Jump to content


Popular Content

Showing most liked content since 06/16/2018 in Posts

  1. 5 points
    While working on a plugin I got the idea of making this small utility: Wheel Of Tools. It allows the user to set up a list of most used tools, which are presented as a circular palette to select from. Available directly within the viewport. Current state of the plugin is only a concept, and I am mostly focusing on the actual tool selection. The part where user does set up the tools still needs to be worked out. For those familiar with PolyGnome, you probably already understand where I will be heading to ... Wheel Of Tools, once fully implemented, will be the base for a new way of interacting for the main part of the PolyGnome plugin.
  2. 3 points
    If you are a modelling purist it is never acceptable, and technically all ngons in final models are modelling mistakes*. Of course clients couldn't care less on the whole, so if it's all about the time, and there are no artefacts in render it is difficult to object, though I still would :) What few people seem to realise is that if you train yourself with a solid grasp of poly modelling and practice for years, then you are often able to model stuff properly in the same time as it takes everyone else to bodge it together any old how. And then the argument for tolerating them rather falls apart... Models without ngons are always better than those that have them because the files are more flexible because they will subdivide predictably in every program the mesh could be loaded into, and you leave that to random chance if you allow the sub-d algorithm to solve them for you - they all do it differently. It's true, booleans don't always break the mesh, and sometimes they definitely are the quickest way of doing stuff. But if you can clear up after them, why wouldn't you ? If you did, you'd be able to use a booled mesh with subdivision and in a whole load of cases where the ngon version would fall apart. But if you don't need SDS or any of the other things that would affect it then there's the argument for leaving it just as it is... Of course ngons while you work are another thing altogether, and can be very helpful while you plan your topology - you just won't catch me leaving them there. I sell my services as a modeller - you can imagine how unprofessional of me it would be to be handing out meshes that contained mistakes, that's all I'll say ;) I am aware that people like me and @VECTOR are the exception rather than the rule, but that doesn't make us less correct ;) CBR *Ok, perhaps twice in a lifetime you might come across a modeller SO good and so uber-aware of how his software translates ngons into quads and tris that he is able to plan and use ngons to achieve specific surface qualities that would be impossible, or significantly detrimental to achieve any other way. In THAT case, ngons are fine :)
  3. 3 points
    Hey Cafe Having enjoyed modelling the last synth so much, I thought I'd do a bigger one for my own amusement. The Sh-09 was an analogue micro Monosynth released in 1980. By 10 years later the digital age had arrived, screaming synth leads got displaced by twinkly dreamy electric pianos, and everything went FM and later PWM synthesis. In 1990 Roland were answering the call for a return to the searing analogue leads and basslines, so released this, the JD-800, a veritable juggernaut of a synthstation, utilising hybrid synthesis so we had the best of both worlds - all the analogue goodness, but combined with digital clarity, and DSP. 24 note polyphony and 16 part multitimbrality made this a powerful pads machine with unprecedented levels of practical slider control over virtually every parameter. To put you in the mood, here's the top patches from that... Rendering-wise, there's been some challenges to this one - getting the anisotropy right took a while, and I had to make all the panel graphics, which took rather longer :) But whereas my little monosynth had been sitting in a studio gathering dust, or getting bashed about on the road, this one is new out of the box and on display in a synth museum... Physical, Adaptive, Automatic, 5%, 3, 3.5, 3, AO. No GI. 21 mins (wide shot), 32 mins (close shot). SDS throughout, 100% quads CBR
  4. 3 points
    No....not nerds. Given that that they are both 3D artists and musicians is pretty interesting. Both require a mathematics (music is very mathematical), a sense for order, etc. Those are left brain characteristics. To create what they do obviously comes from the right side of the brain. They are not nerds but whole brain thinkers. A real nerd reaction is the one I had: I looked at the mesh, saw the logic of it, and then honestly thought it reminded me of an empty Excel spreadsheet! Now that is nerdy. But I will agree it is another beautiful model. Dave
  5. 2 points
    It's here, but no, didn't get asked as far as I remember. CBR
  6. 2 points
    yeah, my guess would be that there is a slight morph involved to alter the shapes a little as they twist. i wouldn't know any solution though on how to do it without masking stuff in post. as for the slight bulging effect on the seams that could also be done in post. the texturing part puzzles me a bit, looks pretty accurate. those things are very hard to figure out without actually trying it.
  7. 2 points
    There you go. Come on Bob, you know this :) Problem here was that you did inner extrudes to get the border, rather than the box corners you need to keep things sharp under SDS. Here's what it needs to look like for this result... note the center edges I added too, close at the pointy and, and fanned out to match the curve at the other... but you could also done that point without adding these edge loops by just sliding the points either side of it in toward the point. Here's the scene... AT-PT_mesh cbr.c4d CBR
  8. 2 points
    Small quick tip, but very amazing experience... Simply must love it :)
  9. 2 points
    If you couldn't use instance object for whatever reason (for example circles are different size) : Create Master Circle and by XPresso pass only those parameters that you want. Use LinkList node and drop your objects there. Iteration node to go through all elements and set proper amount of iterations Adding HUD elements as mentioned above.
  10. 1 point
    Want to do something i don't usually do, so working on this little piece for myself. eventually the traffic will be animated, still working on the composition. will probably also try a couple of different color themes. Feel free to critique to hearts content :))
  11. 1 point
  12. 1 point
    Cinebench works and gives benchmark scores irrespective of which version of Cinema you want to use. It's just a benchmarking app. See MAXON's website for details : https://www.MAXON.net/en/products/cinebench/ So, it's adequate to reference those scores.
  13. 1 point
    Since you are using TeamRender, it's probably your best bet to contact Maxxon about this issue.
  14. 1 point
    Contradiction in terms ;) CBR
  15. 1 point
    Did anybody ask this on Ask GSG Wednesday? I missed it and can't seem to find a recording of it.
  16. 1 point
    Ah, yes. The viewport can be quite deceiving when it comes to the Color and Multi Shaders. Glad you got it figured out!
  17. 1 point
    Indeed not your screenshot :) The person we'll be ejecting certainly isn't you, my friend :) CBR
  18. 1 point
    I just looked at the screenshot which said 16.11. Nobody spends £4K on software and doesn't update it to the latest version of that, unless they can't ! Yes, that's why stealing Cinema is so out of order - there are perfectly fine free alternatives available ! CBR
  19. 1 point
    Hey, that's a great little tip there! I didn't know that one about the clipping =D Nice little project! I don't want to know what the inside of the 5 would look like when trains and cars all come together ^^ Fun!
  20. 1 point
    Indeed. Like the ones you have on the outside corners... Having a 45 degree meeting of edges (as you get when inner extruding) directs the SDS flow around the corner, whereas box corners with crossing edges tell SDS that corner needs to be sharp. CBR
  21. 1 point
    God bless you, Cerbera. By box corner do you mean without the angled edges an inner extrude gives?
  22. 1 point
    Start with a suitable, editable block. Put some cuts in your top, Extrude Inner, then Bevel. Very simple.
  23. 1 point
  24. 1 point
    I have both a 2013 and an old 2009 (firmware upgraded to a 2010). I render on an Nvidia 1080ti with redshift and it works great. The only reason I have kept the 2009 is for Redshift. However, I just recently saw this: https://barefeats.com/egpu_nmp_titan_vega.html It turns out that MacVidCards, who flash Nvidia cards to work without some strange Mac related issues, came out with a flashed 1080ti that works on the 2013 in an eGPU configuration: http://www.macvidcards.com/store/p110/World's_First_EFI_eGPU_card_for_Mac_Pro_6%2C1.html I know as of 10.13.4 Apple removed TB2 support for eGPUs. But apparently, this company was able to flash the 1080ti to get around this some how. It's not cheap because the card is $1,200 and you need an eGPU enclosure for about $300. But is may be a solution if you want to render with CUDA based renderers. Mark
  25. 1 point
    I think the culprit is probably the metal material. Try turning specular down to 0 in the reflectance channel there... CBR
  26. 1 point
    You will have to make the loft editable to do this. So select it (or a copy of it if you want to keep the original), hit C to make it editable. Then just select the outside polys and drag your material directly to that selection before changing the mapping type to cylindrical in the texture tag and using Texture Mode and Axis Modes to place the graphics exactly where you need them. CBR
  27. 1 point
    It's Ask GSG tonight. Be interested to hear what Chris would say if you asked him there... CBR
  28. 1 point
    If you look better on both "better" results, only difference between them are vertex normals alignment. While "highest" quality has aligned vertex normals correctly, "high" version doesn´t... https://www.dropbox.com/s/kfcqc3az29v09f7/vertex_normals.png?dl=0
  29. 1 point
    @everfresh, if it's not Pose Morph (which I believe you, but I'm not sure I see the drifting), do you have any ideas for how it was actually done? I agree that using Pose Morph for this would be very difficult, but I'm stumped as to how else you'd do it. Maybe it was a pretty rough pose morph that got it close enough and then they did something in compositing to bring them together?
  30. 1 point
    i highly doubt both models have the same poly count and that the shape actually morphs into each other for real. it's very well done, but there are some minor inaccuracies visible... for instance look at the second one around 4-5 sec, where the transition is almost finished. you can clearly see that the shapes drift a bit apart from each other. clear indicator it wasn't done with a pose morph.
  31. 1 point
    Awsome, Thank you very much!
  32. 1 point
    Thank you very much for your reply.I solved the problems already.I tried different combination of checkbox like what you suggested. I also found that there were actually two groups of trees overlapped on each other.
  33. 1 point
    No worries :) Your model is fine, but there is only one surface, not the 3 you actually need ! Corona will be seeing that as hollow (but solid within that) glass. So for this to look like it's filled with cola you need 2 more surfaces, both of which you can make with extrudes from the original, but you will have to take a reasonable amount of care when you proceed because any coplanar faces will break the render and you are working with some seriously tight-ass spaces there ! So, let's assume what you have now is the outer glass... Next you need to make the inner glass, so extrude all polys in an arbitrary amount (with caps). now we need to make the cola, so select the innermost polys of the inner glass you just made, and split them off to a new object. Then, with the polys still selected, get the Normal Move tool, and move them OUTWARDS so that they intersect the inner glass by a small amount. They MUST NOT be coincident though, so make sure they overlap fully contiguously across the model. Once you have that, the chances are your render will look pretty much right out of the box because the geo is giving you the right physics. CBR
  34. 1 point
  35. 1 point
    There are a few things contributing to that I think... 1. That noise type does that a bit anyway when used with displacement, although interestingly not when it is used with bump, parallax bump, or normal maps. 2. You are exacerbating the problem by clamping it, which increases jaggies at the edges 3. SP Displacement wants quads ideally, not triangles, although I'll grant you it makes very little difference in this case. 4. Blur doesn't work on noise so we can't solve it that way. So, whilst it's possible to minimize it, the truth is you can't minimize it much while the camera is that close to it. i do note it looks fine if you just zoom out a bit and lose the clamping ! But I can't think of anything else you could do to smooth that out without somehow blurring the noise. CBR
  36. 1 point
    re priority - make sure there is an actual priority problem before you start fixing it. Things that appear to lag are often OK when rendered (or baked). I usually adjust priority by moving stuff around in the OM (top down order), or priority values lowest = 1st. There's also a shift priority tag. I'm sure you know all this. I haven't seen much info on baking. For purely dynamic stuff I use doc settings, dynamics, cache, bake. For my more complex scenes (eg using randomness), I drag everything into the dope sheet and go functions, bake objects. Then I disable dynamics, xpresso, python, and anything that moves things. I use this method for team render, so all the nodes have identical data.
  37. 1 point
    After double clicking on the Redshift material click on the Edit Shader Graph at the top. This will take you to the Redshift graph which has a ton of nodes with different ways to do this. Import whatever animation or image sequence is to be used for the alpha transition into the RS Texture node. After that this video should give some ideas. https://youtu.be/X1DCumk_Q7s
  38. 1 point
    You will have fewer problems if you switch to Octane shaders.
  39. 1 point
    I'm confused what's going on here. Team Render Server is designed to run on a separate machine (i.e. not one that is running C4D or the render client), and it solely operates to serve render jobs to a whole bunch of render nodes. If you've only got two machines, there's no point in running Team Render Server because you'd only be rendering on one. You should just be running regular C4D on your workstation, and Team Render Client on the second computer.
  40. 1 point
    Press Shift+V and in View tab check Axis Scale parameter (default is 100) (or is hidden, solution in previous post)
  41. 1 point
  42. 1 point
    It is Tachyon renderer, but currently with new name, look very promising...
  43. 1 point
    As Fastbee mentioned, MAXON is slowly making the viewport like that.
  44. 1 point
    With the reaction MAXON got with everyone loving the new DOF and other realtime things in R19 I'd be surprised if they didn't continue to make the realtime better in R20.
  45. 1 point
    The best and simplest way would be SDS poly modelling, starting with a cube to which you add enough vertical segments that you can move them into the curve. But also possible with splines in a sweep. Ideally you need a left view so you can see what the curve is, but totally possible to wing it by eye... This is incredibly simple stuff - suggest you watch any of the basic modelling tutorials... CBR
  46. 1 point
    Kiwi is right. You need to establish your curvature in topology. 8 segments is nothing like enough for this. I'd start with 64 myself, like this... When the topology is this dense there is no need for any transverse control loops to tighten the protruding section because we have given SDS 8 times less work to do and established our curvature immutably in base geometry. Also you should avoid any triangles in any mesh that is used with Subdivision surfaces. Note my quad patch in the middle, which solves the problem of triangles and complex poles at the centre... CBR
  47. 1 point
    @djiadjee welcome to the Café. Not sure how experienced you are with C4D or 3D in general, but modeling cars is no simple task. In fact, I think it's one of the most difficult things to model in my opinion. Don't forget that many techniques that work in other apps, ie Max and Maya, will also work in C4D. There are MANY more videos for those programs. LOTS of practice and I wish you luck. Post your progress.
  48. 1 point
    Lols :) Ditto over here. I used to spend every Saturday in the local music shop when that first came out. If you are a synth man you might enjoy what I'm doing next - the new and properly space-age successor to this one, the JD-XA ! CBR
  49. 1 point
    Think C4D default render is CPU based, so external graphics will not do much. Unless you go GPU based rendering, R19 ProRender or Octane, Redshift, Cycles or something.
  50. 1 point
    Thank you, used that method, and little bit of modeling and sculpting to get this deformation effect.