hvanderwegen

Did you turn off Weld Points in the symmetry object? Turn it on if you did.

It's a classic extrude "problem". Extruding towards the inside geometry will create non-manifold geometry, and it needs to be cleaned up or worked around (as shown by Bezo). Sketchup solved this with its "manifold" extrude, and lately software like Max and Blender have begun to adopt this type of extrude tool, which makes an operation like this a doddle to execute. With a bit of luck the C4d developers will implement the same option in an update. It's not perfect, and may create n-gons, but nonetheless a very useful extrude tool. Works like this:

They did booleans the hard way in Blender - much easier and faster when BoolTool is activated: Select one object, select the next, CTRL numpad minus. Done. And it hides the cutter object's geometry and replaces it with a wired box. Anyway, interesting video. It really shows how incredibly spoiled we are with the current crop of 3d apps compared to Tron times...

PS PPI (not DPI: only used for print) has no real meaning. It doesn't say anything about the actual resolution. Only pixels count.

One thing Max has going for it is the sheer brutal performance of its viewport. Great for archviz in this regard. My own experience with Max: It's the 3d software I learned after leaving the Amiga platform behind. From Sculpt/Animate 3d, to Imagine, then C4D (Amiga), Lightwave (Amiga), switch to PC and Max, followed by C4d again, Lightwave, a bit of Max, a very short stint in Softimage, some Modo, then Blender. At work it's a mix of Blender and C4D (with some Modo). I actually quite liked Max at the time. The current release still has that somewhat clunky feel to it (tried it last year). It is no longer my cup of tea, but a good workhorse. The GUI is indeed no comparison to the likes of C4d. Nor to Blender, in my opinion.

One more thing to consider: how many watts are the power bricks in your older machines? A 500 watt should suffice for the 1060 or even a rtx 2060. A 1080ti requires a 600 watts power supply. If the costs of a GPU render license are an issue, perhaps investigate Cycles, LuxCoreRender, and ProRender. All are free, but would require you to send your scenes to Blender and set up materials there for GPU rendering. Free unlimited render nodes as well. Octane Prime is also free for 1 GPU, but also Blender tied. Anyway, can't argue with four free GPU options - but it will mean having to install Blender. A fifth GPU renderer becomes available in that case: Eevee. Incredibly fast for animations. But I understand if Blender may not be acceptable in your workflow. Cycles 4D is a perpetual license and nicely integrated in R20. And quite affordable at $290.

I Ah, my mistake: I thought your intention is to equip all of three machines with a 1660? Because in this case the money is better spent on a new machine for the same money. Or do you intend to equip only 1 of your older workstations?

That motherboard is pretty old. Purchasing 3 1660 supers would cost you around $1000 American. For $949 you can get a pretty fast modern i5 rig equipped with a RTX 2060 card that would give you really nice rendering speeds compared to those combined 3 older machines, because RTX /Optix GPU renderers will render your scenes twice as fast as GTX ones. A machine like this would also have 16GB, and with a GPU renderer that supports combined CPU/GPU rendering it will gave you even faster render times. And no need to bother with setting up multiple nodes/licenses. An optimized GPU renderer such as eCycles speeds up rendering once more by 100% (only in Blender). But Cycles4D is a pretty nice integrated GPU renderer for your R20 as well. Quite affordable compared to Octane. Anyway, in my opinion it is just not worth the money, effort, and time to equip your three Q6600 machines with 1660GTX cards. And a new computer would have many more advantages as well. If you are working on a more current workstation, then the choice is simple; forego your plan to invest in GPUs to create render nodes out of your old Q6600s, and invest in a good RTX2070/80 card instead. Especially seeing that the current 2080 prices are plummeting due to the 3080 launch. Or better, if your machine supports it, invest the same $1000 in a RTX3080 card (you will have to be patient though, because they are sold out everywhere). That will dramatically speed up your rendering. PS GPU rendering will blow your mind. Physical Render is rather slow in the first place, and modern GPU render engines rip through most renders at unbelievable speeds, in particular when some GPU-enabled denoising is used on top of that. It is great for animations. Anyway, just my two cents. I was able to keep doing work on my ancient 13 year old machine thanks to a GTX1080, which renders pretty fast. Currently looking (waiting) to get a 3080 - once they become available.

Ah, IT Crowd...

I learned a neat trick two weeks ago: use virtual walkthrough while recording the camera movement in realtime, add key samples to keep the pauses in movement, then smooth the curves a couple of times to create a nice looking smooth movement. Very quick and neat trick.

Yeah, I agree. It might not be quite obvious from my response, but I prefer polygonal modeling myself (since I tend to do much more hard surface models and more stylized cartoony stuff). For high-end realistic organic characters sculpting is the way to go - which is what I tried to say. Anyway, I find the two methods are more often than not complementary, and I may switch between the two while modeling. As you say, good knowledge of poly flow is essential during the retopo as well. I love to switch to sculpt mode after modeling some hardware and put in dents, or deforming/wrecking parts. Much more free flowing than pulling verts...

No worries! Half the work is thinking it through before starting to model something. I suggest checking out the pushing point books as well. In my opinion some of the best material out there about classic "old-school" subdivision and polygonal modeling technique. I say "old-school" because nowadays almost everyone switched to Zbrush or sculpting to create characters, while only relying on polygonal modeling to start a base mesh to work from (if at all). *edit* Someone posted this video demonstrating their process to model a dragon using the traditional "old-school" polygonal modeling approach. Different software (Lightwave) but still quite informative and fun to watch. In truth almost no-one would do it like this anymore, and instead rely on a sculpting workflow followed by a retopo. But it is still a viable technique for less realistic stylized characters like this star character.

@CerberaYou are of course correct. In my rush I omitted the step to dissolve the edges to create quads only. I would also add one loopcut in the face to reduce the poly stress caused by the mouth. Looks better now. Flow should be more natural and follow the face lines, but I did not want to spend more time on it. I am not convinced that a pure all quad process is required in all cases. Tris can be used, especially when they are part of sections that remain hidden or part of a flat/static area that is not affected by (animated) deformation.

I have seen alternative cooling solutions from vendors here and there. According to Linus the ram and cpu temps are pretty much the same as running a 2080ti card with the official card.

A worthy upgrade from my GTX1080 - if I can get my hands on one. I will await the 3090 results.

In the end it doesn't really matter. Some things are faster in one app, other things in the second app. What is important though is to understand poly flow in subd modeling. William Vaughan's Pushing Points books are a great resource for beginners in this regard. These books are application agnostic. http://pushingpoints.com/v2/the-pushing-points-topology-workbook/ http://pushingpoints.com/v2/the-pushing-points-topology-workbook-volume-2/

It's funny, because I asked about a grid fill function four years ago. Saves so much time in many instances.

The steps are more or less the same in any app. Add a sphere. Delete the top and bottom two rows. Rotate by 9 degrees in H axis. Here is the hard part in C4D: there is no grid fill, so it must be done manually. with bridge and cut. First delete the bottom half, and add a symmetry object in the XZ axis. Use the bridge tool to fill in the rows. Followed by to cuts. (Line cut tool) Now scale in the Z axis. Set the Axis to Root before you do (this scales the half sphere from the mid axis) Select these faces and scale win the x and y axis (green corner widget) Add a subdivision surface. Now apply the symmetry object before continuing (otherwise you will have mouths on both sides 🙂 The rest is the same. Temporarily turn off the subd object. Select these faces: First perform a bevel to keep the sharp mouth edges before an extrude. Followed up by a simple extrude. Before the extrude turn on the subd surface object again. Now select both loops (the very thin loop at the edge of the mouth (zoom in) and the inner mouth loop. This is where having the grid fill command would have saved some more time, because those grid lines were curved, and the mouth takes less adjustment to get right. Anyway. Use the scale tool in the x and y axes and adjust both until you have the overall mouth scale that you want. Then select the top or bottom mouth polys and scale in the x axis to create the smile. Keep adjusting until you feel it looks right. Then add the spheres for the tongue and eyes. Apply materials (I left the inside selection of the mouth out of this example). Et voila. Pay attention when closing the holes: I made a small mistake in the lower left area, which results in bad geometry and two mini holes in the mesh. I did this very quickly and did not pay attention.

The trick is to create a 20 sided sphere with 6 rows: the star character has 5 sides X2, but you actually need twice as much to maintain the rounded sihape after adding a subdivision object.. 1hen delete the top and bottom points and rotate by 9 degrees (360/20/2). Perform a grid fill to fill the top and bottom areas. Followed by flattening the sphere and selecting these poly strips: Followed by scaling the selection. Scale the individual selections to fine tune the roundness. Add a subdivision surface at this point to guestimate the overall shape. This shape may be molded further in shape. The mouth is a simple inset and extrude action. Then scale, drag, etc.faces, edges and points to form the shape. Finally, the tongue and eyes seem to be separate objects. The eyes squashes spheres, the tongue a sphere molded to form a tongue shape. Not quite right yet, and it needs more finesse to get it closer to the original's shapes, but it should get you going. You could also use sculpting now to mold everything in better shape and smooth things out.

It really depends. Currently I myself am looking to upgrade an almost 13 year old machine (i7 920), and the only reason why I did not have to do this sooner, is because I rely on a relatively new GTX1080 to render in GPU render engines like Cycles, ProRender, Octane, and LuxCore. As a C4D user your native GPU renderer is ProRender, but C4D's version is not that mature. The internal classic render engine is purely CPU-based, and aging. If you have access to Redshift, the choice would be simple: go for GPU rendering. If you decide to stick with the old classic renderer, a fast CPU is your only choice. A good GPU may however find good use in overall daily use: it can vastly reduce video conversion times (Handbrake + CUDA), and is essential for a fast viewport in C4D. NEVER GO WITH ONBOARD and 3d apps if you can help it. And Nividia's new RTX3070 GPU is only going to cost $499, which is an absolute steal (it is faster than a 2080ti!). Then again, the cost of current 2070 and 2080 cards are nose-diving, so perhaps a second-hand card may give you both a fast CPU AND a fast RTX20XX card. I can only speak for myself: I will upgrade to a VERY fast 3090, and save money by picking a slower Ryzen CPU like a 3950. (Yes, still much more expensive than what you would have in mind I guess, but I hope you understand my reasoning: I could go with a 3970 which was my original plan, but that new RTX3090 card is really a steal compared, and will have applications for me beyond GPU rendering).

I like the new Sketchup-like extrude. All new features in 5 minutes: https://www.blender.org/download/releases/2-90/

I also have World Creator, and a traditional sphere projecttion will cause those issues, as Cerbera explained. It really depends on what you want as a final result. A lot can be faked: for example, clouds can be great help to hide texture pinching or seams. An approach that I like is "Blended Box Mapping". The basic theory is explained in this original article from 2006: http://www.neilblevins.com/cg_education/blended_box_mapping/blended_box_mapping.htm The author used Max, but any 3d app is capable of producing the same result. For quick planet generation and displacement maps as generated from apps like WC this is a very simple and quick method. I created a seamless heightmap in WC, and exported it. For example: I create a sphere in the 3d app. I prefer rounded cubes to avoid the triangular pinching at the poles when displacing. It creates a more even geometry. It is ESSENTIAL to use a rounded cube for this technique. Then I add the displacement map, and set it to box mapping. Now, here is the step that may be VERY simple to achieve, or rather elaborate, depending on your 3d software. I have done the blend maps in the past manually, and while it is not too difficult, it still takes some time. Since I tend to work mostly in Blender nowadays: in Blender's box mapping a nice "Blend" option is available to blend the seams. One slider, and BOOM! No more seams, no need to UV map this thing. While C4D does not offer a similar blend option, it is still very much possible with the fall-off shader: https://www.cineversity.com/vidplaylist/cinema_4d_quick_tips/blended_box_mapping_with_flat_projections_with_the_falloff_shader This will give you a similar (identical) result. Should be possible with Redshift as well, but I don't have access to it, so you may have to experiment. Here's the end result. No displacement pinching at the north pole (or south pole). As always, there are disadvantages to this method. The primary disadvantage is that the heightmap will be mapped on four sides, and not quite usable for real maps of real planets. There are ways around this, but too finicky in my opinion. Spherical mapping with hiding tricks (like the aforementioned clouds) work better in those cases. And singular unique surface features will have to be blended as a second layer in your shading tree/nodes, or placed as a blended mesh. As Cerbera wrote: there is no one "best" technique. I find that this technique works well for generated "fantasy" planets (and complex objects which require a seamless repeating texture without the need to UV map anything).

AMD Radeon Prorender 2.5 is released for Blender. Now works in 2.83. https://www.amd.com/en/technologies/radeon-prorender-blender PS The release of Blender 2.9 is delayed to next week to fix additional bugs and for more polishing.

I agree mostly. The Blender Foundation's seems to listen better to studios nowadays: the LTS (Long Term Support) initiative is great, allowing studios to work on long-term projects while staying in one stable version that is bug fixed for two years. The current version 2.83 already received 5(!) LTS updates since it was released last June. I noticed a marked improved stability since release 1 (and Blender is on the whole very stable running as it is). And version 2.9 is about to be released (next week) with a quite long list of interesting new features and improved edit mode performance, but also nice little GUI improvements across the board. The usability of the outliner, for example, is improved once again. One thing I have noticed over the past few years is the DELUGE of add-ons for Blender. It is intimidating and impressive at the same time. I see ex-Modo users create addons to bring Modo functionality to Blender. I see ex-Max users create addons to add features they had in Max. The community is hard to beat. Anyway. A c4d to Blender plugin would be nice to have. Are you aware of the copy and paste addon/plugin that allows for copy and pasting of geometry between various apps? https://github.com/heimlich1024/OD_CopyPasteExternal I've used it at work for seamless and quick switching between C4D, Modo, and Blender while working on projects (before Covid, that is). At the risk of repeating myself, in my mind licensing is the major hurdle right now for MAXON. Secondary hurdle is the rather slow development of C4D in the past five up to ten years. It is lagging behind and relying too much on third-party developers to fix the gaps, which rubs more salt in the financial wound, so to speak. Hopefully with R23/S23 development will have picked up.

@ShrikeI realize, after re-reading my previous response, that it sounded a bit passive-aggressive. I apologize for that. Your perspective is interesting, even if I might not completely agree. I have a tip for your Blender colleagues: have them install HardOps and DecalMachine. With Hardops installed hard-surface modeling becomes pure joy, and incredibly fast and efficient. And it has a complete non-destructive workflow (including non-destructive parametric primitives!). It will allow them to work as they work now, but non-destructive (or not) and productivity will shoot through the roof. Also includes a nice easy circular array, and a free version of KitOps for quick kitbashing. Decalmachine is also an incredible useful plugin for hard surface modeling. Both plugins revolutionized my workflow, and they are inexpensive. With some practice modeling speed increases 5 up to 10 fold (I kid you not).