I'm getting similar results to you when importing into UE4, so I will have our guys look into this further.
Yea,I see the same triangle artifact when doing 1 smoothing groups(no tangents and binormals) in toolbag 2.0/UE4.2. I'm not a programmer but the tangents and binormal create by UE4 seem to be a bit extreme in a couple of spot compare to the ones that Maya generate/exported(check my post at the bottom of the link):
I was wondering, I suppose there isn't a way to animate a camera to do something that's not a turntable is it? I think I read somewhere that you could import them from Max/Maya but after that I couldn't find more info about the topic
I was wondering, I suppose there isn't a way to animate a camera to do something that's not a turntable is it? I think I read somewhere that you could import them from Max/Maya but after that I couldn't find more info about the topic
This is not something that is currently supported, but we would like to add it in the future. More of a long term goal though.
Put a small dynamic color swatch next to these sliders (and possibly other sliders, if appropriate) that gives an RGB equivalent of the value from the slider.
I'll sometimes create a material I like in Toolbag that I would like to replicate on a smaller part of a model. It'd be nice to be able to directly sample RGB values instead of approximating or doing math.
Put a small dynamic color swatch next to these sliders (and possibly other sliders, if appropriate) that gives an RGB equivalent of the value from the slider.
I'll sometimes create a material I like in Toolbag that I would like to replicate on a smaller part of a model. It'd be nice to be able to directly sample RGB values instead of approximating or doing math.
Well, there is one for specular intensity/color already, set intensity to 1 and use the color picker, than you can copy/paste the hex value directly into PS.
For gloss, gloss is generally a greyscale linear space value, so the math is really simply, just 255*X, so 0.9 = 230. Though I can see how it would be nice to just copy a hex value instead, so I'll add that to the feature request list.
Edit: Actually for gloss, no math is required at all, just use the numeric value from TB in the brightness input in PS.
Hmm, looks like I was making a mistake by using Photoshop swatches for some color picking I was doing. That would explain some weirdness translating certain things between Toolbag and PS. Guess that simplifies things, though I definitely need to customize my swatches now that I've realized just how janky they are.
I really hope you guys get Photoshop style curves in there (LEEEEEEEEEE DO EEEET). That functionality would be really sick to be able to use to tweak each texture map in-engine as well. Substance in Unreal allows you to do this and it's a huuuuge time saver to not have to go back to Photoshop.
Not sure if this is has already been brought up, but the skin shader's normal smoothing does not seem to be masked by the texture map - it seems to apply normal smoothing to the entire mesh, not just the areas intended to be 'skin'
Not sure if this is has already been brought up, but the skin shader's normal smoothing does not seem to be masked by the texture map - it seems to apply normal smoothing to the entire mesh, not just the areas intended to be 'skin'
Any news on getting a keyable camera?
Like not even maya camera imports or anything just a separate animated camera to render off for things like environments because I can not get the right shots with just turntables. It's very difficult to get nice animated fly throughs with just turntables.
Thanks, I am aware of handplane (Alec and Luke, the developers, are swell guys as well). However, the issues we have with tangents are most likely issues with our code, not necessarily something a content converter like handplane can help with.
Any news on getting a keyable camera?
Like not even maya camera imports or anything just a separate animated camera to render off for things like environments because I can not get the right shots with just turntables. It's very difficult to get nice animated fly throughs with just turntables.
Sorry, no updates on the animation front. Its something we really want to add but have not yet had the time to start working on yet. We've got some really exciting stuff in the pipe though, promise!
I was wondering if someone could please confirm for me whether sky lights have the ability to cast shadows?
My current scene consists of two meshes, a character and a base mesh. No shadows are being cast on the base when the scene is only receiving light from the sky light preset. I've tried scaling up and down both meshes but the result is the same. For me shadows are only cast on the base when dynamic lights are added.
I was wondering if someone could please confirm for me whether sky lights have the ability to cast shadows?
My current scene consists of two meshes, a character and a base mesh. No shadows are being cast on the base when the scene is only receiving light from the sky light preset. I've tried scaling up and down both meshes but the result is the same. For me shadows are only cast on the base when dynamic lights are added.
Yes that is correct, the sky/image based lighting does not cast shadows. You need to add dynamic lights (by clicking in the sky editor or placing them by hand).
Unfortunately it would be much too slow to cast shadows from the sky, as every pixel essentially represents a point of light and we would need to cast shadows from each, which would be the equivalent of having thousands/millions of dynamic lights.
Try to think of the sky light as ambient diffuse and specular lighting.
Ah yes, that article was written for Toolbag 1.0, and the sky light system was a bit different then. Previous, there was 1 preset dynamic light baked into the sky system, now you can simply click anywhere on the sky preview image and add as many dynamic lights as you need.
Unfortunately the light setups shown in that tutorial are not available anywhere (and if they were, they would only load in TB1, not TB2). Though you should be able to recreate them for the most part in TB2, as the basic concepts are very similar.
Lee Devonald and I spend an hour and a half going over the skin shader, eyes, hair, and lighting/general character setup. You can re-watch the livestream on the youtubes:
Nice. Hope 2.05 will be released soon.
Few comments on vid:
1. Translucency and shadows, why not to premultiply them? (i.e shadow will be affected by translucency and vise versa...)
2. Fuzz map look like plain fresnel to me, but actual velvet shader works slightly different (well, maybe just my imho)
3. Cornea is perfectly smooth, (it's a lens!), if it's not - You should go to ophthalmologist
2. Hair shader... http://amd-dev.wpengine.netdna-cdn.com/wordpress/media/2012/10/Scheuermann_HairSketchSlides.pdf You've seen it (can secondary highlight be based on albedo map? blurred)
4. You're trying to implement PBR shader w/o physical camera? Hmm...
Nice. Hope 2.05 will be released soon.
Few comments on vid:
1. Translucency and shadows, why not to premultiply them? (i.e shadow will be affected by translucency and vise versa...)
This is a bit beyond my knowledge, but I will pass the question on to our tech guys. EditL I asked about this and they weren't sure how to respond, could you expand on this a bit?
2. Fuzz map look like plain fresnel to me, but actual velvet shader works slightly different (well, maybe just my imho)
Yep, the fuzz effect is essentially diffuse Fresnel. We are working on a cloth specific specular BRDF which will be completely different, and you will be able to use both at once.
3. Cornea is perfectly smooth, (it's a lens!), if it's not - You should go to ophthalmologist
Right, this is why its important to have a normal map for the bottom layer, so you can get the rough detail of the iris, and another smooth layer for the wet coating on top.
Though it looks like they don't have any surface variation around the cornea in the outer layer normal map, I must not have picked up on this the first time I read it.
Not currently, but you can tint the secondary specular color with a colored value that is similar to the albedo, which is what we did for the hair shader in the video.
Can you give me some more details on what you mean by physical camera, and how you think it would be helpful in TB2? I'm not familiar with that phrase in this context.
This is a bit beyond my knowledge, but I will pass the question on to our tech guys. EditL I asked about this and they weren't sure how to respond, could you expand on this a bit?
Translucency and shadows are not tied by now, that cause lighten areas in shadows and shadows in translucent parts.
Light generates shadow map
Light generates 'translucency' map (kind of brightness map, not the one in textures)
Premultiply
Since translucency is a 'blur' map it'll allow the light 'leak' into shadow areas and they'll became bright. And vice versa, since shadow even being blurred in complete black places still be black - there will not be any visible translucency (glowing ears).
Can you give me some more details on what you mean by physical camera, and how you think it would be helpful in TB2? I'm not familiar with that phrase in this context.
Seems like MT uses 'physical' camera behavior, sorry.
Sorry, no updates on the animation front. Its something we really want to add but have not yet had the time to start working on yet. We've got some really exciting stuff in the pipe though, promise!
Looking forward to that
I would love to bring my characters to life in marmoset It would be awesome if I could create animations (bonedriven / morphtargets / camera movement) in max/maya and simply load it into marmoset
I watched the shader setup video and it looks like the textures created by Yuri are all horizontal. Is this the preferred way as in the earlier video (Toolbag 2 beta test) Marie-Michelle mentioned Cry engine favors vertical layout.
I am just doing the hair maps for my character now and any info would be warmly received.
As my model is a game spec model I only intend to have a diffuse, specular and normal (possibly) for the hair. Thank you
So this is partly an engine specific thing, but with regards to setup in Toolbag, it actually doesn't matter.
You should generally try to aim for having all of your hair planes running in the same direction, but what that direction is is up to you!
Crytek prefers vertical strips because of how they handle their screenspace alpha blur (which is done horizontally in UV space), they also have their anisotropic direction locked to match vertical strips. In toolbag we favor no alpha blur, and the freedom to set your own anisotropic direction.
TL:DR - set up your strips so they're all in one direction, then adjust the anisotropic direction slider to match!
Don't know if anyone has posted about this issue, but parallax mapping doesn't work properly with Max tangents. Has anyone else ran into this issue. I use it on my eye setups and usually I just switch the tangents to Marmoset and invert my green channel. It would be nice to have that working with max tangents though.
Just wondering if there is a way to render out an image with depth of field enabled and transparency enabled that gives a soft transition in the dof blur on a transparent background?
Just wondering if there is a way to render out an image with depth of field enabled and transparency enabled that gives a soft transition in the dof blur on a transparent background?
Yeah that should be doable, make sure you've enabled transparency in the image export options. Also, use psd.
If the skybox background is bleeding through, try using a simple color instead, and try to match it roughly to the background color you intend to comp in.
1. Can you up the maximum size for the shadow map on directional lights, it's looking very low res when you get close.
2. The transform gizmo is super confusing with all the arrows/handles, is there a way to shrink the size of it or just simplify it like Maya. It's really tricky placing directional lights because the gizmo points one way, but the light points the other way.
3. Can the rotation transform tool support negative values, this thing is gimble locking all over the place and randomly changes it's values when you click off of it.
4. The transform menu keeps closing when I click off of a light, I think this might be a bug.
One more feature request, can you make the textures update on save so I don't have to alt tab over to it when using dual monitors, this is in Maya now and it's a huge time saver.
1. Can you up the maximum size for the shadow map on directional lights, it's looking very low res when you get close.
I'll add this to the suggestion list but I'm not sure what can be done there. I've talked to the guys a bit about this and the shadow resolution is already extremely high. We will probably need to do something other than increase the shadow resolution to solve this.
2. The transform gizmo is super confusing with all the arrows/handles, is there a way to shrink the size of it or just simplify it like Maya.
3. Can the rotation transform tool support negative values, this thing is gimble locking all over the place and randomly changes it's values when you click off of it.
I'll add these to the list as well.
Edit: Actually, I think if you understand how the widget works it will make more sense.
Negative values are supported (you can enter them manually, when rotating with the widget they reset after 360 though, but from a practical perspective then do not lock there, the counter simply starts over), when you rotate in the viewport, you need to do so in a screen-space fashion (like modo), so if you drag left-right only, you will hit a point where it no longer moves. Try rotating with a circular mouse motion around the widget control instead, you can free rotate indefinitely then (only exception is at acute angles, it sort of falls apart then). You can also move the cursor further away from the widget center to get more precision while rotating (again like modo).
Looking at Maya (2008, may have changed in newer versions?) it actually appears to work the same way. Only difference is the counter doesn't start over past 360.
It's really tricky placing directional lights because the gizmo points one way, but the light points the other way.
I'm confused by what you mean here, can you elaborate?
4. The transform menu keeps closing when I click off of a light, I think this might be a bug.
Also being able to adjust hue and saturation of skies would be huge, I often find a sky setup I like value wise, but then it's too colourful or the wrong colour for my scene.
One more feature request, can you make the textures update on save so I don't have to alt tab over to it when using dual monitors, this is in Maya now and it's a huge time saver.
Yes this would be very handy. I've asked if this is feasible and it looks like it is, so hopefully we can add this at some point.
Don't know if anyone has posted about this issue, but parallax mapping doesn't work properly with Max tangents. Has anyone else ran into this issue. I use it on my eye setups and usually I just switch the tangents to Marmoset and invert my green channel. It would be nice to have that working with max tangents though.
Verified this as well, got it logged in the bug system, thanks for the report!
Basically, you have a standard lookup table that people can load into Photoshop, do post processing on, and then load into the engine. This lookup table adjusts the scene colors to match the post processing done in PS.
In theory, the effect of any adjustment layer available in Photoshop should be reproducible by using a color lookup table that has been modified exactly the same way as a photo or screenshot or whatever.
Basically, you have a standard lookup table that people can load into Photoshop, do post processing on, and then load into the engine. This lookup table adjusts the scene colors to match the post processing done in PS.
In theory, the effect of any adjustment layer available in Photoshop should be reproducible by using a color lookup table that has been modified exactly the same way as a photo or screenshot or whatever.
Yes, so currently we do not support a cavity input with substance designer content. So you will have to save out a map for that and load it into the cavity map slot manually.
Do you know if SD has a standard cavity type output? If so it shouldnt be too hard for us to support this.
Hey EarthQuake, sorry for the late reply
I don't think they have a standard cavity output although they do have a curvature output (Is that anything useful for outputs?)
I think I will just stick to exporting cavity maps out manaually.
Joe, looks like undo also closes the transform menu of lights, which is super annoying.
Yes, it looks like the transform visibility state isn't saved, I've logged that as a bug here, which would cause the undo bug as well (undo relies on saving the entire state of the app and loading it back up, essentially).
I am loving Marmoset toolbag 2, but where is the render view settings present in the first version that lets you show in realtime only certain visual elements like (shadow, AO, normals, specular, etc.) It be really handy if you could get the feature back in following versions of toolbag 2
Maybe a bit far out feature request.
What if we where able to save states, like I setup a scene save it as a state try a few different things, that doesn't work out then I can go back to the saved state.
I dont think that helps at all, the render view settings present in v 1 is like a render pass for toolbag and lets you filter out the effects of your render so you can even enhance it further in post. it's a tweaking mechanism not as a safe haven in case you mess up your work.
Replies
Thanks for the info EQ :thumbup:
Yea,I see the same triangle artifact when doing 1 smoothing groups(no tangents and binormals) in toolbag 2.0/UE4.2. I'm not a programmer but the tangents and binormal create by UE4 seem to be a bit extreme in a couple of spot compare to the ones that Maya generate/exported(check my post at the bottom of the link):
https://answers.unrealengine.com/questions/14375/normal-map-workflow-with-xnormal-is-not-working.html
Anyway, thanks for looking at the files and it's good to know it's not something i'm doing wrong
This is not something that is currently supported, but we would like to add it in the future. More of a long term goal though.
Ok, thanks for clarifying. It would be awesome indeed to be able to do that, so I'm glad you're thinking of adding it
Put a small dynamic color swatch next to these sliders (and possibly other sliders, if appropriate) that gives an RGB equivalent of the value from the slider.
I'll sometimes create a material I like in Toolbag that I would like to replicate on a smaller part of a model. It'd be nice to be able to directly sample RGB values instead of approximating or doing math.
Well, there is one for specular intensity/color already, set intensity to 1 and use the color picker, than you can copy/paste the hex value directly into PS.
For gloss, gloss is generally a greyscale linear space value, so the math is really simply, just 255*X, so 0.9 = 230. Though I can see how it would be nice to just copy a hex value instead, so I'll add that to the feature request list.
Edit: Actually for gloss, no math is required at all, just use the numeric value from TB in the brightness input in PS.
Thanks, I've logged that in our bug system here.
I'm not a great Marmoset Toolbag user myself, just like work with UI, so, here is my reskin for Marmoset Toolbag 2.04.
It may be not really great for work (not very readable due to low contrast), but i've made it for myself and think someone else may be interested.
There are two versions of it, neutral gray and slightly warm (seen on screenshot).
Only .tga files and readme in archive.
I've only included files I've painted, I guess it's legitimate. Based on trial version, hope there is no difference with purchased one.
Marmoset_Toolbag_2.04_skin_NeutralDarkness_r01_(by_S-ed).zip - 13Kb
https://mega.co.nz/#!kwQnBKhQ!TUQ-F2OB1_GI7HMUWfo8NHunBItaEIKWna-ANGZn5P8
Marmoset_Toolbag_2.04_skin_WarmDarkness_r01_(by_S-ed).zip - 13Kb
https://mega.co.nz/#!U8A1kDRD!aJy_mDRRjTlyFA486IpVf0Sthf1ZQuA2BT5iJ-5l_S0
p.s.: there are few bugs with UI that I was unable to bypass so I cheated in some places, hope it'll be fixed later
Relating normal map spaces, there is great tool:
http://www.handplane3d.com/
Moreover it's free now! (sound like an ad? X))
Like not even maya camera imports or anything just a separate animated camera to render off for things like environments because I can not get the right shots with just turntables. It's very difficult to get nice animated fly throughs with just turntables.
Cool, glad to see someone put the re-skinable UI feature to use!
Thanks, I am aware of handplane (Alec and Luke, the developers, are swell guys as well). However, the issues we have with tangents are most likely issues with our code, not necessarily something a content converter like handplane can help with.
Sorry, no updates on the animation front. Its something we really want to add but have not yet had the time to start working on yet. We've got some really exciting stuff in the pipe though, promise!
My current scene consists of two meshes, a character and a base mesh. No shadows are being cast on the base when the scene is only receiving light from the sky light preset. I've tried scaling up and down both meshes but the result is the same. For me shadows are only cast on the base when dynamic lights are added.
Yes that is correct, the sky/image based lighting does not cast shadows. You need to add dynamic lights (by clicking in the sky editor or placing them by hand).
Unfortunately it would be much too slow to cast shadows from the sky, as every pixel essentially represents a point of light and we would need to cast shadows from each, which would be the equivalent of having thousands/millions of dynamic lights.
Try to think of the sky light as ambient diffuse and specular lighting.
Also I was wondering if any of these lighting setups are available as sample scenes anywhere? http://www.marmoset.co/toolbag/learn/character-lighting
Thanks again!
Ah yes, that article was written for Toolbag 1.0, and the sky light system was a bit different then. Previous, there was 1 preset dynamic light baked into the sky system, now you can simply click anywhere on the sky preview image and add as many dynamic lights as you need.
Unfortunately the light setups shown in that tutorial are not available anywhere (and if they were, they would only load in TB1, not TB2). Though you should be able to recreate them for the most part in TB2, as the basic concepts are very similar.
[ame="
I plan to put together a higher quality 1080P version soon as well.
Few comments on vid:
1. Translucency and shadows, why not to premultiply them? (i.e shadow will be affected by translucency and vise versa...)
2. Fuzz map look like plain fresnel to me, but actual velvet shader works slightly different (well, maybe just my imho)
3. Cornea is perfectly smooth, (it's a lens!), if it's not - You should go to ophthalmologist
2. Hair shader... http://amd-dev.wpengine.netdna-cdn.com/wordpress/media/2012/10/Scheuermann_HairSketchSlides.pdf You've seen it
4. You're trying to implement PBR shader w/o physical camera? Hmm...
This is a bit beyond my knowledge, but I will pass the question on to our tech guys. EditL I asked about this and they weren't sure how to respond, could you expand on this a bit?
Yep, the fuzz effect is essentially diffuse Fresnel. We are working on a cloth specific specular BRDF which will be completely different, and you will be able to use both at once.
Right, this is why its important to have a normal map for the bottom layer, so you can get the rough detail of the iris, and another smooth layer for the wet coating on top.
My setup is based on this paper: http://advances.realtimerendering.com/s2012/activision/Jimenez-Separable_Subsurface_Scattering_and_Eye_Rendering(Siggraph2012).pptx the eye specific stuff starts around slide 100 or so.
Though it looks like they don't have any surface variation around the cornea in the outer layer normal map, I must not have picked up on this the first time I read it.
Not currently, but you can tint the secondary specular color with a colored value that is similar to the albedo, which is what we did for the hair shader in the video.
Can you give me some more details on what you mean by physical camera, and how you think it would be helpful in TB2? I'm not familiar with that phrase in this context.
Thanks for the feedback!
- Light generates shadow map
- Light generates 'translucency' map (kind of brightness map, not the one in textures)
- Premultiply
Since translucency is a 'blur' map it'll allow the light 'leak' into shadow areas and they'll became bright. And vice versa, since shadow even being blurred in complete black places still be black - there will not be any visible translucency (glowing ears).Seems like MT uses 'physical' camera behavior, sorry.
Looking forward to that
I would love to bring my characters to life in marmoset
I watched the shader setup video and it looks like the textures created by Yuri are all horizontal. Is this the preferred way as in the earlier video (Toolbag 2 beta test) Marie-Michelle mentioned Cry engine favors vertical layout.
I am just doing the hair maps for my character now and any info would be warmly received.
As my model is a game spec model I only intend to have a diffuse, specular and normal (possibly) for the hair. Thank you
You should generally try to aim for having all of your hair planes running in the same direction, but what that direction is is up to you!
Crytek prefers vertical strips because of how they handle their screenspace alpha blur (which is done horizontally in UV space), they also have their anisotropic direction locked to match vertical strips. In toolbag we favor no alpha blur, and the freedom to set your own anisotropic direction.
TL:DR - set up your strips so they're all in one direction, then adjust the anisotropic direction slider to match!
color corrected
wow
Yeah that should be doable, make sure you've enabled transparency in the image export options. Also, use psd.
If the skybox background is bleeding through, try using a simple color instead, and try to match it roughly to the background color you intend to comp in.
1. Can you up the maximum size for the shadow map on directional lights, it's looking very low res when you get close.
2. The transform gizmo is super confusing with all the arrows/handles, is there a way to shrink the size of it or just simplify it like Maya. It's really tricky placing directional lights because the gizmo points one way, but the light points the other way.
3. Can the rotation transform tool support negative values, this thing is gimble locking all over the place and randomly changes it's values when you click off of it.
4. The transform menu keeps closing when I click off of a light, I think this might be a bug.
That is all.
I'll add this to the suggestion list but I'm not sure what can be done there. I've talked to the guys a bit about this and the shadow resolution is already extremely high. We will probably need to do something other than increase the shadow resolution to solve this.
I'll add these to the list as well.
Edit: Actually, I think if you understand how the widget works it will make more sense.
Negative values are supported (you can enter them manually, when rotating with the widget they reset after 360 though, but from a practical perspective then do not lock there, the counter simply starts over), when you rotate in the viewport, you need to do so in a screen-space fashion (like modo), so if you drag left-right only, you will hit a point where it no longer moves. Try rotating with a circular mouse motion around the widget control instead, you can free rotate indefinitely then (only exception is at acute angles, it sort of falls apart then). You can also move the cursor further away from the widget center to get more precision while rotating (again like modo).
Looking at Maya (2008, may have changed in newer versions?) it actually appears to work the same way. Only difference is the counter doesn't start over past 360.
I'm confused by what you mean here, can you elaborate?
I'll look into this, thanks!
Can you elaborate on this? What sort of color lookup tables, for what purpose?
Yes this would be very handy. I've asked if this is feasible and it looks like it is, so hopefully we can add this at some point.
Verified this as well, got it logged in the bug system, thanks for the report!
It would be similar to the LUTs you can use in UDK: https://docs.unrealengine.com/latest/INT/Engine/Rendering/PostProcessEffects/ColorGrading/index.html
Basically, you have a standard lookup table that people can load into Photoshop, do post processing on, and then load into the engine. This lookup table adjusts the scene colors to match the post processing done in PS.
In theory, the effect of any adjustment layer available in Photoshop should be reproducible by using a color lookup table that has been modified exactly the same way as a photo or screenshot or whatever.
Ok cool, thanks.
Hey EarthQuake, sorry for the late reply
I don't think they have a standard cavity output although they do have a curvature output (Is that anything useful for outputs?)
I think I will just stick to exporting cavity maps out manaually.
Thanks for the response anyway man
Yes, it looks like the transform visibility state isn't saved, I've logged that as a bug here, which would cause the undo bug as well (undo relies on saving the entire state of the app and loading it back up, essentially).
Is there any documentation regarding character hair setup and creation in toolbag 2? Love this caesar fan art!
What if we where able to save states, like I setup a scene save it as a state try a few different things, that doesn't work out then I can go back to the saved state.
What I was talking about was a separate request that I think will be useful.