@pior Manual copy and paste operations work for moving objects (plus most of their assigned data elements) between different versions from 2.9x all the way back down to 2.79. Not sure if or how it could be automated but the basic functionality is there.
The only drawback is some of the newer data flags (like mesh display settings) may be stripped if the destination version is too old. Most of this can be fixed by going into mesh mode and just re-enabling anything important that was turned off. For closer versions like 2.9x to 2.9x it's mostly seamless but there may be edge cases where certain data elements don't transfer.
@FrankPolygonjust to be clear : are you saying that on your end, you can ctrl-c an object from, say, a running 2.91 session, and have it ctrl-v successfully in a session of another version, like 2.80 ? If yes I am very curious about that because I don't think I've ever been able to do so, at least since the days after 2.79.
I can definitely copy/paste between two sessions of a given version, even weeks apart (which confirms that the copy buffer is actually a file stored somewhere on disk) ; and I can also open any 2.8+ file with any 2.8+ version, either going up or going down, and the same goes with append. But ctrl-c ctrl-v always pastes in the last ctrl-c-ed object from the version currently active, even if the copied object is weeks old.
Would you mind clarifying ? Maybe we're on different systems (I'm on Win64).
@pior Correct. On my system (Windows 7 Home Premium 64bit) it's possible to open multiple versions of Blender, select object(s) from any of them then copy and paste them into other versions. Just tested this functionality again to confirm with all the versions that are currently installed on this system. All the Blender installs are 64 bit and some of them are portable installs.
2.79 to 2.8 works as expected.
2.8 to 2.81A works as expected.
2.81A to 2.82A works as as expected.
2.82A to 2.83 works as expected.
2.83 to 2.90 works as expected.
2.90 to 2.91 works as expected.
2.91 to 2.79 works as expected.
The mesh object was able to make a full cycle through all those versions with edits made in each version. Going forward though the versions there shouldn't be any major issues. Any data flags that reference newer features tend to be removed or ignored so going backwards can cause minor issues below 2.8x.
Copy and paste functions work as expected regardless of whether or not the source instance is running. However when running two versions side by side on different monitors the destination window needs to be active (click anywhere in the viewport) otherwise it will just paste the object(s) back into the other version if it's still the active window.
Blender writes a "copybuffer.blend" file to the temp folder whenever object(s) are copied. This information can be pasted into another instance provided they're both using the same temp folder. If each version is referencing a different temp folder location they cannot share the data in file. Here's a couple links that document issues related to changes in the defualt temp locations.
Blender's auto save function also dumps incremental saves into the temp folder based on time intervals and program exits. So if you close something without saving do not open the program again, just go to the temp folder and grab the latest "quit.blend" or "filename_autosave.blend" file.
Absolutely ! Thank you for taking the time to test things out and write it down. I've got it all up and running now, going back and forth without a hitch. Lesson learned : editing the default temp path is not the wisest thing to do
Hey guys - trying to get into procedural texturing with Cycles for character presentation. Do you know any resources that go over the node and light setup for relatively simple (afaik) clean looks like this?
I think I have the basics down, in regards to where to click to make things happen, but I was researching other artists to find tips they picked up to save time, look out for some pitfalls, etc.
Made a tutorial on creating a 3 layer blendshader in Blender. This will allow you to replicate the material blending workflow from Unreal, so it might be useful for some of you. More images on my artstation, tutorial video below. All textures by Quixel.
Is there any non-manual, non-tedious way of converting face sets from Sculpt Mode to some other face attribute ? I don't really have any preference as to which, as long as it allows to select the resulting regions, somehow. It could be randomly assigned materials, automatically assigned vertex sets, all face sets split as respective parts, or even border edges marked as sharp ...
Of course I am aware that one can hide/unhide face sets, switch to Edit Mode, and select visible faces from there. But that's precisely what I want to avoid as it gets very tedious very, very fast. One should be able to process dozens of face sets in one go - doesn't matter if the end result is a single object with some face attributes, or multiple split objects ...
FWIW at this time the most user-friendly and fastest way of doing so I've found consists using Face Sets > Extract Face Sets, one by one, which requires to switch in and out of Sculpt Mode each time. And then recombining the resulting objects into one single object, and removing the solidify modifier.
I suppose that it may be possible to do something similar automatically, hiding all Face Sets except one (there has to be some kind of index somewhere ?), going to Edit mode, Select All, and separate selection - then repeat for all face sets ?
(To be absolutely clear ... the only reason why I am even using Face Sets in the first place is because of the awesome smooth border feature, which is truly a great way of cleanly defining areas on a model, even at medium density. Really great stuff - it's just that this clean area information is really tedious to extract and work with outside of sculpt mode).
(Also FWIW ... the above approaches (exploiting the Extract tool, and/or hiding and splitting) can be made just a little bit faster by doing so by non-adjacent Face sets ; and then splitting up by loose parts once in Edit mode. It makes the whole thing about 2x faster)
in 3ds max there is a way to update your mesh below the skin modifier stack, then have the deformed mesh update clicking 'always deform' on and off - kind of like a resset xform.
HasBlender got something similar? I basically modelled my characters legs a little bit bowed, so straightened them out by editing the verts, but they are still bowed when the armature is active
Hey there Ruz - as a matter of fact this should be the behavior by default as skin weighting in Blender is just stored directly as vertex groups values. It's a bit alien at first because of how simple and "straight to the metal" it is, but it is actually very powerful as it allows for all kinds of non-linear editing tricks that would be hard to pull off in max/maya.
The one thing that can get in the way of that is morph targets - they too are also very powerful, but their UI is a bit error prone as one can end up modifying a morph as opposed to modifying the base model without realizing it and that can be frustrating Maybe that's what's going on in your case ?
Hello, I'm very very new at rendering images and animations with Blender, but I'm running into an unwanted effect while trying to use cheap fog created via the volumetric scatter shader node, using Eevee - I have a point light in the center of the scene with a very low intensity to subtly enhance the darker areas so I don't need to have the entire world lit uniformly using the background shader node, thus loosing out on convincing shadows.
The problem I have is the ambient point light doesn't play nicely with the fog, and shows up as an ugly glare/lens flare kind of effect, as demonstrated here:
Sure thing, feel free to send over a file if needed. It's also possible that you may have a rest pose different from the pose the model is actually in.
(EDIT: After further reading the documentation I'm not sure this is possible atm with the current feature set + attributes)
Anyone familiar with the new Attribute transfer node in geometry nodes, how can I transfer my custom normals pre-boolean to my after boolean geo? (All procedurally)
Its not working because it has artifacts here, the idea is to have your original primitive normals to always be preserved, keeping all booleans within nodes. I created a vertex group "Vert1", but I'm having problems in understanding how to properly set the system.
Does anyone know how to edit the shades of white (!) of the text from the NLA editor ? The very subtle variation in text color is the only way to distinguish the currently selected object, hence I'd like to make it stand out way more.
Does anyone know how to edit the shades of white (!) of the text from the NLA editor ? The very subtle variation in text color is the only way to distinguish the currently selected object, hence I'd like to make it stand out way more.
In the preferences: Themes -> NLA Animation -> Theme Space List (scroll down, it's at the bottom)
In the sidebar in the Item tab. You need to have vertices selected though. And the mesh needs to have a weighting applied. Or this panel will not show. It shows in edit mode and in weight paint mode. In Weight paint mode you need to be in vertex selection mode and have a vertice selected. It will not show with face select mode. To select a vertice in weight paint mode hold down ctrl and click.
Is there any obvious flaw with my topology that would cause these ugly warpings? This is the back of a seat that needs to be flat. I've flattened out components, lining them up with the same axis. I also ran a cleanup to remove stray components. I'm new to Blender and any help is appreciated.
Below, the highlighted blue edge is the surface with the warps in it. This is unsmoothed and without subdivisions. Both of those reduce, but don't do away with the wonky shading issues.
@Camille_Meehan Looks like inconsistencies in the mesh shading properties. Try recalculating the normals, marking the faces with shade smooth, clearing sharp edges and setting the per-object auto smooth value to something above 45°.
Below are some links to the official documentation that covers each operation:
@Camille_Meehan Looks like inconsistencies in the mesh shading properties. Try recalculating the normals, marking the faces with shade smooth, clearing sharp edges and setting the per-object auto smooth value to something above 45°.
Below are some links to the official documentation that covers each operation:
Heya - That's not my point. By wondering about facebook posts by random people and also by not linking to the actual source of the statement, you are greatly increasing the chances of confusion on the topic. It's now easier than ever to get access to information, so why not making the small effort of contacting the BF to get *actual* clarification on what you are wondering about ? Also who knows if by "Blender Add-ons" they mean the ones shipping with the app by default, or third-party ones.
My point overall is that by just "discussing" things (posting on FB, forum threads, and so on) without any solid statement from the actual source, no one really gets any wiser since everything can lead to misinterpretation. So why not go to the source instead, and *then* share your findings ?
"Everything you share and bundle with Blender scripts has to be GPL compliant. If you have copyrighted parts, these should be sold and shared separately. Everyone has the right to share such Blender script (like Blender itself), that is not piracy."
Now that the 3D industry is constantly developing their tools to work with non-destructive workflows (Substance Painter being the excellent example of that)... the weight painting workflow is still VERY destructive and mistake-sensitive in native tools of Blender, Maya, Max etc.. It saddens me that ngSkinTools are probably the best solution to that... except it's only for Maya, and there's no equivalent for Blender.
I even asked the developer of ngSkinTools, if he's ever going work on a Blender version, and the answer was just a firm "No."
Do you really wonder? The GPL is simply quite a hurdle. Just watch one post above. You run into the problem that too much people might give away your work for free. And that competitors can simply grab the code from your addon or fork and use it for his own addon or fork. Just look at E-Cycles and K-Cycles, and what a heated discussion came up then. That's quite a problem when your goal is to earn money with your hard work. And this problems will remain since this is the very nature of GPL and copyleft. Forking and developing at the base of the previous code is what it was made for.
The curious part is that half of the addons at Blender Market doesn't mind about breaking this license. And nobody cares. As stated above, bundling images and documentation with the addon would mean to put them under GPL. Yet, nearly every addon claims copyright on the documentation and images. Some of them even at the code.
And then there is the GPL itself, where just one thing is clear: that nothing is clear. There are not enough court cases to make clear if a GPL rule is wishful thinking or really law. Bundling non GPL content like documentation is such a thing. The one says yes, the other says no. Another reason why you usually keep your fingers away from everything GPL ...
Hi everybody. I'm learning to make UVs and bake and I facing the problem when I see visible seams on planar surfaces. I'm trying different ways to unwrap chamfered meshes. I've created a cube with chamfers (1 segment bevel) and baked beveled cube (with 9 segment bevels) on it. To make UV map denser I split every cube's face like this:
When I look in LookDev mode the shading of the baked model looks completedly fine, but when I start to zoom out I start seeing shading artifacts on planar faces:
Here are my bake settings:
Is it something wrong with my bake? Will I have the same artifacts in game engines?
The model is in the attachment in case you want to check it.
this is probably one of the worst uvs ive seen for a cube but anyhow the problem youre having is that the uv shells are packed to tightly, and as soon as the mipmaps kick in you will start to see artifacts - see: http://wiki.polycount.com/wiki/Edge_padding
this is probably one of the worst uvs ive seen for a cube but anyhow the problem youre having is that the uv shells are packed to tightly, and as soon as the mipmaps kick in you will start to see artifacts - see: http://wiki.polycount.com/wiki/Edge_padding
Thank you for your feedback. How would you unwrap a cube like that? Why do you think that my UV is bad?
Because the gap between the uv patches is too small. Neighbour pixels bleeds into the uv patch. That's the bleeding part that kio is talking about. Leave at least two pixel space that is not covered by an edge between the UV patches.
And because every single face is mapped as a single UV patch. As much as necessary but as few as possible. When you can remove a seam without to run into distortions then you should do it.. Coplanar faces that doesn't distort should not be split into single UV patches. I could map this cube with one single UV patch. But would use six single uv patches here. Not more.
This can be done with mark seams and then do an unwrap. I would suggest also to test the method conformal. It can give better results than angle based with geometric shapes, but no guarantee.
Because the gap between the uv patches is too small. Neighbour pixels bleeds into the uv patch. That's the bleeding part that kio is talking about. Leave at least two pixel space that is not covered by an edge between the UV patches.
In blend file I've attached above I'm using 4K texture, so there is more than 2 pixels space (actually there are at least 10 pixels between islands). So I suppose the problem is not in pixel bleed? I will double-check that.
And because every single face is mapped as a single UV patch. As much as necessary but as few as possible.
Actually not every face is mapped to it's own UV island. Here is a screenshot with bigger margins:
I can pack it something like this:
In this case UV islands occupy 65.4% of entire UV area. If a split every big face into 4 pieces then I can pack it like this:
In this case UV islands occupy 83.4% of UV map area. In both cases margins between islands set to 1% of UV width. So I assume that the latter pack is better in terms of texel density. I understand that it will slightly increase polygons count but let's assume that I want to sacrifice polycount in favor of higher texel density. In that case does my UV layout make sense?
This can be done with mark seams and then do an unwrap. I would suggest also to test the method conformal. It can give better results than angle based with geometric shapes, but no guarantee.
this is exactly how I did it. And after that, I packed it with UVPackmaster 2.
I still think this could be a bleeding problem. But that's easy to test. Increase the margin and look if the problem goes away.
You use imho ways too much uv patches. As told, i woud not use more than six here. Three would also be enough. Image number two looks best to me. The empty UV space would not worry me too much. You could also use a single texture for every UV patch In case everything else fails.
The problem here is, it looks like you can save some mapping space when you cut the faces into so much pieces. But you forget about the margin. You need to take this empty space into account too. And then the texture resolution that you win isn't this much anymore. And not to forget, more single patches means that you get more visible seams that you need to fix up.
The rule remains: as much as needed, as few as possible.
I still think this could be a bleeding problem. But that's easy to test. Increase the margin and look if the problem goes away.
Yes looks like increasing margins indeed helps. When I increased them from 0.3% to 1% I have to zoom out more (that is model gets smaller on screen) to start noticing these artifacts. But it appears that this problem happens only in EEVEE. In Cycles there are no shading artifacts.
I think we get closer to the problem. Eevee is a realtime engine. This means i expect that it has some sort of mip mapping. Means when you zoom out then the texture resolution gets automatically lower. And then it starts of course to bleed since the distance between the edges goes below two pixels.
Hm, i cannot find explicit mip mapping settings though. There are some textures settings in the preferences in the viewport tab. Anisotropic filter. Turn it off and have a look if this helps.
I would nevertheless suggest to improve the mapping: No seams means no seam trouble
Texture filtering can be set per texture in Blender (yay!). Go into node editor - select the texture node for the normal map and in the first dropdown that by default says 'Linear' instead pick 'Closest'.
Also btw. for proper normal map display you'll want to make sure that color space is set to 'Non-Color', default is 'sRGB'.
Texture filtering can be set per texture in Blender (yay!). Go into node editor - select the texture node for the normal map and in the first dropdown that by default says 'Linear' instead pick 'Closest'.
Yep, it does the trick! Thanks a lot!
@Tiles Thank you! Now I get a bit better understanding of how real-time rendering engines operate.
Oh and thanks for the hint that this is done per texture. You never stop learning.
This is one of the cool little things about 2.8+. Everywhere else I am aware of you either get no option to turn off filtering or there's a global switch (like in 2.79 and earlier) that renders every texture in view razor sharp looking in a headache-inducing way.
2.8 and above proportional edit set connected after every stroke sucks! Is there another way to fix this? I'm doing hair strands at the moment and I miss this feature.
Replies
I can definitely copy/paste between two sessions of a given version, even weeks apart (which confirms that the copy buffer is actually a file stored somewhere on disk) ; and I can also open any 2.8+ file with any 2.8+ version, either going up or going down, and the same goes with append. But ctrl-c ctrl-v always pastes in the last ctrl-c-ed object from the version currently active, even if the copied object is weeks old.
Would you mind clarifying ? Maybe we're on different systems (I'm on Win64).
[Edit] Nevermind ! The error was between the chair and the keyboard. I had different temp folder paths in the two versions.
https://blender.stackexchange.com/questions/140407/unable-to-copy-objects-between-versions
Lesson learned : editing the default temp path is not the wisest thing to do
I think I have the basics down, in regards to where to click to make things happen, but I was researching other artists to find tips they picked up to save time, look out for some pitfalls, etc.
Is there any non-manual, non-tedious way of converting face sets from Sculpt Mode to some other face attribute ? I don't really have any preference as to which, as long as it allows to select the resulting regions, somehow. It could be randomly assigned materials, automatically assigned vertex sets, all face sets split as respective parts, or even border edges marked as sharp ...
https://blender.community/c/rightclickselect/Sxgbbc/
Of course I am aware that one can hide/unhide face sets, switch to Edit Mode, and select visible faces from there. But that's precisely what I want to avoid as it gets very tedious very, very fast. One should be able to process dozens of face sets in one go - doesn't matter if the end result is a single object with some face attributes, or multiple split objects ...
FWIW at this time the most user-friendly and fastest way of doing so I've found consists using Face Sets > Extract Face Sets, one by one, which requires to switch in and out of Sculpt Mode each time. And then recombining the resulting objects into one single object, and removing the solidify modifier.
I suppose that it may be possible to do something similar automatically, hiding all Face Sets except one (there has to be some kind of index somewhere ?), going to Edit mode, Select All, and separate selection - then repeat for all face sets ?
(To be absolutely clear ... the only reason why I am even using Face Sets in the first place is because of the awesome smooth border feature, which is truly a great way of cleanly defining areas on a model, even at medium density. Really great stuff - it's just that this clean area information is really tedious to extract and work with outside of sculpt mode).
The one thing that can get in the way of that is morph targets - they too are also very powerful, but their UI is a bit error prone as one can end up modifying a morph as opposed to modifying the base model without realizing it and that can be frustrating
The problem I have is the ambient point light doesn't play nicely with the fog, and shows up as an ugly glare/lens flare kind of effect, as demonstrated here:
Anyone familiar with the new Attribute transfer node in geometry nodes, how can I transfer my custom normals pre-boolean to my after boolean geo? (All procedurally)
Its not working because it has artifacts here, the idea is to have your original primitive normals to always be preserved, keeping all booleans within nodes.
I created a vertex group "Vert1", but I'm having problems in understanding how to properly set the system.
All the best,
https://ubisoft-mixer.readthedocs.io/en/latest/
Has anyone given this a try?
Any feedback? Any collaborative 3d tool to recommend other?
Does anyone know how to edit the shades of white (!) of the text from the NLA editor ? The very subtle variation in text color is the only way to distinguish the currently selected object, hence I'd like to make it stand out way more.
In the preferences: Themes -> NLA Animation -> Theme Space List (scroll down, it's at the bottom)
Anyone know where the vertex weight panel is in 2.93? Weights not groups.
It shows in edit mode and in weight paint mode. In Weight paint mode you need to be in vertex selection mode and have a vertice selected. It will not show with face select mode. To select a vertice in weight paint mode hold down ctrl and click.
That fixed its little red wagon!
Below, the highlighted blue edge is the surface with the warps in it.
This is unsmoothed and without subdivisions. Both of those reduce, but don't do away with the wonky shading issues.
https://docs.blender.org/manual/en/latest/modeling/meshes/structure.html#properties
This adds 3 simple operations for multiple object UVing. It lets you sync / set / create the active UV map, delete and set as active render map.
Hope you guys find this useful! it's free
- Ton Roosendaal, Twitter.
The curious part is that half of the addons at Blender Market doesn't mind about breaking this license. And nobody cares. As stated above, bundling images and documentation with the addon would mean to put them under GPL. Yet, nearly every addon claims copyright on the documentation and images. Some of them even at the code.
And then there is the GPL itself, where just one thing is clear: that nothing is clear. There are not enough court cases to make clear if a GPL rule is wishful thinking or really law. Bundling non GPL content like documentation is such a thing. The one says yes, the other says no. Another reason why you usually keep your fingers away from everything GPL ...
I'm trying different ways to unwrap chamfered meshes. I've created a cube with chamfers (1 segment bevel) and baked beveled cube (with 9 segment bevels) on it. To make UV map denser I split every cube's face like this:
When I look in LookDev mode the shading of the baked model looks completedly fine, but when I start to zoom out I start seeing shading artifacts on planar faces:
Here are my bake settings:
Is it something wrong with my bake? Will I have the same artifacts in game engines?
The model is in the attachment in case you want to check it.
And because every single face is mapped as a single UV patch. As much as necessary but as few as possible. When you can remove a seam without to run into distortions then you should do it.. Coplanar faces that doesn't distort should not be split into single UV patches. I could map this cube with one single UV patch. But would use six single uv patches here. Not more.
This can be done with mark seams and then do an unwrap. I would suggest also to test the method conformal. It can give better results than angle based with geometric shapes, but no guarantee.
I will double-check that.
Actually not every face is mapped to it's own UV island. Here is a screenshot with bigger margins:
I can pack it something like this:
In this case UV islands occupy 65.4% of entire UV area. If a split every big face into 4 pieces then I can pack it like this:
In this case UV islands occupy 83.4% of UV map area. In both cases margins between islands set to 1% of UV width.
So I assume that the latter pack is better in terms of texel density. I understand that it will slightly increase polygons count but let's assume that I want to sacrifice polycount in favor of higher texel density. In that case does my UV layout make sense?
this is exactly how I did it. And after that, I packed it with UVPackmaster 2.
You use imho ways too much uv patches. As told, i woud not use more than six here. Three would also be enough. Image number two looks best to me. The empty UV space would not worry me too much. You could also use a single texture for every UV patch In case everything else fails.
The problem here is, it looks like you can save some mapping space when you cut the faces into so much pieces. But you forget about the margin. You need to take this empty space into account too. And then the texture resolution that you win isn't this much anymore. And not to forget, more single patches means that you get more visible seams that you need to fix up.
The rule remains: as much as needed, as few as possible.
Hm, i cannot find explicit mip mapping settings though. There are some textures settings in the preferences in the viewport tab. Anisotropic filter. Turn it off and have a look if this helps.
I would nevertheless suggest to improve the mapping: No seams means no seam trouble
@Tiles
Thank you! Now I get a bit better understanding of how real-time rendering engines operate.