Previous >>1025688Big Cat edition>NEWS: 5.2 is out
>>1030378Gotta say I'm looking forward to playing with that thin paper shader thing.
why is this general so god damn dead
>>1030399>>1030399I got depressed about 3D because the Blender animation/rigging improvements I wanted got postponed again.
>>1030399New version's out but we testing only this weekend
Why does this hair curve flatten out 2/3's of the way to the tip? Only mods are interpolate and clump. I wanted it to maintain a consistent taper across the curve, not collapse in on itself.
>>1030406That's a lot of vertex for such a curve, wouldn't it work with a Bezier and 3 points? Otherwise, try Ctrl+T with proportional edition on the flat area
>>1030378>Still no texture paint layers, no sculpt layers, no animation layers, no add custom meshes, no rigging Nodes, no retopo, no android port ....The only interesting shits this release are the new physics thing and the loop tools port.https://www.youtube.com/watch?v=g4OXlrxqIx0
is it just me or is the camera turning differently in the viewport? feels so weird
>>1030419the lack of layers is insane. it was insane 10 years ago, it's maddening now. none of what they're doing with texture painting even feels serious if they don't get that sorted first
>>1030410Sure it could be done with a lot less vertices, it could be done with 4; but it doesn't look as good even with subdivisions on in render settings. I don't know about bezier, i'm trying to create hair strands as opposed to a solid object. Also tilt just tilts it to the side, it doesn't actually fix the flattening issue. The image i posted is a side view of the hair.
>>1030423>>1030406I added a bunch more hair curves and instead trying combing them into the shape I want. Turned out much better, I don't think it's possible to make the shape I want with just 1 curve.
>"Look at our new animation tools!!">NLA editor is still dysfunctional
How do I make expanding topology without just using 3:1 squares? I wanted to just practice topology flow and not make a whole model, and legs are particularly troublesome. (If I were to make the whole model, my goal would be to simply make a sculpt then project the mesh onto it)Is there a better way to do this? I colored the edge loops to show how they create extra vertices and likewise close to match how much space the body part takes up, but am not sure about the flow of the torso to the sides of the legs on the very outer edge
>>1030437also need help if this is good chest topology flow.
>>1030437>>1030441I don't mean this to be disrespectful but the topo is fairly poor in both examples. Best advice I can give is to go to smutbase and download a model that is well rated, such as an Overwatch one. Then study their topology. At most you should only need 1 pole in the area where the torso meets the hips/legs. If anything you will be decreasing the amount of edge loops as you go down the leg, not increasing.
>>1030449No disrespect taken, I'm trying to figure out how good people use topology flow to expand volume. I've just always struggled with thighs twisting but I think I got it somewhat
>>1030423>>1030406I know how to do it with a custom made node. Not using the premade nodes. It involves quite a bit of math. Here, I whipped it up: https://files.catbox.moe/csk5lz.blendIt's probably not *exactly* what you're looking for. But it should at least put you on the path to figuring it out. If you have any questions, just ask, and I'll answer to the best of my ability.
I have a fortnite model and a seperate detailed foot model I want to join together how the hell do I do this
>>1030478Click the foot, then shift-click the fortnite model. CTRL-J to join. This usually screws with the shape keys, so be aware
>>1030479Don't think I get it, but both models have their own rig (The feet model is just the ass to the feet)so when I remove the legs from the fortnite model and place the new ones it doesnt connect just stays in place while it turns completely pink
>>1030480There is no quick method that can be taught in a 4chan post. It requires a lot of steps. You seem to lack the fundamentals. So you have to go find some youtube tutorial where they break this kind of thing down for you.
>>1030481Fair enough, my only problem is just finding the right tutorials
>>1030478>>1030482look for some headhack tutorials as that's basically what you're doing but with feet
>>1030476The problem with having so many poles is you're going to see a lot of pinching and generally unwanted deformation if you're animating or posing this model. Additionally there are way more verts than there need to be on the legs, arms, etc.
>>1030477Wow anon thank you, that's really impressive stuff. I'll have to spent some time checking this out as I am a beg when it comes to geonodes. What nodes can be used for adjusting the clumps?
Where can I find someone who can help finish my music video, I think I need another eye and someone with some more experience
>>1030480What you're asking to do can be done, it's just that you have to frankenstein both the armatures and the meshes together, in the proper order. How to actually do it (we're blind and can't see your models, armatures, so save a backup)>delete leg bones of the fortnite model ARMATURE in edit mode>delete the upper torso bones of the leg ARMATURE in edit mode (if it has any)>Select the leg armature>shift-click select the fortnite armature>ctrl J>In the new frankenstein armature, parent the leg bones to the hips/torso of the fortnite model>click the leg MESH>shift-click the fortnite MESH>ctrl JThen you edit the new frankenstein mesh and sew the vertices together. This may be difficult and time consuming. I'm inferring you don't know yet how the armature moves the mesh around by what you said about joining the meshes together. When a bone moves, it has a vertex group of its name, and tells those vertices to move. Those vertices only respond to that command if it has the armature modifier applied to it, with the specific armature in the modifier, with bones named exactly the same as the vertex groups. The legs mesh has vertex groups associated with the bones of the old armature. When you move the legs of the fortnite armature, it is probably moving legs that have no command over the vertices. By frankensteining also the armatures, the bones of the leg armature are named the vertex group they command. Keep in mind this is not the ideal way to do this, and would be better to remodel the legs from scratchKeep in mind this is not the best way to do it
>>1030523This did work but still had the same issue where it seems both models turn into one big texture. Thank you though I'll see if I can figure this out, this is mostly for personal projects so I have time to learn.
>>1030537Oh forgot about the pink thing. That's very likely just missing textures (videogame characters have a variety of texture images)
>>1030512>What nodes can be used for adjusting the clumps?This simple question led me down a rabbit hole. I have a solution to that. But I also want it to be a good and tight solution. Not a rush job. So I'll get back to you on that.
>>1030399All the shitposters and government bots are finally abandoning 4chan cuz there's no point anymore.
You guys have any tips for me? I'm a huge scrub and I'm struggling to figure out how to get these pants rigged to the model right. It seems like the pelvis and feet are locked in but the legs do whatever they want. I'm sure I'm missing something obvious so any extra eyes or video suggestions would be appreciated.
>>1030594You could abandon Film Maker Studio and rig with current ways instead.
>>1030595If I want to make stuff for 20 year old game, I have to use the 20 year old method.
>>1030597Learn to let go, they will never update it, they don’t care about old games and fans are gone.
>>1030600Ok, thanks. No need to keep replying. Any other anons actually know 3d and how to help with the thing I asked?
Checking things for an answer, but I have my own question, since it's come up more than once for me. Is there a way to cut out geometry using a texture? Like an alpha image. Kind of like the 'trace image to grease pencil' function, but with edges on a model. The use case in this instance is a lot of models with transparent parts needing to go in a game that won't support transparency. So it's a lot of manual editing, otherwise.
>>1030594Does it at least pose right in blender? I don't see any errors in the group names, unless it's a compiling thing, I don't know why it would do that.
What's the best blender version a laptop with intel hd 520 graphics could run assuming i only strictly model in low poly, no pbr materials and little to no lighting? blender 5 chugs my entire machine without rendering anything
>>1030594there's a porn creator named Gifdoozer who has a big archive of animation/character etc tutorials. one of the videos he made is taking clothes and applying (and rigging I think?) them to the character. I think the archive is normally behind a patreon paywall but it's probably floating around somewhere if you take a few minutes to look. the effort of digging for that specific video might be more than finding some random youtube video but I know for a fact he covers what you are talking about
>>1030641Yeah his archive's on /aco/
patiently waiting for a crack of the neXus addon. shit looks insane, I really want to play with all the physics
>>1030399Been busy working on stuff, but can't share it here :( also not a fan of 5.2 so I went back to 4.5/5.1.1
>>1030476I don't know much about human models in Blender but I keep seeing people modeling them from scratch. Isn't there an add-on already that can make humans in any shapes or proportions, from realistic to stylized? I know about MakeHuman but it really looks dated
>>1030399it's a slow board, the last post was an hour ago and this thread is still on page 1.anyways my contribution to the thread is a simple edge glow tutorial. If you enjoy it please share your own compositor tips/tutorials so we can give pizazz to the people
worth noting I already had strong rim lighting for this scene, and the effect I was going for is quite subtle, but it works to separate the character from the flat background and add a splash of color without modifying the existing lighting setup
>>1030698>>1030699>example picsunfortunately got nothing to provide myself but i hope to see the pic once its done. wherever that might be
>>1030698>Needs nodes and lots of undocumented information to get edge colorsLook what they need to match Maya, Houdini and Cinema 4D simple tool that existed for years.
>>1030701I appreciate the interest and I'll share more once it's complete and after catbox gets its shit together since I don't feel like catching a nsfw vacation on a censored nsfw image. Cryptomattes are pretty useful but another tip is that the masks they generate should be adjusted with the dilate/erode node NOT the scaling/transform nodes. Way more direct control of the mask and I probably should have used those instead of in my original example where I added a fog glow onto itself to increase the amount of glow
>>1030698That is THAT rabbit, right?
>>1030419>>1030421They're currently working on these things in the roadmapIt seems like 2026-2027 will be a huge jump for Blender, like they will finish a part of the animation overhaul, or something.
>>1030419>no rigging NodesThey have a bone info node now. It's still not enough. But it's a start.
>>1030719>More nodeEnough with the nodes, it was supposed to be VFX tool and nothing else. The project is not built for painting, animation or other stuff.
>>1030419>no android portWe almost had one for Ipad OS, which would be a damn sight easier to take around than a whole laptop. They were too focused on babying it up for pen-only use and had no funds.
>>1030719>They're currently working on these things in the roadmapI've seen the mock-up that Bretch posted last week for the texture layers and I know animation layers and rigging nodes are in the workboard of Sybren, I also follow Jonas Holzman and the android port, all this could come in the next release, in the next 5 years or never, nobody is working in to improve the add mesh with the asset system or new retopo tools.
>>1030719>they will finish the animation overhaulI don't know about that, anon. If they wanted to fix the animation system and have it be useable, there's quite a few things that could have been fixed already. The NLA editor is how animation layers are implemented atm, and it's got a few quirks that make me want to use S2FM instead, then transfer the file to blender for rendering.For layers to work, they must be additive. The problem with rotation and scale, is that W for quaternion is default 1, and scale is default 1. When you do additive layers, they stack on top of each other, so the scale is too much, and the quaternion rotations break. The solution is for each action you create to be offset by - 1; for example a flexing muscle action has 0.1 scale, which it then ADDS to the default of 1. The problem is, now the action itself will have a shrunken bone muscle and be broken itself, so if you need to go back later to tweak it, you need to change the keyframes back to their regular 1 mid point, tweak, then subtract 1 from the whole thing. To fix it, the NLA could just subtract 1 from all scale values rather than me having to go into each separate action and doing it. If they were serious about animation, this would have fixed years ago. The second issue is how unoptimized Blender is. I think they're bullshitting when they say "look at millions of vertices on this screen!", When a 2008 TF2 map with 500k verts slows blender to a halt, WITHOUT the animation system playing. The camera and screen trace system itself is dysfunctional for this purpose as well. I have spent over a year trying to debug why this occurs, to no avail, and other complicated scenes that I make suffer much worse. 1/2
>>1030594It looks like you have no hisanimation add-ons installed, he has a plethora of tools to do exactly what you're trying to do, including snapping cosmetics to the character with correct weight painting>>10307412/2The screen trace method of only rendering what's in frame of the camera cannot be true, or at least, isn't implemented the way it's supposedly sold as. The new ambient occlusion system forces users to use GI approximation, which breaks in any animation scene. It suffers from the same issue that ambient occlusion does, where it does not occlude unless the object is a decent bit into the camera frame. This is particularly frustrating in any cloistered environment. I had a work-a-round for Blender 4.0, but it no longer works in 4.5+.Somehow the camera can see everything in the viewport enough to slow animation down, yet the camera somehow cannot see a bright orange wall just out of camera frame. I've been forced to render the scene at x2 resolution with a different FOV, then trim only the center, but it's not a perfect fix. Not to mention, EEVEE starts pitching a fit about shadows if you have idk, more than 4 lights in one scene. Once again, I don't believe them when they say the camera only renders what's in frame; it often exceeds RAM shadow limits for objects out of frame and far away. (I've literally tested this with 5 light bulbs in a closed room behind the camera, it breaks the lighting)>just use irradiance volume bakingThis works to an extent, and go ahead and call me filtered, but I could never get EEVEE next to ONLY consider irradiance volumes for world lighting; it will always try to do half GI approximation, or world lighting at all
>>1030744One more thing, I don't want to yap too much. I just don't believe it when people say animation will finally be good in Blender. For a variety of rigging reasons, animation UI and layers, rendering settings, it requires so much work and trouble shooting. These issues aren't unfixable, but they requiring fixing any time you need to do something new. The new changes in 5.2 seem...superfluous? The rename and duplicate function is nice for making bones, but you can't also add them to a collection in the same action. The one cool new feature is half-functional out of the box. There's odd additions such as "the animation player can now go back to the start frame after it plays" and "If you parent an object to a bone, you can slide between the head and tail!". I have no clue how anyone would ever needs these. The NLA editor is still broken, why are they adding stuff like this?https://www.youtube.com/watch?v=cERh1l2RtfUYou can even tell this is oft-asked question by other blender users, at 12:01. "Where are the animation layers?" Indeed. Blender is an amazing free program, and if more people gibs monies to them, it would be better, but the devs have not focused on animation or rigging for the past couple years, they've focused on their soft-of-kind-of raytracing engine. I'm guessing that's where they're getting their donations and big users from, people who use Blender to render furniture, kitchens, for websites and so forth.
>>1030745I know, but I have hope that things will eventually improve. According to their roadmap, work is planned to start in August;https://code.blender.org/roadmap/https://projects.blender.org/blender/blender/issues/154504>The work is planned to start in August 2026.>Non-linear editing with strips (2027 or later)I'm beginner in animation though, so I cannot comment on what u said, I just hope it will improve cause all I care is about animation, since I can't animate futa porn on maya without making all my own models from scratch, so this is all I have. I also like world building, but I think Blender's modeling tools are very ok at that now days right. Also, I can't donate cause I have a credit card debt already.
>>1030745as someone who is just now dipping their toes into animation, are there any essential addons you recommend that will make life a lot easier?
>>1030378>tfw all of the chads talking about 5.2>tfw thought 4.0 was still the latest one>tfw I'm still using 3.6Bros am I missing out? Be real with me here.I don't like have to search for modifiers because I'm too retarded to remember the names of most of them and need the list to remind myself what I'm even looking for. That's pretty much what put me off using 4.0.
>>1030749Uninstall Blender, get another program and learn from that one. It’s been told millions of times, Blender doesn’t do animation well. You don’t get the same experience as a professional animator in Blender. The developers of Blender Foundation will do something and always do something bad to the animation tools you require.
>>1030752>I don't like have to search for modifiersAre you talking about add-ons? The new extension system is definitely better and the newest version remembers your setup so you don't have to go and reinstall add-ons after every upgrade. I think v5 is a big improvement over v3, but I also don't use Blender other than for basic meshes to 3D print.
>>1030754Nah I mean the modifiers not the add-ons. I know they all/most have hotkeys, but I only remember the subdiv one.Remembering the setup is pretty cool though. I think every time I've upgraded it I've just copy and pasted all of my add-ons into the new folder. Or at least that's what I did with the scene setup that I prefer so I assume I did the same with add-ons.
how do I apply a bone custom shape so that when I append the character to another file it doesn't bring the custom shape objects along?
>>1030757Link instead of appending
Topkek
uhh bros... After biting the bullet and dishing out $60 on retopoflow finally, I was reading through the docs here https://docs.retopoflow.com/v4/general.html and under terminology they mention retopolgoy. What did they mean by this?
>>1030775Typos happen.
>>1030511>>1030694I think I fixed a lot of the 5 poles. I need some inset faces for...reasons>too many verticessorry, I need those, and it's industry standard to have 60,000 vertices. Mine is 40,000, the valve characters from deadlock more than mine. If I can game with 12 of those characters on my screen at once, blender should be able to handle it.>aren't there base models?There are, and they're ugly
>>1030780Whoa... how did you animate that? It looks really good. I have to know the secret.
>>1030752Eevee has been getting more accurate lighting, and the new texture cache improves render times across their benchmarks (I haven't played with it much myself), I do like the new thin wall option on materials and it makes planes like glass or leaves calculate faster and more accurately disperse the light. You can still create solid objects for things like windows but the difference is pretty minimal imo and the performance is significantly worse because it has to calculate light rays bouncing around the inside of the object rather than a single efficient calculation as it passes through a flat plane. Also I'm pretty sure you can add the modifiers you like to use to quick favorites and skip the searching altogether. Overall I'd say it's worth it to upgrade to 5.2 for the new nodes and performance+lighting improvements alone, especially if you're using both Eevee and Cycles as the raytracing in Eevee looks nice and is pretty quick
>>1030781curve modifiers that blend in to control shapekeys. For the vibrating, just a sine wave. For the boob bounce, two shapekeys (left right) with values going to negative 1, sine and cosine for left and right.The bra exploding was quick and dirty, just a duplicated bra mesh that swaps places with the original bra, quick effects-> explode, used cloth brushes in sculpt mode, moved the bra to follow a falling motion
>>1030784solid work it looks great! IDK if you'd like the idea but her not having eyes reminds me of the anon guy. not sure if it'd work with green skin but keep up the development
>>1030784A sine wave you say. That gives me an idea.
>>1030764rigify applies custom shapes without creating new objects, is it just internally hiding them?is there no way to apply the bone's custom shape and then delete the original object?
>>1030804Rigify does create the widget meshes and puts them all in a collection called WGTS_rigname that is then disabled, same as basically anyone using custom shapes.
Made this with paint 3D and then blender for materials. Really like how it looks.
I know jackshit about modeling, but i was able to whip out this cheese, i just have no idea how to:1. Make the Insides of the cheese wheel have more faces so they can be sculpted2. How to add some random surface holes and a 3d texture to make it look a bit more "real"Any tips?
the topology is kind of a huge mess, you should have added more loop cuts before inserting in the booleans, fixing it now would take forever. it's much much faster to just do the model again.also, if you're using booleans to create the cheese holes, make the added/subtracted object low poly and only after the boolean do you subdivide it. you have hole with a very high amount of points which is further messing up topology
>>1030886meant for >>1030856I somehow deleted the reply
Is this possible to do with auto rig pro? adding weapon bones and making them export via ARP is simple, but with all the existing drivers and constraints I think it'd break a lot.https://www.youtube.com/watch?v=mXKXIYUa79wIf it is not recommended then I suppose I need to make my own rigs and ARP is for characters without things that mess with the existing constraints/bones.
>>1030476>>1030441Hello againI need help with this head shape. I swear I measured everything for golden ratios with a reference, but I can't get how much the face should jut out. She either looks like she has fetal alcohol syndrome, or if pulled out too much, like a chimpanzee
>>1030895It'll look fine once you add the hair
Blender Animation course from TOAnimate dropped here if anyone wants it>>1030874
>>1030895>but I can't get how much the face should jut outWell... Let's see what the side profile looks like.Here are a few tipsFor the top of the head: The back cranium should be wider than the front forehead. I can see that in your top view, you have it backwards. You made the front wider than the back.Ears are super important for judging the overall shape of the head. Block in some ears at leas. Even if you don't want to get into the details of the ear, you still need to see them to help you assess the shape of the head.Similarly, lash lines are important for framing the eyes. Again, even if you don't want to do them in detail, even a simple blocking of lashes will help. At least the top lashes. Lower lashes are less important.
I have an eyeball mesh for a model where a perfect sphere will not fit the socket without poking outside the skull. So I used a lattice to deform it. The problem now though is, how do I animate with it now? If I turn the eyeball won't it just show the ugly hidden deformation?
>>1030895What is the purpose of this model? Are you going to animate with it or...? Your topology for the face, from what I can see, is fine. But you have waaaay way way way too many quads on the body. Unless you're rocking a nice gpu like a 5070 or higher, animating with that is going to be super sluggish
>>1030930My 10 years old laptop animates stuff like that without a hitch. if you have perf troube rather than actual polygon density the problem is usually that your character's modifier stack is making subdiv go through CPU, so just use your simplify settings to set max viewport subdiv to 0 which will bypass it
MPFB is starting to be okay but where can you get more clothing and hairstyles? Everytime I look for these I can only find those from the owner's website and they suck
>>1030889please reply
>>1030930>why do you need that many vertices?reasons in the last image, anon. >the face topology is fineI don't agree, or at least, when I look at professional 3D studios, they have much better topology flow. The side of the bridge of the nose flows into the underside of the eyebrow ridge, I want to do stuff like that. I also don't want the cheeks to look all twisted, and the lips curl outwards for topology flow, so on and so on. I plan on rendering in EEVEE anyways, it only takes about 2-4 seconds per frame which is fine imo. >>1030899>>1030920I took the advice on the cranium, now I need to model ears (I tried doing blockout but still suck at topology flow). Shit is difficult, because cat ears are actually half-spoons that THEN fold onto themselves. I want to be able to rig the ears later if needs be. Any advice would be appreciated
>>1030950forgot to ask, I want to avoid 6 poles, particularly with the nose, yet the nose needs to be a 5 pole to jut out. After doing that, I want a supporting loop by insetting the faces so there's a nice sharp line where the nose ends. However, this creates a 6 pole; you can see it in the original image with the pink nose and white loop going around. Is it fine for armatures and subdivisions?
>>1030929You should be able to rotate the eyeball inside the lattice, so it keeps the correct deformation. This could result in the iris/pupil changing shape as the character moves their eyes, but in a front-to-back squash it usually looks fine so long as they don't look too far to the side. If you've squashed the eyes side-to-side, the effect will be more pronounced and obvious with even slight movement.You could try making the eye texture get projected from an empty or something, so it's always kind of the right shape, but a sufficiently deformed eyeball will still become obvious at some angles. You can probably make some kind of weird set-up with custom normals and virtual surfaces so that it works out, but I wouldn't know where to start with that.An alternative to deformed eyes is to simply have the problematic geometry disappear. You can do this in several ways. For example, you could add an invisible object to the characters head, covering where the eye would protrude and boolean difference they eye mesh. Or you could have geometry nodes delete any faces which aren't facing the front of the eye-socket. Or do the same thing with the shader, mixing in an invisible material wherever the normals are facing in a different direction to an invisible control object attached to the head.
>>1030929>>1030972I tried doing it a few different ways, and I think the best solution is a shader which just makes everything except the front of the eye totally transparent.I used an empty and some math to define where the front of the eye is (you can't just use the eye's own coordinate system, because it needs to be independent of the eye, but you could use the face for the coordinates and a hand-drawn mask or something if you want to fit the eyelids exactly.The bottom bit is just a basic shader for the eye itself; replace it with whatever you would normally use.
>>1030974Actually, you might need a somewhat more complex setup if you're going for seriously giant eyes. One set of coordinates for the eye transparency, and another for the pupil, because when your eyes are so big that they are overlapping each other and emerging from your head, the front isn't going to be inside the eye's outline, and rotating them so that the pupils are within the visible eye will make the character look wall-eyed. So you need to have a pupil coordinate system (probably driven by an empty or something) which rotates around the middle of the eye-socket rather than the eyeball.Yet another alternative is to simply fill the eye-sockets with a holdout material and render the eyes separately before pasting them in with the compositor.
Is there a way to render two scenes at once? Like, imagine a portal from the game Portal or something, where there is a hole in reality you can see through to a different place. It's easy to composite two renders together, but can you do it so that you only have to render once, saving on rendering all the bits beyond the portal that would be obscured by the 'front' image, and maybe being able to see the effect as you are editing rather than having to plan it all out and only see what it looks like after rendering and compositing everything.
Is there any way of making a wireframe / skin modifier type effect which doesn't have completely dogshit geometry? It seems like it should be possible to just turn some lines into tubes that connect OK, but when I actually tried in geometry nodes it was an absolute nighmare.It's easy enough to break a mesh into lines, then write a recursive function which spawns a profile object at the centre of each line and use a raycast to project the vertices of that object along the path of the line. What got me stuck is that then you need to have objects for the raycast to hit, based on all the other lines (or, even better, just the ones which share a common vertex with the line in question in the original mesh). It seems like a plane which ran along the normal of the line (or at least some direction perpendicular to the tangent) in one axis and a kind of average of the tangents of the two lines in the other would do the trick. I would also need to figure out a way to stop planes from random other lines in the structure just blocking the incoming rays before they hit the planes at the junction. I think it could work, but getting all the pieces right is proving to be a headache.
>>1030978In cycles there's a portal BSDF that just does it out of the box.beyond that, rendering once is sorta easy, you make two scenes in your file and use both as compositor inputs, and when you render the scene that's containing the compositor it'll render the other one too. You'll need some constraint to keep the cameras in sync of course.saving on rendering though, not really. Since it's still a composite it'll need to render both in full. At best you can use geometry nodes to replicate the scene behind the portal (create a plane, add geometry nodes, points on mesh(face)-> instances on points->realize instances to copy the scene as one mesh) and then to clip it with "delete geometry" nodes or booleans set to remove stuff outside the camera field of view.
In geometry nodes, is there some way to pick points on a curve which share a position with points on some other curve? I want to spawn some instances on points which share a position, but I can't figure out how to build a node structure which picks a point on one curve and then finds all the points on some other geometry which share that position.
>>1030977>>1030974>>1030972I don't know if I'm wording this correctly but could you "animate" the eye by setting up a driver connected to a mapping node in shaders?
>>1030989You could, but that isn't really the easiest or most intuitive way to do it. Mapping to an empty object and driving the empty with a bone or something is way more straightforward.My fucking navigation widget keeps re-setting its colors to green for Z and blue for Y instead of vice versa, even though I've saved the latter as my startup and preferences. Is this a known bug?
>select hierarchy doesn't work on bones :^)fuck youwhat is the use case for shift clicking down a list of bones and only getting the topmost one. i want everything faggot
>>1031005>what is the use caseok, the use case is preserving their hierarchy when i parent them all to something else. i still don't see why it's so hard to range select EVERYTHING in the outliner though
Sorry if this is old news, but what the fuck is up with Blender's directions in the 3D viewport?Positive Y is 'back', with positive X on the left and negative X on the right.Positive X is 'right', with positive Y on the right and negative Y on the left.Same on the other side; 'front' puts positive X (left) on the left when you're supposedly facing the front of the object.But if I'm looking at someone's back, their left should be on my left, and their right on my right. If I'm looking at their front, it should be the other way round.It's like they changed the naming conventions between axes, so in one case they refer to the front / back / left / right of the object, and in others to that of the observer.
How to do rust (or grime, water, etc.) steaks procedurally? I assume use edge detection, ambient occlusion, and normals to figure out the initial placement of rust patches, but then I'm not sure how to do the streaking. At first I thought maybe just randomly displace bits of the vector coordinate downwards, but although that makes a kind of streak, the top is also displaced, so the streak soon detaches from its rust patch. So maybe some kind of repeat node setup, where it keeps blurring a bit, then combining the result with the previous result using multiply or something?Anyone have a guide for exactly how to do that? Or know of another way to achieve a similar effect?
>>1031014It’s called PBR a different image format file and design for repeating 3D graphics.Here is a video explaining it: https://www.youtube.com/watch?v=PjGCtnEDDeUHere is how to make one: https://www.youtube.com/watch?v=GItohAbyzCM
>>1030950>>1030951>The side of the bridge of the nose flows into the underside of the eyebrow ridgeIf that's really what you want, then force it too.Clear out all of the faces that don't support that flow, connect the edges that you need to make that flow work, and then patch or fill up the remaining space.You've probably read or watched topology guides, and they talk about "loops". That's essentially just creating a circle of faces around a feature. The most obvious loops are around the eyes. Because your eyes are essentially just holes, so you can just loop around the eye hole a few times in order to get good deformation there. Similar deal with the mouth. You It's not overly complicated, because the mouth is just a hole. Create as many loops around the mouth hole as you need to make it look nice and smooth.Loop around the most broad shapes first, and then connect them will some filler faces if you have to. The objective is to try and get away with as many clean loops as you can, before you're forced to connect the loops with patch work faces. If you're clever enough you can make loops overlap in a way that they're both clean. doing so usually results in creating 5 poles. 6 poles are the results of 2 points turning on the same vertex. It's not inherently bad. Sometimes you just need loops where you need them. Poles be damned. Around sharp corners and indentations, a 6 pole won't be a problem for shading. And as long as the loops are clean, it won't be a problem for subdivision of armatures either.Though with the case of your nose, you can probably get that sharp look by manually painting a normal texture around the nose. You don't want to rely too much on topology for small details like that. Or else your mesh will get messy and difficult to manage. Rather, the mesh should get the broad shapes, and then when you're ready to texture, let the normals handle the smaller things.
>>1031015nta, but thanks, I learned something.
>>1031015Maybe I didn't explain myself well, because that doesn't answer my question at all. I know how to texture in general. What I want to know is if there is a procedural method to quickly generate a specific feature, which is the long streaks of discoloration caused by rainwater flowing over a surface, dragging particles of rust and dirt into near-vertical lines below certain areas.
I figured out a very basic steak-generating set-up, but I think this might be a dead-end, because I can't figure out how to combine it with masking in order to have the steaks start within a masked area but then extend beyond it. Just putting a mask over the effect restricts the whole steak to inside the mask. Ideally, it needs to be able to take a second mask to control horizontal spread too. It should also have some method of narrowing the streaks as they go down, which seems tricky to implement in this paradigm.
I know you can use reference objects (e.g. empties) to place textures in specific areas by using the texture coordinate node's object output. That's helpful for placing markings, stickers, etc. on things.Is there a way to do something similar with the coordinates centred on arbitrary points on a mesh (e.g. those created in geometry nodes with distribute points on faces, or elements of the mesh with a certain attribute)?I guess it's possible to do a Voronoi-style pattern by simply selecting points in space and then comparing the position of each element to the nearest point and saving the relative position as a named attribute, which can then be called as the coordinates for a texture, but that won't allow overlapping. I guess you could just make sure the areas are spread out fairly thinly and simply repeat the process several times, but that seems inelegant.
I'm not sure I understand the premise here. You have curves that are overlapping? Something like pic related? And you want to pick all of the points that overlap, so you can place instances on them?
>>1031034Essentially, yes. I had a series of curves which had been formed from a mesh, so at every point which was previously a corner of the mesh there were multiple points, all with the same location but connected to different curves. I wanted to spawn instances in such a way that their position was dictated not only by the curve their point was part of, but also by all the other curves that had a point which was also at the same location.It turns out I just needed to know that the Sample Curve node existed, because it did exactly what I needed.Now I just need to wrap my head around the trigonometry of actually placing all these instances facing the right way.
I found this image, which seems like it could be helpful, but I don't understand the math involved. What purpose does the cross product of the normal and the tangent serve?
>>1031039I'm glad you found your solution. But I'm still curious what you were/are attempting.>I wanted to spawn instances in such a way that their position was dictated not only by the curve their point was part of, but also by all the other curves that had a point which was also at the same location.What does that mean? If the points are all overlapping, then they all share the same position in space. So what does sampling all the other curves accomplish exactly? And if they points overlap, then wouldn't you be creating 4 overlapping instances? Is that what you want? I'm confused.
>>1031041I'm trying to build a set-up which takes a mesh and turns it into an open trellis of pipes (think scaffolding). The pipes need to be cut at appropriate angles to fit together without looking all fucked up, and I think the easiest way to do it is to spawn a bunch of planes at the end of each pipe, then raycast the profile from the mid-point into the planes, which should result in a vaguely appropriately-shaped end (I still need to figure out the whole merging them together thing; one step at a time). In order for the planes to be placed at the correct angles, they need to know which other curves are involved in the intersection, so that they can use the curve tangents.It's probably not the most efficient way to do things, but any other solution seems even more technical.
What's the best way to git gud at UVs?Also, does any fucker actually post useful guides as just text and pictures any more? Seems like everything is a video now, which is actually really slow and difficult to learn from compared to a simple webpage which would take half the time to create.
Do you guys think this brick wall was done with a mesh for all the bricks/gaps or was it just simple planes with displacement/height maps?
>>1031044>What's the best way to git gud at UVs?Can you be more specific?
>>1031049I want to significantly improve my skills with using UV mapping, because they are currently rather poor. I understand the basic concepts of mapping a mesh's faces onto a 2-dimensional image space, but the fine points of how to do it well escape me. I want to learn when and how to use seams appropriately, how to get the faces lined up UV space so I can do things like apply procedural patterns to them with predictable alignment on the model, how to work with multiple UV maps, and especially how to use geometry nodes to create UVs as needed in a procedural manner.What specifically prompted this was that I'm struggling to implement some effects procedurally and have come to the conclusion that I need to have a lot of overlapping instances of the texture applied over the surface of the mesh. So I figure that I can use geometry nodes to identify appropriate patches of surface, unwrap each of them to a UV map (I'm not sure whether it is practical to have a few hundred UV maps on one model, or whether I will need to find some way to get a lot of separate patches onto each map), then use the UV coordinates to lay out a unique feature for each patch, then layer all of those together in the shader, and finally use the result to mix shaders and get the final effect. However, I think there are several gaps in my knowledge of UV maps which will make it difficult or impossible to do all that without first gaining a better overall understanding of them.
>>1031050Maybe someone else can help you but as it stands I'm struggling to grasp what it is, specifically, you are trying to do. But you can always try prompting Youtube and searching for a video.
>>1031048Real mesh since it's for a movie and you splurge on quality there. And the gaps look too deep to be pure height map trickery.
>>1031048It definitely looks like height to me, look at the cutouts for the grates on the bottom. Also the rounded edge of the wall between the windows. The only thing I'm confused by is the grout or whatever the seam where the green window panes meets the brick. Makes it look like the brick is mesh but it can't be.>>1031054The thing is extra faces doesn't instantly mean better quality. And when you have 2 dozen buildings in a street shot you definitely don't want detailed brick mesh. A strong texture setup will look 10x better and render 100x faster, meaning less work for better effect.
>>1031043Ohhhh... Interesting idea. I might work on it myself as a test. I might be able to do it. Raycasting is likely not even necessary.
I want to implement a material for an outline using the inverted hull method which inverts the colour of anything seen through it. Cryptomatte doesn't seem to be the solution, because it gives you only the shape of the object or the entire material, but I need to distinguish the inner surface from the outer. Green screen won't work unless I render twice, because I need to actually see what is behind the material.
>>1031064It seems that if you have anything transparent in EEVEE, it fucks up all the outputs from that material, even AOVs. So you can't have a see-through material which outputs its presence to the compositor, because adding transparency in any way makes it all invisible. So I guess rendering twice is the only way. Not the end of the world, but a pain in the ass.
I want to record the amount of light hitting each part of a mesh. I think this is something called something like light-map baking, but the modern internet is so fucked that it's basically impossible to search for a simple guide on how to do things and the blender manual is unhelpful unless you know exactly what you are looking for. How do I just create an image which records a light level hitting each part of a UV map?
Anyone got a guide on how to do procedural tank tracks? I figured out how to place wheels and make a mesh which fits around them, then instance the track links and connectors on the faces and verts of that, but there are a few details I'm still struggling with:How to get the drive sprocket to line up with the tracks reliably. All the tutorials I've basically just say to rotate it by hand until it looks right. Is there a better way?How to make the tracks sag slightly where they aren't supported by either the wheels or the ground. Convex hull and shrinkwrapping both give basically straight lines between wheels. I could always go crazy and run a cloth simulation, but that seems like overkill.How to rig the road wheels. They don't generally move straight up and down, but instead swing around an offset pivot (the torsion bar), so it seems like they need an armature to handle them. But that means building the armature and rigging by hand, rather than just having a node setup figure it out for me. Is there some function which can work out how much to deflect a circle around another circle so that it is all above a surface?
Is there some way of just copying what this thing does?
>>1031072I think this actually requires programming or if you want to manually do it like in the movies you gotta learn graph editor and curve to the best possible results.
>>1031074Why not use the thing?
Any advice on how to create hair spikes like on anime characters? I think I can upload a reference later, but every tard on youtube uses curve/mesh garbage that isn't optimized in vertex count, but also, doesn't connect with other pieces. I guess what I'm asking is how to make thick pizza slices connect to one another, the 'crust' halfway down the middle of another slice, and so on
>>1031077Come on dude if you search Youtube for this you will get dozens of step by step guides. You didn't even bother searching before you wasted a reply slot for this
>>1031077If I understand right, I think what you need to do is visualise a surface where the hair transforms from a load of cones / pyramids into a solid mass, which exists quite a bit outside the character's scalp. So probably something like extruding the scalp, sculpting that into the first few inches of hair, then extruding parts of that into the spikes. That seems rather tedious for several dozen spikes, but it might be possible to do something with geo nodes to make it easier. Possibly just poking faces (extrude with zero offset, then scale face top to zero, and merge by distance) and displacing with a voronoi or something.If you want to animate it and have the spikes be somewhat flexible without clipping through each other, that's going to be an additional challenge. You might be able to do it with a cloth sim, with a bit of stiffness in the material and some pressure inside.
Is it possible to do jiggle physics simulations in which both breasts and cloth interact? Like, simulating bra pressure. The closest i have got to it is the clothes interacting with the breast cages as if they were solid
>>1031081>The closest i have got to it is the clothes interacting with the breast cages as if they were solidwhy do you guys hate posting your examples, why must we respond to you based on text alone
>>1031054>Real mesh since it's for a movie and you splurge on quality theredead wrong, quality comes from high quality textures and rendering>the gaps look too deep to be pure height map trickerydisplacement bump maps would blow your mind
>>1031080I'll try what you said later tonight. Here is an example I snuck during my lunch break. Instead of a long cylinder, The sculpting uses less vertices and naturally wants to snap into my preferred shape, but I can't divine how to connect them. I will try using voronoi though, hadn't though of that.
Is this stroke method broken for anyone else in 5.2?
>>1031077Well, I thought it seemed easy, so I spent a few hours trying to implement a geometry nodes set-up which would give acceptable results, and it ended up being way harder than I thought. There might not actually be an easy way to do a large number of hair spikes with acceptable topology.The base of the spikes need to connect seamlessly, but they also need to be irregular, because the hair is typically made of different sizes and shapes of spike.So the obvious method is to use a mesh with faces sized and shaped for the various spikes, then somehow extrude the spikes from them.I thought that using For Each Element to isolate the faces one at a time, spawn a curve on each one, then Curve to Mesh using the original face as the profile would neatly solve the issue. Sadly, getting the original face to function as a profile curve is beyond my skills, because it has to be re-aligned so that it fits the curve direction. The best I could manage was getting it facing the right way, but spun randomly on the axis of the normal.My next best idea was extruding repeatedly. If you just want to go straight out, this is fine, although getting the taper right is much harder than with the curve method. If you want to have any shape at all though, it's very difficult. You need to somehow define a gradient space and move and rotate the face at each iteration. I think this is probably possible, but too much effort for me to test it and make sure.The next approach is using simulations. In this case, you make simple straight-out spikes, then add a cloth or soft-body simulation to them to manage the irregularity. I tried some basic experiments, but I don't really understand the physics system well enough to have a good go at it.The other obvious approach is simply modelling each spike yourself. This gives fine control, but is tedious (maybe not as tedious as figuring out how to do it the 'easy' way though), and wouldn't be as re-usable as other approaches.
>>1031085Maybe generate the hair on a flat plane and use something like this to fit it to the head:https://www.youtube.com/watch?v=Nj5a4blpcAk&t=1327s
>>1031085Picture a UV sphere. Rotate it so that the poles are about where the character's ears are. Delete the triangles around the poles and roughly half the other faces, leaving you with a 'cap' of quad faces roughly corresponding to where the hair should be.Now you only need to model about three or four spikes to form one segment. Mirror it and then duplicate and rotate it until you cover the scalp. If you did it right, the bases will all be touching close enough to use the weld modifier.This requires your character to have a fairly spherical cranium, but that's fairly standard for anime characters. If the actual head doesn't fit, just expand the hair into a spherical shape (you can use the cast modifier to bring vertices in line with this) and then start the spikes a few cm from the scalp (this is actually how most anime characters with spiky hair seem to have it).If you want a bit of randomness in your hair length and/or shape, you can either spawn it procedurally, tweaking the geometry each time you place it, or add some kind of vertex group or something to it and feed that into a node setup with a noise texture or something.
neurotypicals see nothing wrong with this
>>1031090A geonode rotation is an euler rotation which is a xyz vector, so implicitly convertible
When does Blender get good?
>>1031099Around 2.49
How the fuck do I animate a bulge in a object? Like a bug crawling under skin and you can clearly see the define shape of said bug? Image is sorta what I want to go for but as said a more defined shape.
>>1031101Depends how fussy you are about topology.Here's a basic geonodes setup which stretches one object around another, but the results look pretty bad unless the stretching object is very high res and the target object is quite rounded.In general, for this effect you are probably best off with a mobile patch of skin which follows the bug, because you will want to have much higher resolution there and be performing a bunch of actions on it. Blend the edges into the base model by transferring normals.https://www.youtube.com/watch?v=hgvYtwc5mUM
>>1031101If you've got shit-hot hardware, you might be able to brute-force it by making a solid, very slightly downscaled copy of both objects, then wrapping a high-resolution skin of the larger object over both and running a cloth sim. Get the settings right and the skin will stretch like realistic skin, bringing all the textures and everything along with it in an acceptable way. Get the initial bulge by clipping the stretcher through the stretchee.
>>1031102Bro I see you posting geonode solutions here all the time I think, how do you get so good at them? It feels like learning katakana, like it's impossible to just learn them, you can only learn them when you have a use for one
>>1031107Not that guy, but as someone who initially bounceds off geonodes the key is starting with simple small stuff and building on it. Also favor tuts that explain step by step how they build stuff, to get a feel for how semi-complex actions should be thought about
Should I even bother with sculpting in blender or should I just use zbrush?
>>1031115Try it
>>1031107I honestly don't think I'm actually that good at them. I've seen far more complex and impressive examples. Most of what I post is only half-way to the actual solution.But I would say that the tutorials which got me to my skill level were the Udemy Ultimate Blender 3D Geometry Nodes course by Alex Cordebard. He's an annoying fag, and you will produce a lot of ugly shit, but eventually a combination of repeated use building knowledge of what tools are available and how to use them and the general procedural mindset will sink in and you'll at least be able to use them with some confidence.Some basic coding knowledge doesn't hurt. Remember that nodes are basically a visual programming language. Likewise, the more comfortable you are with basic geometry, the better.If I had to give one specific insight which helped me more than any other, it's that Blender reads the nodes right-to-left. It starts at the output and works towards the input trying to find the values to put into the output.The other part is just how you get to Carnegie Hall. Specifically, debugging and problem solving will teach you more than anything else.If you get stuck, read the manual, ask here, ask Stack Exchange, or ask your favourite AI. I find basic-bitch Claude Sonnet 5 Medium to be shockingly good at debugging from just describing the issue in plain English.
I am trying to uv unwrap this human model which has seams down the sides of the torso. The front unwraps just fine, symmetrical, but when I unwrap the back side the left shoulder is squished for some reason. I've tried deleting half the model and mirroring it just to be sure and still every time I unwrap the back part of the torso, the uv island is has a squished left shoulder and isn't symmetrical. Anyone know wtf is going wrong?
>>1031128Post a picture
>>1031128Which unwrapping algorithm are you using? Are you unwrapping the whole model all at once, or one section at a time? Can you post a picture of the result?
>>1031130>>1031129the problem ended up being some fuckup with the butthole, i have no clue what specifically but putting it on its own island separate from the torso fixed the problem
I'm trying to learn the animation tools, and I'm so confused. I don't get the relationship between actions, slots, and strips. Do actions go into slots? Or do slots go into action? Which one is the container, and which one is the contained?What's with this "push down" button in the NLA interface. Why am I "pushing" actions "down" into strips? Can't I just create an empty strip, and then fill it with any action I desire?I just spent the last hour trying to figure out why pushing down an action didn't create a new strip on top of the others. Rather, it created a strip next to and slightly offset from an already existing strip. After some clicking around, I figured out that the strip on top is only a single frame long. If I edit the action of that strip to be longer, then attempt to push down again, the new strip properly appears at the top as it should. So it appears to be the case that having a single frame strip at the top of a stack creates some kind of conflict that misplaces newly pushed strips.Except... that's not true. Because I just created a fresh file, and tried to reproduce this problem, but I can't.I don't know what I'm doing. I need a turtorial that explains like the core concepts. What does what and why. I'm just clicking and guessing, and then opening up the manual and getting mad.
>>1031134Just buy a mocap or use AI and forget about it
https://x.com/Jackson_Boof/status/2085082694573002969this will help my /3/ career
>>1031134Historically, actions were all the linear animation data for a given, well, action for a single object. This poses problems when you want two objects to act in concert, like a guy and his cool animated weapon, or a shader qttributr that should change during the animatikn or whatever. Of course you could do separate actions for each object and name them so you recognize them but it's nothing formal. Or another typical approach is to have all the involed objects and attributes controlled by a single control rig (usually an armature) which is the one with the action. Action slots are the solution to all that.Thus, now, an action contains linear data for *one or more* objects. It has slots, you assign an object per slot, and the data goes in there. See slots as folders for the animation curves. A handshake action has left guy slot and right guy slot. You can swap the objects and if they have similar rigs it'll work out, and so on. So when you're using anxaction with an object, you're really using an action and a slot.To get strips you should peobably play a bit with the video editor. Drop a movie clip in the editor, a strip containing it is created. Cut the strip in two, now you have one strip for each half, but there's still only one movie clip, each strip is a window into it. Put a strip on top of another one and it masks it. Same for action strips and actions.Main difference is there's always an automatic top track for the object's current action, out of the NLA, that is your normal linear blender animation workspace. Since it's always on top the act of creating a strip from that action is "pushing down" the action
>>1031137I watched this guy's video a little earlier https://www.youtube.com/watch?v=ArrJe4c9ar8and I think understanding is beginning to dawn. Figuring out that actions are tied to objects cleared some things up.Pushing down feels like a gimmicky way to create strips. At least for now it does. Maybe I'll change my mind later when I'm more experienced. But for right now, it feels like they made the push down button to look cool, rather than actually be useful.You still have to push down each slot individually. In my mind, I thought that the action would contain all of the slots. And then you edit the action, which automatically edits all the nested slots.But if I'm understanding this correctly, even after creating the action, you still have to individually push down the slots you need for the object.The thing that's really bothering me, is that even though the slots are the things that are truly being created an edited, both the track and the strip inherit the name of the action. Shouldn't they inherit the name of the slot?I think I figured out the weird push down behavior from earlier. It had to do with the fact that my action has two slots. One slot starts at frame 1 and goes to frame 120. The other slot starts at frame 21 and goes to 140. I offset the second slot, because I needed the first keyframe to start a little bit late. And then loop it. It worked. I got a clean loop that works in tandem with the other slot. But when I pushed down the offset slot, it didn't make a new track, since the next track down ended on frame 1. They didn't overlap, so they were placed next to each other. At least, that's my guess.I'm still confused. Whining about it is part of my process. Please give me more tips if you have them.
>>1031102Would that node create something like this old effect?
>>1031140You can do that with a lattice
>>1031140It would take a couple of minutes to re-create it and find out. But basically, yes, if you shape the object which causes the bulge right. There are also many other ways of doing it, some of which are probably more efficient and/or look better.
>>1031141This raises another point about getting good with nodes; if you want to really learn a lot about nodes, force yourself to use them all the time. Don't accept that there are easier and possibly better tools for achieving things. Do everything in nodes for a few months and you will learn a lot far more than if you only used them when you need to.
I have a model with two different meshes for hands that I'd like to be able to switch between. One normal and one with some claws and stuff. How do I do the UVs so the textures blend in with the wrist/arm? I know how to do it with just 1 hand model obviously but how do I do it with 2 that have different meshes?
Are there any sites with gumroad/superhive stuff for free?
>>1031155high quality example of what I am after
>>1031155Depends on the specifics, I think. The 'cheating' method would be to just model the claws and so on as a separate mesh and give them their own UV map. You can use tricks like data transferring the normals to make the blend really seamless. If you need to have some parts of the old mesh disappear, you can do that by making them just fade to total transparency.There probably is some way to do what you want 'properly' but I'm afraid I don't know it.
>>1031158Is simply moving the seam to the transition point between the normal and spooky texture not an option?
>>1031138>But if I'm understanding this correctly, even after creating the action, you still have to individually push down the slots you need for the object.The NLA tracks are per-object, and each strip itself refers to a slot within an action, which is why you push them down individually. Basically remember that actions and slots don't refer to objects, it's the other way round. Objects are basically merely interfaces to play back or edit actions.As for the automatic naming, it's likely partly a holdover from when slots didn't exist, but also since tracks and strips are contained within an object the slot name will be the object's name most of the time, so it's redundant to include in the strip. Normally within a track stack you'd be combining different actions together, not different slots of the same action.But also no, the things that are really being created and edited are still the actions, the slot's basically just a filter on top of it.>But for right now, it feels like they made the push down button to look cool, rather than actually be useful.No, it's really handy because that means you can just edit your animation as if you weren't in the NLA and directly push down when it's good enough then immediately get started on the next action. If you don't like it you can just add a new action strip referring to your action or an empty one through the menu though.
>>1031170You sound you know a thing or two. Can you help me create cyclic animations, where the keyframes don't sit on the first and last frame? Because the animation has overlapping elements to it, at times, I feel it's necessary to offset keyframes. so not starting on frame 1 or 0. But say frame 20 for example. And then because the animation has to loop, I need the last keyframe to interpolate seamlessly into the first keyframe. But blender doesn't like this. It automatically assumes that cyclic animations should have a keyframe at the first and last frame. But I don't think it has to work that way. As long as it can tell where the first and last keyframes are, it can measure the distance between them, and interpolate properly. It shouldn't be an issue.Do you have a solution for this?
>>1031182Any walk cycle tutorial will tell you that but if you want the animation to be perfectly cyclic then the keys thatare meant to loop must be on the frame after last
>>1031182>>1031187Oh and offset keframes work the sane only further behind or ahead of the start/end
>>1031157i saw some on kemono
>>1031195what year did you download them? Kemono has been down for some time
>>1031201Kemono's up with the right proxy
>>1031203kill yourself vagueposter
>>1031204I tried jumping out the window but my fall was cushioned by that big pile of freely downloaded gumroad content~
>>1031206>acting smug instead of giving any details incredible larp underage anon, thank you for sharing your neurosis with the class
What's the best method to model AFV hatches? Specifically, how do you manage the blending between the hinges and the round hatch? Is it possible to do it in a parametric / procedural workflow, or do you have to actually get into the nitty-gritty of solving topology for each individual hatch design?
Is there an 'easy' way in geometry nodes to turn the outer edges of a square mesh grid into a circle? Or bring the edges of a circle into evenly spaced edges around the perimeter of a square? Or just bridge between two edge loops?Only way I can think of even approaching it would be to convert both edge loops into poly curves, then iterate around each spline segment spawning quadrilaterals with corners match the two ends of the segment and the two ends of the corresponding segment on the other curve. But there doesn't seem to be any way to guarantee that the two curves will actually line up so that the first segment of one is closest to the first segment of the other.I guess the other way is to just work out where each point is and needs to be moved to, but that's crazy levels of trigonometry if you want the number of points to vary.
Is this where I can hire someone?
>>1031236Nah, you want https://www.vgen.co
>>1031235Kind of managed it by just using trigonometry.Make sure circle and grid have the same number of edges (i.e. circle edges must be divisible by four).Repeat as many times as you have edges on one side of the square:>Draw a quadrilateral with points>First point is starting corner of square, plus an offset to bring it down the square an amount equal to iteration count x total length of the square's side / number of iterations>Second point is exactly the same (use a group) but at +1 iteration>Third point is point on circle with angle equal to starting angle (depending on which corner you started on) plus a modifier of 90 degrees x iteration divided by number of iterations; calculate position with simple trig: x = cos(a)*r y=sin(a)*r>last point is same as third, but for +1 iterationNow you have one quarter of the shapeDuplicate and rotate three timesAdd rest of the geometry back inMerge by distance
Good gpu for blender?my 3070 is nice but I feel with me getting into it more and more, the render times have been really rough for animations and renders in general. I see people say a 4070ti/5070ti would be best, what do you all use
>>1031240a single 5070 won't cut it if you're trying to do this beyond simply looking at models. a entry tier hobby setup would unironically be 5090 dual or single.
When dealing with a mesh or multiple that are stacked together or overlapped, is there an effective way to select and erase any faces that aren't visible from the exterior? I know box selecting in Solid Shading will only target the visible faces, but it's only from the one perspective, and I want to get every angle.I guess I could flood the scene with bright point lights, bake an AO image, and use the total black from that for the deletion.
>>1031234I think the hinge is the easy part, the little windows are harder imo. I'm a shitter but i'm sure the smartest way to do it is with some kind of normal trickery, like one of those extensions that blend normals onto surfaces. I personally would just brute force it because i don't know those extensions but i think my position is wrong anachronistic. There's also bevel shaders or just straight up intersecting geometry depending on what you're doing
If I want a object on a bed and the blanket of the bed deforming from the weight of the object moving what would be the best way to approach that?
>>1031250Strictly speaking, cloth simulation. But that requires time, good hardware, and a lot of effort finding the right settings.If you just want the blanket to indent a little, quicker to do it directly. What I'd do is make a small, flat (invisible) shape with the same 'footprint' as the object on the bed (i.e. it is the shape of the parts which touch the blanket). Parent it to the object or whatever you are using to move the object. Then make an invisible copy of the undeformed blanket and shrinkwrap this shape to that. Now you can add a geometry nodes setup to the blanket which uses proximity to the shape to move points downwards. Float curve will allow natural looking deformation.There are more elaborate methods which will give you wrinkles and so on, but at some point you reach the level where a high-resolution cloth sim would actually be easier.
>>1031250Didn't the latest edition of blender add some kind of simulation node to geometry nodes that is basically this exact use case
Anyone here good at rigging? There's a melina headhack someone made that has all the bones and skinning intact but the actual setup to link the head and body and make you able to move face erc isn't done. I have no idea how rigging works but would want to use it for animation.If anyone knows how to do it I'd be gratefulhttps://f95zone.to/threads/general-butch-models-collection-2024-02-29-g_v.183033/post-20820672
>>1031253I'm not signing up to your email collecting scheme.
>>1031237It's all anime shit
>>1031256If you didn't want anime shit you wouldn't be asking /3/
>>1031254If you've never used f95 you're a zoomer retard and should go back
>>1031253you're in luck. Shortly after you posted, a guy commented in that f95 thread saying he fixed some of the rigs and wants permission to post it to smutbase. Also, I will never not recommend taking the time to understand a bit of rigging. It's not nearly as daunting as lazy people make it out to be. A lot of it is just time consuming so most people tap out when there's not a 1-click easy rig solution, but that's why it's important to understand workflows and build a checklist of what you need to do. It'll take a month of study to grasp pretty much everything and less than a year before you're practically an expert
>>1031263nice, thanks for mentioning it. yeah i really should but it's hard to find the time. i work at a game studio but all the rigging and animation i just pass off to another person after i model characters. hopefully when things get calmer i can sit down to learn it
>>10312652/3 of these are completely free, and the 3rd is just extra tips/tools for the weightpainting process. Definitely enough info to get you going if you ever get a weekend to learn the workflow... Best of luck to you and worth noting I've never paid for blender studio's subscription so I can't vouch for the paid content as worth it or not especially if you're working on video games and not solely in blenderhttps://studio.blender.org/tools/pipeline-overview/asset-creation/rigging
>need a part for a project>'hmm... I'm going to need like four different variants of this part, and I don't want to spend an hour re-building it each time'>use geometry nodes>takes 12 hours to makewhy do I do this to myself?
>>1031284 if you only made it to specifically make those 4 variations then I'm sorry for your loss but at least you had fun making it right? >built geonodes asset that excels at a very specific niche purpose >never have to rebuild it again as long as you don't have multiple storage drive failures
What are some good methods of making things look like cloth without actually using the cloth physics system? Ideally compute-lite.Squashed Voronoi with smooth F1, distance fed through a float curve, used for displacement and bump gives a vaguely cloth-like wrinkle effect, but only looks at all good if the mesh density is fairly high (typical face size smaller than each Voronoi cell) and the object is tube-shaped.
>>1031087>>1031077I worked on something similar to this before. >>1028032Using geometry nodes, I got as far as creating spikes that seamlessly conform to the base mesh. That much is doable. To take it further beyond is likely possible. But I'm not sure I have the capabilities either. Looking back at the node group now, I forgot some of the functions.The broad concept, is that you want to convert the base mesh(or scalp) into the profile for the curves. Think about it like this: if you draw a line from one edge to another edge, it will always be flush with the face.(ignoring the hidden triangulation of the face)Now imagine drawing a circle around the curve. Each edge that you crossed while drawing, should have a point on it. And if you can connect those points with edges, you will have a flush profile.I kind of forgot how I managed that. But I did. You can start by having the base mesh sample the nearest curves. When a mesh samples a point cloud, it creates a vague sketch of a voronoi pattern. Each section of the mesh knows which point it's closest to. Creating the notion of boundaries where points meet halfway. And the thing about those boundaries, is that often they cross through edges. One side of an edge is within one boundary, and the other side of an edge is within the neighboring boundary. These edges that exist in both boundaries, are the edges you want to convert into points to create your profiles.Because an edge can sample the position of both points, it can subtract one point from the other to get the normal of the boundary. And with that normal, you can find the position where the boundaries intersects with the edges.(I used higgsas's "line plane intersection" node to do the math for me)There is a lot more to this that I'm omitting for brevity's sake. But I hope that will give you some insight.My node can generate 1,500 spikes in 500ms.(think light yagami haircut length) It's not good for real time. But it's usable.
How to make actually good tank tracks? Ideally as procedural as possible.Wheels aren't that hard, although making them procedural without distorting features like bolt heads and circular cut-outs is tricky.The track is difficult. Can't just array and fit to curve, because that causes distortion. Instead need to make a curve, resample curve, curve to mesh to make a ribbon of faces, then instance links on faces and connectors on points. Actually creating each link probably means modelling by hand rather than procedural, because they vary so much. Resample curve squashes things a little, but links can be slightly scaled in one dimension without it being noticeable.Generating the curve to place the links is another challenge. Easy enough if the track is taut and on level ground (i.e. entirely convex around the wheels). A bit more complex to fit to terrain. Haven't figured out how to make it sag in a plausible way without breaking out the simulation stuff.Suspension is a nightmare. Multiple different kinds, but the main one in the modern day is torsion bar. Torsion bar doesn't go straight up and down, but swings on an arm. Got to figure out how far to swing each arm to keep the wheel above ground. Fortunately I don't need to animate this, so I can just do it by hand if absolutely necessary.Drive sprocket would be a headache if animating, but fortunately able to just do it by eye for now.
