a post-process is better since they are shaders optimized for doing screen-space effects (if I remember correctly the blendables are merged together).
Note also that doing it with the HUD wouldn't allow you to distort the scene easily (like simulating a lens distortion) and you will probably get a large amount of overdraw which can be very expensive.
Can you tell me how the Road decal works, or point me to the help file?
Here is a video of it in use. I noticed the artists applied it to rocks, fence and other objects.
[ame]www.youtube.com/watch?v=MlgvfEicdwU[/ame]
Also how does he keep the camera open bottom right? I find when I click off the camera it disappears
Landscape Splines
https://docs.unrealengine.com/latest/INT/Engine/Landscape/Editing/Splines/index.html
Cool beans, thanks
Im using 4.3.1 btw
Wow thats great news. I can definitely see how useful behaviour trees are gonna be.. but these crashes have been making me extremely frustrated.
EDIT: I got it to compile, the issue was because of my plugin folder, but the editor still crashes when i try to set an ENUM type key in the decorator.
I have mostly-implemented something similar to whats in the Ambient Occlusion Field pdf (in unity), and plan to use on mobile devices in very small controlled scenes.
I noticed that the updates dont happen immediately in the UE4 video which makes me think its not a direct deferred splat style application.
I thought the distance field AO was just pre-calculated AO used with dynamic lighting.
It can be. It is a much higher quality version of the subsurface model, so you should use it sparingly as it has serious performance implications.
Not fully dynamic, if they are similar to the technique in the infamous 2 paper then they would be a offline bake process to generate the fields and then those would be splatted into the frame buffer dynamically, similar to deferred lights. I suppose the fields could be generated pretty quickly with dx11 using direct triangle voxelization and compute shaders. With my implementation I use a many shadow map approach to approximate voxel visibility and average geometry direction, which can be done is about 20ms with very high quality (400 shadow maps).
It's actually the reverse, pre-integrated is cheaper.
https://docs.unrealengine.com/latest/INT/Engine/Rendering/Materials/MaterialProperties/LightingModels/index.html#preintegratedskin
So I'm running into this problem right now and I can't find anything in the documentation about it. Searched Google and it sent me to this thread.
How would you use more than 4 masks in a Color ID? I have a object with 6 materials (and I'm sure I will run into objects with more) and would like to figure out how to do the masking properly without having multiple Color ID maps for 1 model.
I was playing around with that the other day and created a material function that gives you 8 channels (R,G,B,C,M,Y,K,W). There might be a better/more efficient way but here's what I cooked up:
https://answers.unrealengine.com/questions/85570/static-lights-lose-colored-specular-on-metallic-ma.html
Basically there's a big visual gap between static and stationary lights, as static lights lose the high quality deferred lighting specular of stationary lights. I have a hack where I use a dummy sphere with an emissive shader with my light, so that the glowing sphere gets captured in a reflection capture actor and gives a nice specular to metallic surfaces, even with a static light. Basically, the light source itself gets captured, not just the light's interaction with the rest of the environment.
I'd like to set up a construction script blueprint to automate creating these emissive dummy spheres but I can't figure out how to hide the dummy geo during gameplay. Disabling the "Actor Hidden In Game" attribute also hides the geo from the reflection capture.
Any ideas?
Another option might be to do some magic in the material such that it will no longer render when in game. Something like a masked material with a dynamic toggle so you can hide it in game.
Any particular reason why you can't use the stationary lights?