Will's Tips

135

Comments

  • Oso3DOso3D Posts: 14,414
    edited September 2016

    Surfaces have an absorption/translucency element and a reflection element.

    The big difference between, say, a mirror and your skin is that your skin absorbs and passes through a very small amount of light and, more importantly, your skin is bumpy from large scale (moles) to extremely fine scale (cell surfaces).

    Imagine taking a scouring pad to a mirror and what it would look like; that's essentially what roughness _is_. Glossy roughness (or the inverse of glossiness) is nothing more than a code short-hand for 'there's a bunch of tiny bumps on this thing.'

    In practical terms, glossiness means how broad or tight the highlights are.

    Glossy weight is how bright the highlights are, and glossy color (or specular color) determines what color the highlights are (there are subtle differences, I think, but that's good enough).

    Now, technically, any non-metal should have uncolored specular/glossy color. Only metals do funky shifting of reflected light. I occasionally break this rule with velvet and other surfaces that do weird things with light that aren't otherwise something a shader can handle.

    With PBR Specular/Glossy, most non-metals should have a fairly dark specular; clothes, skin, and rocks don't typically have strong highlights. They should also have fairly high roughness/low glossy unless they have unusually polished/wet/glistening surfaces.

    With PBR Metallicity... there is debate. Personally I prefer to keep glossy color pure white and glossy weight at 1, and then lower reflectivity (which, IMO, more accurately captures what you are doing). Other people argue this. You can get roughly similar results lowering glossy weight or glossy reflecitivity, but sometimes these adjustments don't work the same with different lighting.

    So, if a surface looks too plastic, chances are the glossy is too bright. In 3DL, you can lower glossiness, specular strength, or use a darker specular color. For regular skin, personally, I like Glossiness of about 20%, Specular Strength 25%, and a medium to dark-gray specular color. Glossier skin, increase some of that.

     

    Also, if you want more 'realistic' metals in 3DL, I recommend using Reflection; set it to Raytrace, then adjust reflection strength and maybe give it a little blur. If you go with a pure metal you might even go 100% reflection strength and ignore diffuse/specular entirely. Metals may also have reflection color, given the oddness of some metal types (like gold tends to reflect a more yellow-orange). To be really accurate you might also want to look into Fresnel, but that gets into weird territory.

    Just remember if you do this, you need a backdrop to be reflected properly.

    (Also, raytraced reflections slow things down)

    Post edited by Oso3D on
  • On the topic of UV maps, in case this was the question:

    What is a UV Map?

    Well, first, I'm not an expert in this. Buuut... essentially, a UV map tells the renderer where, on the figure, different bits go. The translation of what, in a square image, goes where, on a very bumpy figure, can vary a lot.

    With figures, there are different UV maps for the same figures. Practically speaking, this means certain parts of the figure will be in different 'places' on the map. So, a V4 eyebrow is pointing to a different place than a V5 eyebrow.

    The reason for this is to focus detail in areas considered important to the figure. For a dramatic example, consider a humanoid rat figure, where the part of the face around the nose is pushed forward and larger, and things like ears are tiny. If you don't adjust the UV Map, the snout may end up looking pixelated and lower quality, while the ears have unnecessary detail; it may be worth creating a new UV map where more of the texture space is focused on the nose area and a smaller region for the ears.

    Now there is a lot of debate whether something like a UV map for a specific figure, like Stephanie, is really worth bothering with. That point is debated; theoretically a unique UV mesh can highlight details that are more prominent. Personally I'm not sure it makes that much of a difference, but hey.

     

    UV map is a 2 dimensional representation of a 3 dimensional object. Remember those paper models, like http://panzercombat.com/paperModels/SU100_1_100.jpg, where you start with 2d plan or picture of it and than cut away the excess paper, apply the glue and than fold it into 3D paper model. UV map (and making one) is sort of reverse from that, you start with 3D object and than make cuts or seems on it and than unfold it into 2D image.

    So, when somebody puts a 2D image, texture map over UV map it tells the render engine that this and this pixels from the image go onto this and this "geometry square" on the 2D UV map, which in turn is this and this "geometry square" on the 3D model.

    As for UVs for specific figure, its a quality thing. 2D texture and 2D UV map and 3D object are in mutual relationship. If somebody morphs (its happening more or less in some areas of the figures even with just posing the figure) the figure thus changing squares (they are bigger now or smaller or not cubes like before but squares) on the 3D figure, UV map is still "telling" render engine that still the same pixels go on the still the same square on the UV map which in turn is now differently shaped area of the 3d model and we get texture pixelization (same pixels on bigger area) or squishing (same pixels on smaller area) or stretching (same pixels on directionally differently shaped area). Hence the need for UV map adjustments to the specific figures.    

  • Oso3DOso3D Posts: 14,414

    Converting Iray to 3DL:

    Use a similar shader with the general right look. Duplicate the object, convert new copy to similar shader. Set diffuse color, bump, displacement, maybe Normal (if it has that). Probably done at this point.
    Otherwise, duplicate the object, new copy apply Daz shader or UberSurface or whatever. Set texture maps. If it's shiny, glossiness 90%, specular strength 100%, specular color light (195 or higher if it's REALLY glinty). If it's stone/wood/skin/etc, a safe choice is glossy 20%, specular strength 25%, specular color medium/dark gray.

    For metals, use Reflection (raytraced). Set reflection color to glossy color (if there is any). Metallicity x 100 -> Reflection strength. Glossy roughness x 100 -> Reflection blur.

    With Daz Shader, pick a light model that's appropriate. Plastic/Metal/whatever.

    Skydomes are pretty easy -- set diffuse strength to 0%, put image in ambient, set to 100%. Sometimes I've found it useful to set ambient strength to 200%. Go into Display parameters and 'casts shadow: Off'

     

    Examples to follow.

  • Oso3DOso3D Posts: 14,414

    Iray and then 3DL. The hardest part was adapting lighting, really. I also found that doing 'UberSurface2 Conversion' preset actually did the job perfectly and I didn't have to play with the figure at all. The ground took a little more work; ended up going neutral gray specular color, 15% glossiness, 20% specular strength.

    Amusingly, the Iray render was faster than the 3DL one, and I think the 3DL one looks more realistic...

    Conversion example1 Iray.png
    1080 x 1080 - 2M
    Conversion example1 3DL.png
    1080 x 1080 - 2M
  • The ground and sky look more detailed in Iray. At least to me.

  • CybersoxCybersox Posts: 7,976
    edited September 2016

    Iray and then 3DL. The hardest part was adapting lighting, really. I also found that doing 'UberSurface2 Conversion' preset actually did the job perfectly and I didn't have to play with the figure at all. The ground took a little more work; ended up going neutral gray specular color, 15% glossiness, 20% specular strength.

    Amusingly, the Iray render was faster than the 3DL one, and I think the 3DL one looks more realistic...

    I'd say that the difference is more in the lighting setup than in the renderers or shaders.  You've got a lot harsher light setup going on in the Iray version and it looks like the exposure's been kicked up across the board to show detail in the shadows, a combination that the human eye would seldom actually see. Plus, it also looks like the light temperature is off... there's way too much blue in both the key and fill to match the colors shown in the sky.  I agree that the the 3DL version works better because the lighting is more naturally balanced and dim, so the shadows go black as the eye would normally perceive them against a backlight.   All that said, I do think that the Iray shader is a lot busier in terms of textural detail, something that the extra exposure in the Iray version makes more blatant, but a lot of that could be fixxed by adjusting the scaling.    

    Post edited by Cybersox on
  • jardinejardine Posts: 1,100

    this thread shaping up to be the 'missing manual' for daz studio.  or 'the idiot's guide' to...

    thanks, will!

     

  • IvyIvy Posts: 6,793

    Amusingly, the Iray render was faster than the 3DL one, and I think the 3DL one looks more realistic...

    Agreed, 3DL does look realistic.. nice work Will

  • AlienRendersAlienRenders Posts: 707
    edited September 2016

    As per request, here is an explanation of Discontinuous UV.

    To understand what these are, let's first examine what happens when we use continuous UV's only. This means that each 3D point has EXACTLY one UV coordinate associated with it. So for each 3D point, you can get the colour at its uv coordinate in the image you are using to texture your object.

    Let's start with a cylinder. I've removed the top and bottom "circular" polygons to simplify this example. 

     

    Now, let's produce a UV map for it. 

    Notice something odd? Our image has a single black bar on it. Yet we see two black sections on our model in the 3D perspective view. What happened? Well, since the image is 2D, the app had no choice but to squish the cylinder against the screen folding the cylinder onto itself in the same way you would crush a paper cup that had fallen over. So there are two sets of overlapping polygons in our UV map. Let's take a closer look.

    I've highlighted the two polygons that overlap the rest of the polygons in the UV map. In other words, these two polygons occupy the exact same area on the texture as all the rest of the polygons put together. This is why the black bar appears twice. Let's make it more obvious what's going on. I'm going to move the center UV coordinates in the middle of those two polygons. See below. This is not what we want. We want the left edge to wrap around back to the right edge.

    One thing we can do is split the geometry and create a seam. We wouldn't actually physically move the polygons away from each other, but I'm doing this to show what I'm doing. In a real case scenario, you'd keep the polygons next to each other. We create a new UV map (because the old one is no good). This is still a continuous UV map. It looks good... but there's one problem. It doesn't actually look good on objects with smoothing. You will see a clear seam where the two split polygons meet. This is because the normals won't point toward the user, but slightly away because those points are now only attached to ONE polygon each. We'd like the normals to use the average of both polygons, but that can only be done if a point is included in two polygons. A normal dictactes how light will bounce off of it and will affect your render. This is why it's important. 

    So how do we get a proper UV map? You can actually WELD your points together (welding means turning 2 or more points into one) and your 3D modeling software should be able to create discontinous UV's for you automatically. In fact, there are usually tools that do all this for you anyways. But splitting the geometry first is actually a good technique to force your 3D app to create (unwrap) your 3D model the way you want. Then you can weld your points back together.

    So below, I've welded the points back together and you'll see it looks correct. The normals are correct. But notice the highlighted points in the UV map in the bottom right panel? I've selected two UV coordinates. But they BOTH belong the SAME 3D point. One UV coordinate is on the left edge of the image. The other UV coordinate is on the right edge of the image. In Lightwave, discontinous (secondary UV coordinates) appear in red (apologies for using a red image in the background). This is how you loop images back onto themselves in 3D. While your 3D software may do much of the work for you, this is very useful to know when something goes wrong. You won't be scratching your head as to what's happening.

    I hope this helps understanding discontinous UV mapping and what your 3D software is doing when it unwraps closed volumes when creating UV maps for you.

     

    sc1.png
    1907 x 1074 - 93K
    sc2.png
    1907 x 1074 - 121K
    sc3.png
    1907 x 1074 - 121K
    sc4.png
    1907 x 1074 - 124K
    sc5.png
    1907 x 1074 - 104K
    sc6.png
    1907 x 1074 - 120K
    sc6.png
    1907 x 1074 - 120K
    Post edited by AlienRenders on
  • Oso3DOso3D Posts: 14,414

    Another Iray tip:

    Don't use Displacement. Seriously, just... don't. It sucks enormous resources to make it work decently, and that's effort better spent elsewhere. You can morph an object with mcasual's Elevate script ( https://sites.google.com/site/mcasualsdazscripts/mcjelevate-reva ) if you must change a surface.

    Iray works better if you make more use of Normals.

    'But Will!' you may exclaim, 'I don't have a Normal for this!'

    Aaaah... well, thankfully it's actually pretty easy to MAKE one. Seriously, make one, plop it in whatever directory the item in question is in, and use it from now on.

    You can generate a decent Normal pretty easily from either a displacement or bump map using Photoshop (Filter > 3D > generate Normal map, note that it has to be a RGB image; you may need to switch the mode of grayscale image)

    Sometimes Bump maps are not very good. Frankly, some of them are godawful -- just grayscale versions of a diffuse map with shadows and nonsense in it. Not much you can do with that. But if someone bothered to make a displacement map, that map is a great source to make a Normal.

     

    If you lack Photoshop and anything else handy, there's a cool 'turn things into Normals' converter online at : http://cpetry.github.io/NormalMap-Online/

    Note that you need Nvidia card, I think, to make use of that site.

     

     

  • JimbowJimbow Posts: 556
    Don't use Displacement. Seriously, just... don't. It sucks enormous resources to make it work decently

     

    Sounds a bit unfair. It all depends on the polygon resolution of your base mesh (the higher the rez the lower your displacment subdvision needs to be) and how long you're willing to wait for the presampling. MCJ's push script just does what iray's displacement does as if you were displacing at base resolution with no subdivision at all, but gives you the luxury of a morph slider. I find pre-empting the need to do high subdivs in the displacement settings by increasing the SubDivision Level, under the figure/prop General tab, before the render kicks in reduces the need for high displacement subdivs, and bear in mind you don't have to use only Catmark SubDiv algorithm. I don't have a fancy setup, but I'm not coy about using displacement provided I go into it with my eyes wide open.

  • ArtiniArtini Posts: 6,337

    @AlienRenders: Thanks for the valuable tips. Does that mean, that Discontinuous UV means UV2, which I approach pretty often in Unity3D models?

     

  • Richard HaseltineRichard Haseltine Posts: 71,490
    edited September 2016

    I have been told that using the material SubD division settings should be avoided - it effects the whole object even if it is turned up on only a single surface so it doesn't save polygons over the object SubD levels, and is much less efficient (at the stage of geenrating the mesh for Iray) than the object-level SubD so is more likely to overload the GPU. While it is probably better to use normals where possible, remember that they are not compressed in Iray regardless of the compression settings so they can still potentially eat a lot of RAM.

    As Jimbow says, it depends on the mesh resolution - and the distribution. If your model has as ingle polygon floor and more polygons for the skirting boards or other details then dividing enough to get visible grout lines for the floor from displacement is also going to divide the already higher-polygon areas, leading to a vastly inflated polygon count; on the other hand if the mesh resolution is already even, or if you can split out the area you wish to displace (for example using the Geometry Editor) then the problem wil be much less severe.

    Post edited by Richard Haseltine on
  • Oso3DOso3D Posts: 14,414
    edited September 2016

    Sure, it depends on the mesh.

    In my experience, though, it's a lot of work to find out if it's going to work this time, and you will very often spend a lot of time only to conclude 'ugh, no, let's do Normals.'

    Good point about distribution.

     

    Ok, I'll amend my tip. Try Normals first. If that looks good to you, stop. ;)

     

    Post edited by Oso3D on
  • JimbowJimbow Posts: 556
    Ok, I'll amend my tip. Try Normals first. If that looks good to you, stop. ;)

    LOL yes

  • Oso3DOso3D Posts: 14,414

    Iray skin tips:

    Take the standard Iray skin converter result.

    Remove top coat. Remove any glossy color/weight maps, make weight 1, color white. Make most skin Reflectivity .3, Roughness .7 (adjust up/down for wetness), lips Reflectivity .4/Roughness .5 (again, adjust depending on how wet you want lips)

    Remove any translucency weight maps, but make sure translucency color maps are diffuse maps (or sss maps if the skin has them).

    Switch Base Color Effect to Scatter & Transmit.

    Set SSS Reflectance Tint to a very light blue (to taste)

    Increase SSS Amount to 1

    Change SSS Direction to .5

    Here's a before/after. From bottle tan to human!

     

    comparison2.png
    1080 x 1080 - 1M
  • vwranglervwrangler Posts: 4,302
    edited September 2016

    Iray skin tips:

    Take the standard Iray skin converter result.

    Remove top coat. Remove any glossy color/weight maps, make weight 1, color white. Make most skin Reflectivity .3, Roughness .7 (adjust up/down for wetness), lips Reflectivity .4/Roughness .5 (again, adjust depending on how wet you want lips)

    Remove any translucency weight maps, but make sure translucency color maps are diffuse maps (or sss maps if the skin has them).

    Switch Base Color Effect to Scatter & Transmit.

    Set SSS Reflectance Tint to a very light blue (to taste)

    Increase SSS Amount to 1

    Change SSS Direction to .5

    Here's a before/after. From bottle tan to human!

     

    What does removing the glossy weight maps do? And what's the benefit of positive numbers over negative numbers for SSS direction? (I do remember the debates from early in the Iray implementation, but only that they happened, not the content.)

    Also, I thought translucency weight maps and SSS maps were the same thing.

    Post edited by vwrangler on
  • Oso3DOso3D Posts: 14,414

    Theoretically, glossy weight maps would set certain parts of the skin to be 'more glossy' than others. If you really are committed to doing that, I would move the maps to Glossy Reflectivity.

    See, again, in hazy theory, most substances are purely glossy. Where that stops happening is because either the surface absorbs light, or the light is reflected more scattered. Reflectivity handles (IMO liberally) that adequately.

    SSS direction determines how light scatters. If light is pointing right at you and something is in the way, a positive number increases how much is scattered while light coming in from the sides are not.

    A positive value for skin mainly means that you are more likely to notice scattering with backlighting than light bouncing off the skin. It's... debatable, but I've liked the results.

    As for SSS maps.. it depends. Basically, Eisa and I think Belle (V5? 6? Something like that) have a translucency color map where certain areas are 'pinker' than normal, like the ears and lips. That's worth keeping.

    But translucency weight maps have odd and contradictory results, and aren't really needed if you are using translucency color maps. So, chuck 'em.

     

  • dracorndracorn Posts: 2,307

    Will - I have an idea and don't know how to go about doing it.  Not sure if the geoshell will work.

    I want to create a sword that has a woman's face just above the pommel.  I don't want it static - I want G2F's expressions, and in fact, I want it to look like a face that I had made for G2F using morph dials.

    So, do I shove and squeeze G2F's face into the slot and make the rest of her invisible?  Or is there some way to copy that face and place it in the sword, and then create some of my own morphs for expressions?

  • dracorndracorn Posts: 2,307
    Jimbow said:
    Don't use Displacement. Seriously, just... don't. It sucks enormous resources to make it work decently

     

    Sounds a bit unfair. It all depends on the polygon resolution of your base mesh (the higher the rez the lower your displacment subdvision needs to be) and how long you're willing to wait for the presampling. MCJ's push script just does what iray's displacement does as if you were displacing at base resolution with no subdivision at all, but gives you the luxury of a morph slider. I find pre-empting the need to do high subdivs in the displacement settings by increasing the SubDivision Level, under the figure/prop General tab, before the render kicks in reduces the need for high displacement subdivs, and bear in mind you don't have to use only Catmark SubDiv algorithm. I don't have a fancy setup, but I'm not coy about using displacement provided I go into it with my eyes wide open.

    Gorgeous render!

  • Oso3DOso3D Posts: 14,414

    I think your best bet would be the 'put her face there and make most of her invisible', and possibly flatten her head/face a little and use collision so the face 'smears' out. Maybe even some sort of mask to help her blend in. Mmm.

    There's no easy way to have a poseable face otherwise, that I can think of. (I think there are some special posable marble busts in the store, but I'm not sure how much of a difference they'd make)

  • vwranglervwrangler Posts: 4,302

    Theoretically, glossy weight maps would set certain parts of the skin to be 'more glossy' than others. If you really are committed to doing that, I would move the maps to Glossy Reflectivity.

    See, again, in hazy theory, most substances are purely glossy. Where that stops happening is because either the surface absorbs light, or the light is reflected more scattered. Reflectivity handles (IMO liberally) that adequately.

    Ah, OK. Not terribly committed either way, just trying to understand.

    SSS direction determines how light scatters. If light is pointing right at you and something is in the way, a positive number increases how much is scattered while light coming in from the sides are not.

    A positive value for skin mainly means that you are more likely to notice scattering with backlighting than light bouncing off the skin. It's... debatable, but I've liked the results.

    OK, now I am a bit confused. A negative number would lead to ... less than no scattering? It seems like you could have values down to zero, and then you're done. I think I'm not understanding quite what the settings do, relative to each other. What little I remember of the past discussion was that the numbers were about direction, as you say, but that discussion seemed to indicate, as I recall, that positive numbers essentially said there was no scattering, because the direction was wrong, and it had to be internal, and therefore negative.

    As for SSS maps.. it depends. Basically, Eisa and I think Belle (V5? 6? Something like that) have a translucency color map where certain areas are 'pinker' than normal, like the ears and lips. That's worth keeping.

    But translucency weight maps have odd and contradictory results, and aren't really needed if you are using translucency color maps. So, chuck 'em.

    Ah, OK. I don't think most of Genesis 3 (or Genesis 2, for that matter) has translucency color maps. The FWSA/FWArt characters do, I think, and Adrienne G3 by Nathy and Kurt by DTG from elsewhere. For Genesis 2, if you get Reinterject, you can have translucency/SSS color maps for G2F and V6; you'd have to do map transfers to get those to other characters, but that would be workable.

    Somewhat off-topic: If you want something with a huge variety of maps to tinker with, I recommend picking up the Dark Beauty G3F merchant resource from here. I picked it up during one of the earlier sales and basically sat there looking at all the maps, wondering what on earth you'd even DO with all of them.

  • Oso3DOso3D Posts: 14,414
    I haven't looked at g3 stuff much, no interest. As for direction... I don't know, but I've found one of the somewhat reliable ways to get godrays through fog is have a light source pointed mostly at the camera and then set a very high positive sss direction (like .99)
  • 3Diva3Diva Posts: 11,017

    Iray skin tips:

    Take the standard Iray skin converter result.

    Remove top coat. Remove any glossy color/weight maps, make weight 1, color white. Make most skin Reflectivity .3, Roughness .7 (adjust up/down for wetness), lips Reflectivity .4/Roughness .5 (again, adjust depending on how wet you want lips)

    Remove any translucency weight maps, but make sure translucency color maps are diffuse maps (or sss maps if the skin has them).

    Switch Base Color Effect to Scatter & Transmit.

    Set SSS Reflectance Tint to a very light blue (to taste)

    Increase SSS Amount to 1

    Change SSS Direction to .5

    Here's a before/after. From bottle tan to human!

    Wow! That looks great! :D 

    I had to try this out. 

    Here are the results of the settings I've been working on:

    Here are the results with your suggetions/ tweaks:

    I really like it! It as a more natural look and gets rid of the orangy tint. Thanks, Will! :)

  • Looks good, @divamakeup!

  • 3Diva3Diva Posts: 11,017

    Looks good, @divamakeup!

    Thank you! :D 

    I do miss seeing a bit more of her veins though. I think I'll do a bit more tweaking and see if I can figure out how to make her a bit more translucent or have her veins show a tad more. Although it's probably more natural/normal to not see them as much. 

  • Oso3DOso3D Posts: 14,414

    You might want to try making her less translucent, maybe, so more of the skin shows.

    Another approach would be going into an image editor and making a more vivid version of the diffuse maps as a translucency color map... maybe.

     

  • RAMWolffRAMWolff Posts: 9,655

    Iray skin tips:

    Take the standard Iray skin converter result.

    Remove top coat. Remove any glossy color/weight maps, make weight 1, color white. Make most skin Reflectivity .3, Roughness .7 (adjust up/down for wetness), lips Reflectivity .4/Roughness .5 (again, adjust depending on how wet you want lips)

    Remove any translucency weight maps, but make sure translucency color maps are diffuse maps (or sss maps if the skin has them).

    Switch Base Color Effect to Scatter & Transmit.

    Set SSS Reflectance Tint to a very light blue (to taste)

    Increase SSS Amount to 1

    Change SSS Direction to .5

    Here's a before/after. From bottle tan to human!

     

    Is this for PBR Specular/Glossiness or PBR Metalicity/Roughness?  

    I'm using Specular/Glossiness but trying to give this a go...

  • Oso3DOso3D Posts: 14,414

    Metalicity. I THINK it should work much the same in Specular, though with Glossy roughness reversed.

    Instead of Reflectivity, alter Specular Color; it should be fairly dark for most non-metals.

     

  • RAMWolffRAMWolff Posts: 9,655

    OK, I'll give that a try!  Thanks Will! 

Sign In or Register to comment.