Although I had been working with Substance Designer for some time, I needed a little more convincing to buy Substance Painter. I already had 3DCoat. In addition to the fantastic sculpting, retopo and UV options, I already had a fully-fledged toolset for PBR texturing. Or so I thought. Of course, that changed abruptly after I tried Substance Painter for the first time and got to know the great procedural workflow. But one wish remained unfulfilled for many years. In 3DCoat I was always able to use textures along a path and work very precisely. I always missed that in Substance Painter. Well, that has changed.

New updates to the Substance3D products have been released, bringing with them long-awaited features. In addition to a complete Path Tool Set, some very helpful presets have been added to Substance Painter. 3DCoat still remains one of my favourite tools for all the other possibilities. Here I will go through various application examples. Let’s start with Substance Painter.

At the beginning

So I’ll start as usual by creating a new project and adding a 3D model. As soon as we add a painting layer in Substance Painter, we also have access to the various painting tools at the top left of the user interface. There is now a new icon at the third position from the top which looks like the tip of an old fountain pen. We can now use it to paint paths on our model by placing them point by point on the surfaces and thus defining the path length. To end a path, press the ESC key or the Enter key.
Initially, all points are automatically smooth Bezier points without handles. However, they can be converted into corner points by double-clicking on them. Alternatively, you could also select several points and convert them using the corresponding button in the top toolbar. In a subsequent update, the user was also given the option of adjusting the handles on the splines.

As a result, we now have the path that uses the brush that was selected when it was created. As long as we are in Paths mode and the path is active, I can click on one of the brush presets in the library and then use this brush with this path. The Paths tool doesn’t work with all brushes, but there are many existing brushes that can be used with the new interactive spline paths. What I like most here is that I can reopen the project tomorrow and adjust each point of these paths again. So we have a non-destructive workflow. That’s how I like my procedural projects. I had rarely used the Paint Layers before because they are, to put it crudely, destructive. Of course, I don’t mean that they destroy my work. I rather mean that they have not been procedural so far.

What I used to paint was then fixed in my material and if I wanted to make changes, I had to delete these paint layers and then create them again. Today, I can also select the spline path option at any time and then make adjustments later by changing the path or the brush used with it afterwards. I can add more points by clicking on a path between the existing points or delete points by selecting them and deleting them with the Delete key.
Of course, new brush presets have also been added that work specifically with the new paths. So we now have seams with different selectable thread patterns, pure tension pattern folds caused by hidden seams or other influences, zips in various designs, welded seams and a pipe rack tool which creates something similar to a crumpled tube.
When the Draw Paths tool is active, I see the word Paths in the top left of the 3D view window. This is a menu in which I can select my desired path for later adjustments. The paths can also be copied, duplicated or deleted in this menu. I like to use this to duplicate existing, fully defined paths, move them as a whole and then just reposition the points.

Practice!

So, enough theory. I have created a few models especially for this article and will upload one of them to a new project. After the usual rituals and baking the high poly information onto my low poly model, I assign a metal material to the main body and a material from the leather box to the pouch. As this is a robot-like model, welded seams or rivets could fit very well here. Both can now be applied much more easily and precisely thanks to the new path tools.
To create the weld seams that connect the legs to the body, I simply roughly place the points for the path where I estimate good positions. I also prefer to place fewer points to start with rather than too many. The first path is created and if the weld seam preset Metal Welding has not already been selected, I assign it now. This is followed by the fine adjustments until the sizes, dimensions and distances look realistic. Each of these presets has a variety of setting options. For the Metal Welding material, I can specify the welding method in a list. Here we can choose between Gas Standerd, Arc Standard and Gas Weave. I can also use the Shape Irregularities slider to access the quality of the weld seam and the strength of the relief and discolouration at the edges of the seam.

Designer

Although we have a fantastic and very helpful tool in Substance Painter, my secret love is still Substance Designer. There used to be external add-ons from creative people that you could buy and work with some kind of paths. Since this update for the whole Substance 3D family, paths and splines have been added to Substance Designer. This now offers us so many possibilities that I would actually have to write several articles about it.
Two new categories have been added to the libraries in Substance Designer. One contains the spline nodes and the other contains the path nodes. The spline nodes can be divided into four sub-categories: Create, Merge, Modify and Render. In the Create section we have three, actually four, options to add a spline to the graph.
Spline (Poly Quadradic) can almost be compared to a Nurbs Spline. You can define the number of points between three and ten and these then influence the shape of the spline like magnets. If I wanted to try to create a sharp corner using this method, I would actually have to position two, or even better, three points very close to each other.
From this point of view, this node is helpful if you want to create flowing shapes. However, I can assign specific properties for each point. In addition to the height and thickness, I can also set the smoothness. This means that a value of 0 for smoothness would also produce a very hard angle.

Spline (Cubic) consists of a start point and an end point, both of which have a handle to form the spline in between. This makes it a very practical tool for simple paths. Later I will go into how each individual point of a spline can have individual multiple properties. This is very practical and also helpful, but can lead to a confusing number of spline points and at some point can be difficult to control and quickly become confusing. So it’s like always in 3D: as much as necessary, as little as possible. For this reason alone, the spline (cubic) is a good alternative for simpler splines.

Circles

Spline Circle is already self-explanatory. A circle is created. However, this is a shape that we can use in many variations. One possibility would be to combine several splines into a more complex shape. But I’ll come back to that later.
When I started this article, I was still thinking about going in a Celtic direction in terms of design. So you could quickly and easily create a stylised sun with the spline circles. To do this, I need two centred spline circles. One small and one large circle.
With the Node Spline Bridge (2 Splines), I can create any number of connections between two splines. As a result, I only get the splines of the connections and the two circles are no longer included.
Many of the tools that we know from the previous graphical nodes are also available as spline variants. We can use the spline warp node to distort these splines using black and white noise.

Thickness Node

As we have not yet given these splines any information about their width, we can do this retrospectively with a Spline Samples Thickness Node. As this example is a circular arrangement, I use a paraboloid shape to achieve a wider shape in the lighter areas than in the darker areas. To generate a graphical shape from these thin splines we use the spline mapper Grayscale. For now, we only use the upper settings in this node.
We can use the Segments Amount to make the resulting shape appear more angular or softer. For the effect of a Celtic symbol, it can be a little less angular and so seven segments are enough for us here. With Start and End we could define from where to where the shape starts and ends at the splines.
We utilise the full range from 0.0 to 1.0. For Thickness Mode, we leave the setting at From Spline because we have already defined which areas should be thicker with the previous node. We then set the desired width using the slider below. As we are not adding a colour map here, we do not need any UV information and the shape does not matter at the moment. The default setting of Plane therefore fits very well.
The inner area of the sun now looks a bit tattered. Which makes sense, because the spline warp has also done its work there. However, we still have the small circle from the beginning of this construct. From this, we go into a new spline fill node and, because the circle is still too small for me, I add a classic edge detect and use the edge width to create a larger ring. If I set the Invert to True in the Edge Detect, I now have the option of using a Blend Node to remove this white ring with Subtract from this tattered inner area.

Celtic

Since my brain was still stuck in the Celtic design world at the time, let me briefly show the benefits of Spline Poly Quadratic with the creation of a classic Celtic band. This often involves several lines that sometimes run on the left and sometimes on the right, crossing over and under each other.
In this case, four lines are involved. So I start with the first spline, which runs from the bottom left via the centre to the top right. In the Spline Poly Quadratic settings, I can define at the top how many points I want to use and further down the exact position of each individual point. This allows me to set each individual point to its mathematically logical position.
Once I have drawn all four splines in this way, I can merge them with several spline append nodes. This is followed by the spline mapper Grayscale used earlier. This time, however, I add a pattern to the colour input. As a pattern that runs from left to right works best here, the pattern was created using a transformed uniform colour map and an edge detect.

To ensure that everything now looks like the Celtic template, I have to return to the Spline Poly Quadratic Nodes. So I double-click on the spline mapper node to see its result in the 2D preview and single-click on the first spline poly quadratic node to adjust its settings. Above the section for the positions I have the Rollout Points Properties, in which I can set the height of each individual point. This makes it easy to define the over and under course of the spline.

Rivets

Even though paths now allow us to create any shape, the previous workflows, i.e. shapes using a combination of basic shapes and effects such as warp, swirl and many more methods, can still offer an easier way. This is where the Mask to Paths Node helps us. We can use it to convert shapes and black and white patterns into paths.
The nice thing here is that we can not only use shapes with hard outlines, but also shapes with gradients, i.e. with soft edges. This node also offers us a slider for the exact positioning of the path on the gradients. This allows us to determine whether the path is placed closer to the white area of the shape or closer to the black area.
This is a practical application if you are too lazy to draw all the paths in Substance Painter for every single seam or section of rivets. With a few simple tricks, this can be quickly automated in Substance Designer.

As this Distance Node penetrates from the white areas into the black areas, I add an Invert Grayscale Node beforehand. Now I can set the distance so that I can define an even gradient to the inside of each UV island. In the past, I would have used a Histogram Scan Node or a Threshhold Node to create a new sharp-edged black and white mask.
However, as the new Mask to Paths Node also works with gradients, I can save myself this extra effort and use it to my advantage. In the Mask to Paths Node, which now follows the Distance Node, I can now use the slider to set exactly where I want to place the path within the gradient to the edge of the UV Islands. Thanks to the procedural setup of the Substance Designer workflows, I can also adjust this at any time afterwards.
As is so often the case, not all nodes always match all other nodes. For example, we cannot distribute other objects along a path; this is only possible with splines. A Paths to Spline converter node is therefore attached to the Mask to Paths node and we can use the Scatter on Spline to distribute any patterns we have to this spline.

As usual in Substance Designer, there is also the subdivision in the nodes as a colour version or as a black and white version. The black and white version is sufficient for this specific workflow. We only want the distribution of the pattern as a greyscale pattern for the height relief information.
There are now three outputs from the Paths to Spline Node, Spline Coordinates, Spline Data and Splines Amount. We could now make each individual connection manually or we can temporarily use the shortcut “2” to link all outputs and splines at once. With the shortcut “1” we would then be back to the individual link per input and output. As the names of the outputs already indicate, they provide different information that we can utilise in other workflows. The scatter on spline node, on the other hand, has considerably more inputs. So we could also insert a background here, to which the information created here is then added. You could, for example, also use the fabric height information here and then use it to add this rivet structure. Similar to some tile generators, we also have inputs here to mask the pattern to be created, to scale it and to assign height information to it.
Then there is the obligatory pattern input, which we can use to insert one or more self-created patterns for distribution. But we don’t need this here. As with all tile generators, we have our usual basic shapes that we can use. And what could be better for rivets than the paraboloid template. The settings of the Scatter on Spline Node offer a wide range of options, of which I will only go into those that are important at this point in the workflow.

Spacing

The Shape Spacing slider is definitely important for us here. We use it to control the spacing of the distribution of the paraboloid shape on the spline. In addition to the usual settings in tile generators for position, rotation and scaling as well as the option of a random factor for all of these transformations, only the basic scaling is important for us in this case. This allows us to set the size of the rivets to a credible dimension.
It would be nice if the scatter on spline node also output a mask. But this is not so important here, as we can quickly extract it from the output using a histogram scan node or a threshold node. This allows us to blend these rivets over the other height texture and mix the colour assignments as well as define other areas, such as the possible folds in the fabric caused by the rivets.

The octopus

At some point, the idea of an octopus took root in my brain, which I would use as the basic concept for this article. I create the head of the octopus in the classic way with the shape nodes Parabolid and Capsule and scale and position several copies of the basic shape of the skull and the round eye shapes. The eye opening, on the other hand, is a rather curved shape and would need several deformers to define it precisely. Alternatively, I could also create a shape in an SVG node. However, this is not quite as easy to control. Partly because I can’t define Beziers in SVG. So that would be another task that is much easier to accomplish with splines. Here I could use the splines (Poly Quadratic) or also splines (Cubic). However, it is important to me here that the respective end points and the start points lie exactly on top of each other.

There are various methods for connecting several splines together. The Spline Append Node creates a common spline object, but still allows each individual spline to be edited completely. I can use Spline Connect to connect the respective beginnings and ends of the individual splines.
If one of the splines is now edited, the start of the other spline remains attached to it and moves with it. If I want to create a long path consisting of several splines, I access Spline Merge List and should also pay attention to the direction in which the respective splines were drawn. For my purposes here, a spline append node followed by a spline fill node is enough to create a closed shape. This is then subtracted from the eyeball as usual with a Blend Node and Subtract.

Cubic Quadratic

For the first version of the octopus, I basically used the same method as for the sun created earlier. The only difference was that I started with a small cubic quadratic spline, duplicated it and distributed the points of the duplicate around the edge of the image to create a larger outer shape. This was followed by the Spline Bridge Node, a Spline Thickness Node, the Spline Warp Node and finally the Spline Mapper.
However, this central motif on the backrest did not fit in with the Celtic pattern I had been holding on to until then. And so I saw the arms of the octopus and the idea of the repeating pattern with the folded arms was born. Of course, I need more control for this than the spline bridge node offers me.
So I create a separate spline for each arm for the octopods that are to make up the pattern and simply connect each spline node to the next. This is another method of connecting several splines together. To make my life easier, I set the appropriate thickness per point for the first spline I create so that the top point is the thickest and the following points are thinner. Now I just have to duplicate this first spline several times and adjust the respective positions of the points.

For the exact positions and mutual retention, I attach a spline mapper node to the end of this node strand and use a normal transform 2D node and a height blend node to create the appropriate segment for tiling. It is important here that I do not use the colour output from the spline mapper node, but the height output. Now I can adjust the position and height for each individual point in each individual spline so that the arms can cross over and sometimes run over and sometimes under. It sounds complicated at first, but this setup makes it quite quick and controlled.
I decided to keep the original octopus created by Spline Bridge as the central element. I just want to enhance it a little. As already mentioned above, I can also add a texture as colour input in the spline mapper. For the first spline mapper, I create a horizontal gradient Linear 2 and scale it so that the light area is in the centre of the arms and the dark area is towards the edge. In this way, I can also assign the rounding of a cylindrical shape to the arms. If I set the mode in the spline mapper to ‘Draw Single Spline’, I can select one of the input splines based on its ID. This makes my graph a little more complex, but I can assign each tentacle its own individual UV twist.

The nice thing about Substance Designer is that many developers put a lot of thought into how events could be linked together. In this particular case, when the bridge splines were created, each of them was assigned an individual spline ID. And this is exactly what we can access here.
So I duplicate the spline mapper node together with the colour input and adapt the latter by adding suction cups at the bottom with the help of a tile generator. To make optimum use of this new colour input, I add a curve node to the twist curve input in the spline mapper. This allows us to adjust the curve and the uvs rotate around the arms accordingly. Of course, the shape in the spline mapper must be set to Cylinder so that we can twist the Uvs accordingly.
Now I duplicate the spline mapper and the curve node seven times and can thus create different variations of the respective UV mappings per tentacle. These eight tentacles can now be merged with several blend nodes. Here I select the spline ID output as the mask. As this carries further transparency information with it, I first have to send it through a levels node to obtain a clear white mask. Now we have a spline mapper with the UV-adjusted colour input, which looks interesting but has lost its three-dimensionality. However, since we created fake height information for the splines in the original spline mapper node earlier, we can merge this as compositing. In the end, I create a new blend node and place this fake height information with multiply on the UV colour information. The upper layer with the fake height information can also be influenced by the layer transparency.
Just to complete the workflow, I would like to mention that I run the result of the central octopus through a threshold and enlarge it with an edge detect and a blend to create a mask that blends this pattern into the general background pattern. But these are workflows that you are probably already familiar with in Substance Designer.

Conclusion

Here it is. Our Art Deco armchair with a customised pattern, which I can adjust and change procedurally at any time. I guess you can see between the lines why I’m so excited about the new paths and splines in Substance Painter and Substance Designer. These workflows that we went through here are also just one of the many tips of the iceberg of what can be done with these new tools in the future.

Autodesk has announced the release of 3ds Max 2024, which offers several new features. A technology preview now enables the use of OCIO colour management in Max. A new Boolean modifier with OpenVDB meshing method enables alternative calculation methods that are independent of the initial geometry. The array modifier has also been updated to provide a new phyllotaxis distribution technique and a progressive transformation mode. A new controller allows the addition of multiple transform controllers as layers of overlaid animation data. In addition, the slate material editor has been reworked and the material editor now includes a new compound node. The list of modifiers has been improved and now includes a search function. The default mirror axis of the symmetry modifier has been changed to X to better suit artists’ workflows. There are also improvements to STL import speed, automatic smoothing and retriangulation for Edit Poly and Editable Poly. 3ds Max 2024 comes with Substance 2.4.10 and Arnold 5.6.0.1.

Here are the full release notes.

I’m more of a “user” than a “technically gifted” person. So I’ve put together and built all my last workstations together with a good mate who, unlike me, really knows his stuff. Today it’s about testing whether this can also be a customised workstation, assembled in an Nvidia Studio certified company? So people who are familiar with such things put together the right components in joint consultations and ordered the package from Schenker XMG.

Nvidia 4000?

At the time of testing, the Nvidia 4000 cards were not yet available – but the hard work grinds slowly, which is why we are looking at the 3090 version here. However, our colleague Finkbeiner will be taking a look at Houdini, AI cores and the generation leap at the next available opportunity when test devices are available – which is why we only touched on these features very briefly in our test.

What’s coming from XMG?

The large parcel was delivered by DHL and I was glad that they delivered on the first attempt this time. In the large box, surrounded by several air cushions, were the actual box with the workstation, as well as the boxes for the Nvidia card and motherboard.

In contrast to the packaging for the graphics card, the box for the motherboard was very heavy. One look inside later and I found a bag with a lot of cables and backplates that had to be removed to give the graphics card access to the outside world. Next to it was the manual and a USB stick containing all the important drivers.

The workstation itself was in another box and additionally secured with foam inserts. Once you have finally freed it completely, you can see a sticky note – a request to remove the additional transport protection inside the workstation before starting it up for the first time. Someone has thought of this, because as a DAU user, you would immediately look for the nearest power connection to quickly see the new device in action.

My previous computers usually had screws on the back that had to be removed in order to slide the side panel backwards. With this Fractal Design Torrent Compact Black Solid case, however, I found no such screws at the back. A simple, careful attempt to simply slide the side panel backwards did not produce the desired result either. In a video I found, the explainer simply pulled the wall away to the side and fortunately also showed a place where the side panel could also be secured with a screw.

On the “back” of the motherboard, the rear panel was secured with a screw, but not on the front – this was also the side from which the transport lock could be removed from the computer.

Under the bonnet

This gave me the opportunity to have a look around the inside of the workstation. None of my previous computers had as much free and empty space inside as this workstation. The cable management was also exemplary. Nothing hangs around anywhere. Everything is connected where it makes sense and attached to the housing.

And where you have already opened it, you can also see the screw that secures the side panel. It’s not only with 3ds Max that looking helps! At first I even assumed that the large hard drive wasn’t in there at all, but this later turned out to be a mistake on my part. With such an interior, the airflow for cooling has a clear path to do its work.

However, it was also lucky that I had had a look inside, because two fastening clips for the fans were no longer in their correct position and would therefore not really fulfil their task safely. This probably happened when the transport securing device was pushed into the computer – or DHL is rallying again.

External values

The case itself has a clean look with the power button at the top centre front surrounded by two USB 3.0 and one USB-C ports. There are also two sockets for a headset and microphone and a small button on the other side for a forced restart if the machine gets too bogged down in a task.

Inner workings

At the rear, the Asus Prime Z690-P D4 motherboard offers two USB 2.0 ports (Type-A), one USB 3.2 Gen 2 port (Type-A), one USB 3.2 Gen 2×2 Type-C port, two USB 3.2 Gen 1 ports (Type-A), one Realtek 2.5Gb Ethernet port, one optical S/PDIF output, five audio jacks, as well as one HDMI port and one DisplayPort, neither of which we need because of the RTX 3090 at the centre of the workstation.

Inside, in addition to the aforementioned mainboard and graphics card, an Intel Core i9-12900 processor, which is kept at the right temperature by a Noctua NH-U12A CPU cooler, and two 32 GB DDR4-3200 Corsair Vengeance LPXs provide the thinking power. The power supply is regulated by an 850 Watt Seasonic Focus Plus power supply unit. The supplied system was rounded off with a 1 TB M.2 Samsung 980 PRO and a 2 TB M.2 Samsung 980 PRO as well as an 8 TB Seagate Ironwolf Pro HDD with 7,200 rpm, which was hidden in the upper area under the top cover. All in all, this sounds like a really good combination on paper

It’s alive!

But now: connect the power supply and keyboard, mouse and monitor and you can finally press the button to start the computer for the first time. You can hear a slight breathing noise, but otherwise the workstation is pleasantly quiet. There are few things worse than sitting next to a hissing and groaning monster. Let’s see when I start to plague the box a little.

All the basic components I’ve plugged in so far, such as the mouse, keyboard and monitor, were recognised straight away and I want to have a look around the new system. However, Windows 11 first wants to get to know its environment and users and has presented me with a checklist that I should work through first. As this was my first direct encounter with Windows 11, I had to be careful – clicking “Next” without reading the texts would have quickly landed me with an unwanted subscription to Microsoft Office and a Game Pass.

Tests

I’d love to jump straight into my favourite programs now, but I should run a few tests first. After all, this article is about finding out whether I will continue to assemble my own computers in the future or whether I can rely on assembled systems from experienced companies.

So I put together the usual programmes. These are Cinebench from Maxon and V-Ray Benchmark from Chaosgroup. In just fifteen minutes, I’m already smarter. While the Cinebench test was still in progress, I had already shot up to second place. Somehow this is a source of pride, even though it wasn’t even me who assembled the system. At the end of the test, the system remained in second place in Cinebench R23 with a score of 27217.

The V-Ray benchmark delivers three different tests. The V-Ray Render test scored 18039 Vsamples, the V-Ray GPU CUDA test resulted in 2054 Vpaths and the V-Ray GPU RTX test finished with 2681 Vrays.

To test the speeds of the installed drives, I used the Aja system test. The 1TB M.2 Samsung 980 PRO NVMe and the 2TB M.2 Samsung 980 PRO NVMe had a read speed of 5429MB/sec and 5384MB/sec on a 1G file and the 8TB Seagate Ironwolf Pro HDD had a read speed of 255MB/sec. The write speeds were, in the same order, 4350 MB/sec, 4191 MB/sec and 234 MB/sec. The fact that the large hard drive is significantly slower was of course to be expected from the design. After all, it is only supposed to store the data.

Warranty and advantages

I wouldn’t give myself a three-year warranty, but I get one here. If a problem were to occur with this system within the next 36 months,Schenker XMG wouldpick up the computer, repair it and return it. It also comes with a fully installed operating system, which I still have to personalise, of course. All the drivers for all the built-in components are already set up, but they may still need to be updated to the latest version. In my case, I then installed the latest Nvidia Studio driver before I started the real tests with my favourite programs.

Studio tools?

While the latest version of 3ds Max and Adobe Substance Tools is loading, you can also try out what the supposedly clever “Nvidia Studio Tools” are all about. “Broadcast” sounds practical, and in a test call with the dishwasher running and children shouting outside the house, the person opposite won’t notice a thing. And the resource monitor jumps up by 2% in the system – you can almost carry on rendering in the meantime! The “Background Replacement” works without a hitch, eats up another per cent of system resources, but a white wall is no great challenge.

Compared to the old machine, the “additional load” on the system is ridiculous – there it was just under 20% more, and a few stoppers were also added. So if you often give webinars and courses, you can get a plus in professionalism for free, or, depending on your disposition, save yourself the hassle of cleaning up. Omniverse-Suite and Canvas tests will follow soon!

Everyday life in Max

Finally, I can turn my attention to why I bought such a machine in the first place. One of my main programmes is 3ds Max and I had just created a scene on my work computer with a procedural object with well over 1.5 million polygons. On my work computer, consisting of an AMD Ryzen Threadripper 1920X 12 processor, 64 GB RAM and an RTX 2080, I can actually work quite well and smoothly with this scene. However, as only 10 to 100 FPS are achieved in some cases, there are stutters here and there. Of course, my work computer is older than the test system in all components, but I would still like to see if there is a significant difference.

The scene loaded quickly and was also easy to edit. The FPS hardly ever dropped below 100. I was also able to leave the “Show Endresult” button switched on for this more complex modifier stack and easily make adjustments further down in the stack, which I can then also see interactively in real time.

Everyday life in Substance

A working project with a size of 2.3 G has developed in Substance Painter during the training that has just been completed. It still behaves well on the work computer. But I would like to see how this project behaves on the workstation that is currently available for testing here.

Even loading this rather heavy project went smoothly. Zack and the project is open. I was also impressed by the layout of the preview images for the materials and brushes. On my work computer, I sometimes have to wait many seconds for all these preview images to appear one by one. On the test system, this happens almost in real time as I scroll down the palette. Editing the project also ran really smoothly. Substance Painter often has to think for a short time before finally displaying the result of the procedurally created masks and textures in the 3D view window. Even on this workstation, the thinking bar was occasionally briefly visible, but to talk about seconds here would be an exaggeration. It is really fun to work with this programme on this system. It’s also nice to switch on the higher values in the Substance Painter Display Settings so that everything looks really good without having to worry about the computer going down on its knees. On my workstation, I usually leave the resolution of a project at 2048, as I can specify higher resolutions when exporting the textures anyway. On this workstation, I was able to work just as fast as 2048 even at the highest resolution.

Everyday life in 3d Coat

I didn’t have a current heavy scene in 3D Coat, but that didn’t stop me from putting together something with many millions of polygons. So I connected various objects in voxel mode, created technical objects with the sketch tool, opened a few existing scenes and carried out various operations. So far, everything has run smoothly and I haven’t yet managed to make the computer sweat or stutter. The fan has also not become louder in any of the tests so far. So it is pleasant to work with.

And what does it cost?

In terms of price, this solution was similar to our self-assembled system from two to three years ago and I didn’t have to assemble anything. However, as hardware and prices change all the time, I put the components installed here directly into a shopping basket with one of the component suppliers and came out at around 4055 euros. However, I was then informed by my mate, who knows his way around, that even now the prices are fluctuating greatly and that there has just been a significant drop in prices. Apparently I missed something with crypto? Was that important?

Of course it would be cheaper to buy the individual components, but then it’s my knowledge and my time to put everything together myself. I would have to install everything from the operating system to the drivers myself. In my particular case, that would even be someone else’s work and time, namely the person who knows the ropes. Apart from that, I’m sitting here in front of a system that was put together by certified professionals and simply works straight away in all areas.

Conclusion

Even as a non-technician, but a pure mouse pusher or actually rather a tablet scratcher, here is a solidly built box at a competitive price – and even if I don’t know the difference between DDR4, i9 K/S/xyz, the “Nvidia Studio sticker” helps. Without gaming lighting, but with plenty of power reserves and a system optimised for “content creation” (as all artists and 3D/video people are apparently currently referred to), you can start working straight away. Of course, you can certainly also play games on the machine, but I much prefer to create content than play with it.

All the points that we criticise are down to companies that are only peripherally involved with the machine – DHL and Microsoft respectively.

When it comes to UV unwinding, not every user is enthusiastic. Many even dislike it. So I belong to the rare group that really likes the meditative work and order of good unwinding. Making the right cuts and creating a beautiful, barely distorted unfolding is only part of the job. Once you have defined all the segments, they still have to be distributed as efficiently as possible in the UV frame. This then resembles a kind of slow, deliberate Tetris.

In the courses I give, I am often asked why it is so important to have a good layout for UV processing. In principle, the square that we see in the UV editor represents a bitmap on which the textures for the respective areas of the model are located. This can be a 4K map, but for games it can also be a 512 x 512 texture or even smaller. This means that every pixel on this layout that does not cover part of the geometry is a wasted pixel that we could potentially have used to provide a better and sharper texture for our object. That’s why it’s important to find a layout that leaves as little open space as possible. Whilst this sounds like a great change from crossword puzzles or Sudoku, it can also be quite time consuming.

UVPackmaster still leaves the first part of the work to us, i.e. setting the cuts and preparing the respective UV islands, but helps us with the second part of the task to be performed. The programme is specially designed to distribute the UVIslands as well as possible within the UV limits.

The add-on itself is available for 3ds Max and for Blender. In 3ds Max, it is used as a modifier that is placed over the UVW unwrap. It can be applied to one object and to several objects at the same time.

The handling of this modifier is not complicated anyway. However, if you simply want to arrange your UV islands neatly, it could hardly be easier. One click on the large Pack UVs button will satisfy the majority of users. The programme then packs the various contiguous segments into the UV frame quite efficiently, and usually much better than 3ds Max’s own UV packer would do.

Each time the “Pack UVs” button is activated again, the previous calculation is discarded and the software takes the initial UV layout as the basis for the next calculation, which may have slightly different results. If you really want to finalise the UV layout, the modifier stack must be collapsed. Of course, this sounds nonsensical in a procedural workflow. However, you could do this with a copy of the original and place a UV unwrap modifier on it, which can then be copied onto the original.

But first, let’s look at the settings you can make to ensure that the plug-in does the best possible job. In the second rollout, entitled “Packing Devices”, you can define which hardware is to be used. UVPackmaster can use the CPU and GPU for the calculation. Normally, the box is already ticked to use all available resources.

Further quality settings can be found in the “Basic Options” rollout. In addition to defining which map channel should be used, a slider can be used to set how many threads should be made available. Basic Margin, i.e. the desired distance between the UV islands, is also specified here. As this distance is added to the islands before the calculation and all islands are then scaled to fit into UV frames, it is not possible to define a really precise pixel value. A more precise calculation of the distance can be achieved with Pixel Margin in the “Advanced Options” rollout. The distance between the islands is set using the “Pixel Margin” slider, and the distance to the edge of the UV tile is set using the “Pixel Padding” slider. As soon as the Pixel Margin value is greater than 0, Basic Margin is ignored and this more precise method is used. Here too, the calculation accuracy can be increased using the “Adjustment Time” slider if required.

The most important slider is probably the one for precision in the “Basic Options” rollout. The higher this value is set, the more precise the calculation will be. Of course, this also increases the calculation time. However, the default value of 200 is usually sufficient. If you want to work with a very small distance between the islands, you can increase the value. A value of over 1,000 is rarely recommended and also leads to significantly longer calculation times. On the other hand, a value that is too low can also lead to inaccuracies such as overlaps. The pixel margin settings ensure that the islands have enough space between them and room to breathe in relation to the UV edge. In the Pixel Margin options, you can define how much space should be kept between the islands and towards the edge. If the “Rotation Enable” checkbox is ticked, the algorithm can also rotate the islands in order to stow them better. Below this, the angles at which the respective island can be rotated can be set. The default value allows rotation in 90-degree increments. Smaller values may allow better utilisation of the UV area, but also increase the calculation times.

If you want to squeeze out a few more pixels of optimisation, you can select the “Enable Heuristic” checkbox in the “Heuristic” rollout. This causes the add-on not to use the first solution straight away and to find other, perhaps better distributions. To ensure that UVPackmaster does not search itself to death, a slider below it can be used to set the time in seconds for which new distribution options should be searched. If the slider is left at 0, the programme will search for the best solution for eternity until the user presses the Escape key to end the search and wants to see the result. However, the use of Advanced Heuristic is only recommended by the author for a model with few UV islands. The advanced search will use a different algorithm for positioning and rotating the UV islands.

When packing the individual islands, they usually have to be scaled to fit into the desired areas. By default, UVPackmaster scales all UV-Islands proportionally to each other, with a few exceptions. It is therefore also important to ensure the appropriate proportionality between the UV islands in the underlying UVW unwrap modifier. If an area is to be given more texture resolution, these UV islands must be scaled accordingly in the UVW unwrap in comparison to the other UV islands.

Pack to Others

From time to time, a situation arises in which parts are added to a completed model or some areas require a new layout. You can then try to distribute these islands manually in the existing layout or have everything repacked and thus lose the previous layout. This is not always a bad thing. However, if you have already prepared some textures or baked textures, you would like to lose as little work as possible. This is where the “Pack to Others” option in the “Advanced Options” rollout comes into play. In this case, only the selected islands are inserted into the existing layout. Only the workflow for this feels a little bumpy to me. You have to select the desired UV islands in the underlying Unwrap UVW modifier, then add the “UV PackMaster” modifier and execute this function. In addition, these selected islands are scaled to fit into the available free spaces and are therefore given a different texel density than the rest. As practical as this function can be, you should think carefully about when to use it.

As we are already on the subject of scaling and if you have already scaled your UV layout so that it corresponds to a defined texel density, you can tick “Fixed Scale” in the “Advanced Options” rollout and the UV islands will not be scaled when packing. Of course, this restricts a few freedoms for the distribution. However, if the “Tiles” option has also been selected in Packing Mode, UVPackmaster can pack the islands onto the next tile if the “Fixed Scale” option is switched on and there is no suitable space in the respective tile.

The topic of UDIMs has become increasingly important in recent years. Where a multi-sub material was used in the past, several UV tiles are now used for one material in order to achieve a higher texture resolution. In UVPackmaster, in the “Advanced Options” rollout under “Packing Mode”, you can set whether all UV-Islands should be distributed on one tile or on tiles. You also have the options “Group to Tiles” and “Group Together (Experimental)”. For simple distribution to tiles, the user can set directly below how many tiles the UV-Islands should be distributed to and below how many tiles make up a row before the next row of tiles is populated.

The “Group to Tiles” function offers interesting options. You can define the distribution of the islands to the various tiles according to different criteria. The option by material will distribute the islands to different tiles according to their material or sub-material in a multi-sub-material workflow. Possible advantages of this are that one material may also have transparent areas and another may not. If all islands are now wildly distributed across all tiles, it can happen that you have to add an additional opacity map, although only a tiny component on this UV tile would need this map. You can also select the Object, Mesh Part or Similarity category to define the distribution across the various UDIM tiles.

If you want to define this completely yourself, you can also set the “Pack Selected” option in Selection Mode and build a modifier stack in which a few islands are selected in a UV unwrap modifier, followed by a UV packmaster modifier that distributes the selected islands on the UV tile. In the “Packing Box” rollout, you can define the tile on which you want to place these islands. Then comes the next UVW unwrap modifier, in which further islands can be selected, and then another UVPackmaster modifier on top, which places the now selected islands. Yes, it’s a bit convoluted, but it’s a way to get there.

Often several UV islands are placed on top of each other to save space and thus achieve a better texture quality. This can be the mirrored half of the model or just a few parts such as tyres, screws or similar. Of course, this only works as long as you don’t use dirt or other clearly recognisable patterns or want to bake information from a model onto it. To cut a long story short, there is a pull-down in the “Advanced Options” rollout called “Lock Overlapping”, in which you can define whether the overlapping of islands should be ignored and they should simply be freely distributed next to each other or whether overlapping islands should be retained as a unit. A distinction can be made between Exact and Any Part. In the first case, the bounding box must fit exactly; in the second case, a partial overlap is sufficient to keep the islands together. The nice thing about these two options is that you can also clearly specify that you want to retain overlaps but separate accidental overlaps.

Whilst most textures are square, it is not always practical to have such a square texture. A snake, for example, would have to be divided into several segments to fulfil the requirements of a square texture. This would result in an unnecessarily large number of cuts in the UV unwinding, which in turn could cause problems with the continuity of the textures. Therefore, the proportions of the tiles can be defined in UVPackmaster. The desired dimensions of the texture to be used can be clearly specified in the “Texture Dimensions” rollout. On the one hand, there is the more time-consuming manual method of defining the corresponding proportions in the UV unwrap modifier below and unchecking “Lock Dimensions” and “Adjust UV Map to Texture” in the UVPackmaster modifier above and entering the corresponding pixel values. Alternatively, you can shorten this process by unchecking the “Lock Dimensions” box, defining the pixel values for the height and width and simply ticking the “Adjust UV Map to Texture” box.

The lowest rollout “Statistics” then shows how many iterations were calculated and how much time was spent on them. More informative for the user, however, is probably the indication of the efficiently utilised area in the UV layout created. This information is also displayed in a small extra window after the packing process.

You can also display the layout of the UV unwrap modifier below in the topmost rollout and thus use this tool to analyse how much of the area has been used effectively. It is a pity, however, that the analysis in the “Statistics” rollout only shows the average of the total area used instead of per tile. But that is already complaining on a higher level.

Group

Another nice thing about UVPackmaster is the licence model. Even though more and more companies are joining this ugly rental model, this is not the case when purchasing UVPackmaster. You pay this very low price once and have a lifetime licence that also includes all future updates. The other big players in the industry should take this as an example. We customers are happy to pay the price for a good tool if our work is not held hostage.

Conclusion

All in all, I am very happy with this tool. As a meticulous person in some digital areas, I often spend a lot of time trying to spread my UVs as well as possible. In my tests, I noticed that I often achieved a similar efficiency of space utilisation as the UVPackmaster did. You might think that I wouldn’t actually need this tool. But what I invest a lot of time in, this tool does in just a few seconds. So if you would like to have a little more privacy or simply want to use your time for other tasks, you could consider buying this tool