Frequently asked questions:Q. Where do I load my diffuse map?
A. Your diffuse map should be loaded into the albedo slot.
Q. Where do I load my roughness map?
A. Your roughness map should be loaded into the gloss map slot. If you're using a roughness map where white = most rough and black = most glossy, click the invert checkbox.
Q. Does Toolbag 2 support animation?
A. Not at this time, though we do plan to add animation support in a future update.
Q. How do I use transparency?
A. Transparency can be enabled in the transparency module in the material editor, select a transparency mode from the dropdown. Your transparency mask must be placed in the alpha channel of the diffuse/albedo texture.
Q. Does Toolbag 2 support soft alpha blending?
A. Not exactly. Toolbag 2 current supports 3 types of transparency, all of which are order-independent (which means they do not have sorting/popping issues as standard alpha blending generally does):
- Dithered, which is best for hair or other objects that need semi-transparency and soft fading. It will look noisy in the viewport but renders out much smoother. Render double size to improve quality even further.
- Cutout, standard on/off alpha-testing, black = transparent while white = opaque
- Additive, additive blending is great for glass and wet or secondary layers on top of other materials (eyes for instance). Additive blending works by adding the specular reflections onto the render. Does not work correctly with the depth of field effect.
Q. Does Toolbag 2 have a refraction shader?
A. Not currently, however, this is something we will likely add in a future update.
Q. How do I scale my mesh?
A. Per-mesh scale is not currently supported, though we do plan to add it in a future update. Global scene scale can be adjusted in the scene object properties.
Q: Is there a floor or shadow casting object feature?
A. Not currently, but we do plan to add this in a future update.
Q: Can I hide the light GUI?
A. Yes, hit CTRL+U, or toggle show guides in the view menu.
Please feel free to post any support questions directly in this thread, or make a new thread in the Marmoset sub-forum.
Replies
I go to Render, and I switch to Oculus but the device doesn't connect to Marmoset, I don't see anything in my led display... Any help?
First off, yes this is the correct place to ask. Secondly, unfortunately we do not yet support the second gen Oculus, we need to get our hands on a devkit.
First thing that pops into my head: Do you happen to have thin straps modeled where you're seeing the issues? There are some problems with the shadowing system when you have faces on either side of an object that are very close together. In that case, you could try plumping those bits up with a push modifier (or whatever the equivalent is in your app).
I hope you guys will get yout DK2 soon, because is way better than the first one!
I triangulated the mesh in Maya 2013, exported it as an FBX 2013 with normals, tangents and binormals. I baked in xNormal 3.18.8 with the binormals override enabled.
On the screenshot attached : the left object is just imported with the tangent space setting being empty, on the right it's the same mesh but with the tangent space set to mikkt. I didn't duplicated the object, I imported it twice.
You can see some artifacts on the right object on the top of the little cube faces. The middle edge of the quad is sometimes visible and the reflection seems not planar.
That is working as expected, if TB2 finds bi-normals/tangents in the file it will load them automatically, and the tangent space will show up as blank. Its a little confusing that it doesn't show "custom tangents" or something though.
When you select another space, it recalculates the tangents, so its not surprising that the results change at that point. Meshes baked in XN w/o imported tangents look great when using the XN/mikktspace tangent option.
So just leave the tangent space blank if you want to read the bi-normals/tangents from the file.
I just got into trouble with the transparency (as always). I tried to make a glass with texture on it but the additive blending won't use the alpha from the diffuse. Is this normal or am I making something in the wrong way?
Cheers,
Fisher
Can you post a screenshot of what you've got and a reference photo of what you're trying to accomplish?
The way I do texture variation with glass is to add the variation to the spec and gloss maps.
No, if you want some opaque areas as well what you can do is duplicate the mesh, and give it a new material with dithered blending and use the alpha channel to mask where the transparent and opaque areas are.
I'm writing a tutorial specifically on this which should be out soon, but for now you can check out the diving helmet content which is set up how I've described above. Files: https://dl.dropboxusercontent.com/u/499159/divinghelmetcontent03.zip
This is what I wanted but with more dirt:
This is the result:
Not the prettiest but it is acceptable. The only major problem I couldn't solve was the artifact that happens when a light casts shadows.
With shadows:
Without shadows:
It is hard to describe properly so I uploaded to Dropbox. Here
Also, this is my first proper work with baked normals and hand made textures so if you happen to have any critique or comment I would be glad to hear it.
Btw, your diving helmet looks absolutely stunning, learnt a few neat tricks from it. Can't wait for the tutorial!
Thanks EQ!
alpha test, alpha blend, blend-add etc. are all different blend modes that affect the entire material. this isn't just a marmoset limitation, i'm pretty sure.
http://www.marmoset.co/toolbag/learn/glass
Edit: LOL I thought that I get instant email alerts from the subscribed threads and not daily, I thought you wrote the post just now. I should go to bed now. :poly105:
Hey I hope you don't mind that I post this here.
First off, you make some very good points about how I approached the glass, and you're right it doesn't make much sense that the glass would be totally transparent. I think the changes you made make a lot of sense so I've posted that here as an example of how to do glass with various opacity.
Would it be advisable to use in this slot a map generated by xNormal's normal-to-cavity-map tool? Or to stick with a baked cavity map?
As far as I understand, a baked cavity map is more physically accurate. However this takes far more time than converting normal map to cavity map. I've used normal-to-cavity maps in the past, but that was as a mask for overlays that became part of the diffuse map used in a non-PBR rendering environment. Since Marmoset's cavity slot uses the map in a different and more accurate way, I'm wondering if there would be a definite benefit to sticking with baked maps.
Thanks!
Edit: whoops not sure if I posted this in the right forum. Admin, feel free to move to FAQs or something if more appropriate
If the map is baked or generated later from the normal map it shouldn't make too much of a difference. The more important thing is that the map is mostly white, with darker values to represent the cavities. Certain apps like crazy bump can give you a "cavity or detail map that isn't really correct for this input, where its grey with bright content for the convexity and dark for the cavity.
Basically, your cavity map should look something like this:
Image from this tutorial: http://www.marmoset.co/toolbag/learn/pbr-practice
Generally I will make the cavity map from the normal map, after extra details have been added to the normal map in 2d, this way you catch all the detail that may not be in the bake.
edit - it's there, just very slight changes. No real effect on the main highlight, just the outter edge of the ggx falloff seems to get tighter. Just curious if that's correct or not.
you can look forward to actual area lights in the next update, and they work properly regardless of the reflection type!
as you written:
"Additive, additive blending is great for glass and wet or secondary layers on top of other materials (eyes for instance). Additive blending works by adding the specular reflections onto the render. Does not work correctly with the depth of field effect."
no mater what i do i cannot find the ability to only add only dirt to the base texture. it seems no mater if cut the area with alpha it adds something to the base (i tried different colors of the non dir area but still no luck, either its to dark or to bright.)
Additive blending works be adding (add means to make brighter in this context) the specular pass from the material. The diffuse can be included too, but it will add (make brighter) as well. This is useful for transparent materials like glass (see more about that here: http://www.marmoset.co/toolbag/learn/glass).
For solid decals like you're trying to create here, you most likely want to use the dithered blend mode instead of additive blending. Dithered will look a bit noisy in the viewport, but renders out cleaner. You can render an extra large image to reduce noise further (there is more about this in the tutorial link above as well).
I hope this helps.
I just update to the 2.06 build, but I get a very weird default startup scene. I have attached the screenshot of the startup with a model and a rendered version of that model. They look totally different. Look like the startup is way darker than it should be. I was told that it is the brightness/ gamma issue, but I don't know how to fix it.?
I have updated my graphic card driver. I'm using GeForce 555M. driver version 344.48
the model is from tf3dm.com
Thanks guys
If I were to use Marmoset in the future, would this software require a similar edge padding or can I reduce that ?
Kinda related:
Can I reduce the edge padding for Quixels software if anyone knows the answer to that for it's viewer to be compatible with Marmoset as well.
It is easier for me to ask this question then to spend days baking out textures only to find that they do not have adequate edge padding.
As a general rule you should use as much edge padding as possible in all textures. There's a fairly simple reason for this: Mipmaps!
As you zoom in/out, just about every engine out there (including toolbag, although we have a checkbox to disable this) will render different mipmap levels, the higher the level the lower the resolution of the image, at a low enough resolution, without padding you'll start to see white/black seam lines on your model.
It's good practice to keep your padding in, regardless of the final rendering software.
I find that 16 pixel edge padding is the minimum to use when rendering in 3ds max with vray if it is a 4096x4096 texture.Anything less and I will get the white seam issue.
I was hoping that if I migrated over to Marmoset, that I might not need as much edge padding for hard surface models.
This isn;t for game art or any engine so mip mapping ingames isn;t a concern for me.
So if I understand you right, I should just stick to 16 pixel edge padding as I am already doing.
Also was looking around and there doesnt seem to be any hard/fast rules for edge padding on textures sizes 4096 and above that can be applied to any rendering engine that isn;t game related.
Thanks in advance.
There are actually two issues here.
The larger your texture, the more space you should have between your uv islands. This is because as the model is viewed from farther away, the texture resolution is sized down and what was 8x8 pixels is now 4x4 pixels, one mip down and 4x4 pixels is now 2x2, etc. At some point what was a healthy amount of space between uv islands results in two uv islands blending together at the edges and causing seams. This is UV spacing.
The second issue is pixel padding, when baking normal maps, ao, etc you want a lot of pixel padding. The reason for this is similar to the above, when those edge pixels blend you want them to blend to a similar color, not to the white or black of the background (also you should avoid white or black backgrounds for the same reason). This sort of padding has no adverse affect on anything, except for the few seconds it takes to calculate while baking or to add in photoshop when you are done with your texture (xnormal plugin is good here). Pixel padding added by your baker/plugin has no negative effect on anything, so you can use a really high value, I usually use like 64 or something so most of the empty space is filled in.
Again as Lee says, you can disable mip-mapping on a per-texture basis in Marmoset Toolbag, but it isn't recommended to stop using uv spacing or pixel padding.
There was a bug with the new area lights feature when used in combination with the Dota shader and Blinn-Phong where the specular exponent (gloss) value for dynamic lights was locked to max glossiness. This will be fixed in the 2.07 update, but we decided to release the updated shader file if anyone wants to patch the problem manually in the mean time.
Download the shader file here: https://dl.dropboxusercontent.com/u/499159/dotashaderfixtb206.zip
Installation instructions:
For Windows, your install directory will typically be C:\Program Files\Marmoset Toolbag 2\. If unsure, right click on the shortcut to Toolbag in the start menu, select properties, and then click open file location.
For OS X, your install directory will typically be Macintosh HD\Applications\Marmoset Toolbag\
Any issues, feel free to post in this thread or contact support@marmoset.co
Hmm, try baking 16 bit (tif) and then converting to 8-bit (tga, png, etc) and see if that helps.
Tried it but it's still the same. It has improved after I baked it from highly subdivided cube but there are still problems. I wonder if it's repairable or is it the best you can get from normal maps.
edit: somehow making a new mesh fixed that. Oh well.
Yes, go to the scale mode (ctrl+r) and use the central scale gui element.
If you want to type in scale numerically to a specific value you will have to enter it 3 times.
As you can see, the reflection seems to be clipped and the clipping line move as I move the lamp down. I double-checked my scale and units and they seem to be correct, the whole lamp is around 50 cm high. Then I tried to activate or deactivate the backface culling but it doesn't change anything.
Sorry if this problem has already been reported here.
Unfortunately this is a technical limitation with screen space reflections. The reflection content is generated from what is visible in the frame, so areas of the mesh that are not visible an any particular angle, like the underside of that section on your lamp there, will not show up in the reflections because the rendering doesn't "see" that area.
Screen space reflections are really fast and look good in a lot of situations, however, what you're seeing here is the technique's Achilles heel.
You could try setting up a different angle to limit the issue. Sorry, I don't have any sort of solution to offer you other than that.
With a proper lighting and a wood texture underneath the glass top, this should be less visible. I'll make sure to take screens at a grazing angle.
Thanks for the answer.
Thanks.
[ame]www.youtube.com/watch?v=_Ef75xW1eiU[/ame]
if you don't mind emailing the scene over to support@marmoset.co for debugging that'd be awesome, and if you don't mind us seeing the source mesh as well. I'll try to reproduce this over the weekend.