Aquila Kytori

Adding this just to illustrate where to load up the Physics model in the mesh uploader and where to change the Physics shape type to Prim ...................... AS mentioned above you will need to create a separate simplified building from prims to act as collision surfaces. After following Reys instructions given in the first part of her answer on how to create a physics / collision model using solid prims, not setting any of the prims faces to 100% transparent, (where possible it will be a simplified version of the visual building. Only having prims where you really need to have collision surfaces) you will have two "buildings" to use the Mesh Generator on. The first will be the Visual building and the second will be the Physic building. Use the Mesh Generator to convert them so that you will have two separate .dae files. To import back into SL : 1. Upload as usual the Visual Building.dae into the SL mesh uploader. 2. Next you will need to tell the mesh uploader which .dae file you want to use for collision surfaces. To do that open the Physics tab of the mesh uploader and in Step 1: load up the Building Physics .dae file and then in Step 2: hit the Analyze button. In the preview window you should then see your physics model as a collection of pastel coloured blocks. See screenshot below. Upload the building and rezz it out on the ground. By default the Physics shape of a mesh object is a Convex Hull. Think mesh wrapped in cling film. thats why you cant get inside because all the openings are covered over. To get your building to use the Physics model that you custom made for it you need to Edit your rezzed Building, open the Features tab in the build floater and change the Physics Shape Type from the default Convex Hull to the Prim . Now you can get inside Note that to the right of the Physics Shape Type drop-down there is a little eye icon , click on this and it toggles Firestorms "Show physics" option. The Collision surfaces will usually show as blue surfaces on the selected object. With this enabled you don't need to run around physically checking all the collision surfaces of your building you can just view them instead. Note in the above example the Physics cost of the object is alot higher than the download cost causing the final LI cost to be higher than it could be if the physics cost were lower. If this is a problem then there is an alternative way of creating your collision model using prims but its slightly more complicated. It involves setting some of the prims faces 100% transparent as shown in the second part of Reys answer, (the images of the prims with coloured faces) before using the the Mesh Generator and then NOT using the Analyze button in the Physics tab of the mesh uploader. Using this other method will quite often reduce the physics cost by quite alot. If you think you really need to lower the physics cost then we can explain this in a little more detail. Edited to add : In the above example the number of prims use in creating the collision building was 14. After using the Analyze button in the mesh uploader it should have created 14 Hulls instead of 22 ! 1 Hull for each prim used. this would almost half the Physcs cost. One reason this can happen is if you have overlapping prims in the Physics building or prim faces that are touching other prims. This can cause the uploader to create more hulls than necessary when it analyzes the .dae file you gave it. To help avoid this, when making the physics model leave a little space between each prim. A few centimeters ( 5 - 10cm) should be enough to avoid confusing the mesh uploader when it analyzes the model. Edited again: I think my answer here has become a bit confusing , Rey is saying "do not click the Analyze button" and my answer is saying the the opposite "in Step 2: hit the Analyze button." I will try to explain : There are 2 types of Physics model you can create, Box type and Planes type. Sometimes the Box type is called Hull type physics. Sometimes the Planes tpye is refereed to as triangle type Physcs. With the Box type physics you create your collision surfaces with boxes (prims, cubes). Using as few boxes as possible. When loading up this type of physics model to the SL mesh uploader you do need to click on the Analyze button in the Physics panel. The uploader will then "analyze" you physics model with the goal of it creating a collection of hulls that will be used for collision surfaces. In my example even though I used 14 boxes in my physics model after analysing the uploader decided that there needed to be 22 Hulls to have the correct collision surfaces. Thats why I later suggested leaving enough space between each box so that the uploader will find it easier to "analyse" and hopefully arrive at the same number of Hulls as Boxes used in its creation. Note: Each Hull the uploader creates has a fixed physics cost of 0.360. Lots of Hulls (boxes = higher Physics cost). With the Planes type Physics the collision surfaces are created using simple Planes. Again the rule is to use as few as possible. Fortunately the Mesh Generator ignores prim faces set to 100% transparent so to create a Planes type physics model using prims all you have to do is set the 4 faces of each prim that join the inner wall surface to the outer wall surface 100% transparent. When loading a Planes type physics model to the mesh uploader you don't click on the Analyze button. The result of not hitting the Analyze button is .......... you are asking the uploader to leave your physics model alone and you get a collision model exactly as you intended it to be. Thats a good thing . And there is another benefit. For large models like buildings the Physics cost is usually lower than the physics cost using box type physics. For Planes type physics (not Analyzed) a large part of the physics cost calculation is the average size of the planes used in its creation. The larger the average planes size the LOWER will be the Physics cost ! Keeping your Physics model as simple as possible using large planes (prims) and avoiding as much as possible small planes will, if done correctly always result in the physics cost being lower than the Download cost. Long story short, DO as Chinrey suggested. Use the Planes type physics for buildings and don't click on the Analyze button in the mesh uploader. I am sorry if this now all seems so complicated when really it shouldn't be. I considered deleting this post completely but I think some of the screenshots may be useful so will leave it as it is for now .

An example of how to use a single vertex to increase the Bounding box dimensions of a very small mesh object can be found here : For an object that needs to have its Bounding Box dimension increased along more than 1 axis you would have to place the single vertex somewhere to a corner outside the original BB.

arton is right again In the screenshot below, in green are the default units for Blender (and Avastar). When first opening the JewelCraft addon, at the top of the panel is a button to Set Units : ......if you hit that Set Units button it will change Blenders Unit Scale to 0.001 : so now when we add an avatar it is scaled by .001 making it 1000 times to small, (upper body X length 1.79 mm instead of 1.79m) : So to avoid this problem don't use the Set Scale button at the top of the JewelCraft panel : Instead to add a 1mm size jewel just set the jewels size to 0.001 : and all will be well again :

Check in the Outliner panel if the Spot light's little camera icon "disable in render" is enabled .

The Default, (leaving the option blank) if I remember correctly uses the LOW Lod model for Physics. If this low LoD is relatively complex (think smaller objects) then the Physics cost can be alot higher than the download cost. Example of a small plant, (see the build floater for the custom LoD tri counts). Default Physics model (nothing selected) the Physics cost dominates the download cost Download = 0.236 Physics = 3.480 Land impact = 3.480. and when rezzed : Same LoD models but this time using a single triangle for Physics : The physics cost is under control. Download = 0.236 Physics = 0.200 Land Impact 0.500. When rezzed and Physics shape type is Convex hull or Prim the Physics weight remains at 0.2 Agreed Lets forget the triangle idea ............... . (but leave the User defined option.......................... )

Confusing? yes but worth keeping. My two cents : First thoughts were "What's the difference between From File and User Defined. Surely they are the same thing?" Also having them one above the other at the bottom of the list doesn't help. Perhaps have Load from file (instead of From File, as this would be consistent with the wording used when choosing the Source of the LoD models in the Level of detail tab) at the top of the list. Does anyone use the High, Medium, Low or Lowest options especially now there are the new ones? If yes aren't those covered with Load from file ? So with the LoD options removed next would be Triangle because that seems to be a popular choice to have, followed by Cube and Hexagonal Cylinder. (Cylinder added because that's what it is, less confusion). And finally User pre-defined instead of User defined. At the moment when hovering your mouse over any one of the options it displays the same "LOD to use for physics shape" in the info field. Perhaps you could make use of this information field to show a link to the relevant Beq's tech blog page or have a few words describing what it is ? And why is Choose one . . . shown twice ?

Arton again got there first lol and notes something important :

Earlier in the year the Firestorm Devs made some improvements to the mesh uploader and one of them was adding 3 new options when selecting which Physics shape to use. From Beq's techy blog : https://beqsother.blogspot.com/2021/06/summarising-next-improvements-to.html " As of the next Firestorm release, creators will find 3 new options when selecting a physics shape. 1) The cube. 2) The hexagonal cylinder. 3) the user-defined mesh. " User-defined means that if for some reason or other you often find yourself needing the same basic Physics shape for different models you can now have this same particular custom shape readily available as an option in the mesh uploader. For example, lets say you like to use a triangle based pyramid shape for many of your projects : 1: Create the triangle base pyramid model and save it as a .dae file somewhere on your hard-drive. 2: Next time you open the SL mesh uploader go to the Preview settings options and paste in the location of the tri_ pyramid_Phys.dae file. 3: Open the Physics tab and now whenever you choose the User Defined option the triangle based physics shape will be automatically loaded as your physics model. Choosing to use the User Defined option and not having it linked a .dae file on your hard-drive will be the same as leaving the "Pick a physics model" field blank. The Physics cost then varies from model to model (depending on complexity ?). And in the Features tab of the inworld Build floater > Physics Shape Type, you wont have the Prim option.

Yes I had Animesh enabled in the Build floater when rezzed and the size/scale remained the same as it was in Blender. If you would like me to check, you can send me your bat and I will try re-scaling it as I did the dragon and see what happens..

I just tried with the original dragon mesh and in two steps the dragon were resized and it kept its new size when rezzed and set to animesh inworld. Example with the dragon: 1. In Blender Object mode Select the dragon then Shift select the bones. Now you can rescale them as needed. In this example they were scaled to half size using S key to scale followed by typing 0.5 and LMB to validate. (S + 0.5). 2. Next the new scale has to be applied. : With both items still selected use Ctrl + A to open the Apply menu and choose the All Transforms option. Export the dragon as usual using the SL+open sim rigged Presets. In the SL mesh uploader > Rigging I had both Include skin weights and Include joint positions checked. Below is a screen- shot of 3 dragons of different sizes set to animesh and keeping their correct new size when animating: If it works for dragons then it should work for bats 🙂

@Quarrel Kukulcan yes I believe that it is your version that VK has shared with me and we are using here.

The other way of manually assigning or removing vertex weights weights: 1) Enable the Vertex selection (masking) option. 1b) Note now that below the vertex list, the options to Assign, Remove, Select etc are now available. 2) Select the vertex (or vertices) that you want to work with by holding down the Shft key and LMB to select the vertex. 3) From the list of vertex groups select the relevant vertex group. 4) Check that the weight value is set to 1.000 (100%) then hit the Assign button. 5) The selected vertex (vertices) should now pop back into the correct position(s). 5a) Note that in the left hand side panel that the vertex weights are now displayed with the vertex group and the vertex weight value. Also if you had more than one vertex selected when the assign button was used it will be the last vertex's weights (the active vertex) that will be displayed.

OK, if you like we can take it one step at a time and try to work out what it is that is different from my Blender settings and yours. I have saved a Blend with the viewer set up as in the screen-shot below. I have added the link for the download in your DM's. It should open just like in the screen-shot. Note that 1a: Vertex selection/masking is not enabled. 1b: Brush settings, Weight = 1.00 and Strength = 1.000. 1c: Auto Normalize is enabled. 1d: mhandMiddle1Left is the vertex group that is selected. This was not selected directly from the list of vertex groups but by holding down the Ctrl key and hovering the mouse over that bone in the 3D viewport and left clicking on the bone. In the screenshot it is the vertices that I have indicated with the orange arrows that should be assigned 100% to the mHandMiddle1Left bone. But at the moment they are not weighted to any vertex group. So step 1: will be to see if you can paint over those vertices so that they snap back into there correct positions. As I said before the .blend file should open almost exactly like in the screen-shot. I want you to try painting over those vertices and see if they snap back into position. To do that hold down the Left mouse button and drag the orange circle (paint brush) over each of those vertices. You may have to do it one vertex at a time. If they snap into position I would like you to then select the next bone (mHandMiddle2Left) by holding down the Ctrl Key and selecting it with the LMB. Then painting over the remaining "sharp" vertices. They to should snap back into position. If you do not succeed with this then take a screenshot and post it here. If all is good then try continuing on to the next step which is selecting all the vertices and Limiting each verex to a maximum of 4 vertex groups etc .

Thought of this thread when I came across these two videos today :

Thanks Quarrel, I have modified the relevant screen-shot .

In SL a single mesh object, when rezzed inworld is composed of 5 different models. 4 LoD (level of detail) models, High, Medium, Low and Lowest and a separate Physics model. When uploading your car object into the High LoD slot you have the choice then to let the Uploader auto generate the 3 lower LoD models or you can upload your own custom model in each of those lower LoD slots. In the Physics tab you also have a similar choice for the Physics model. Auto generated lower LoDs for something like a car will always be bad but if you don't care about how they will look then simply manually reducing the tri counts using the sliders for each LoD slot can reduce the Li cost a lot. Ideally you should be optimizing your High LoD model in your 3D editor before uploading, reducing the tri count to what you think is the minimum and also creating your own lower LoD models so that you have full control overh both the way the car will look as you zoom in and out and and the Download cost ( LI , "prim" cost). A car model should consist of a collection of different objects, for example the outer shell object, wheel objects, door objects and the interior of the car object. Each should have their own lower Lod models ...................... getting complicated now :). Here is an example from an earlier thread: Read the text and note the tri count for the complete car (15250) and the tri counts for the shell of the car (3550) and for each of its different lower LoD models, (2350, 218, 24,) If I remember correctly the final LI cost for the complete car was around 45 when rezzed. ( the car shell by itself would have been ...........a little less than 20 LI)

With vanilla Blender the option can be found in Weight painting mode, with all the vertices selected > Weights > Limit Total : And to Normalize the weights for all vertices : again with all the vertices selected, Weights > Normalize All In the operator panel uncheck Lock Active : And when weight painting enable the option Auto Normalize :

Stop whinging ! (complaining all the time). You have done all this before with the Dragon. If you would just sit down and watch a weight painting tutorial or two you would know how to sort out this kind of problem already. I am sure we said the same thing 2 or 3 months ago. So with that said a little recap : That is because those vertices are not assigned to any vertex group. They have no weighting at all. The clue is when you rotate a bone further up the chain, for example the mShoulderLeft we see that some vertices are left behind. To illustrate this, in the image below I have rotated the mShoulderLeft upwards (green arrows) and we see the "sharps" or "spikes" as you call them (red arrows). These are vertices that were left behind when rotating that shoulder bone. Left behind because they are not assigned and weighted to any vertex groups. For some reason or other they were not weighted when we parented the mesh to the bones using the Automatic weights option. How to fix this? 1: With mShoulderLeft still rotated upwards we look carefully at the mesh around the area where the vertices were left behind and we see that they should move along with (be weighted to) one or other of the two bones that are indicated with pink arrows. ( mHandMiddle1Left and mHandMiddle2Left ) 2: Select the first bone, mHandMiddle1Left (pink arrow in next image). Select the Paint brush tool and set the Brush settings as shown (outlined in green). Tip: when weight paining control the effect (influence) of the brush by having the Weight at a value of 1.000 for adding weight or 0.000 for removing weight. and nothing in between. To vary the influence of the brush vary the Strength value. Normally when weight painting the strength value should be a lot less than 1.000 so that you are adding or removing weights gently, slowly adding/removing influence as needed. With the brush Weight at 1.000 and the Strength at 1.000 paint over the vertices that belong to the selected bone (indicated by the pink arrows). they will (should) "magically" pop back into position 3: Select the second bone (mHandMiddle2Left) (pink arrow) and paint over the vertices (orange arrows) that should be weighted to this bone. They to will pop back into position. Done Unfortunately not 😒 Many of the vertices of the bat belong to more than 4 vertex groups right? but we know that for SL, 4 vertex groups per vertex is the maximum! You are now using the Avastar addon for Blender which I don't have but I believe there are tools in that box that should be making your life easier and one of them is ....... ...........a method to ensure that no vertex is influenced by more than 4 vertex groups. (and also I believe another tool for checking that all vertices belong to at least one vertex group). So next step for you should be to find out how to apply these tools I would imagine it would mean something like selecting the bat and re-parenting again with auto weights but this time ensuring that no vertex is assigned to more than 4 vertex groups. If after fixing the problematic vertices the model is all over the place because of rotating bones when weight painting you can reset to the original pose again by : in Pose mode, select all the bones, open the Pose menu > Clear Transforms and choose the All option.

Bat4.1 joined : I was getting: Bone Heat Weighting: Failed to find solution for one or more bones. After following Quarrels advice to Merge by Distance with a distance setting of around 0.0061m which removed 206 vertices and then again Ctrl + P > With Automatic Weights the mesh was animating approximately along with the bones. You need to be a lot more careful how you align the individual bones to the mesh some of the bones look like straw thrown on a roof Overlapping bones like that are not a good idea. Start by deleting all the bones you don't need then spend time positioning the bones correctly. Doing that will probably save you some of the frustration of manual weight painting. When positioning the bones on the model change the bone Display As from the default Octahedral to Stick. That way you may find it easier to position the bones more accurately.

I just realised that your image shows that you are in Object mode ! You have three windows (meshes) but only 1 object , correct? Note that Alt S only works in Edit mode. 🙂

Doing a google search for how to extrude out edges while keeping a constant offset (the opposite to Face > Insert > Offset Even ) I learn that this can be done by enabling the Mesh: Edit Mesh Tools add-on that comes shipped with Blender. Edit > Preferences > Add-Ons > do a search for Mesh and from the search result choose the Mesh: Edit Mesh Tools. Now, amongst other things there will be some new options in the Edge menu. https://docs.blender.org/manual/en/dev/addons/mesh/edit_mesh_tools.html The relevant one for here is Ctrl E for edge menu > Offset edges > Extrude.

I may be answering a question you didn't ask because I understood casements to be a border around the window shape but now I am not so sure. 😕 Anyway this is one way how to extrude an edge out and keeping a constant offset from the original edge using ALT S, scale along face normals. In Face select mode select the outer ring of faces for each window. Shft D to Duplicate the faces. With the new faces still selected use Alt S to scale them along their faces normals. In the operator panel enable the Offset Even With the faces still selected Alt N to open the Normals menu and choose the Flip option. This is necessary before using the Bridge tool. To fill in the space between the original window and new faces, in Edge select mode select each pair of edge rings then open the Edge menu with Ctrl E and choose the Bridge Edge Loops tool. If necessary select the back edge loop and delete it.

Hmmmmm ? Just checking ........ rotations for Plant are also at 0° ? The the next thing to do is share your Blender file. If you don't feel happy about that then make a simplified model of your plant object, for example, replace the leaves with 3 or 4 triangles. Keep the same dome object in the file then you could upload the new file to this Blender file sharing site : https://pasteall.org/blend/ Of course do a test upload to SL of this new triangle leaf model before sharing

From the screenshot of both objects in Blender it looks like the dome has the correct height but y (and x ?) dimension(s) are too small. This shouldn't usually be a problem because the uploader will stretch the lower LoD models to fit the bounding box dimensions of the high LoD model except that in the preview window of the mesh uploader the Low LoD model looks rotated and squished ? In the SL mesh uploader Upload Options tab, with the scale at 1.00, make a note of the x y z dimensions of your plant object then do an new upload but this time the Low LoD dome object as High LoD model and check to confirm that both the plant object and Low LoD objects dimensions and orientation are same. If the orientations are different in the preview window then check again that for both objects in Blender > Object mode > N (Property) panel > Transform > Rotation values x, y, and z are all set to 0° . (and, les important that scale values x, y and z are all at 1.000 ).

Chinrey has already explained that all you need to do is use the high LoD model in the High, Medium and Low LoD slots then create a Lowest LoD model by making a copy of the High LoD and removing as many faces as you can being carefull to keep the remaining mapped to the texture. And now arton has suggested other methods, 1: Using a bottom less cube for the Lowest LoD model, creating a new much smaller texture and map the cubes faces to it. or 2: Rework the original texture to leave space for the faces of the bottom less cube. This would also mean a little editing of the original High LoD's UV's. Example : Texture modified (in Gimp) to add ( cut and paste) small imposter images that will be mapped to the faces of the cube : Cube for Lowest LoD model, UV unwrapped and the faces mapped to the little imposter images that were added to the modified image texture : Example Blend file with packed image texture : https://pasteall.org/blend/3235adc051d6450dacc39d722fe52314 In the SL mesh uploader : Load up the High LoD model as usual. For the Medium and Low LoD slots use the Use LoD Above option. Load up the Cube in the Lowest LoD slot. For Physics, in Step 1: you can use the Lowest LoD model option, and Analyze in Step 2. Note that you can save image textures in the .blend file by checking the File > External Data > Automatically pack Into Blend option before saving.