Some ideas for Clothroom version 2018



  • I'd love for the clothroom to add the ability to rotate the camera around during a simulation. I've seen videos of that plugin that that guy over at rendo made for dynamic simulations and I really like that he could rotate the image while it was progressing through the sim.

    Maybe SM should just pay that guy so I can make all my hair dynamic as well. :p

    I'd also like to see a real time simulation mode where you can gently move a character at your last frame and the clothing will follow and fall where it should since it's constantly running the dynamic simulation. It would probably be extremely CPU intensive but if the computer could handle it, it might be a less frustrating way to work with dynamics.



  • @johndoe641
    I assume 'that guy' is Gérald Contesse, the maker of VirtualWorldDynamics.
    Sure here is no reason why the camera, if it is not animated of course, would not be open to be moved while a simulation is going on. Not sure why Poser locks a camera that is not animated. Bulllet Physics live simulation does allow the camera to be moved.

    @bopperthijs
    Sure a sort of 'Marvelous Player' addon would be a perfect solution. I think the mechanism used by DSON could be used here also.
    Allow me to paint a scenario. I presented this few times before on different locations, among which the mourned RuntimeDNA forums. Since these are down the drain now this may be a good occasion to bang the drum again.

    Here goes:

    • user selects figure to dress and loads garment from library.
      -> System in the background calls the corresponding MD garment file and pops up a 'room' representing an MD 3D screen with the selected figure as avatar in 'A' pose and loads the (.zpac) file containing the MD definition of the garment.
    • User uses the 3D tools to fit the garment around the avatar. Some form or automation of this will be fine but initially w can do without. If the pose is standardized and the garment is not too far off in size fit in practice is a matter of seconds. Note here I mean REAL refit, like picking real clothing from a rack and trying it on. If you try a tailoerd fit shirt on a potbellied figure gues what the fabric between the buttons opens. Simulation runs until converged or interupted by user.
    • User poses figure and garment follows pose. Beware that the limitations as in the cloth room apply here? Pose changes must be such that body parts do not pass through one-another for example. Also pose changes can be arried out only as fast as the simulation on your system allows. That should be a lot faster than the Poser Cloth Room but still slower than when doing only animations. Here is where mesh refinement could come in (not an active feature in MD yet): using a low-res version of the garment for the posing phase, offering speed, and hi-res for the final phase.
    • At some convenient point user moves back to the Pose Room to do his normal thing. The simulation runs in he background, like BulletPhisics Live Sim. It starts automatically when pose is changed.

    A reason to use Marvelous Designer is that it is to my knowledge the only format available now that clothing can be produced from a medium cost level tool. Sure it is not cheap but not a price level that canot be recovered from sale of products worth that name.
    The price of the garment items can be good because the versatility of the product is high. They can be realisticaly fitted to any morph of figure.

    The player window as desribed can be a win-win situation for content makers who can develop new markets; for Smith Micro who has new shine for Poser, and for Marvelous Designer because their product has found a new application.

    The MD format is binary and proprietary, not published. Poser constitution calls for open formats, and that is why the addon format would be most suitable.

    <steps down from soapbox and makes himself scarce while hell breaks loose, LOL!>



  • @F_Verbaas Just out of curiosity have you asked Marvelous Designer about a creating a plugin for Poser or DS? I'd like to know what they think, Truth is, especially from their licensing set up, they strike me as rather territorial and might see Poser as a threat or at least an encroachment on their territory.



  • @eclark1849
    Oh yes I did.
    I think it is unlikely they see Poser as a threat. They on purpose rely on applications like Poser for all animation. Requests to provide dials for avatar posing were turned down for that reason.
    They are complementary on purpose.



  • @F_Verbaas said in Some ideas for Clothroom version 2018:

    • effect of bias. The stiffness range of the fabric in the cloth room in practice is somewhat limited to cover the full range of fabrics and the effect of bias cut is not taken into account. The orentation of the grain of a fabric relative to the dominant load direction has a large influence on the drape of the fabric.
    • differentiation in properties in weft and warp direction. Cloth Room uses one figure only.

    Ironically the stretch and fold values are sent in with a U and V variant, however Poser seems to set these to the same values. I did (in PP2012) try setting them with python but it appeared not to work (might revisit, will let you know if I find anything)

    I would add for my part, expose the whole of the cloth sim via python, currently to set values (fold shear etc) you have to run a script for each object / dynamic group in your sim. I can select which sim I want but not which object / group to affect, really annoying)

    Would definitely like real elasticity, yes I know it does it to a certain extent but not really successfully.



  • I think simplifing the whole cloth room would be a good idea. Although dynamic clothes give the best results but most of the people prefer conforming clothes, because it's a lot easier to set up.
    I think spliting the clothroom in two sections like a player and a creation room would benefit the use of dynamic clothes. User could use the player section and content creators the other one.
    i encourage a library of preset clothsettings, even I, who have done thousands of simulations, am struggling with the settings. Sometimes I just alter a setting to get a better result, without even having a special fabric in mind.

    Speaking of special features, there are some that comes to my mind:

    • elastic, like said before, but also (really) stretching fabric like lycra, jersey and rubber.
    • buttoning and unbuttoning during the simulation, also unzipping and zipping. ;-)
    • immediate draping, it has to be much faster. From what I have seen with MD, it should be possible.
    • much more refinement, cloths are just too coarse at the moment, smaller folds and details.

    Best regards,

    Bopper.



  • @bopperthijs said in Some ideas for Clothroom version 2018:
    ..

    i encourage a library of preset clothsettings, even I, who have done thousands of simulations, am struggling with the settings. Sometimes I just alter a setting to get a better result, without even having a special fabric in mind.

    And, if they could implement that in a friendly way, vendors could sell Cloth Presets. They could include really customized presets in their dynamic clothing, sell "packs" of presets, play with settings and sell some really weird presets (anti-gravity-rubber-hurricane-suit... or something).

    Some sort of plug-and-play functionality will help to popularize the Cloth Room. "Power users" won't need it, necessarily, but even they would like to play with something like that. But, new users need to be happy with their purchase in every way that it's intended to be used and since the first-glance style of Poser is certain "button mash your way to art" with preloaded library content, it's a no-brainer to include that luxury and functionality in other parts of the program.

    Regardless, though, of any increase in ease-of-use of Poser, it's still a very, very, advanced program, capable of all sorts of nifty things that other programs just can't do as well.The New User either needs an easier way to enter into learning about more of Poser's features or they have to have a heck of a lot of willpower to go from "novice" to "Super User." No reason SM can't stack the deck in favor of the latter while STILL making significant technological improvements to keep their long-time Super-Users happy.



  • @morkonan said in Some ideas for Clothroom version 2018:

    @bopperthijs said in Some ideas for Clothroom version 2018:
    ..
    And, if they could implement that in a friendly way, vendors could sell Cloth Presets. They could include really customized presets in their dynamic clothing, sell "packs" of presets, play with settings and sell some really weird presets (anti-gravity-rubber-hurricane-suit... or something).

    Part of the problem is that presets only work for a specific mesh, if the mesh is more dense then what simed like satin might look more like rubber, to have presets they would have to be tied to a specific mesh / product, it's why I moved away from PhilC's presets long ago.

    Amanda



  • @amethystpendant said in Some ideas for Clothroom version 2018:

    Part of the problem is that presets only work for a specific mesh, if the mesh is more dense then what simed like satin might look more like rubber, to have presets they would have to be tied to a specific mesh / product, it's why I moved away from PhilC's presets long ago.

    Amanda

    It might prevent people from putting together generic sets for cloth presets to distribute, but if a specific group of settings works great for the cape I want to distribute, it would help allow me to distribute them together so the end user can get the cape to drape properly. Right now it seems like everything loads with the default setting.



  • @amethystpendant said in Some ideas for Clothroom version 2018:

    Part of the problem is that presets only work for a specific mesh, if the mesh is more dense then what simed like satin might look more like rubber, to have presets they would have to be tied to a specific mesh / product, it's why I moved away from PhilC's presets long ago.

    Amanda

    I'm not sure I follow you - A "preset" would only be a list of values for the available dynamics settings. It's not geometry specific.

    I do understand that not all topology is the same, though, and that can contribute to one preset's suitability for a material/mesh over another. So, if a preset chosen was "cotton" but the mesh wasn't suitable for that, being low density, then it wouldn't really "look like" cotton. BUT, I suppose it would be as "cottony" as that mesh could get. :)

    Hmm... We do have a good sub-d process available, though. I haven't checked, but will the Cloth Room work on a sub-d mesh, properly? (ie: Instead of using the base mesh for calcs, use the sub-d verts instead?) If so, wouldn't that help to solve the problem?



  • @morkonan Yes SubD does appear to be honoured in the sim, although as most of my meshes are quite dense I find that subd'ing the result is better than running the sim on the subd'd mesh



  • Presets representing a certain type of fabric can work only if the 'null state' of the mesh is known. Properties of a fabric are non-linear so it matters where you take the origin.
    also about 'null' states:

    • If the mesh has an angle between two facets, is this then a bending of the fabric, that is, should the sim try to bend it back, or is this a fold 'pressed' into the fabric or maybe is it a topstitch?
    • Likewise, if in the mesh the angles between the edges connecting to one vertex do not add up to 360 degrees (as they shoul for flat fabric) must the sim then try to correct this and add a radial stress component, or is it part of a preform 'steamed' in the fabric?

    The good thing in MD clothing definitons is that at least information about the unstrained state and the information about fold lines or flat press of ends of darts or other shaping is in the garment definition file. Non-linear properties are not (yet) there but I am pretty sure they are working on that.



  • Some people may wonder why these 'non-linear properties' I mention are so important, and why it is so difficult to the 'the' properties for cotton or nylon.

    To take the easy case of strain, take a strip of fabric and pull, as an example let me quote from a book Structure and Mechanics of Textile Fibre Assemblies, edited by P. Schwartz,
    0_1486757123859_Knipsel.JPG
    0_1486757471436_Knipsel.JPG
    0_1486757531101_Knipsel.JPG

    For real fabrics the nature of the curve can look different, but it is
    hardly ever straight.
    0_1486758242942_Knipsel.JPG
    (from: STRESS - STRAIN CURVE ANALYSIS OF WOVEN FABRICS MADE FROM COMBED YARNS TYPE WOOL, by
    VÎLCU, Adrian et al.)

    Simulation engines approximate the behaviour as if it is a pure spring. The same curve looks then as follows:
    0_1486757990954_Knipsel.JPG

    The behaviour in one simulation step can be approximated linearily, so one can define a stiffness (-steepness of the line) for each step. Spring constants A, B and C can give some form of approximation for situations around their intersection point with the curve, with situation for stiffness A and C of course being way better than for stiffness B, but even B for its situation way better than a single stifffness for the whole range.
    0_1486759654723_Knipsel2.JPG

    Knitted fabrics of course are by naure interlocked springs and show a more friendly behaviour. Stiffness can with good accuracy be expressed by a simle formula.
    0_1486760106540_Knipsel.JPG
    (from: Dynamic Elastic Behavior of Cotton and Cotton / Spandex Knitted Fabrics by Senthilkumar Mani, PhD, Anbumani N.)

    There is more fun in mechanics of fabrics and in their simulation, but I hope this gave you a taste.
    For simulation of the behaviour of light we have gone from very simple shader models via raytracing and make-do's like ambient occlusion to present Physics Based Rendering. For fabric simulation we may have to go a similar way.
    See the good side of it. There willl be room for the artist to make a difference.



  • @F_Verbaas said in Some ideas for Clothroom version 2018:

    Presets representing a certain type of fabric can work only if the 'null state' of the mesh is known. Properties of a fabric are non-linear so it matters where you take the origin.

    This is why in some dynamic clothing the mesh is not "pre-draped" in any particular shape, but is present in a shape suitable for a draping algorithm. For instance, a skirt mesh, intended to be dynamic, will be just a big circle, centered on the hip section. In that case, everything is left up to the algorithm acting on the mesh.

    For surface effects, the Cloth Room can designate areas to be "soft" or "hard" decorations, following the simulation or not, but, I agree, it takes a great deal of preparation to accommodate more realistic surface features and to prevent those from detracting from the realism of the sim. That's especially true of clothing that has been pre-sculpted for "folds" or even for "fitting" to a figure, which is very difficult for the Cloth Room to deal with in a "purist" sort of way. Fitting a dynamic dress to V4, with her ample chest, to a figure that is less well-endowed and getting an accurately realistic simulation of a particular cloth type is... very difficult without first deforming the cloth, manually.



  • @F_Verbaas said in Some ideas for Clothroom version 2018:

    Some people may wonder why these 'non-linear properties' I mention are so important, and why it is so difficult to the >...There willl be room for the artist to make a difference.

    Above all, this is what everything needs to be focused on. Without that sort of support or advantage, new content won't be reliably available. SM does not "produce new content" for its product on a normal basis. Instead, they rely on others to do that. There is nothing at all wrong with that approach, but it has to actually work and needs to be a reliable effect. That won't happen unless the opportunity exists for artists to do that!

    Poser's Cloth Room has definitely been an advantage for it for a long time. That advantage is being reduced, however, in recent times. So, its development and advantage needs to be regained and, as you suggest, that's one way of doing that. Another, as I mentioned, is to make it more accessible and user-friendly for novice users as well as to open up another product/marketing strategy for content producers, even if that is a relatively small niche/effect. The mere noticeable appearance of such things would have a positive effect on consumer's perception of the Cloth Room, itself.



  • Adding to the stew:
    [a paper with an idea for easier manipulation and initial draping of clothing.](link http://www-ui.is.s.u-tokyo.ac.jp/~takeo/research/cloth/index.html)
    This could help to keep the entry easy.



  • @morkonan
    You are totally correct about problems when fitting existing pre-formed clothing to less curvacious figures.
    Here grading and size come in. Not only in geometry but also in texture. A one-inch stripe fabric pattern will remain to be a one-inch stripe regardless of the size of the garment cut from it.
    Like ready to wear clothing in a shop, a choice in sizes can become a selling argument.



  • @fverbaas said in Some ideas for Clothroom version 2018:

    Here grading and size come in. Not only in geometry but also in texture. A one-inch stripe fabric pattern will remain to be a one-inch stripe regardless of the size of the garment cut from it.
    Like ready to wear clothing in a shop, a choice in sizes can become a selling argument.

    Very true. This is one of the advantages of materials-based textures - They will not, necessarily, result in "texture stretching." Texture stretching due to changes in the geometry after it has be UV'd are a common issue in Poser when it comes down to refits and morphs. As it is, that's unavoidable, really, unless everything could have a materials-based-only texture on it and that's just not feasible and, from the normal user's point-of-view, undesirable. (Materials-based texturing can be extremely complex as everyone knows and very difficult to use to get a desired look for many types of textures.)

    As it is, for Poser and many, many, other applications, texture stretching (UVMap stretching on geometric objects that are texture mapped) is an issue. Most of the time, it's not too terrible. But... if every "fabric" behaves like stretchy rubber stuff, we'll not escape that problem. :(



  • @morkonan said in Some ideas for Clothroom version 2018:

    Texture stretching due to changes in the geometry after it has be UV'd are a common issue in Poser when it comes down to refits and morphs.

    One more reason to do refits by re-draping rather than by stretching/shrinking; basically fitting it on as if when trying on ready-to-wear clothing in a shop or theatre dressing room.
    If the change from donor to target is such that in real life you would need to do cutting and stitching to make fit, recourse has to be made to manufacturing tools: Scissors and sewing machine in real life and originating app or geometry app in the 3D world.
    In Marvelous Designer (sorry to mention again) UV map is updated automaticaly if you re-make the clothing (and use seamless textures).

    Grading and fitting of patterns usually is done along certain Rules of mathrmatical transformation. If these Rules are applied to geometry and UV mapping a reasonable kind of re-sizing could be obtained without re-meshing and keeping the texture free of distortion.
    The [Lutterloh System, AKA Goldene Schnitt](lhttps://lutterloh-system.de/ is an example. Resizing is done by moving the edges of key points to and from a specified center point over a distance set by the desired size. Provided the UV map is uniformly scaled and free of strain, the UV co-ordinates used in the image node could be tweaked in the material room. Size so becomes part of the shader. The edge lengths of the facets in 3D space can be scaled accordingly. A simple simulation in 10 or so steps would re-size the garment. From there on, draping would give the desred fit.

    In real life fashion industry this approach has fallen in disuse because the fit was not as good as modern comfort requires. Our 3D characters are less pecky about comfort, so the mehod may be good enough for resizing 3D garment models.
    image distortion-free. hese Rules could be applied to geometry and to UV mapping . Some of them could be applied in the material room.



  • Continuation from above post;
    Basically this means a parametric 'morph' for the UV map, given in a polar co-ordinate system.

    For those scared by the referred page in German: there are videos on he site showing the method. Imagine the work is done on the UV map. I found one reference in English.
    The below gives a typical pattern for this method:
    0_1486900808687_Vest1938.JPG
    The number gives the distance on the Goldene Schnitt Ruler. This image shows the top only as it comes with the kit. Below the '8' a cm based ruler is to be attached, at 8 cm, so the scale 0 to 8 is continued.
    0_1486901025842_$_1.JPG
    The figures on the right give circumference of chest of he figure in cm. The coresponding positon is held at the cross in the pattern and measurement is made in the dirction of the pattern.
    I used this a few times for making clothing for Roxie, for example for this vest:
    0_1486901355730_Vest1938_Roxie.png
    I think I should try the above method for a refit from Roxie to Pauline, LOL!

    One more thing: A typcal 'problem' area for refitting is the breasts. Question is of course which bra you want the figure to wear under the clothing.