the shadows for the smaller grass is looking really strange. it looks like the shadow bias is a bit of and the resolution might be a bit low or something, what ever it is, its taking away some of the amazing realism thats present in all theother screenshots.
You're using a skylight with a cubemap and the ambient cubemap in the PPV? Have you looked at anything using the Subsurface Profile shading yet? The ambient cubemap option doesn't seem to work well with it and relatively low values end up blowing out the skin. Just a warning
Thanks all leleuxart I have a cubemap in PPV and sometimes in Skylight as well, but I tend to use a cubemap intensity of 1 and use very little skylight intensity because flat ambient light is absolutely useless. For some reason using a cubemap in Skylight doesn't exactly work as IBL so that's why I try to stay away from Skylight. I haven't paid attention to character skin so far when using a cubemap in PPV. I'm gonna take a look and if cubemaps don't work there then we have a problem. But I might use a different shading for skin if that's the case, because dropping the IBL makes the entire game world look rather cheap.
Is There any twitter/facebook page about the game i can follow?! Any new on how we can expect a trailer? ^^ I'm a big fan of WW2 And this project blow my mind
Is There any twitter/facebook page about the game i can follow?! Any new on how we can expect a trailer? ^^ I'm a big fan of WW2 And this project blow my mind
Search ourghostsofwar of Twitter. A trailer is inevitable of course but we're still working on the game map. Thanks.
Since they haven't updated this, from what I was told by others I worked with on the project, they decided to start choosing when they wanted to pay for artists work, and some artists dropped from the project because of this.
Since they haven't updated this, from what I was told by others I worked with on the project, they decided to start choosing when they wanted to pay for artists work, and some artists dropped from the project because of this.
To be specific, we started turning down rushed work that wasn't worthy of being used in a professional project. We asked someone to update their workflow because what they turned in looked like really bad and they quit on the spot. So yes, if someone turns in junk that isn't able to be used in our project, and upon review they're unable, or unwilling, to improve submitted work to a usable quality, we're not going to accept it. If upon review, their work measures up to the quality we've showcased in this thread thus far, we're happy to pay.
JasonHeckmen said: To be specific, we started turning down rushed work that wasn't worthy of being used in a professional project. We asked someone to update their workflow because what they turned in looked like really bad and they quit on the spot. So yes, if someone turns in junk that isn't able to be used in our project, and upon review they're unable, or unwilling, to improve submitted work to a usable quality, we're not going to accept it. If upon review, their work measures up to the quality we've showcased in this thread thus far, we're happy to pay.
Although you aren't just paying for the delivered work, you're also paying for the time the artist invests in it.
And if they've invested hours of work into it, they do deserve some pay, though if it's unusable, then not the full amount. Of course it does depend on the terms agreed to, but I'd never accept work from someone who would be ok with not paying me at all if the completed work wasn't as desired. It gives the employer too much power in the relationship; the work can be rejected for fickle or intangible reasons. Which is not what I'm saying you guys have done; just my take on it in general.
I don't like making assumptions as I don't have the full picture but from the sounds of it you don't seem to be hiring well enough because you said you started turning down rushed work , which to me gives the impression that this has happened more than once.
If you hire unreliable and unprofessional freelancers , subpar quality will be expected.
So here's two suggestions:
1. Fesearch the freelancers first and see if they can
separate themselves from professional work and are able to handle and
apply critique 2. Milestone Payments : Once the task is completed ( Blockout , high poly , texture whatever it is ) , the feedback you provided is completed and the freelancer moves onto the next step , you send a payment and tick off the completion. I'm guessing you pay all at the end at this time IF it seems to fit you.
This would make both parties happy. You are happy that work is completed to how you like it at the quality that you like it. The freelancer is happy that he's gotten rewarded and can move onto the next step.
* So I've finished reworking the landscape shader network. It runs a lot faster due to having much less texture samplers now. * More than 2/3 of the roads are placed. * Landscape is fully painted (1st pass). * While we aim for a dynamic time of day cycle, for myself I've made 4 lighting presets because I don't like how unreal handles time of day so with my own lighting presets I can post some nice screens. * I've started working on some impact craters. For this I used a custom alpha for sculpting the depth and then used a single decal on top of it for normal details. I'm also thinking about creating craters as static meshes and place them on the terrain then lower the terrain under so they wouldn't clip. This will require much more manual work, but wouldn't require a decal per crater (which are very expensive in unreal). Below is my custom alpha + decal approach. I didn't get to work on the full material yet, it's only normal in the decal. What do you guys think?
Looks really good, what kind of terrain res are you using to get that amount of detail?
It might be nice to have some little bits of dirt dotted around the outside of the crater to simulate bits of dirt that have been thrown outwards from the blast, would add some variation to the shape too.
they are doing something similar with the terrain (digging animation, terrain being lowered, mesh/decal being put in place then things fading into each other). Watch it at .25 speed and it becomes pretty clear.
Project looks great, been following it for some times, always enjoy seeing the progress you guys are making.
Looks really good, what kind of terrain res are you using to get that amount of detail?
It might be nice to have some little bits of dirt dotted around the outside of the crater to simulate bits of dirt that have been thrown outwards from the blast, would add some variation to the shape too.
All the detail is coming from the decal normal map. Terrain surface is just having a very smooth hole. Thanks for the feedback I'm looking into it.
@Flight, Watched the video, very helpful indeed. In Frostbite decals can affect terrain displacement so basically as decal spawns terrain is automatically lowered by material displacement. In unreal however, decals can't have displacement mapping. I'm doing a test mesh to see how that ends up looking as well.
Here's a quick mesh test. I think this has more potential than the previous approach.
I'm going to focus on creating a gooder crater in world machine and use that along with a proper shader and also add a muddy/burnt ground layer to terrain shader to paint under and around it and hopefully that'd be good enough. With some debris meshes around it as well.
I think the texturing on this one works a lot better than the previous image, it works better being a bit darker overall. the black charring in the middle is cool
Here's the result with a proper low poly mesh + baked normal blended with 2 material layers + masking for soot. Texture ended up being too busy though, but that's and easy tweak.
4.20 adds the ability to affect the terrain heightmap through render targets, so you could in theory have your crater meshes being rendered as their own height and offset the terrain underneath to simplify the process a bit.
4.20 adds the ability to affect the terrain heightmap through render targets, so you could in theory have your crater meshes being rendered as their own height and offset the terrain underneath to simplify the process a bit.
Thanks for letting me know about that.
Here are some quick viewport print screens to show some progress.
Really great job on the foliage work, again! The overall environment work is really good. Look forward to see a bit mroe of houses, bunker and stuff like that! Any vehicule footage already?
But i gotta say i'm still not really convinced by the gameplay i like the idea of sandbox survivng in 1941 Normandy but building?? I dont know. Could make more sense if you're a paratrooper who's goal is to srvive as long as possible in enemy lines, still gotta find a way to justify possible PvP tho... And forget about building bunkers out of nowhere, but traps and temporary woodcamps or something
Just wondering how you profile your landscape materials and approach optimizing it. Is looking at instructions count enough, or do you break it down further. If instruction count is the most important thing, how many instructions is this running on currently?
Just wondering how you profile your landscape materials and approach optimizing it. Is looking at instructions count enough, or do you break it down further. If instruction count is the most important thing, how many instructions is this running on currently?
Hello,
Instruction count is irrelevant for a landscape material as cost varies depending on how many layers you have painted on each individual landscape component so it's mostly shader complexity view that matters and what impacts that the most is the number of textures used for each landscape layer so we just channel pack everything to keep that number as low as possible.
Albedo (AO pre-multiplied) Normal (RG for Normal, B for Displacement)
Roughness and Specular are driven from Albedo. It's not 100% correct in PBR theory but also UE4 isn't rendering things as in real life so we have to bridge the gap and make it look good.
So in most cases it's just 2 texture samplers per layer. For our farm floors we just share 1 Albedo and 1 Normal for 4 farm layers but color them differently through material instance.
B} Experimenting with dynamic HDRi sky (The last of us style, thanks to Tilman Milde for demonstrating the workflow).
C) Cooked some ground materials using Megascans Mixer.
D) Decal shader for puddles is using 3three Vector3 nodes to generate 3 brown colors, and blended together using a heightmap to create some depth illusion. No normal/roughness maps. Just an RGB mask for different puddle shapes. Masks for the decals were created from photos by simply using Threshold and/or Select Color Range in Photoshop plus manual cleanup to create cutouts.
E) Adding some marketplace content to project (Optimizing/Customizing them takes a lot of time though). F) Added background planes.
Currently working on: A) Improving proceduraly distributed ground cover. B} Optimizing landscape shader and grass types. C) Throwing trash out of the project. D) Updating farm crops meshes.
As @tadpole3159 mentionned it there are few elements that cut the immersion here, I'd also go with removing trash bags and blue tarps. Anywya for sure right now it's really off with its time. Also not sure about the tire but I guess it could work once in a while.
Mailbox seems off too but that's a tinier detail. (off as I really don't think that's the kind you find (found) in Normandy)
Good work on all the rest, crazy lighting of yours as usual.
RAVX, Thanks. tadpole3159, Thanks for the explanations. Plastic bags are part of a trash pack from UE4 Market, my intention was to replace their textures with something like sacks but I'm not sure if that'll look right either. I'll have a look at link you posted and see what I gather. Thanks again! Ask, Thank you! another caveman, I have a strong desire for those tarp pieces haha. I should do something about them to make them look right so I can keep them in for the release.
Generally I haven't had much information about what kind of stuff people were used to throw in their backyards 70 years back, and was expected WW2'ers to point some stuff out, tips from you guys are very helpful there. Thanks!
And here is new overcast lighting preset + some late night level design.
Your landscape work is absolutely amazing. Like, really... seriously. My jaw is on the floor.
As for some CC, I'm not sure what it is but the human-built components and architecture don't quite match up with the organics for me. They kind of pop off the screen in an isolating way. Too much contrast in the shadows maybe? AO too strong? I'm not sure, and maybe you're not complete with shaders for all the materials and maybe some post adjustments and color grading still aren't totally dialed in? The buildings especially just don't seem to sit right just yet. They also could probably use a little more real-world detailing I think (cracks, slightly more irregular corner stonework, thickness in the roof tiles, headers/lintels above doors/windows, rusty metal flashings, ground dirt pass, etc.). Maybe you just haven't gotten there yet... Hope some of my thoughts are helpful
Anyways, I'm incredibly excited to keep following your progress. That last shot on your Jul-21 post... incredible.
aclund3, AO as you pointed out is somewhat too strong at some spots, I'm really struggling there because DFAO is generally very unpleasing feature to work with. It's like if you try to tone it down on trees then you've lost all the nice AO on those small shacks. Usually doesn't look well on all assets. Some get nice AO and some other get too dark depending on the mesh complexity. I should still try to get closer to that sweet spot between structures and natural elements. I haven't played much with post yet, might be able to lift some of the ultra darks from there.
I intended to add some mesh decals to house pieces to ground them better but they are modular and I couldn't find a solution to add them in without having to duplicate the meshes. For example we have lots of 2 story houses but we want mesh decals on first floor walls only near the ground. I haven't found a solution for that yet.
Thank you for all the feedback, I'll look into improving it all here and there on houses and other parts.
One thing you could do to reduce DFAO contribution if you really need to do so is use the distance to nearest surface plugged into a one minus overlayed onto the albedo. Definitely not ideal, but Unreal has terrible controls over DFAO, and that should help counter the AO contribution just enough as long as you don't apply it too intensely.
You might just edit the Distance Field Indirect Shadow Min Visibility and see if that does the trick. You need to change the setting on an actor that's already in level though.
You might just edit the Distance Field Indirect Shadow Min Visibility and see if that does the trick. You need to change the setting on an actor that's already in level though.
That requires meshes to be movable and also level should have baked lighting. We currently don't have any indirect shadows due to not baking the lighting. That feature relies on volumetric lightmaps. Besides that, meshes painted with foliage painter don't give access to that option, it's only for static mesh actors unfortunately.
I'll try the distance to nearest surface method you mentioned. I doubt the shader complexity would be worth it though. Will post results.
Replies
You're using a skylight with a cubemap and the ambient cubemap in the PPV? Have you looked at anything using the Subsurface Profile shading yet? The ambient cubemap option doesn't seem to work well with it and relatively low values end up blowing out the skin. Just a warning
leleuxart I have a cubemap in PPV and sometimes in Skylight as well, but I tend to use a cubemap intensity of 1 and use very little skylight intensity because flat ambient light is absolutely useless. For some reason using a cubemap in Skylight doesn't exactly work as IBL so that's why I try to stay away from Skylight. I haven't paid attention to character skin so far when using a cubemap in PPV. I'm gonna take a look and if cubemaps don't work there then we have a problem.
A trailer is inevitable of course but we're still working on the game map.
Thanks.
To be specific, we started turning down rushed work that wasn't worthy of being used in a professional project. We asked someone to update their workflow because what they turned in looked like really bad and they quit on the spot. So yes, if someone turns in junk that isn't able to be used in our project, and upon review they're unable, or unwilling, to improve submitted work to a usable quality, we're not going to accept it. If upon review, their work measures up to the quality we've showcased in this thread thus far, we're happy to pay.
Although you aren't just paying for the delivered work, you're also paying for the time the artist invests in it.
And if they've invested hours of work into it, they do deserve some pay, though if it's unusable, then not the full amount. Of course it does depend on the terms agreed to, but I'd never accept work from someone who would be ok with not paying me at all if the completed work wasn't as desired. It gives the employer too much power in the relationship; the work can be rejected for fickle or intangible reasons. Which is not what I'm saying you guys have done; just my take on it in general.
If you hire unreliable and unprofessional freelancers , subpar quality will be expected.
So here's two suggestions:
1. Fesearch the freelancers first and see if they can separate themselves from professional work and are able to handle and apply critique
2. Milestone Payments : Once the task is completed ( Blockout , high poly , texture whatever it is ) , the feedback you provided is completed and the freelancer moves onto the next step , you send a payment and tick off the completion. I'm guessing you pay all at the end at this time IF it seems to fit you.
This would make both parties happy. You are happy that work is completed to how you like it at the quality that you like it. The freelancer is happy that he's gotten rewarded and can move onto the next step.
Hope some of this is helpful
AMAZING! This actually looks more like a photography!
* So I've finished reworking the landscape shader network. It runs a lot faster due to having much less texture samplers now.
* More than 2/3 of the roads are placed.
* Landscape is fully painted (1st pass).
* While we aim for a dynamic time of day cycle, for myself I've made 4 lighting presets because I don't like how unreal handles time of day
* I've started working on some impact craters. For this I used a custom alpha for sculpting the depth and then used a single decal on top of it for normal details. I'm also thinking about creating craters as static meshes and place them on the terrain then lower the terrain under so they wouldn't clip. This will require much more manual work, but wouldn't require a decal per crater (which are very expensive in unreal). Below is my custom alpha + decal approach. I didn't get to work on the full material yet, it's only normal in the decal. What do you guys think?
It might be nice to have some little bits of dirt dotted around the outside of the crater to simulate bits of dirt that have been thrown outwards from the blast, would add some variation to the shape too.
If you look at this video from Battlefield 5 (0:27 sec to 34 ish, apparently i can´t link to the right timestamp
Project looks great, been following it for some times, always enjoy seeing the progress you guys are making.
Thanks for the feedback I'm looking into it.
@Flight, Watched the video, very helpful indeed.
In Frostbite decals can affect terrain displacement so basically as decal spawns terrain is automatically lowered by material displacement. In unreal however, decals can't have displacement mapping.
I'm going to focus on creating a gooder crater in world machine and use that along with a proper shader and also add a muddy/burnt ground layer to terrain shader to paint under and around it and hopefully that'd be good enough. With some debris meshes around it as well.
Here are some quick viewport print screens to show some progress.
But i gotta say i'm still not really convinced by the gameplay
Just wondering how you profile your landscape materials and approach optimizing it. Is looking at instructions count enough, or do you break it down further. If instruction count is the most important thing, how many instructions is this running on currently?
Instruction count is irrelevant for a landscape material as cost varies depending on how many layers you have painted on each individual landscape component so it's mostly shader complexity view that matters and what impacts that the most is the number of textures used for each landscape layer so we just channel pack everything to keep that number as low as possible.
Albedo (AO pre-multiplied)
Normal (RG for Normal, B for Displacement)
Roughness and Specular are driven from Albedo. It's not 100% correct in PBR theory but also UE4 isn't rendering things as in real life so we have to bridge the gap and make it look good.
So in most cases it's just 2 texture samplers per layer. For our farm floors we just share 1 Albedo and 1 Normal for 4 farm layers but color them differently through material instance.
Sharing some progress.
A) New lighting preset!
B} Experimenting with dynamic HDRi sky (The last of us style, thanks to Tilman Milde for demonstrating the workflow).
C) Cooked some ground materials using Megascans Mixer.
D) Decal shader for puddles is using 3three Vector3 nodes to generate 3 brown colors, and blended together using a heightmap to create some depth illusion. No normal/roughness maps. Just an RGB mask for different puddle shapes. Masks for the decals were created from photos by simply using Threshold and/or Select Color Range in Photoshop plus manual cleanup to create cutouts.
E) Adding some marketplace content to project (Optimizing/Customizing them takes a lot of time though).
F) Added background planes.
Currently working on:
A) Improving proceduraly distributed ground cover.
B} Optimizing landscape shader and grass types.
C) Throwing trash out of the project.
D) Updating farm crops meshes.
Apparently people used to line bins with news paper, food waste would go onto the compose heap and glass would be collected to be reused
https://simplelivingwithkids.com/2015/07/28/what-did-people-do-before-plastic-rubbish-bags/.
Since your using megascans maybe download this
https://megascans.se/assets/pffhdyp2
then buy an old timey news paper from the correct time and place as your game in set and blend the textures with a multiply or something
https://www.britishnewspaperarchive.co.uk/
tadpole3159, Thanks for the explanations. Plastic bags are part of a trash pack from UE4 Market, my intention was to replace their textures with something like sacks but I'm not sure if that'll look right either. I'll have a look at link you posted and see what I gather. Thanks again!
Ask, Thank you!
another caveman, I have a strong desire for those tarp pieces haha. I should do something about them to make them look right so I can keep them in for the release.
Generally I haven't had much information about what kind of stuff people were used to throw in their backyards 70 years back, and was expected WW2'ers to point some stuff out, tips from you guys are very helpful there. Thanks!
And here is new overcast lighting preset + some late night level design.
As for some CC, I'm not sure what it is but the human-built components and architecture don't quite match up with the organics for me. They kind of pop off the screen in an isolating way. Too much contrast in the shadows maybe? AO too strong? I'm not sure, and maybe you're not complete with shaders for all the materials and maybe some post adjustments and color grading still aren't totally dialed in? The buildings especially just don't seem to sit right just yet. They also could probably use a little more real-world detailing I think (cracks, slightly more irregular corner stonework, thickness in the roof tiles, headers/lintels above doors/windows, rusty metal flashings, ground dirt pass, etc.). Maybe you just haven't gotten there yet... Hope some of my thoughts are helpful
Anyways, I'm incredibly excited to keep following your progress. That last shot on your Jul-21 post... incredible.
aclund3, AO as you pointed out is somewhat too strong at some spots, I'm really struggling there because DFAO is generally very unpleasing feature to work with. It's like if you try to tone it down on trees then you've lost all the nice AO on those small shacks. Usually doesn't look well on all assets. Some get nice AO and some other get too dark depending on the mesh complexity. I should still try to get closer to that sweet spot between structures and natural elements. I haven't played much with post yet, might be able to lift some of the ultra darks from there.
I intended to add some mesh decals to house pieces to ground them better but they are modular and I couldn't find a solution to add them in without having to duplicate the meshes. For example we have lots of 2 story houses but we want mesh decals on first floor walls only near the ground. I haven't found a solution for that yet.
Thank you for all the feedback, I'll look into improving it all here and there on houses and other parts.
I'll try the distance to nearest surface method you mentioned. I doubt the shader complexity would be worth it though. Will post results.
Thanks.
If it wasn't please check the others around it.
artizan, Thanks.