What is the difference between ArmorPaint and using the shader editor in Blender with masks and procedural nodes with the Texture paint mode? Is there anyone who is currently using it? I gather it has some issues or has it been sorted out? How does this compare to substance painter?
guitarguy00 What's better in Modo? I tried UVing there last year and haven't found any real advantages . Unless I missed something. Packmaster does much better packing imo.
Max "relax" option is totally redundant in Blender imo. Never feel a need for it. But my UV workflow usually almost automatic :
1 set auto smooth , apply scale and edge split modifier
2 split cylinders side edge manually
2 U button with conformal option or angle one depending on subject with correct aspect off
3. run packmaster
4 merge split vertex back.
Nothing to relax usually after . Rarely "stright" button from textools and live UV with pins.
Wish I'd know Python scripting, Would make it one button solution.
Maya's Nightshade does look greatly superior although from a few you-tube videos I saw. As of MAx it's UV is just a try to copy Blender one . A successful one finally
Yeh I was referring to Modo's pack algorithm compared to Max's(not Packmaster though). From what I saw of Modo's Pack function, it did an insanely good job.
Thanks RaphaelBarros . It actually works pretty fine. Foe some reason I couldn't do so when tried it last time in 2.79. So I usually stitched it manually with V .
@melviso check if the faces are truly planar and if the normals are pointing in the right directions and haven't been split. The latter you can do via the overlays.
How do you convert material displacement to actual geometry ? Or feed a material output to the Displace modifier ?
I see this way: https://www.youtube.com/watch?v=McALCOr39rY But it’s very unpractical to work with as you bake it all to re-displace it behind… So quality loss /very unpractical when doing large elements
You can get the shader file on my gumroad for free.
Great tutorial, I like the idea of the mask blending instead of just using vertex paint. I have a few QOL comments you might find useful. - Try the Subdivide operator instead of adding loopcuts - Adaptive subdivision is still experimental, but usually works very well with displacement maps, might increase performance and/or quality compared to a high level of standard subdiv. - You can copy the whole node tree between the two materials Crtl-c, Ctrl-v (even between files), easier than adjusting all the textures. - If you setup the base materials as groups it will give you a couple of advantages. You can use this groups inside the mix material, this way you can easily swap out materials. You can reuse the base material without copying all the nodes and edit them globally etc. Node trees get less complex. Something like like this:
- If you setup the base materials as groups it will give you a couple of advantages. You can use this groups inside the mix material, this way you can easily swap out materials. You can reuse the base material without copying all the nodes and edit them globally etc. Node trees get less complex. Something like like this:
As a super random addition regarding node groups, I recently discovered a way to add otherwise inaccessible extra socket types to groups. The usual way to add a socket to a group is to plug in the input/output socket of a inner node into an empty group input/output socket, creating one of the same type of the source.
The issue starts when you're creating conditional node groups. It's possible to do things like logical int switches and more with math nodes, but the only number socket available is a float. Ideally you'll want integers for this. No shader node currently uses them, but they do exist in the API.
To use them create and rename your nodes group then type the following in the console:
@birb thanks for that, could be useful. I don't understand why there is no way to choose the data type when creating new input or output variables in Blender.
Does anyone know an easy way to see in what collections selected objects are ? Like it was in old Blender? I bought that 3 bucks addon form blender market but it somehow cryptic and doesn't help at all
Does anyone know an easy way to see in what collections selected objects are ? Like it was in old Blender? I bought that 3 bucks addon form blender market but it somehow cryptic and doesn't help at all
While hovering over the outliner, your keymap for show active, then show/hide one level - I've changed my hotkeys so I'm not sure what's default.
That's not as easy as it was in 2.7 although. I found collections pane in N panel. it's kind of easier to work with. Still the checkbox vs eye ys render is making so much confusion I would say the whole collection improvement is an example of "better is enemy of good"
You can get the shader file on my gumroad for free.
Great tutorial, I like the idea of the mask blending instead of just using vertex paint. I have a few QOL comments you might find useful. - Try the Subdivide operator instead of adding loopcuts - Adaptive subdivision is still experimental, but usually works very well with displacement maps, might increase performance and/or quality compared to a high level of standard subdiv. - You can copy the whole node tree between the two materials Crtl-c, Ctrl-v (even between files), easier than adjusting all the textures. - If you setup the base materials as groups it will give you a couple of advantages. You can use this groups inside the mix material, this way you can easily swap out materials. You can reuse the base material without copying all the nodes and edit them globally etc. Node trees get less complex. Something like like this:
Great info, thanks for sharing. I`ll give this a try....
- If you setup the base materials as groups it will give you a couple of advantages. You can use this groups inside the mix material, this way you can easily swap out materials. You can reuse the base material without copying all the nodes and edit them globally etc. Node trees get less complex. Something like like this:
As a super random addition regarding node groups, I recently discovered a way to add otherwise inaccessible extra socket types to groups. The usual way to add a socket to a group is to plug in the input/output socket of a inner node into an empty group input/output socket, creating one of the same type of the source.
The issue starts when you're creating conditional node groups. It's possible to do things like logical int switches and more with math nodes, but the only number socket available is a float. Ideally you'll want integers for this. No shader node currently uses them, but they do exist in the API.
To use them create and rename your nodes group then type the following in the console:
Nice. I can`t seem to properly recreate it, any chance you could upload a demo file somewhere?
@IxenonI Recreate what, the socket or the logic behind the switch? Uploading a file wouldn't help with adding new sockets through Python because you need to do this by hand to alter a specific group, and I no longer have this file anyway.
You need to add import bpy at the top of the file or script. Either write it down in a text file and press Run or copy and paste everything into the python console. If you stick two empty lines at the bottom of the script pasted onto the console it'll execute automatically. I can take some screenshots or make a gif showing how to use a script if you need help.
The contents of that switch group:
The ValMix are equivalent to MixRGB nodes, but with Math. I plugged this into a ColorRamp for convenience and left it as Linear to show it's picking very narrow positions, but you mix different things and get even more exact using the same principle.
When I copy/paste an object I get another unique material . If same material exists Blender just adds 001.
So I usually end up with gazillion of material001, material002 and so on . A long mess of perfectly same materials and nothing could help to get rid of them.
Do anyone know a secret checkbox or something to workaround it?
Maybe in style of 3d max something which usually asks you if you prefer to use scene materials when you import?
When I copy/paste an object I get another unique material . If same material exists Blender just adds 001.
So I usually end up with gazillion of material001, material002 and so on . A long mess of perfectly same materials and nothing could help to get rid of them.
Do anyone know a secret checkbox or something to workaround it?
Maybe in style of 3d max something which usually asks you if you prefer to use scene materials when you import?
Try SHIT+D to duplicate the object instead of copy and pasting.
hello, I have a problem with an object that I made for unreal and that I wanted to import into blender to do some rendering tests, some parts look good but other not.
here are some pictures with, what I did and what it should look like
Set for Eeevee with sun animation, For Cycles you need to set the sunlight to zero since it will get the lighting from HDRI itself . Note also that the light is set to physically correct for sun : 1300W/cm2 .
If you'd like to experiment with shader construction it would b reasonable to set the light back to 1 and exposure to zero in order to see procedural maps in viewport or just use material preview mode.
When I copy/paste an object I get another unique material . If same material exists Blender just adds 001.
So I usually end up with gazillion of material001, material002 and so on . A long mess of perfectly same materials and nothing could help to get rid of them.
Do anyone know a secret checkbox or something to workaround it?
Maybe in style of 3d max something which usually asks you if you prefer to use scene materials when you import?
Try SHIT+D to duplicate the object instead of copy and pasting.
I meant copy/pasting from one open Blender window to another one or appending from another file. Not Shift+D
It just occurred to me that the interface drawing performance in 2.8+ differs greatly between Solid mode and Eevee/PBR mode (called "Material Preview"). In Solid mode, manipulating UI panels is about as smooth as it was in 2.7x ; yet in Eevee things slow down significantly, regardless of the content of the scene and/or what's on display in the viewport.
Of course it's probably possible to throw more CPU at this, but my point is that on the same machine, there *is* a noticeable difference in interface performance between modes, making the whole app feel sluggish.
Is there some setting somewhere that causes this to happen ?
As far as I read Blender uses OGL to draw its interface. Also saw reports that manipulating e.g. nodes in the node view gets slower depending on scene complexity even if only node editor is visible. Probably nothing the user can do about.
Yes, 2.8 feels like a step back in performance. Heck, I went out and bought the fastest GPU I could find to make Eevee bearable. Would have not had any need for it otherwise - probably cost more than a year of Autodesk rental too. And they say open source! Dude - it's free! At least no complaints on the feature side.
I wish old Blender would get the new features of 2.8 . specifically vertex normals tools, bevel baking and ability to unwrap several selected objects at once + recent udim tools . Still hate the new interface. Almost a year after I still can't work as efficiently as before.
Collections are a puzzle wrapped into another puzzle
Lots of important old addons are still not working in 2.8. Like shader GLSL export and Ninja ripper
@thomasp , @gnoop - oh well, at least that confirms that this isn't my machine being the issue here so that's always that
Here's another one : does anyone know of a way to get the flattened out version of a model based on its UVs ... while also having the resulting flat model still having UVs ? If that makes sense. The Textool 1.3 port (as well as the 2.7x version) does the unfloding of the mesh just fine ("Create UV mesh" button) but it does kill the UVs.
Now it is always possible to recreate the UVs by hand by projecting from top view but that is only an approximation (with one workaround being to add 4 verts in the corners of the UVs to act as a frame of reference, or to use the square outline that Textools is creating ... but that doesn't really help if anything is outside 1-1). So if there was a way to do that cleanly without any workaround that'd be great.
@pior I don't want to disparage you from doing that, but what is the use case of that? I tried coming up with one and so far, I'm not coming up with anything.
@Surfa - Ha, I didn't even think of that ! I've never used that feature before but that does make sense. In this case it doesn't quite apply since that would require the vertex data to be exactly the same (which isn't the case here as the flattening causes splits, obviously) but I might be able to leverage that in some cases. I sure wish there was a proper UV projection feature though (agnostic of vert indexes), as I believe both Max and Maya have it. Maybe it's buried somewhere in the Data Transfer modifier.
I believe I have good enough workarounds now. Still I do wonder if there is a more direct way to retrieve/not lose the data. But for now I'll just re-unwrap from top view, it's good enough. Thanks !
@Pior if the mesh still has the same vertex count and vertex order, maybe it can be taken as a shape key for the original mesh. So the shape key just flattens it and the UVs remain the same.
Heya RN - well, the Textool flatten to UV tool already creates the shape key to morph the flattened mesh into the original shape ; but the mesh itself has different vertex data than the original model, since the process of flattening creates splits around each UV island (of course).
So yeah at this time the best option is still to manually UV the mesh from top view, which isn't a big deal at all
What happens if you mark sharp all your UV-seams then apply an edge split modifier first? I'm pretty sure that process does not destroy the UVs of the object but should mean the splits are already present in the mesh before you run the Textool flatten to UV. It works on my really quick test object but with more complicated mesh seems to be that it is fairly easy to miss some of the seams on the model when marking everything as sharp. Especially if you manually UV some sections of the model using projects etc..
The more complex mesh I tested it on does seem to throw up some errors of 4 edges present on the UV Mesh but not on the normal mesh. But I managed to fix that issue by triangulating the mesh first.
So the workflow would go.
UV mesh
Select by bound using textools
Apply Sharp edges
Edge split modifier with sharp edges
Apply triangulate modifier
Use Create UV mesh in textools
Add a UV map to the created UV mesh
Transfer UVs from original mesh to UV mesh
Obviously your mileage may vary. If it seems to work well for most mesh we could always script the process to do in one click.
Well, simply put : baking down floater details modeled just above the flattened mesh/layout.
it's possible to bake floater detail without flattening. The detail should share same UV space ( having UV projected by data transfer) and have invisible side toward target surface. Then be baked as object space normal map after being joined with a target . Not even a cage is necessary.
Projecting across UV seams doesn't work although but I guess it's same with flattening
@pior I think I tried this recently. Just use Textools, create the uv mesh. Then select the uvmesh and the original mesh. Go to Object-Make Links- Transfer uvmaps. Assign the original mesh material to the uvmesh. Should work fine.
Quick question, should I create a full body cloak, not just a cape for a character, from a flat plane position where it will fall onto the character upon simulation, or should I create it "on" the character as if they are wearing it?
And this definitely applies to me because these are the two programs I primarily use for my work. Model in Blender, animate in Max. I'll take a look at it.
@99499 I would go with roughly laid out around the character. You know, as if You would put on a piece of clothing, then blow a lot of air into it. Once that is done, have it simulate falling into place. That is roughly the process Marvelous Designer follows.
Another, more stylized way is to just model stuff normaly/sculpt it on the character and just use the simulation where you need/want it. Mostly on loose parts of the clothes I would wager.
Hi all, Is there a way to stop blender from creating an "Armature" root upon exporting an fbx with animations?
I am trying to use the same avatar in unity with custom animations and animations from mixamo. This is not possible if blender adds this root.
Replies
Closest thing you'll get from this is to pin (shortcut P) the main UV, then unwrap the part again.
We also have the stitch option (select the shared edge between the models you want to stitch and press V), but that would take time in this case.
How does this compare to substance painter?
Thanks !
Full video here:
https://www.youtube.com/watch?v=ccciTF5aZP8
You can get the shader file on my gumroad for free.
I have a few QOL comments you might find useful.
- Try the Subdivide operator instead of adding loopcuts
- Adaptive subdivision is still experimental, but usually works very well with displacement maps, might increase performance and/or quality compared to a high level of standard subdiv.
- You can copy the whole node tree between the two materials Crtl-c, Ctrl-v (even between files), easier than adjusting all the textures.
- If you setup the base materials as groups it will give you a couple of advantages. You can use this groups inside the mix material, this way you can easily swap out materials. You can reuse the base material without copying all the nodes and edit them globally etc. Node trees get less complex.
Something like like this:
The issue starts when you're creating conditional node groups. It's possible to do things like logical int switches and more with math nodes, but the only number socket available is a float. Ideally you'll want integers for this. No shader node currently uses them, but they do exist in the API.
To use them create and rename your nodes group then type the following in the console:
bpy.data.node_groups['GROUP_NAME'].inputs.new('NodeSocketInt', 'SOCKET_NAME')(Replace
A list of all current socket types can be found in the manual. There are other fun types there like percentages, angle and vector acceleration.GROUP_NAMEwith the target group, andSOCKET_NAMEwith your new socket name)I don't understand why there is no way to choose the data type when creating new input or output variables in Blender.
You need to add
import bpyat the top of the file or script. Either write it down in a text file and press Run or copy and paste everything into the python console. If you stick two empty lines at the bottom of the script pasted onto the console it'll execute automatically. I can take some screenshots or make a gif showing how to use a script if you need help.The contents of that switch group:
The ValMix are equivalent to MixRGB nodes, but with Math. I plugged this into a ColorRamp for convenience and left it as Linear to show it's picking very narrow positions, but you mix different things and get even more exact using the same principle.
Try SHIT+D to duplicate the object instead of copy and pasting.
here are some pictures with, what I did and what it should look like
I meant copy/pasting from one open Blender window to another one or appending from another file. Not Shift+D
It just occurred to me that the interface drawing performance in 2.8+ differs greatly between Solid mode and Eevee/PBR mode (called "Material Preview"). In Solid mode, manipulating UI panels is about as smooth as it was in 2.7x ; yet in Eevee things slow down significantly, regardless of the content of the scene and/or what's on display in the viewport.
https://www.youtube.com/watch?v=Jo3tIj7-3nY&feature=youtu.be
Of course it's probably possible to throw more CPU at this, but my point is that on the same machine, there *is* a noticeable difference in interface performance between modes, making the whole app feel sluggish.
Is there some setting somewhere that causes this to happen ?
You can Purchase it here for $19: https://bit.ly/UV_toolset_1
Here's another one : does anyone know of a way to get the flattened out version of a model based on its UVs ... while also having the resulting flat model still having UVs ? If that makes sense. The Textool 1.3 port (as well as the 2.7x version) does the unfloding of the mesh just fine ("Create UV mesh" button) but it does kill the UVs.
Now it is always possible to recreate the UVs by hand by projecting from top view but that is only an approximation (with one workaround being to add 4 verts in the corners of the UVs to act as a frame of reference, or to use the square outline that Textools is creating ... but that doesn't really help if anything is outside 1-1). So if there was a way to do that cleanly without any workaround that'd be great.
I believe I have good enough workarounds now. Still I do wonder if there is a more direct way to retrieve/not lose the data. But for now I'll just re-unwrap from top view, it's good enough. Thanks !
So yeah at this time the best option is still to manually UV the mesh from top view, which isn't a big deal at all
What happens if you mark sharp all your UV-seams then apply an edge split modifier first? I'm pretty sure that process does not destroy the UVs of the object but should mean the splits are already present in the mesh before you run the Textool flatten to UV. It works on my really quick test object but with more complicated mesh seems to be that it is fairly easy to miss some of the seams on the model when marking everything as sharp. Especially if you manually UV some sections of the model using projects etc..
The more complex mesh I tested it on does seem to throw up some errors of 4 edges present on the UV Mesh but not on the normal mesh. But I managed to fix that issue by triangulating the mesh first.
So the workflow would go.
Obviously your mileage may vary. If it seems to work well for most mesh we could always script the process to do in one click.
And this definitely applies to me because these are the two programs I primarily use for my work. Model in Blender, animate in Max. I'll take a look at it.
Is there a way to stop blender from creating an "Armature" root upon exporting an fbx with animations? I am trying to use the same avatar in unity with custom animations and animations from mixamo. This is not possible if blender adds this root.