Thoughts so far after playing a bit more. First of all it's absolutely amazing so far
1 - The object list in the top left corner of the screen is great! Would it be possible to add windows like functionality to the list? Shift does add to selection but CTRL doesn't deselect and selecting an object and then shift clicking another object further down the list wont select everything in-between either.
2 - I couldn't figure out how to delete unwanted objects or materials.
3 - Dragging and droping a decimated .obj from zbrush crashed toolbag 2.0 instantly but worked fine when using the "create" menu option.
4 - I agree with the suggestion to change the "create mesh" menu option to "add mesh"
5 - I too would love having a metalness mask option. And it makes total sense after reading Almighty Gir's explination of the UE4 method.
6 - I couldn't find any place to get anisotropy going. If I want to fake hair I would love love love the ability to fake it.
7 - It would be amazing if there was a repository "Cheat sheet" included with TB2 that tells artists ballpark values for known materials for reflectance and gloss.
8 - The peachfuzz setting in the skin shader may be useful on other shaders for creating a cloth-like sheen effect?
9 - I wanted to ask about some of the things in this link,
A. page 22: Does TB2.0 roughness value effect the diffuse as well as reflection?
B. page 23: Specular tails. I kept trying to get a good reflection on the eyes of a character and find it very difficult. What do you guys currently have in TB2? From the results it looked close to the Beckmann model. This also ties into lights I think ( I watched the presentation from guerilla games on their PBR setup for the new killzone, they said artists tended to exaggerate the gloss values in order to get a more broad highlight from a point light but this lead to all sorts of problems because the gloss value no longer represented the material they were trying to emulate) The solution proposed by guerilla games is to give point lights a "shape" that shows up in reflections. here's the section of the video I'm referring to.
More to come soon! Loving it so far!
Giving feedback on feedback, lols.
1 - seconded!
2 - click the object in the list then press delete? haven't tried deleting a material yet.
3 - i just tried it and have the same issue.
4 - completely agree.
5 - thanks, i'm actually writing a more convincing argument for this... more to come later! that said, i can also understand why they don't want it.
6 - i think EQ said Aniso shaders are coming, which is awesome. also want it for brushed metal... although, in my opinion it would be better to be able to mask aniso on/off in specific areas of a material rather than have its own unique material.
7 - i've got a couple that i'll upload later. they're by no means a complete list but a good starting point.
8 - i can see this working...
9a - because it's realtime, it's highly likely that they're just using lambertian diffuse. roughness wouldn't directly affect it, except for during energy conservation, where the total light between specular and diffuse cannot exceed the amount of light hitting the surface (ie. if you have a very bright specular highlight that reflects 90% of the light, the diffuse will only reflect the remaining 10% and would therefor be quite dull).
9b - it looks like (and to be honest, it's up to them whether they want to be transparent about their math.) they're using a "physicalised" blinn-phong rather than any of the other physical models. it's an aesthetic thing though. i'd certainly love it if they included a variety of physical models for us to play with, blinn-phong, ggx, beckmann etc.
will try to do another 1 day asset today, not sure if i have time though.
So here's why the workflow i suggested before is a good idea, in my opinion (feel free to disagree):
Currently the industry is going through a shift from one workflow to another. one of the great things about the new workflow is that it's possible to have a very efficient and streamlined way of approaching materials. I personally believe that toolbag2 should be a driving force in the simplification process.
Currently in Toolbag2 you can have a specular colour map, and use the intensity slider. what's important to understand here, is that the intensity slider DOESN'T reprisent specular intensity, it reprisents reflectivity. you can demonstrate this by putting gloss to 1, and intenisty to 0.04. regardless of your diffuse colour, you will have a glossy plastic material, and at intensity 1, you have a metallic material. Currently, without a coloured specular map you will only have a white specular value - this isn't necessarily a bad thing.
so in order to differentiate between non-metals, and metals in toolbag 2, currently you should use a coloured diffuse for non metals, and a black diffuse for metals (where possible). the reason it's important to use a black diffuse for metals is that it's technically possible to still have some colour in the underlying diffuse even with the reflectance slider set to 1. because it multiplies with your specular colour (which is always less than 1 if it's a colour) the final result is less than one. and because of energy conservation, the final diffuse is 1 - finalspecular. here's a very simplified math of that:
white = 1,1,1 (or 255,255,255) - this is our diffuse.
gold= 1,8.4,0 (or 255,215,0) - this is our specular.
reflectivity = 1.
so the (extremely simplified) math would go:
diffuse * (1 - (specular * reflectivity)) = diffuse
in numbers:
1,1,1 * (1 - (1,0.84,0) * 1)) = 0,0.16,0 (or 0,40,0)
this is technically incorrect, because metals should have no diffuse as no light can actually enter their surface and scatter. so in this way toolbag2 is not entirely correct as a physically based solution.
that's a technical problem rather than an art problem though. so here's the problem for the artist -
due to the technical problem above, artists will either have to very strictly change their workflow by creating black diffuses for metals etc. or they will be fudging values the way they always used to, which will lead to a lack of progression.
as of right now, artists need to supply two different maps, a diffuse map, and a specular map. here's why i think that's a bad thing -
- Metals have no diffuse and are purely reflective. this means that the colour of the metal in the specular map should always be multiplied by 1. if it's multiplied by less than 1, then it's no longer as reflective as it should be.
- Non metals have varying degrees of reflectivity, but on average they're extremely low, around the 0.04 mark.
this means that when an artist creates a specular map which contains both metals and non metals, anything that is non metallic should be clamped between 0 and 0.04 (or 18,18,18 ). how many artists are truly likely to do this? i think most will work outside of these ranges either out of ignorance or simply thinking it's "wrong". let's face it we're visual creatures and there are many times when our eyes play tricks on us, making us think things are correct when they aren't. so when an artist works outside of that range, and all of a sudden their surface is more reflective than it should be in toolbag, and as a result the diffuse is also darkened due to the energy conservation rule, what do they do? they adjust the slider! but adjusting the slider means now their metals are incorrect! OH NOES!
that's why i believe we should remove the intensity slider, and just have a metallic map instead. if you really want to have a specular map as well (for pearlescent effects and stuff) then go for it. but the intensity slider absolutely needs to go.
and as for roughness/gloss? pretty much the same as above. having the slider there just allows artists to work within an incorrect range of values. personally i think it should go, and artists should learn to make their maps to the correct values (that, and i can't remember a time when i've had it, or seen other people have it set to less than 1).
please feel free to correct me if any of this is wrong. TB2 is {} this close to being perfect (shaderwise). don't look at the removal of those "features" as the creation of a limitation, it's really not. it's actually removing the limitation from artists that the old workflows had inherent.
Thanks for the response EQ. Here's a better screenshot with reflections and bg color changed.
2 requests for a next release( maybe?):
- If I could blur the "mirror" feature, mipmap it somehow. I know I could blur the sky but... you know. Or even have a feature to blur the sky itself! Or just folow the gloss map just as the blinn does.
- Soft shadows or penumbra (The Size/Softness doesn't seem to be doing on the shadows, only on the shading, am I right?)
THerefore I'm still loving this, I'm thinking on doing a whole prop just to test this entirely. Again, thanks for the opportunity. Can't wait for a next release.
EDIT:
I just tested the light editor on the skymap, fantastic idea.
I found a bug it seems. Pressing F9 doesn't take a screenshot to clipboard, I have to go to the menu to do it.
Tested also SSS (terrible model and textures, it's old, but you get the idea.)
Hey dude, if you want blurry reflections, you shouldn't use the mirror specular model. Really the only difference between the mirror model and blinn-phong is that mirror is clamped to the highest possible specular reflection resolution. So just switch over to blinn-phong and adjust the gloss as you like to control the blurriness of reflections.
Soft shadows: We've got some scale issues it looks like, we expect models to come in at at certain size, so the light size function is tied to that, if your mesh is a lot large than the program expects, you may need to manually type in much higher values into the light size. See if that helps, meanwhile we'll look into what we can do re:scale. You can also try scaling down your mesh (or up, you can get an idea of relative scale by moving the mesh around and looking at the units, if you move it halfway across the screen and it says you moved it 0.005 units, its scaled small, if its says 500 units, its scaled large).
Soft shadows: We've got some scale issues it looks like, we expect models to come in at at certain size, so the light size function is tied to that, if your mesh is a lot large than the program expects, you may need to manually type in much higher values into the light size. See if that helps, meanwhile we'll look into what we can do re:scale. You can also try scaling down your mesh (or up, you can get an idea of relative scale by moving the mesh around and looking at the units, if you move it halfway across the screen and it says you moved it 0.005 units, its scaled small, if its says 500 units, its scaled large).
What scale is TB2 setup to expect btw? Is there a way to know beforehand?
EDIT2:
- Is there a way to toggle of visibility of light icons in the viewport? I know they aren't in the final exported image but they can get be distracting when trying to find a good composition.
- PBR shaders are only half of a PBR renderer, I notice there is a manual slider for light falloff on dynamic lights in TB2. Are there any plans to implement a more physically based lighting equation? (quadratic falloff perhaps.)
So here's why the workflow i suggested before is a good idea, in my opinion (feel free to disagree):
...
and as for roughness/gloss? pretty much the same as above. having the slider there just allows artists to work within an incorrect range of values. personally i think it should go, and artists should learn to make their maps to the correct values (that, and i can't remember a time when i've had it, or seen other people have it set to less than 1).
Certainly a lot to think about, Gir, thanks for the input.
We've always considered the sliders as a way to tweak and play with the texture content within the app. So instead of switching to photoshop to adjust the curves on your gloss map, you just mess with a slider in TB and see what works. I can see your point about this leading to bad content creation practices though.
The color pickers for modifying diffuse and specular color are a similar shortcut. You *should* be putting that color in the textures themselves but it certainly is handy to play with it in the app. Would you still want to keep those?
A compromise Earthquake and I came up with was to have a slider or color-picker present only if the texture slot is empty. That way you can still tweak the material on models with no UVs, preview it before you've finished the spec-map, whatever. But the interface is cleaner and forces you to work through maps instead of arbitrary scalars at the end.
Now with gloss, one thing I was initially pushing for was to have two slider values: both a min and a max value for gloss. Blinn-Phong gloss values technically range from 0 to infinity. A very high gloss value being a very high specular exponent, making for a very tiny point of a specular highlight; perfect mirror would be at infinity.
In TB2 we just took an arbitrary "high enough" value and mapped that to between 0 and 1. But you could technically map your gloss map to any range of values. You also only have 256 shades of gloss to work with, so spreading your 8-bit gloss channel out across a huge range may not be desirable. Thoughts?
i think the idea with color/slider when you have no texture is pretty good, it will help previewing certain things pretty quickly.
I couldn't test yet so i have no clue yet, but giving the possibility to copy the color value or slider value over as a shade of grey to photoshop after finding the right one, would be a pretty cool option.
Certainly a lot to think about, Gir, thanks for the input.
We've always considered the sliders as a way to tweak and play with the texture content within the app. So instead of switching to photoshop to adjust the curves on your gloss map, you just mess with a slider in TB and see what works. I can see your point about this leading to bad content creation practices though.
The color pickers for modifying diffuse and specular color are a similar shortcut. You *should* be putting that color in the textures themselves but it certainly is handy to play with it in the app. Would you still want to keep those?
A compromise Earthquake and I came up with was to have a slider or color-picker present only if the texture slot is empty. That way you can still tweak the material on models with no UVs, preview it before you've finished the spec-map, whatever. But the interface is cleaner and forces you to work through maps instead of arbitrary scalars at the end.
Now with gloss, one thing I was initially pushing for was to have two slider values: both a min and a max value for gloss. Blinn-Phong gloss values technically range from 0 to infinity. A very high gloss value being a very high specular exponent, making for a very tiny point of a specular highlight; perfect mirror would be at infinity.
In TB2 we just took an arbitrary "high enough" value and mapped that to between 0 and 1. But you could technically map your gloss map to any range of values. You also only have 256 shades of gloss to work with, so spreading your 8-bit gloss channel out across a huge range may not be desirable. Thoughts?
Hey man,
the colour swatches/sliders only for when there is no texture present is actually a fantastic idea. it's a great way for you to be able to preview how a material "should" look, so you can try to emulate it in your textures, at the very least this would make it a great learning tool!
on gloss:
i had assumed you were using blinn-phong and having "1" set to 255, so you're not actually interpolating from a grayscale texture. i've played around with Beckmann, Trowbridge-Reitz, Cook-Torrence, and Blinn-Phong recently, of them all Blinn-Phong is the only one that allows you to go to an infinitesimal value, all of the rest fit very nicely in the 0-1 range... in fact if you go outside of that range you get some rather bad artifacting. i think that's one reason why i'm against having the slider there... the "big names" in PBR at the moment are all using the 0-1 range inputs and they feel very comfortable to use. there's no need for sliders or anything, you just make your map and put it in and it works. there's no need to adjust that value, it is in my opinion, something that leads to bad habbits from the following point of view:
as an artist, if you're used to creating any old gloss map, and then adjusting a slider afterward to make it look good. and then you go work in a studio where that's a static value that can't be adjusted, you're going to have to adapt to that. whereas if you're used to playing within the specific range of values and get consistently good results, when you go to a studio that does allow you to adjust it, you won't need to. you'll just set it to max and be happy!
also, as i mentioned before, i can't think of a time when i've seen it set to less than max when used with a texture. i can understand the slider being there for previewing materials (like i said in the previous paragraph, great learning tool). but i just don't see the need for it outside of that. again, personal opinion and i'm sure there are others who will disagree.
on a side note. i'd just like to thank you and EQ for taking the time to talk to me here and via pm. it's very humbling and i'm happy to be of any help i can be, even if it's as a constant nag :P
I did a quick test with one of our asset at work. It looks pretty damn good! Although, I'll have to agree that the all "tweaking sliders/colors" part is, at least for me, a step backward from what PBR is trying to achieve from a workflow point of view.
We finally have the tools to create something predictable and accurate, which will (FINALLY) force artists to really edit their textures the proper way, but most importantly, the same way(!). And while I can understand that from a "user-friendly" perspective you want to avoid people to go back and forth between photoshop and Toolbag to get what they want, I think it would be a miss opportunity to help out artists to be "production ready"
I'd personally like a "lite" version of what you guys have with WAY less options. No more check box for energy conservation (should be on by default), no more sliders (as long as you have a texture, like some people already mentioned it) and no more color overrides. Everything should be texture driven, just like it would happen on a real production where you want to have as much consistency as possible.
Quick additional user-interface note : While playing with the different skies, it looks like each time you rotate your mesh in the viewport, you loose the "sky menu" on the left panel. It can becomes really "clicky" when you don't know right away which sky you want...
Overall, really impressed with the engine improvement, you guys did an amazing job with that! I just wish the shader setup would be as next-gen as the engine itself (the simpler the better!).
I'm soo hoping that this will allow me to dump standard rendering for at least 70-80 percent of my needs.
Are there soft dynamic shadows in TB2.0? How about some nice glass example?
Can we see higher res, 8K or something like that, version of this image?
All dynamic light types cast shadows now, and soft shadows at that. Be sure to check out the feature overview here: http://www.marmoset.co/toolbag/tb2-new
We don't have refraction in yet (which is what you'll probably want to use for glass). I've been using the dithered order independent transparency for glass which really isn't at all what its meant for. So unfortunately I don't have any awesome glass examples.
I found a bug it seems. Pressing F9 doesn't take a screenshot to clipboard, I have to go to the menu to do it.
I just tried the F9 copy thing and its working for me here. One issue I get, if performance is pretty low it has a tendency to drop keyboard input for me, so maybe try again, press F9 a couple times to make sure it sticks and let me know if you still have the same bug.
I had a minute to continue testing TB2.0 so I threw one of the titans from sins of a solar empire into marmoset as a test model to play around with. The first image is an old render from mental ray and the second is from toolbag 2.0. Absolutely stunning. If I play with the lights a bit more I probably could almost exactly replicate the first image.
Well done guys! well done! I'm amazed I never thought the day would come so soon when I can barely tell the difference between something rendered in realtime and something that took about 2 minutes a frame to render.
Awesome! I did a similar test a while back with a Modo render I had done a few years ago:
And then the Toolbag 2 render:
Its not a perfect match but I was very happy with it, especially considering how utterly awesome Modo's offline renderer is.
B. page 23: Specular tails. I kept trying to get a good reflection on the eyes of
a character and find it very difficult. What do you guys currently have in TB2? From the results it looked close to the Beckmann model. This also ties into lights I think ( I watched the presentation from guerilla games on their PBR setup for the new killzone, they said artists tended to exaggerate the gloss values in order to get a more broad highlight from a point light but this lead to all sorts of problems because the gloss value no longer represented the material they were trying to emulate) The solution proposed by guerilla games is to give point lights a "shape" that shows up in reflections. here's the section of the video I'm referring to.
TB2 does have area lights which should affect the shape of the highlight. Change the "Size/Softness" of the light.
Regarding the gloss/roughness value sliders. I'm all for a correct workflow but there is still a need to preserve some level of backwards compatibility regarding textures that are non PBR compliant. At least there should be a fallback mode for TB1 scenes so people are not forced to retexture already finished scenes rather than tweaking in TB2 until they get something they feel happy with, but again I think this should be a fallback/legacy option and marked as so somewhere and definitely not on by default.
I like the idea of having the colour picker when a texture isn't present though.
As for the levels of gloss available when using a texture, I would say that a value of 0-1 would make more sense than a an infinite range.
I always use the brightness value of 0-100% in PS rather than 0-255 values so this workflow wouldn't really change anything for me in that regard, though I do hate linear colour pickers, so I wouldn't want that at all. The ability to add hex code and RGB values is super important, even if they are converted internally before rendering. Usability FTW.
TB2 does have area lights which should affect the shape of the highlight. Change the "Size/Softness" of the light.
SWEET.
So Are there any upgrades to toolbag that make it a viable option for environments? It would be cool if it could just read a 3ds max scene and live update it and all the cameras and lights in it.
Playing about with the skin shader tonight, its crazy how much the lighting environments affect it....are there any plans for a whitebox style environment/default 3 point light scene so that you can get the levels just right, or is it a create it yourself kind of issue?
Bug: draging the colour picker circle to the top right corner of the colour display makes the hue and saturation sliders lock into place at the top of their bars till I move the circle again. I dont know if it matters but I'm using a wacom.
Edit* another dabble this morning, transparency in captures is a bit rough, in psd you get a lot of missed pixels in your object, in tga there is a white pixel thickness border to the object, but fewer missed pixels.
Turntable settings for both scene and camera bring the lights with them, is there an option that I am missing just to rotate the model?
Do not have the beta, new features look really nice on your page
2 things however; Now with the more and more powerful presentation of realtime assets, can you even talk about a real preview of game usage ? The normal marmoset seemed pretty borderline to me and while I personally don't give anything about that, I only look at the visuals, but there seem to be many people looking for a higher moral ground in that regard from what ive been reading
Also, dont want to be that guy, but your interface looks not really nice and definitely an downgrade visually. I would assume it is a placeholder, but you write shiny new interface on your page, so I dont know what to think about.
It seems to borrow visuals from the older windows and OSX at the same time, and reminds me of render packages from the 2000s to be perfectly honest
The output images look fantastic, with realtime becoming so strong, Im not sure if I should continue with my offline workflows : )
All dynamic light types cast shadows now, and soft shadows at that. Be sure to check out the feature overview here: http://www.marmoset.co/toolbag/tb2-new
We don't have refraction in yet (which is what you'll probably want to use for glass). I've been using the dithered order independent transparency for glass which really isn't at all what its meant for. So unfortunately I don't have any awesome glass examples.
Thanks.
1. Are you planing to support Alembic? FBX loading in TB1.0 is slow and even in Modo and Houdini FBX is slow to load, a lot slower than Alembic.
2. Are you planing to support Pixar SubD surfaces?
I've had a chance to have a little play with it and it's been fun so far.
Some of these have probably been mentioned or are already planned for later releases:
-Bug: Light widget disappears
-Lights need some visual representation to show direction/radius etc like in TB1
-Visibility toggle in scene tab for meshes, lights etc.
-Option to Lock camera view/position
-Option to pause turntable, rather than stop jump back to the start.
-Scale reference and scale option (like in TB1)
-2 sided option
-Option to save/load materials would be handy too.
I was also getting some thin black artifacts around meshes when I was using image capture (even with high sampling). The image below is from Image to clipboard option.
The skin shader is great, had lots of fun messing around with the settings and the secondary spec options are really cool as well.
Couldn't get any good results with the transparency for the hair (hence the heavy DOF on the hair ), probably just need to adjust my maps though. Looking forward to playing with the anisotropic and other materials and the rest of the goodies you guys have planned.
Hello.
A little, I tried to test.
very cool new features,
I favorite Local reflection&New DOF!
By the way, I can't find the detailmap and chromatic aberration.
Function of it is my favorite in TB1.
I've had a chance to have a little play with it and it's been fun so far.
Some of these have probably been mentioned or are already planned for later releases:
-Bug: Light widget disappears
-Lights need some visual representation to show direction/radius etc like in TB1
-Visibility toggle in scene tab for meshes, lights etc.
-Option to Lock camera view/position
-Option to pause turntable, rather than stop jump back to the start.
-Scale reference and scale option (like in TB1)
-2 sided option
-Option to save/load materials would be handy too.
I was also getting some thin black artifacts around meshes when I was using image capture (even with high sampling). The image below is from Image to clipboard option.
The skin shader is great, had lots of fun messing around with the settings and the secondary spec options are really cool as well.
Couldn't get any good results with the transparency for the hair (hence the heavy DOF on the hair ), probably just need to adjust my maps though. Looking forward to playing with the anisotropic and other materials and the rest of the goodies you guys have planned.
Would you mind posting your textures used on the skin? I'm curious how many maps you used and what your settings are in TB2. It looks really nice I'm planning on doing a skin test soon in TB2 and I'm wondering how close to other workflows the TB2 skin shader is.
-Lights need some visual representation to show direction/radius etc like in TB1
-Visibility toggle in scene tab for meshes, lights etc.
-Option to Lock camera view/position
-Option to pause turntable, rather than stop jump back to the start.
-Scale reference and scale option (like in TB1)
-2 sided option
Nice render afisher, I agree with all of this.
Seems like you already can "save/load materials" there is a arrow besides where you duplicate and create a new material that has import and export inside it.
Nice renders guys. I'm concern about the displacement, is it working better? last time I created pretty obvious open gaps between uvs.
All tessellation in game engines will suffer from the same problem with the mesh breaking apart at the UV borders so I wouldn't expect that to get fixed any time soon tbh. It's down to the limitations of the algorithm that's being used and it's exactly the same in UDK and CE3 afaik. You would need to use fewer UV shells or place them where they would be hidden to remove this problem.
IMHO, It would be better for performance (and faster to code too) if something like parallax occlusion mapping with silhouette was added.
Just messing around with this now, I don't want to post anything because it's just really rough tests and a lot of "what does this do...what does that do?" type of shit. If I can, I'd really like to author something from scratch rather than taking old assets and making them work to really get a feel for the new process and what not.
On BSI, we used a similar rendering method. For our PBR we actually did have coloured diffuse, but it was practically black...very similar to how all of this is set up, just through unreal before it became "a thing." I need to really dive itno the skin stuff, Adam has had better luck with it than me it seems, but I'm sure that's a user error on my part and just not giving it enough time.
A lot of it is just echoing what has been said, I'm finding.
I think it's slick. I like the new format and how it resembles software that most people will be comfortable or familiar with. I like how everything is right there in the hierarchy and easy to see. The drag and drop workflow is awesome (which I think is new), how you manage assets is great and, like mentioned above, PBR is the way of the future. So many places are going in this direction...I think Toolbag2 is a great example of how powerful it can be and presented in a really clean way.
Some gripes:
-I naturally go to "File" to import a mesh because I'm not creating anything. (echo..echo..)
-Naturally trying to use shift to move the skybox, I can get over it..I'm just so used to it as a basic control.
-If there is a way, I didn't see it: Is it possible to just save the model and materials all together like in Toolbag 1 (Save Mesh and Materials)? Without saving the entire scene?
-I miss having representation for my lights. I seem to do more guess work and just wish there was a volume like in Toolbag 1 that showed me my lights in action.
-Could we have a "new" material that is just the skin shader without enabling it through a normal material?
-I wish things stayed selected from the hierarchy when navigating in the viewport. I feel like, I want to move around but still have a light selected to tweak while I navigate around. It seemed that way in Toolbag 1, but I would need to open it up to be sure.
All tessellation in game engines will suffer from the same problem with the mesh breaking apart at the UV borders so I wouldn't expect that to get fixed any time soon tbh. It's down to the limitations of the algorithm that's being used and it's exactly the same in UDK and CE3 afaik. You would need to use fewer UV shells or place them where they would be hidden to remove this problem.
Thanks Metalliandy for the explanation, really helpful so displacement algorithm still sucks, but why in Maya works perfectly? Some days I feel my brain will explode...
Thanks Metalliandy for the explanation, really helpful so displacement algorithm still sucks, but why in Maya works perfectly? Some days I feel my brain will explode...
For offline rendering its all done on the CPU. For realtime its done on the GPU with pretty standard fixed methods of doing it. Pretty much all game engines use the same method to tessellate and displace geometry, because the GPU API has specific features that do it. This is my basic understanding of it.
So yeah, as Andy says, its unlikely we can do anything about it unfortunately. If anyone knows of any realtime engines that do displacement without gaps at UV seams, or any whitepapers etc we would be very interested.
The best thing to do for now is to use as few uv seams as possible, hide uv seams as well as you can (behind other objects like cloths/props/hair is ideal), and using 16bit displacement may help a little as well
-If there is a way, I didn't see it: Is it possible to just save the model and materials all together like in Toolbag 1 (Save Mesh and Materials)? Without saving the entire scene?
Everything is saved in a single scene file now, mate. The file can be moved around on your HDD without breaking now too, which saves headaches
Thanks Metalliandy for the explanation, really helpful so displacement algorithm still sucks, but why in Maya works perfectly? Some days I feel my brain will explode...
Maya and other non realtime displacement solutions use Catmul Clark subdivision rather than realtime solutions such as PN Triangles and Gregory patches. Catmul Clark is a great subdivision solution but it also increases your polygon count by 4x for each level, which obviously makes it less suitable for realtime than more efficient solutions that are tailored towards game art and the like.
Just messing around with this now, I don't want to post anything because it's just really rough tests and a lot of "what does this do...what does that do?" type of shit. If I can, I'd really like to author something from scratch rather than taking old assets and making them work to really get a feel for the new process and what not.
On BSI, we used a similar rendering method. For our PBR we actually did have coloured diffuse, but it was practically black...very similar to how all of this is set up, just through unreal before it became "a thing." I need to really dive itno the skin stuff, Adam has had better luck with it than me it seems, but I'm sure that's a user error on my part and just not giving it enough time.
A lot of it is just echoing what has been said, I'm finding.
I think it's slick. I like the new format and how it resembles software that most people will be comfortable or familiar with. I like how everything is right there in the hierarchy and easy to see. The drag and drop workflow is awesome (which I think is new), how you manage assets is great and, like mentioned above, PBR is the way of the future. So many places are going in this direction...I think Toolbag2 is a great example of how powerful it can be and presented in a really clean way.
Some gripes:
-I naturally go to "File" to import a mesh because I'm not creating anything. (echo..echo..)
-Naturally trying to use shift to move the skybox, I can get over it..I'm just so used to it as a basic control.
-If there is a way, I didn't see it: Is it possible to just save the model and materials all together like in Toolbag 1 (Save Mesh and Materials)? Without saving the entire scene?
-I miss having representation for my lights. I seem to do more guess work and just wish there was a volume like in Toolbag 1 that showed me my lights in action.
-Could we have a "new" material that is just the skin shader without enabling it through a normal material?
-I wish things stayed selected from the hierarchy when navigating in the viewport. I feel like, I want to move around but still have a light selected to tweak while I navigate around. It seemed that way in Toolbag 1, but I would need to open it up to be sure.
I do have to agree that I miss some of the TB1 hotkeys simply because I'm used to them. Like hitting "T" to start and stop the turntables and such.
Here's a few more thoughts too:
- When using ctrl to move the skybox there's no way to reset it back to the default where the "Clear" button in TB1 for the corresponding turntable in the animation menu would usually reset it to the default rotation. Not a huge deal but If I had my model positioned to take advantage of the default rotation and I accidently hit the ctrl key and rotated the skybox it might be annoying to go back if I was OCD.
- There's no far and near clip plane controls on the camera so long objects have the front clipped during turntable rotations.
- I notice adjusting the size/softness barely has any effect on my lights, could this be a scale related issue?
- Double clicking a Toolbag scene file crashes toolbag after it launches. I'm forced to open the program and use the "open file" menu option for it to work.
- I couldn't find the skin material for awhile either. It might make more sense to have a skin shader listed under presets in the create new shader section.
Everything is saved in a single scene file now, mate. The file can be moved around on your HDD without breaking now too, which saves headaches
Maya and other non realtime displacement solutions use Catmul Clark subdivision rather than realtime solutions such as PN Triangles and Gregory patches. Catmul Clark is a great subdivision solution but it also increases your polygon count by 4x for each level, which obviously makes it less suitable for realtime than more efficient solutions that are tailored towards game art and the like.
What about Pixar open Subd? I was under the impression those could approximate Psub limit surfaces with DX11 sparse triangulation.
In the GPU codepath, the amount of subdivision and the density of Bspline tessellation are both independently controllable over the surface. So a part of the model close to camera can have more detail than a part of the same model far away.
Maya and other non realtime displacement solutions use Catmul Clark subdivision rather than realtime solutions such as PN Triangles and Gregory patches.
Hi,
This is in-correct.
I'm one of the people who implemented the tessellation and displacement in Maya and we have completely crack-free, real-time tessellation that is similar to PN Triangles (similar in cost) and we also support hard/soft edges (smoothing groups).
And, to the best of my knowledge, I thought Unreal also has crack free displacement, but I never tried.
From my experiments, the crack free displacement in UE3 seems to create a poly loop between the 2 sides of the cracks, it doesn't look very good but not as bad as a crack obviously.
i'm not familiar with the consequences. is that the full stop already? according to the packaging my graphics card supports some Direct-X from the future.
Will DirectX 10 be available for Windows XP?
No. Windows Vista, which has DirectX 10, includes an updated DirectX runtime based on the runtime in Windows XP SP2 (DirectX 9.0c) with changes to work with the new Windows Display Driver Model (WDDM) and the new audio driver stack, and with other updates in the operating system. In addition to Direct3D 9, Windows Vista supports two new interfaces when the correct video hardware and drivers are present: Direct3D9Ex and Direct3D10.
Since these new interfaces rely on the WDDM technology, they will never be available on earlier versions of Windows. All the other changes made to DirectX technologies for Windows Vista are also specific to the new version of Windows. The name DirectX 10 is misleading in that many technologies shipping in the DirectX SDK (XACT, XINPUT, D3DX) are not encompassed by this version number. So, referring to the version number of the DirectX runtime as a whole has lost much of its meaning, even for 9.0c. The DirectX Diagnostic Tool (DXdiag.exe) on Windows Vista does report DirectX 10, but this really only refers to Direct3D 10.
Replies
Giving feedback on feedback, lols.
1 - seconded!
2 - click the object in the list then press delete? haven't tried deleting a material yet.
3 - i just tried it and have the same issue.
4 - completely agree.
5 - thanks, i'm actually writing a more convincing argument for this... more to come later! that said, i can also understand why they don't want it.
6 - i think EQ said Aniso shaders are coming, which is awesome. also want it for brushed metal... although, in my opinion it would be better to be able to mask aniso on/off in specific areas of a material rather than have its own unique material.
7 - i've got a couple that i'll upload later. they're by no means a complete list but a good starting point.
8 - i can see this working...
9a - because it's realtime, it's highly likely that they're just using lambertian diffuse. roughness wouldn't directly affect it, except for during energy conservation, where the total light between specular and diffuse cannot exceed the amount of light hitting the surface (ie. if you have a very bright specular highlight that reflects 90% of the light, the diffuse will only reflect the remaining 10% and would therefor be quite dull).
9b - it looks like (and to be honest, it's up to them whether they want to be transparent about their math.) they're using a "physicalised" blinn-phong rather than any of the other physical models. it's an aesthetic thing though. i'd certainly love it if they included a variety of physical models for us to play with, blinn-phong, ggx, beckmann etc.
will try to do another 1 day asset today, not sure if i have time though.
Currently the industry is going through a shift from one workflow to another. one of the great things about the new workflow is that it's possible to have a very efficient and streamlined way of approaching materials. I personally believe that toolbag2 should be a driving force in the simplification process.
Currently in Toolbag2 you can have a specular colour map, and use the intensity slider. what's important to understand here, is that the intensity slider DOESN'T reprisent specular intensity, it reprisents reflectivity. you can demonstrate this by putting gloss to 1, and intenisty to 0.04. regardless of your diffuse colour, you will have a glossy plastic material, and at intensity 1, you have a metallic material. Currently, without a coloured specular map you will only have a white specular value - this isn't necessarily a bad thing.
so in order to differentiate between non-metals, and metals in toolbag 2, currently you should use a coloured diffuse for non metals, and a black diffuse for metals (where possible). the reason it's important to use a black diffuse for metals is that it's technically possible to still have some colour in the underlying diffuse even with the reflectance slider set to 1. because it multiplies with your specular colour (which is always less than 1 if it's a colour) the final result is less than one. and because of energy conservation, the final diffuse is 1 - finalspecular. here's a very simplified math of that:
white = 1,1,1 (or 255,255,255) - this is our diffuse.
gold= 1,8.4,0 (or 255,215,0) - this is our specular.
reflectivity = 1.
so the (extremely simplified) math would go:
diffuse * (1 - (specular * reflectivity)) = diffuse
in numbers:
1,1,1 * (1 - (1,0.84,0) * 1)) = 0,0.16,0 (or 0,40,0)
this is technically incorrect, because metals should have no diffuse as no light can actually enter their surface and scatter. so in this way toolbag2 is not entirely correct as a physically based solution.
that's a technical problem rather than an art problem though. so here's the problem for the artist -
due to the technical problem above, artists will either have to very strictly change their workflow by creating black diffuses for metals etc. or they will be fudging values the way they always used to, which will lead to a lack of progression.
as of right now, artists need to supply two different maps, a diffuse map, and a specular map. here's why i think that's a bad thing -
- Metals have no diffuse and are purely reflective. this means that the colour of the metal in the specular map should always be multiplied by 1. if it's multiplied by less than 1, then it's no longer as reflective as it should be.
- Non metals have varying degrees of reflectivity, but on average they're extremely low, around the 0.04 mark.
this means that when an artist creates a specular map which contains both metals and non metals, anything that is non metallic should be clamped between 0 and 0.04 (or 18,18,18 ). how many artists are truly likely to do this? i think most will work outside of these ranges either out of ignorance or simply thinking it's "wrong". let's face it we're visual creatures and there are many times when our eyes play tricks on us, making us think things are correct when they aren't. so when an artist works outside of that range, and all of a sudden their surface is more reflective than it should be in toolbag, and as a result the diffuse is also darkened due to the energy conservation rule, what do they do? they adjust the slider! but adjusting the slider means now their metals are incorrect! OH NOES!
that's why i believe we should remove the intensity slider, and just have a metallic map instead. if you really want to have a specular map as well (for pearlescent effects and stuff) then go for it. but the intensity slider absolutely needs to go.
and as for roughness/gloss? pretty much the same as above. having the slider there just allows artists to work within an incorrect range of values. personally i think it should go, and artists should learn to make their maps to the correct values (that, and i can't remember a time when i've had it, or seen other people have it set to less than 1).
please feel free to correct me if any of this is wrong. TB2 is {} this close to being perfect (shaderwise). don't look at the removal of those "features" as the creation of a limitation, it's really not. it's actually removing the limitation from artists that the old workflows had inherent.
Showing of some soft shadows and a bit of local reflection.
Quick test with a mesh courtesy of Christophe Desse, using substance textures.
Some more feedback/requests:
- global filtering options would be great, so you don't have to set it for each texture of each material manually.
- no Undo/Redo, it's easy to mess up the positions when working with multiple objects and being able to undo viewport modifications would be handy
Hey dude, if you want blurry reflections, you shouldn't use the mirror specular model. Really the only difference between the mirror model and blinn-phong is that mirror is clamped to the highest possible specular reflection resolution. So just switch over to blinn-phong and adjust the gloss as you like to control the blurriness of reflections.
Soft shadows: We've got some scale issues it looks like, we expect models to come in at at certain size, so the light size function is tied to that, if your mesh is a lot large than the program expects, you may need to manually type in much higher values into the light size. See if that helps, meanwhile we'll look into what we can do re:scale. You can also try scaling down your mesh (or up, you can get an idea of relative scale by moving the mesh around and looking at the units, if you move it halfway across the screen and it says you moved it 0.005 units, its scaled small, if its says 500 units, its scaled large).
I second both of these requests!
btw that substance designer stuff translates really well into TB2.
EDIT:
What scale is TB2 setup to expect btw? Is there a way to know beforehand?
EDIT2:
- Is there a way to toggle of visibility of light icons in the viewport? I know they aren't in the final exported image but they can get be distracting when trying to find a good composition.
- PBR shaders are only half of a PBR renderer, I notice there is a manual slider for light falloff on dynamic lights in TB2. Are there any plans to implement a more physically based lighting equation? (quadratic falloff perhaps.)
Certainly a lot to think about, Gir, thanks for the input.
We've always considered the sliders as a way to tweak and play with the texture content within the app. So instead of switching to photoshop to adjust the curves on your gloss map, you just mess with a slider in TB and see what works. I can see your point about this leading to bad content creation practices though.
The color pickers for modifying diffuse and specular color are a similar shortcut. You *should* be putting that color in the textures themselves but it certainly is handy to play with it in the app. Would you still want to keep those?
A compromise Earthquake and I came up with was to have a slider or color-picker present only if the texture slot is empty. That way you can still tweak the material on models with no UVs, preview it before you've finished the spec-map, whatever. But the interface is cleaner and forces you to work through maps instead of arbitrary scalars at the end.
Now with gloss, one thing I was initially pushing for was to have two slider values: both a min and a max value for gloss. Blinn-Phong gloss values technically range from 0 to infinity. A very high gloss value being a very high specular exponent, making for a very tiny point of a specular highlight; perfect mirror would be at infinity.
In TB2 we just took an arbitrary "high enough" value and mapped that to between 0 and 1. But you could technically map your gloss map to any range of values. You also only have 256 shades of gloss to work with, so spreading your 8-bit gloss channel out across a huge range may not be desirable. Thoughts?
I couldn't test yet so i have no clue yet, but giving the possibility to copy the color value or slider value over as a shade of grey to photoshop after finding the right one, would be a pretty cool option.
This car is awesome, Jerc!
Undo is on our list, always the last bitter pill of a feature.
"Look at the cool new thing I added!"
"Great. Now let me undo it."
"What. Why?"
Hey man,
the colour swatches/sliders only for when there is no texture present is actually a fantastic idea. it's a great way for you to be able to preview how a material "should" look, so you can try to emulate it in your textures, at the very least this would make it a great learning tool!
on gloss:
i had assumed you were using blinn-phong and having "1" set to 255, so you're not actually interpolating from a grayscale texture. i've played around with Beckmann, Trowbridge-Reitz, Cook-Torrence, and Blinn-Phong recently, of them all Blinn-Phong is the only one that allows you to go to an infinitesimal value, all of the rest fit very nicely in the 0-1 range... in fact if you go outside of that range you get some rather bad artifacting. i think that's one reason why i'm against having the slider there... the "big names" in PBR at the moment are all using the 0-1 range inputs and they feel very comfortable to use. there's no need for sliders or anything, you just make your map and put it in and it works. there's no need to adjust that value, it is in my opinion, something that leads to bad habbits from the following point of view:
as an artist, if you're used to creating any old gloss map, and then adjusting a slider afterward to make it look good. and then you go work in a studio where that's a static value that can't be adjusted, you're going to have to adapt to that. whereas if you're used to playing within the specific range of values and get consistently good results, when you go to a studio that does allow you to adjust it, you won't need to. you'll just set it to max and be happy!
also, as i mentioned before, i can't think of a time when i've seen it set to less than max when used with a texture. i can understand the slider being there for previewing materials (like i said in the previous paragraph, great learning tool). but i just don't see the need for it outside of that. again, personal opinion and i'm sure there are others who will disagree.
on a side note. i'd just like to thank you and EQ for taking the time to talk to me here and via pm. it's very humbling and i'm happy to be of any help i can be, even if it's as a constant nag :P
We finally have the tools to create something predictable and accurate, which will (FINALLY) force artists to really edit their textures the proper way, but most importantly, the same way(!). And while I can understand that from a "user-friendly" perspective you want to avoid people to go back and forth between photoshop and Toolbag to get what they want, I think it would be a miss opportunity to help out artists to be "production ready"
I'd personally like a "lite" version of what you guys have with WAY less options. No more check box for energy conservation (should be on by default), no more sliders (as long as you have a texture, like some people already mentioned it) and no more color overrides. Everything should be texture driven, just like it would happen on a real production where you want to have as much consistency as possible.
Quick additional user-interface note : While playing with the different skies, it looks like each time you rotate your mesh in the viewport, you loose the "sky menu" on the left panel. It can becomes really "clicky" when you don't know right away which sky you want...
Overall, really impressed with the engine improvement, you guys did an amazing job with that! I just wish the shader setup would be as next-gen as the engine itself (the simpler the better!).
(Oh and btw, I love the new color picker
All dynamic light types cast shadows now, and soft shadows at that. Be sure to check out the feature overview here: http://www.marmoset.co/toolbag/tb2-new
Just for fun I rendered out an 8K version of this image. Linked because its too big to post in the thread: https://dl.dropboxusercontent.com/u/499159/workbenchcrazybig.jpg
We don't have refraction in yet (which is what you'll probably want to use for glass). I've been using the dithered order independent transparency for glass which really isn't at all what its meant for. So unfortunately I don't have any awesome glass examples.
I just tried the F9 copy thing and its working for me here. One issue I get, if performance is pretty low it has a tendency to drop keyboard input for me, so maybe try again, press F9 a couple times to make sure it sticks and let me know if you still have the same bug.
Awesome! I did a similar test a while back with a Modo render I had done a few years ago:
And then the Toolbag 2 render:
Its not a perfect match but I was very happy with it, especially considering how utterly awesome Modo's offline renderer is.
TB2 does have area lights which should affect the shape of the highlight. Change the "Size/Softness" of the light.
I like the idea of having the colour picker when a texture isn't present though.
As for the levels of gloss available when using a texture, I would say that a value of 0-1 would make more sense than a an infinite range.
I always use the brightness value of 0-100% in PS rather than 0-255 values so this workflow wouldn't really change anything for me in that regard, though I do hate linear colour pickers, so I wouldn't want that at all. The ability to add hex code and RGB values is super important, even if they are converted internally before rendering. Usability FTW.
We've got swapable shader models so it won't be an issue to support multiple methods/techniques.
Is shift drag no longer affiliated with rotating the skydome? edit i see its the ctrl key now. interesting change for people used to shift
SWEET.
So Are there any upgrades to toolbag that make it a viable option for environments? It would be cool if it could just read a 3ds max scene and live update it and all the cameras and lights in it.
Bug: draging the colour picker circle to the top right corner of the colour display makes the hue and saturation sliders lock into place at the top of their bars till I move the circle again. I dont know if it matters but I'm using a wacom.
Edit* another dabble this morning, transparency in captures is a bit rough, in psd you get a lot of missed pixels in your object, in tga there is a white pixel thickness border to the object, but fewer missed pixels.
Turntable settings for both scene and camera bring the lights with them, is there an option that I am missing just to rotate the model?
2 things however; Now with the more and more powerful presentation of realtime assets, can you even talk about a real preview of game usage ? The normal marmoset seemed pretty borderline to me and while I personally don't give anything about that, I only look at the visuals, but there seem to be many people looking for a higher moral ground in that regard from what ive been reading
Also, dont want to be that guy, but your interface looks not really nice and definitely an downgrade visually. I would assume it is a placeholder, but you write shiny new interface on your page, so I dont know what to think about.
It seems to borrow visuals from the older windows and OSX at the same time, and reminds me of render packages from the 2000s to be perfectly honest
The output images look fantastic, with realtime becoming so strong, Im not sure if I should continue with my offline workflows : )
Thanks.
1. Are you planing to support Alembic? FBX loading in TB1.0 is slow and even in Modo and Houdini FBX is slow to load, a lot slower than Alembic.
2. Are you planing to support Pixar SubD surfaces?
Some of these have probably been mentioned or are already planned for later releases:
-Bug: Light widget disappears
-Lights need some visual representation to show direction/radius etc like in TB1
-Visibility toggle in scene tab for meshes, lights etc.
-Option to Lock camera view/position
-Option to pause turntable, rather than stop jump back to the start.
-Scale reference and scale option (like in TB1)
-2 sided option
-Option to save/load materials would be handy too.
I was also getting some thin black artifacts around meshes when I was using image capture (even with high sampling). The image below is from Image to clipboard option.
The skin shader is great, had lots of fun messing around with the settings and the secondary spec options are really cool as well.
Couldn't get any good results with the transparency for the hair (hence the heavy DOF on the hair
A little, I tried to test.
very cool new features,
I favorite Local reflection&New DOF!
By the way, I can't find the detailmap and chromatic aberration.
Function of it is my favorite in TB1.
Thank you&sorry my bad english.
Would you mind posting your textures used on the skin? I'm curious how many maps you used and what your settings are in TB2. It looks really nice
Nice render afisher, I agree with all of this.
Seems like you already can "save/load materials" there is a arrow besides where you duplicate and create a new material that has import and export inside it.
in general, i've found the displacement to be ridiculously sensitive.
All tessellation in game engines will suffer from the same problem with the mesh breaking apart at the UV borders so I wouldn't expect that to get fixed any time soon tbh. It's down to the limitations of the algorithm that's being used and it's exactly the same in UDK and CE3 afaik. You would need to use fewer UV shells or place them where they would be hidden to remove this problem.
IMHO, It would be better for performance (and faster to code too) if something like parallax occlusion mapping with silhouette was added.
On BSI, we used a similar rendering method. For our PBR we actually did have coloured diffuse, but it was practically black...very similar to how all of this is set up, just through unreal before it became "a thing." I need to really dive itno the skin stuff, Adam has had better luck with it than me it seems, but I'm sure that's a user error on my part and just not giving it enough time.
A lot of it is just echoing what has been said, I'm finding.
I think it's slick. I like the new format and how it resembles software that most people will be comfortable or familiar with. I like how everything is right there in the hierarchy and easy to see. The drag and drop workflow is awesome (which I think is new), how you manage assets is great and, like mentioned above, PBR is the way of the future. So many places are going in this direction...I think Toolbag2 is a great example of how powerful it can be and presented in a really clean way.
Some gripes:
-I naturally go to "File" to import a mesh because I'm not creating anything. (echo..echo..)
-Naturally trying to use shift to move the skybox, I can get over it..I'm just so used to it as a basic control.
-If there is a way, I didn't see it: Is it possible to just save the model and materials all together like in Toolbag 1 (Save Mesh and Materials)? Without saving the entire scene?
-I miss having representation for my lights. I seem to do more guess work and just wish there was a volume like in Toolbag 1 that showed me my lights in action.
-Could we have a "new" material that is just the skin shader without enabling it through a normal material?
-I wish things stayed selected from the hierarchy when navigating in the viewport. I feel like, I want to move around but still have a light selected to tweak while I navigate around. It seemed that way in Toolbag 1, but I would need to open it up to be sure.
Thanks Metalliandy for the explanation, really helpful so displacement algorithm still sucks, but why in Maya works perfectly? Some days I feel my brain will explode...
For offline rendering its all done on the CPU. For realtime its done on the GPU with pretty standard fixed methods of doing it. Pretty much all game engines use the same method to tessellate and displace geometry, because the GPU API has specific features that do it. This is my basic understanding of it.
So yeah, as Andy says, its unlikely we can do anything about it unfortunately. If anyone knows of any realtime engines that do displacement without gaps at UV seams, or any whitepapers etc we would be very interested.
The best thing to do for now is to use as few uv seams as possible, hide uv seams as well as you can (behind other objects like cloths/props/hair is ideal), and using 16bit displacement may help a little as well
Everything is saved in a single scene file now, mate. The file can be moved around on your HDD without breaking now too, which saves headaches
Maya and other non realtime displacement solutions use Catmul Clark subdivision rather than realtime solutions such as PN Triangles and Gregory patches. Catmul Clark is a great subdivision solution but it also increases your polygon count by 4x for each level, which obviously makes it less suitable for realtime than more efficient solutions that are tailored towards game art and the like.
So if I import a material, then edit that saved material in a second scene it will updated in the first?
Imported materials would act essentially as instances not uniques after import
I do have to agree that I miss some of the TB1 hotkeys simply because I'm used to them. Like hitting "T" to start and stop the turntables and such.
Here's a few more thoughts too:
- When using ctrl to move the skybox there's no way to reset it back to the default where the "Clear" button in TB1 for the corresponding turntable in the animation menu would usually reset it to the default rotation. Not a huge deal but If I had my model positioned to take advantage of the default rotation and I accidently hit the ctrl key and rotated the skybox it might be annoying to go back if I was OCD.
- There's no far and near clip plane controls on the camera so long objects have the front clipped during turntable rotations.
- I notice adjusting the size/softness barely has any effect on my lights, could this be a scale related issue?
- Double clicking a Toolbag scene file crashes toolbag after it launches. I'm forced to open the program and use the "open file" menu option for it to work.
- I couldn't find the skin material for awhile either. It might make more sense to have a skin shader listed under presets in the create new shader section.
What about Pixar open Subd? I was under the impression those could approximate Psub limit surfaces with DX11 sparse triangulation.
From the FXGuide breakdown for OpenSubDiv:
Hi,
This is in-correct.
I'm one of the people who implemented the tessellation and displacement in Maya and we have completely crack-free, real-time tessellation that is similar to PN Triangles (similar in cost) and we also support hard/soft edges (smoothing groups).
And, to the best of my knowledge, I thought Unreal also has crack free displacement, but I never tried.
well DX-10 is a minimum requirement soooo...
Certain DX11 features like tessellation/displacement will require a DX11 compatible video card as well.
Also I think we're 64bit only, so you will need the 64bit versions of Vista, Win 7, or Win 8.