Jump to content

Zed Tremont

Resident
  • Posts

    40
  • Joined

  • Last visited

Everything posted by Zed Tremont

  1. Personally I use already quite some years Firestorm only for sl and that works perfectly fine. Apart from now and then some issue as above mentioned which is solved now too.
  2. Thank you very much @SoRe Nirvana, now and then I happen to run into this problem, now and then it indeed was the name and Maya seem to have somewhere a node that make the name in some way longer. You only see it when you import in another program since just renaming doesn't help. Up to now I could get around it when I extract a part of the mesh, combine it again and after the name changed automatically to polySurfaceXX. Then the name can be changed to whatever you wish if it's not too long (don't know the max number of characters) This time that didn't do the trick, checked the model, exported as .obj, reimported, applied new material (Lambert1) only one face for this model required. Still nothing Changing the debug settings did solve the problem. Anybody any idea why? For complete info, model was zeroed out, (delete history, center Pivot, Freeze transformations). Used viewer: Fs
  3. You indeed can script mesh components to get a moving tail like movement. Project Bento is the better solution though the the bones seems to be confirmed by now before it's really in the "final" version it will take still several months. For what I know of it estimations are 5~6 months.
  4. I have the impression you used 1 texture face for the complete coat. Perhaps as a starter you might consider to use 2 since you only need a small part to be alpha. I ran mostly in this kind of problems where 2 alpha faces are pretty close to each other as with glasses, bottles etc. Since you have a none alpha avi in between it might be a solution to get around this. Good luck
  5. Quote" Non of these tools, nor XNormal bakes reflections/refraction, or any camera angle dependent effects though. Since these are dynamic in real time environments, it's fairly uncommon to bake them anyway. "unquote Fully agree that baking these dynamic effects is uncommen and in principle not logical seen we are in a 3D environment. On the other hand it does is possible to bake the reflections/refractions and specular lights in Maya as well as in SP. Since it's against 3D logics I presume that it's absolutely no priority for allegorithmics or Autodesk to have a real good look in this. None of the major engines Unity or Unreal needs this kind of solution.
  6. Heya, sorry for the late reply and thank you for both your comments. Indeed both are PBR driven and unfortunately sl is still a quite old engine with in my personal opinion rather poor as it regards normal and specularity render, not mentioning the use of PBR's. Also many if not most people haven't got their advanced lighting on which makes that they want see the effects of the maps. It has been mentioned that project sansar would allow the use of some form of substances of and on. Let's say it's wait and see what happens there. Substance painter does have a filter (light baking filter) that allows to bake the specular lights in a diffuse texture. The various options to set your lighting are limited but for quite some materials it works OK seen the limitations of baking in general. It does so in a 360 degrees camera angle which makes it quite interesting for use in sl. Gold however as a pretty though to bake material is not satisfying. I looked through tutorials/info about Quixel, in many ways it's quite similar to SP, I didn't find info as it regards baking. Perhaps it's at once baked/visible on the texture once your lighting is setup? Would be interested to have some info before I consider a purchase.
  7. Thank you for your reply Ivan. I will look in Quixel suit don't have that one... yet. Looked at the materials of the program, there are a lot and seem to be of a very high quality. Have substance painter and designer, ready materials... I'm not so impressed and some of the materials for sale are downright not OK. Still it is a great and useful tool. Q if I may: I'm very sure the problem is that the camera takes a correct bake of whatever is visible. The backside, not at all. How is this with Quixel and/or Xnormal. Will do further testing/painting meanwhile
  8. A Li of 2 for clothing seems ok to me, seen unfortunately a lot worse. Trick in mastering reducing your Li is to play with your LODs in this case. To do this efficient it's best to have nice quads if not it becomes even more tedious then it already is. a tank top can be brought in for less then 1 but you will find that increase the time you work on it considerable. On the other hand it will be better to render for your customers. Just simplify gradually your LOD 2 and 1 (medium and low) often it is not needed to get LOD 0 (lowest), the last one is quite have with even a minimum of tris. Keep an eye when the collapse happens and do not touch the edges of a UV shell. If you do the UV will be totally deformed and your texture will appear quite weird as soon as you go to a lower LOD.
  9. Seems there is not really a ready made solution since till now nobody replied. Seems to be in the nature of the beast. I've painted the lines manually out in Ps and managed to get them invisible while keeping the same result. Still interested and searching for a solution but with little hope. Does anybody has experience with Quixel? How are result with this Ps plugin?
  10. Hi Belli, Can't help it but I have the impression you kept your building as one large mesh. If so I would recommend you to break it in pieces for following reason. 1. you're physics will end up too complex and they're quite lag creating stuff. 2. You can use more texels on your mesh and have a better appearance of you textures afterwards 3. floors are typically done in a separate mesh, will make life a lot easier if you need to make holes for staircases etc. 4. Very good chance that the Li of the building will end up a lot cheaper as a whole then one huge mesh item. When you make physics shapes want want to keep an eye on following. 1. Sorry to say so but I don't see any advantage in using cubes as physics shapes. Why would you use 12 triangles if you can do it with 4. 2. Agree that windows are in most cases neglected, unless you really want to jump out of your windows. Use triangles in physics shapes wise. 3. Personally never had problem with planes "touching" each other and yes I did make some large house already. Combine verts, you can use n-gons and triangles for physics shapes for sl to reduce number of triangles. (ngons = a face with more then 4 sides) (for rl modelling you really don't want ngons in your model and no or very few triangles depending for which engine you model, sl is another story) 4. If you make a physics shape, make sure you make it as large as the most extreem (x,y,z) coordinates. The shape will position itself perfectly where it should be. If not you will have it off position when uploaded with all related problems. 5. Analyse... depends in most cases it is not necessary, in case of holes you sometimes need it to be able to pass through the hole sometimes not. Depends the situation and is best tested on aditi. Hope this helps, you can always contact me inworld if you like. Good luck, Zed
  11. Agree with Arton and Chick Apart from the Li you will see 3 values while uploading your mesh just right of Land impact Download; physics and server Please check those too, since 40 is extremely high guess that it is indeed the physics shape. note that when the density of your physics shape is high you can get the effect that you're Li increases while decreasing the size of the mesh.
  12. Hello, OK updated, saw the thread again today and it was a big mess, sorry for that but as mentioned below my internet was really bad. This thread might be similar to the issue Crovanti has, sorry but my internet is too bad today, the pictures simply don't want to show. Object: Earring (gold with gemstones) I'm only after the bake of the gold part, render setting still have to be tweaked a bit to optimise the effect.Gemstones are irrelevant and only added to get their reflections (red of ruby) All normals/edges are softened (smooth) where they should be threated like this.  Used raytracing, reflection, refraction, jitter (to reduce eventual artefacts), UV map clean and show nice even squares while testing with a grid, final gathering Material: (only for baking purposes not to bring in sl of course) mia_material_X lighting: IBL lighting with B/W gradient + 1 directional light + 2 spotlights Problem: 1. picture of low polygon version bake.  as could be foreseen The edges oposite of the camera/viewer angle are quite visible and are distorted. 2. Made a high polygon version of some elements for further testing, bake shadows was switched off this time. pictures taken from 2 different angles   As can be seen the lines on the oposite of the camera still appear, less heavy but not acceptable. Artefacts increase as the polygons increase, did a further test while still increasing the polycount though artefacts become too problematic then. Adding lights didn't help either and increased the artefacts too. This is keeping me already a long time busy, been googling, watch whatever tutorial I could find on youtube and digital tutors. There are not many available for sure compared baking compared to rendering. Not illogical either not many engines need this anymore. I'm quite convinced That it has to do with the baking, camera settings as such. Focus/priority (simple said) is given to the camera direction. Robin Sojourner confirmed this and said that she has the same issue with MODO too. Eventual solution would be to take multiple bakes and make a composition of the "good" parts. (use of masks and painting it nicely to match) Disadvantages: 1. By moving the cameras your baked reflections change accordingly. 2. Quite a tedious job if you have to do that for complex pieces. Would appreciate it very much if somebody found some way around this issue. I'm slowly out of ideas by now. Thank you very much in advance, Zed
  13. About this topic a lot can be written so kindly consider the text below as a first step practicle "how too" guide. To describe all tricks and threads would make this too long and boring for many. I fully agree with above mentioned and did read the post of Drongle quite some time ago. Really greatful to be informed more indepth. This is my personal workflow which has proven to work pretty well, quite efficient though time consuming to do well too. Modelling something takes the least time, whatever it is. To handle all the rest behind that can be quite tedious and absolutely not my favourite occupation. Making a correct mesh include taking decisions in the balance efficiency. 1. Basic decisions: A. Determine which LOD viewer setting you need/want. Many creators will use 4 as standard. If made for setting 1.5-2 as some insist on will have as consequence that your collapse will happen faster so in many cases your Li will be considerable higher. B. Determine if your mesh is typically used "indoor" or "outdoor". If indoor is it really wise to make a superstrong LOD if it want be seen anyway since you will see the house but not the objects inside. Use your resources wisely... as always balance... C. Consider how avatars will collide with the object, in case of architectural use planes whenever you can instead of cubes (4 triangles instead of 12). Consider if you really jump out of the windows? Collision boxes tend to drive your Li up and can be cause of addng quite a bit to your lag. For items set on a floor eleminate the face down, personally I don't think there is need for a collision there, same for the top in many cases. 2. Workflow (personal of course other system might do the trick too) A. Upload and rez your mesh (in aditi) only LOD3 (high), cam out and see when the collapse happens with prefered LOD viewer setting. I take a screenshot just before the collapse happens. B. Determine which detail is still valid from this distance by looking at your screenshot. Prepare your LOD2, eliminate whatever is not needed but while doing this keep an eye on your UV editor. You really don't want to take edges on a UV shell away. UV maps have to be stable and within acceptance of the original. Upload mesh LOD3 (high), add LOD2 (medium) -> rez and determine again when the collapse happens. C. Repeat steps for LOD1 (low) if necessary, this mostly depends on the size of the mesh and/or if it's a exclusive "indoor" used object. D. LOD0 (lowest) in most cases not needed, the least number of triangles for this LOD will add fast to your Li. In some cases and mostly in the case of very small object it might be eventually considered. E. Physics shapes: as already mentioned make them as simple as possible in most case I just use planes (or a cube of which I delete not relevant faces. You can make Physics shapes visible with Firestorm viewer. Might be other viewers too, don't know activate your advanced menu via preference activate you developer menu -> last check under advanced menu developer -> render metadata -> physics shapes (check it) to return to the "normal mode" just do uncheck again You will see a colour code: 3 main colours are most seen and most important: I. blue: your physics shape is considered as efficient or at least acceptable. Verts are not dense for given size II. orange: verts tend to be dense on at least one spot. You may see that when you decrease the size of the mesh the Li will increase due to the clustering III. red: your physics is too complex, the verts too dense. I've seen some very bad examples of this where the power of the CPU of the server was completely consumed. You have a serious risk on lag. Few of those on a sim are in my eyes really bad. As mentioned above this is not a pleasent job to do BUT: 1. you're meshes are as efficient as can be for a given geometry while keeping in mind the detail you want to obtain in the high level. 2. You will barely notice a collapse on the mesh for the given predetermined viewer LOD setting. 3. While keeping an eye on your UV map while preparing the LODs you will have a correct setup of of the applied textures on that distance. As said above the topic is large, influence of several parameters have to be taken in account. Experience will give you a good feel of what you can do and what is to be avoided. With experience you will/might take the coming steps in mind while making your topology as well as your UV maps/shells.
  14. Hi Devriv, A bit late since it's already solved but perhaps a tip that might help in future. You can make a physics shape visible with Firestorm viewer (perhaps in other viewers too) activate your advance menu in preferences -> choose the last option in this menu to activate the develeper menu. Developer -> render metadata -> "check" physics shapes to undo this just uncheck it again. You will see mesh physics shapes in blue, orange or red blue: simple physics shape which is good Orange: too complex, physics shapes can create lag since the calculations of collission boxes are complex. Physics shapes do not like dense clusters of verts. In case of orange you might find out that when you decrease the size of the mesh the Li increase because of this density. Red: downright bad and best to restart, test have showed that these can "consume" the CPU of the server pretty bad.
  15. Thanks for the replies people, really appreciate it. And yes Arton, I indeed changed my naming convention from stuff like BottleCarrierLOD1 to Bottle_Carrier_LOD1 as common practice in computer graphics. Also since substance Painter requires a _low and _high extension. They put you on a downright wrong track with the "materials" error. Did a try and no errors anymore, still have to do the full upload but don't expect problems anymore. Thank you very much.
  16. Zed Tremont

    Face Issue

    Verixx, Noted you also use Photoshop. If you save as a .tga, personal preference of me there are two ways to save in this format. 1. 24 bits 2. 32 bits If you do not wish a alpha channel save as 24, with alpha channel 32. Each channel will use 8 bits Meaning R(ed) 8 bits + G(reen) 8 bits + B(lue) 8 bits = 24 bits R(ed) 8 bits + G(reen) 8 bits + B(lue) 8 bits + alpha 8 bits = 32 bits Simple said if you would save a a 32 bits you will force a alpha channel even if this wouldn't be immediately visible inworld. You can verify this in sl by pressing crtl + alt + t, alpha will show red
  17. Hello Drongle, thanks for your comment. I try to check those lines tomorrow, sorry been a long day today. I will try the reset and see what that does too. Haven't tried that yet. With highest i mend officially LOD3 then. I get the error message when I want to bring in the very first high polygon model iow it can not be related to a difference between names, lambert materials etc of LOD2-1 or 0. Printscreen of upload trial: https://gyazo.com/914b058b9f38d7566647ea0a72d0036b
  18. Just for info and follow up. This model has not over +21K tris/material, the whole mesh hasn't got that many anyway. (as per ref Drongle in one of the threads) Exported as a .obj file to remove all eventual "residues", imported and exported again. Passed the model to a friend and asked to try to get it uploaded, same problem. Almost all threads I've seen are about same number/name for materials in LOD1 and following. Since the problem occurs while trying to get LOD1 uploaded this is not the problem. In hypershade I always prepare the same coloured lamberts which I use over and over in the same project. Other items in these scene/project didn't had any problem at all.
  19. Thanks for your reply Chic. Had to mention before that I did go over the posts before sending mine. Weird thing is that all I've seen at least are about uploading LOD2 and further. I get it when I just want to upload LOD1, so goes wrong from the start without being able to do anything extra. Always bring in LOD1 in first to determine when my collapse happens, then depending this parameter start making manually the other LODs.
  20. Hi, This drives me nuts, made already lots of models and got them imported in sl. Sometimes with my share of weird problems with texture faces, LODs and name it. This though is a new one for me. 1 mesh 5 material (different coloured lamberts as done already tons of times created in Maya 2016 viewer Fs, never had a problem solved by using the LL viewer till this date When I try to upload LOD1 (high) I get following error "Material is not a subset of reference model" Geometry is correct checked it again and again Replaced the materials with just 1 grey lambert... still same problem Tested on other grids and came in perfect... Am I missing something here, bit on the end of ideas so some help/ideas would be very much appreciated.
  21. The principle of painting meshes is for programs like Photoshop, Gim and Paintshop Pro quite the same. Really not hard to get the basics, paint within the lines of the UV shells, if there is a ambient occlusion available apply it as the last layer and blend it in. Ambient occlusions are generated by the 3D program the modeler uses, he/she will add those to the items you got. Following tutorials I made for Phostoshop some time ago. They're done for Photoshop but as said the principles are the same for all this style of programs.
  22. Lovely Jobly Rotaru, I didn't knew that it was related to the selecting of the meshes. Up to now I simply brought in the complete (in the .dae file) combined element on LOD1 with the lowest possible physics and replaced the meshes one by one. For complex items it was the best way to get the positioning right without having to place them all manual. Of course I will keep an eye on this one and give it a try. Thanks, Zed
  23. Yes you where right, it's indeed the above mentioned bug. I tried a few things: 1. Upload a mesh where more elements where combined in a .dae file but not merged in the 3D program -> failure the moment you try to add LOD2, the viewer "crosses out" the 2 LODs 2. simply make one texture face instead of the 3 I had and all went perfect. Nasty little bug it is and made me loose quite a bit of time. Thanks for the help
  24. Alright something that drives me nuts for 2 days now, really appreciate help. Project: window box with curled wrought iron, I do hope I can attach a picture later. Workflow: made LOD1, 2 and 3 manually and the related physics shapes (2 planes for each = 4 triangles) Uploaded the combined meshes with only LOD1 and physics to none for easy positioning inworld Uploaded the seperate elements with LOD1, 2, 3 and the physics shapes. 2 of the 4 elements went perfect The poles on the side are creating issues. The collapse of the "final" poles is a lot faster and not acceptable then the primary combined uploaded mesh even though there are 3 LODs present while in the first one only LOD1. I know these are rather high in polys for gaming but really did want to have the detail in them LOD1: 2376 LOD2: 462 LOD3: 147 exceptionally let the viewer do this one. LOD4: down to minimum 4 size: <0.082, 0.081, 1.594> Texture faces: 3 I am sure and it's verified that the model is correct made, made already smaller banisters where the collapse is totally correct. As a test I uploaded in an opensim environment, all was perfect there. Discussed with people regarding this, Tammy Moyet said that this is a sl problem and that only solution is to increase the triangles in the lower LODs. Uploaded again with LOD1 and 2 as same LOD3 on 462 which gives of course a crasy Li... result the same Will try further by uploading the elements combined with the correct LODs. If anybody has some ideas regarding this will be most glad to hear something, it's driving me quite nuts. White pole at the right side has only LOD1, left side LOD1, 2 and 3 left pole collapse much earlier then the other
×
×
  • Create New...