1) Is there a way to do volume lighting in Toolbag 2? Ideally something that also does light occlusion where geometry gets in the way.
Not currently, but I'll add a vote for it to the suggestion list.
2) When I'm manually reordering/organizing items in the list under the Scene tab, how do I drag item B under item A without making it a child of item A. I like that you can create parent-child relationships, but sometimes I just want to shuffle things around for organizational purposes without parenting them.
This can be sort of finicky but you can do it. You need to drag to right in between the two objects, and the little UI element should snap to the upper level in the heirarchry when you have the position right.
Recently i was doing some extended testing of new Toolbag2 one thing noticed is lack of something that would act as a floor shadow catcher. Transparent shader/plane that would only catch shadows/occlusion that would work good with alpha.
Is it somewhere and I just miss it may be? or is there a way for doing it manually somehow?
Part 2, covering the practical aspects of art creation coming soon.
great tutorial, thanks.
would love to see a tutorial on skin shading at some point. right now all the samples i see seem to have a greenish tone to the skin and i am also having a hard time getting the right skins colors.
@ MM The green is because the skin shader is placing the opposite color of your sub-dermis in the lit areas. Not sure there is a way around the green tones other than hammering back your saturation to tone it down. It's probably more complicated than that, but that is what I am observing.
@ MM The green is because the skin shader is placing the opposite color of your sub-dermis in the lit areas. Not sure there is a way around the green tones other than hammering back your saturation to tone it down. It's probably more complicated than that, but that is what I am observing.
if you mean desaturating the diffuse map then no that does not help.
from what i have seen the green tint is a result of the combination of subdermis scatter color (or opposite it) and the diffuse in skin. so by default if we have a redish scatter color then the opposite blue-ish color mixes with the yellow on the skin to give a greenish tone. i could be completely wrong but so far this is what i observed.
only thing that helps is using white scatter color or at least very desaturated red for the scatter color. the more saturated red color in scatter the more greenish tone on the skin given that the skin has a generic melanin tone on it. but using white scatter color we dont get any nice redish scatter any which is also required for skin shading.
if this is part of some sort of energy conservation then perhaps a slider to control that would be nice.
@ MM - From what I observe it places almost the opposite color of your subdermis in the lit areas of your model. For example with intensity of the subdermis jacked up, a yellow subdermis color, and white diffuse it places blue in the lit area. With red it places a teal color there. I was saying a more desaturated subdermis tends to help a bit and if you put more saturation in the diffuse that seems to cancel it a bit.
Here is a screen of what I am seeing. Like I said I am not sure there is much of a way around the green when you have a red subdermis.
like i said, desaturated scatter color gives a desaturated scatter which is not something you would want in a skin shader. saturating the diffuse to compensate for that does not work. you might think it works but so far to me it doesnt seem to work at all if any. saturating the diffuse only saturated the entire diffuse globally which is not what i want, i only want saturation on the scatter areas.
@ MM - Oh no don't get me wrong. They don't solve the problem in any way. It helps with the greenish a bit, but creates other problems as well. I would want it to just place saturation in only the scatter area as well. I was just saying I am not sure there is a way to do that currently.
Recently i was doing some extended testing of new Toolbag2 one thing noticed is lack of something that would act as a floor shadow catcher. Transparent shader/plane that would only catch shadows/occlusion that would work good with alpha.
Is it somewhere and I just miss it may be? or is there a way for doing it manually somehow?
Yeah, this has been missing for a while!
Other than that, been trialling the new toolbag for a couple of days now and love it. Job well done, guys.
MM: Andres, the guy who wrote the skin shader in toolbag, and the one in skyshop (which is almost identical) wrote a tutorial for the skyshop version. You can find it here: http://www.marmoset.co/skyshop/learn/skin-tut
I'm not sure if this will solve your issues, so let me know if you're still having problems after reading it. I'm out of the country for now but I may be able to get you a more detail response sometime soon.
Will TB2 ever be the price of one?
If not put TB1 back on sale and I will buy it because TB2 is just too much for me
I am sorry but Toolbag 1 is no longer in active development, as such we have no plans to offer it for sale. Current Toolbag 1 licences will continue to function, however maintenance will be very minimal (only fixing major bug fixes relating to crashing etc).
Toolbag 2 is, in our opinion at least, a massively improved product, and we plan to add many more features for free with updates in the future. If you haven't yet, I recommend trying out the trial version.
We may run sales for TB2 at some point, but I can't give you any specific information about that, its best to follow our facebook page https://www.facebook.com/marmosetco for information about sales and other news.
@ Joe, i checked out that tutorial and i see the same problem in the final images of the model in there.
this paragraph goes into the reason:
Similarly, energy conservation is performed per color component. Youll notice if more red light is being scattered out the sides and back of the model, less of it is leaving the lit, front side, leading to a cyan hue. This hue can be alarming at first but is present in human skin as well. It is also more exaggerated by a pure white diffuse map. The cyan-ness can be adjusted with the saturation of the subdermis color. The less saturated and closer to white or grey the subdermis is, the more evenly the different color components will be scattered.
however i don't think this is the correct way to make the skin shader work/look even though theoretically it is following real world physics.
the end result is not realistic may be because the shader is following one rule of physics and not all of them which would be impossible.
here is the final result in the tutorial, as you can see the diffuse looks too desaturated and the lit area still has quite a bit greenish/bluish tone to it.
i just found a quote from Andreas explaining this.
This is due to energy conservation in the scattering. Basically, when you set your skin to scatter red light more than green and blue, the red light gets blurred out farther but also appears dimmer as a consequence. The red component of light energy is spread out evenly across a wider area where as not-red light, green + blue, is narrowly concentrated. The result is, at the fringes of light hitting a surface you'll see scattered, subdermis-color light, but in the middle you'll see more of white minus subdermis color.
This effect looks more alarming with a white or grey model. Once you put a peachy skin tone in the albedo, it will actually "absorb" and neutralize some of that cyan tint. It does make the subdermis map a little unruly to work with at first.
You'll notice this effect if you enable energy conservation with specular color as well. Specular color will tint the diffuse color response in an inverse-color fashion.
i understand that this is based on real world physics. BUT lets keep in mind that everything here and everywhere in any art is always fake. some more fake than others. no rendering system is truely physically correct since that is completely impossible.
so may be we can bend some rules to make the end result better ?
i dont know what would be a solution to this but may be start with a slider for energy conservation for the subdermis scatter ?
i understand that this is based on real world physics. BUT lets keep in mind that everything here and everywhere in any art is always fake. some more fake than others. no rendering system is truely physically correct since that is completely impossible.
so may be we can bend some rules to make the end result better ?
i dont know what would be a solution to this but may be start with a slider for energy conservation for the subdermis scatter ?
Gonna second this. I would rather have it so I can control the results when it isn't working out.
the end result is not realistic may be because the shader is following one rule of physics and not all of them which would be impossible.
Yeah :-/, I can't say you're wrong about this.
A big part of the problem is you have a photo of skin for your albedo, lit evenly with white light from all angles to make for a nice texture. But all that omni-directional light is already scattered in your photo.
This wrap lighting and energy conserving tint is an attempt to deconstruct that scattering and simulate what kind of shading from a single light direction could cause the skin to look like your photo under an omni-directional light source.
You could probably come up with better approximations of this deconstruction, but conserving energy and making sure the sum of all lighting adds up to 1.0 is a big part of what makes IBL skin work. Without the conservation, weird things start happening as your sky lighting gets more uniform and ambient.
i dont know what would be a solution to this but may be start with a slider for energy conservation for the subdermis scatter ?
As practical advice: I'd suggest desaturating your subdermis color more as a way to control the conservation. If your subdermis color is white/grey, all wavelengths of light are scattered equally and you don't see any tint either way. This will make your skin shading softer without changing the appearance of your albedo [diffuse] texture at all.
I've also had better luck matching the subdermis color closely to the hue of the albedo texture. I'm not sure why this is, maybe the negative of it combined with the albedo hue cancels nicely into whites instead of greens?
All that being said, this is a good topic of discussion. The skin shaders in both Toolbag and Skyshop could use more work. //TODO.
Not sure if this is intentional or a bug. When I load my spec map it automatically sets it to srgb color space, but everything else to standard. If it isn't a bug, what is the reason behind it?
Not sure if this is intentional or a bug. When I load my spec map it automatically sets it to srgb color space, but everything else to standard. If it isn't a bug, what is the reason behind it?
This is not a bug, but simply the default behavior.
Normal maps are always in linear space (sRGB off)
Gloss maps are usually linear as well (but this may vary depending on your workflow/engine)
Albedo maps are usually sRGB/gamma space
Specular maps are usually gamma space as well
So depending on where you load a texture into, these are the default settings. If you mistakenly load a normal map into the albedo channel, gamma space gets "stuck" to that texture, and its important to remember to change it back to linear in the per-texture options, vise-versa for loading albedo maps into the normal slot.
What I mean is a library or collection of that data online. I don't really have the resources or time to be able to shoot that kind of photo reference for all my projects.
What I mean is a library or collection of that data online. I don't really have the resources or time to be able to shoot that kind of photo reference for all my projects.
Unfortunately no, there are bits and pieces here and there, generally reflectivity values for raw metals are easy to find, but there is no easy to access database really.
It looks like it has reflectivity values for a variety of materials, but not albedo or roughness/gloss values so it may be of limited used. Generally, everything that isn't a metal reflects in a small range of like 0.02-0.04 or so, while metals are in the 0.9-1.0 usually.
Gloss and albedo simply vary a good deal, so I find it really helpful to find reference close to the specific type of material I want to use and start there. Again Megascans is great because it has a decent variety of real world materials, though it isn't officially out yet so its not particularly helpful. I tried to include a wide variety of materials there, so you can sort of extrapolate and come up with logical values there by finding similar ones on the chart.
I found an extensive database at some point but it used some proprietary file format and I couldn't really make sense of it. Can't remember what it was called now.
Can you tell me which links you're having trouble with? The link now goes to the temp Magescans site (will go to the final one when its live).
Oops, thought you guys were testing the posted link, but it turns out he was saying that it needed to point somewhere else and I didn;t read carefully. (It works now and probably did back then.)
Can you tell me which links you're having trouble with? The link now goes to the temp Magescans site (will go to the final one when its live).
Unfortunately no, there are bits and pieces here and there, generally reflectivity values for raw metals are easy to find, but there is no easy to access database really.
It looks like it has reflectivity values for a variety of materials, but not albedo or roughness/gloss values so it may be of limited used. Generally, everything that isn't a metal reflects in a small range of like 0.02-0.04 or so, while metals are in the 0.9-1.0 usually.
Gloss and albedo simply vary a good deal, so I find it really helpful to find reference close to the specific type of material I want to use and start there. Again Megascans is great because it has a decent variety of real world materials, though it isn't officially out yet so its not particularly helpful. I tried to include a wide variety of materials there, so you can sort of extrapolate and come up with logical values there by finding similar ones on the chart.
I found an extensive database at some point but it used some proprietary file format and I couldn't really make sense of it. Can't remember what it was called now.
Yea that is generally how I approach it. I was just wondering if there was data out there besides Quixel's mega scan stuff that could help with this. I really appreciate the sample that you posted, it helped narrow down the guessing game a lot more.
On a side note, I noticed that they have the spec of everything have a tint of color, not much, just a couple steps. I am wondering if there is any way to observe this, or if you have ever observed this in a photo. I always try to keep a close eye on my reference but I have never seen any evidence of this. Do all non metals have a subtle hint of color in their specular that I cannot see?
Thanks for your help as always EQ. I appreciate it.
This is some awesome info, thanks! I'm quite new to the PBR workflow and wanted to learn more about it.
I was reading Etienne's (Slide) article in this month's 3D Artist in Smiths. When I read how much detail he goes into, e.g. Different values for each real world material like plastic, etc. I just burst out laughing in the middle of the store, feeling a little noobish and realising how much new stuff there is to learn
Hey, guys! I have a nasty self shadowing problem in Toolbag2. Here on the first screenshot I designated the problem areas. This straps have a little bit of thickness and in the most areas opposite faces are deleted, however in some areas they are not because I need them to be there and it's what causing this nasty self shadowing issue.
What's more strange is that if I'll hide the body of the car which is the biggest part of the mesh, most of the self shadowing instantly get fixed except for some areas
Turning Contact Refinement on and off for the light source do not help. Ofcourse I can just simply delete all the opposite faces on the straps (this fixes the issue), but then I'll loose some of the shadows that they are casting, especially on the wheel and sleeping bag rolls which is not good. Any ideas what's going on here and how this issue can be solved?
@ Yury - Yeah it's an issue that has been there since release. One way to get decent results is to separate the object and import it, then check off backface cull and cast shadows. You don't get cast shadows from those pieces, but they won't look oddly dark. Alternatively if you need the shadows and don't care about tri count, you could just shell those pieces.
Is there a way to get orthographic views, top, right side, front etc ? For example, it seems like setting up a full scene with object, lights etc could get rather complicated just by using the 3d view. Sorry if this exists, I just downloaded it...
By the way, thanks for putting this out, it's amazing.
I have a quick question reguarding AO and the sky light. It seems as though if you don't have a sky light you will not get AO between two seperate objects. This causes things to seem floaty. I made a test scene to illustrate what I am talking about.
With the skylight turned on: without the skylight and a point light casting shadows.
@ Yury - Yeah it's an issue that has been there since release. One way to get decent results is to separate the object and import it, then check off backface cull and cast shadows. You don't get cast shadows from those pieces, but they won't look oddly dark. Alternatively if you need the shadows and don't care about tri count, you could just shell those pieces.
Could it be fixed with some tweaks to shadow depth bias multiplier if one exists in MT 2
@ Yury - Yeah it's an issue that has been there since release. One way to get decent results is to separate the object and import it, then check off backface cull and cast shadows. You don't get cast shadows from those pieces, but they won't look oddly dark. Alternatively if you need the shadows and don't care about tri count, you could just shell those pieces.
Thanks. I'll try shell modifier. For the presentation purpose I don't care that much about tri count really, but I woul love to have proper shadows on my final renders :poly142:
I hope this problem will be fixed in the future releases though
Hey I've tried a few times to use a Substance file in Marmoset and it only kind of works...
What are the mappings to auto-place textures? I figured out Smoothness/roughness = gloss, but my Metallic map is ending up in the Albedo slot for some reason? I know how to use these things in Substance, so I think something's wrong with Toolbag
I can't find any documentation about this!
edit: I can give you some suggestions on how to improve it, but I'd like to know the facts of what is there first...
Replies
This can be sort of finicky but you can do it. You need to drag to right in between the two objects, and the little UI element should snap to the upper level in the heirarchry when you have the position right.
The bug I am referring to is the one I wrote about in this post.
http://www.polycount.com/forum/showpost.php?p=1972767&postcount=551
This was/is still occurring on 2.02
Thanks,
Eric
Is it somewhere and I just miss it may be? or is there a way for doing it manually somehow?
http://www.marmoset.co/toolbag/learn/pbr-theory
Part 2, covering the practical aspects of art creation coming soon.
great tutorial, thanks.
would love to see a tutorial on skin shading at some point. right now all the samples i see seem to have a greenish tone to the skin and i am also having a hard time getting the right skins colors.
if you mean desaturating the diffuse map then no that does not help.
from what i have seen the green tint is a result of the combination of subdermis scatter color (or opposite it) and the diffuse in skin. so by default if we have a redish scatter color then the opposite blue-ish color mixes with the yellow on the skin to give a greenish tone. i could be completely wrong but so far this is what i observed.
only thing that helps is using white scatter color or at least very desaturated red for the scatter color. the more saturated red color in scatter the more greenish tone on the skin given that the skin has a generic melanin tone on it. but using white scatter color we dont get any nice redish scatter any which is also required for skin shading.
if this is part of some sort of energy conservation then perhaps a slider to control that would be nice.
Here is a screen of what I am seeing. Like I said I am not sure there is much of a way around the green when you have a red subdermis.
Yeah, this has been missing for a while!
Other than that, been trialling the new toolbag for a couple of days now and love it. Job well done, guys.
If not put TB1 back on sale and I will buy it because TB2 is just too much for me
I'm not sure if this will solve your issues, so let me know if you're still having problems after reading it. I'm out of the country for now but I may be able to get you a more detail response sometime soon.
This is on the todo list.
I am sorry but Toolbag 1 is no longer in active development, as such we have no plans to offer it for sale. Current Toolbag 1 licences will continue to function, however maintenance will be very minimal (only fixing major bug fixes relating to crashing etc).
Toolbag 2 is, in our opinion at least, a massively improved product, and we plan to add many more features for free with updates in the future. If you haven't yet, I recommend trying out the trial version.
We may run sales for TB2 at some point, but I can't give you any specific information about that, its best to follow our facebook page https://www.facebook.com/marmosetco for information about sales and other news.
this paragraph goes into the reason:
however i don't think this is the correct way to make the skin shader work/look even though theoretically it is following real world physics.
the end result is not realistic may be because the shader is following one rule of physics and not all of them which would be impossible.
here is the final result in the tutorial, as you can see the diffuse looks too desaturated and the lit area still has quite a bit greenish/bluish tone to it.
i just found a quote from Andreas explaining this.
i understand that this is based on real world physics. BUT lets keep in mind that everything here and everywhere in any art is always fake. some more fake than others. no rendering system is truely physically correct since that is completely impossible.
so may be we can bend some rules to make the end result better ?
i dont know what would be a solution to this but may be start with a slider for energy conservation for the subdermis scatter ?
Gonna second this. I would rather have it so I can control the results when it isn't working out.
A big part of the problem is you have a photo of skin for your albedo, lit evenly with white light from all angles to make for a nice texture. But all that omni-directional light is already scattered in your photo.
This wrap lighting and energy conserving tint is an attempt to deconstruct that scattering and simulate what kind of shading from a single light direction could cause the skin to look like your photo under an omni-directional light source.
You could probably come up with better approximations of this deconstruction, but conserving energy and making sure the sum of all lighting adds up to 1.0 is a big part of what makes IBL skin work. Without the conservation, weird things start happening as your sky lighting gets more uniform and ambient.
As practical advice: I'd suggest desaturating your subdermis color more as a way to control the conservation. If your subdermis color is white/grey, all wavelengths of light are scattered equally and you don't see any tint either way. This will make your skin shading softer without changing the appearance of your albedo [diffuse] texture at all.
I've also had better luck matching the subdermis color closely to the hue of the albedo texture. I'm not sure why this is, maybe the negative of it combined with the albedo hue cancels nicely into whites instead of greens?
All that being said, this is a good topic of discussion. The skin shaders in both Toolbag and Skyshop could use more work. //TODO.
This is not a bug, but simply the default behavior.
Normal maps are always in linear space (sRGB off)
Gloss maps are usually linear as well (but this may vary depending on your workflow/engine)
Albedo maps are usually sRGB/gamma space
Specular maps are usually gamma space as well
So depending on where you load a texture into, these are the default settings. If you mistakenly load a normal map into the albedo channel, gamma space gets "stuck" to that texture, and its important to remember to change it back to linear in the per-texture options, vise-versa for loading albedo maps into the normal slot.
PBR: And you can too!
Thanks, should be fixed now.
nope
It works for me.
try cross polarization photography.
What I mean is a library or collection of that data online. I don't really have the resources or time to be able to shoot that kind of photo reference for all my projects.
Can you tell me which links you're having trouble with? The link now goes to the temp Magescans site (will go to the final one when its live).
Unfortunately no, there are bits and pieces here and there, generally reflectivity values for raw metals are easy to find, but there is no easy to access database really.
You could try this: http://refractiveindex.info/
It looks like it has reflectivity values for a variety of materials, but not albedo or roughness/gloss values so it may be of limited used. Generally, everything that isn't a metal reflects in a small range of like 0.02-0.04 or so, while metals are in the 0.9-1.0 usually.
Gloss and albedo simply vary a good deal, so I find it really helpful to find reference close to the specific type of material I want to use and start there. Again Megascans is great because it has a decent variety of real world materials, though it isn't officially out yet so its not particularly helpful. I tried to include a wide variety of materials there, so you can sort of extrapolate and come up with logical values there by finding similar ones on the chart.
I found an extensive database at some point but it used some proprietary file format and I couldn't really make sense of it. Can't remember what it was called now.
Oops, thought you guys were testing the posted link, but it turns out he was saying that it needed to point somewhere else and I didn;t read carefully. (It works now and probably did back then.)
Yea that is generally how I approach it. I was just wondering if there was data out there besides Quixel's mega scan stuff that could help with this. I really appreciate the sample that you posted, it helped narrow down the guessing game a lot more.
On a side note, I noticed that they have the spec of everything have a tint of color, not much, just a couple steps. I am wondering if there is any way to observe this, or if you have ever observed this in a photo. I always try to keep a close eye on my reference but I have never seen any evidence of this. Do all non metals have a subtle hint of color in their specular that I cannot see?
Thanks for your help as always EQ. I appreciate it.
I was reading Etienne's (Slide) article in this month's 3D Artist in Smiths. When I read how much detail he goes into, e.g. Different values for each real world material like plastic, etc. I just burst out laughing in the middle of the store, feeling a little noobish and realising how much new stuff there is to learn
What's more strange is that if I'll hide the body of the car which is the biggest part of the mesh, most of the self shadowing instantly get fixed except for some areas
Turning Contact Refinement on and off for the light source do not help. Ofcourse I can just simply delete all the opposite faces on the straps (this fixes the issue), but then I'll loose some of the shadows that they are casting, especially on the wheel and sleeping bag rolls which is not good. Any ideas what's going on here and how this issue can be solved?
By the way, thanks for putting this out, it's amazing.
With the skylight turned on:
Could it be fixed with some tweaks to shadow depth bias multiplier if one exists in MT 2
Thanks. I'll try shell modifier. For the presentation purpose I don't care that much about tri count really, but I woul love to have proper shadows on my final renders :poly142:
I hope this problem will be fixed in the future releases though
What are the mappings to auto-place textures? I figured out Smoothness/roughness = gloss, but my Metallic map is ending up in the Albedo slot for some reason? I know how to use these things in Substance, so I think something's wrong with Toolbag
I can't find any documentation about this!
edit: I can give you some suggestions on how to improve it, but I'd like to know the facts of what is there first...
can someone explain to me what the "sRGB colour space" option is doing in the materiel options?
I assumed that all normal maps use RGB colour space however my assets seem to behave weird when this is checked.
My spec map appears dead and the normal maps don't behave as expected.
When turning it off things seem to look better.
An explanation to what is happening or if I may be doing something wrong would be really helpful.
Thanks everyone.
http://youtu.be/Qe10ExwzCqk
Dota 2 shader support is now in, as well as a few more features and a bunch of bug fixes.
Full change log here: http://www.marmoset.co/toolbag/history
http://vimeo.com/87235066
This video was recorded before we got a few things in, like VMT loading, so that isn't shown.
Do you guys at least see things like "Dota 2 Specular", and "Dota 2 Diffuse" under Reflectivity, or Diffusion?
true i did noticed but not the template as showed in the video...