Avatar

Greetings!

Welcome to Scifi-Meshes.com! Click one of these buttons to join in on the fun.

Badges

McC

About

Username
McC
Joined
Visits
45
Last Active
Roles
Member
Points
373
Posts
704
Blender
Blender
GIMP
GIMP
Inkscape
Inkscape
Krita
Krita
Badges
8
  • So, uh... how attached are you to this here forum?

    No strong attachment to the forum software. At the end of the day, the user experience is much more about how the site admins have customized the UI than it is what's going on under the hood, so whatever makes the most sense for you is the thing I support!

    Definitely would not like to see a "mobile-first" forum package, though! If I want to interact with the community on mobile, that's what Discord is for. I feel like forums should be aimed at the desktop experience first, possibly with support for a mobile experience, but not as the focus. Maybe that makes me a dinosaur, but there's definitely a clear demarcation for me between "stuff that's mobile" and "stuff that's not" and forums fall squarely into the latter category.

    No strong feelings about Markdown vs BBCode. Both have their strengths, though Markdown does feel like a more widely-used option these days.
    • Custom theme: As long as there's something vaguley darkmode-like, with blue accents and not a lot of overhead, I'm happy. (Currently using the Blue Compact theme, for context)
    • Gamification: Don't care
    • Liking posts and hooking that up to the gamification above: Don't care (strongly). We've chatted about this in the past, and I kind of wish there was a way to sort of "bump" a thread with a "like" (in quotes because it needn't be an actual "like" mechanism) to indicate "Hey, this thread is active and something cool got posted that I wish to provide accolade to, which will count as activity in the thread" without requiring the user to post a one-sentence comment like "That looks great!" to achieve the same goal. But you very rightly also pointed out so big potential gotchas with this approach, so...not sure what the right answer is!
    • App icons in profile: I do feel like this is helpful, especially displayed near the user's avatar or actual comments, to provide context.
    • Social icons in profile: Don't care.
    • Commissions open icon: Not specifically relevant to me, so abstain.
    • Thread thumbnails: Definitely! Though a more automatic way of handling it might be handy? Having to specifically attach an image to the first post and all that, which one needs to keep updating, is by no means odious, but since I host my own images, it means I have to upload not just to my server, but also remember to go back and edit the first post to update what's attached there, which in turn cruds up your server. Perhaps giving the thread submitter some way of specifying the thread thumbnail in each subsequent comment they make, with a URL or an upload, and give the most-recent one precedence?
    • Thread prefixes: Definitely useful, but by no means a dealbreaker.
    • Tagging: Nice-to-have, but also usually self-evident from the post itself.
    • A whole bunch of user data improvements: No strong feelings
    • Sketchfab and Marmoset embedding: No strong feelings
    • Necropost tags and warning no one seems to read: No one ever reads warnings, but these are still good to have.
    • Antispam measures: Definitely good to have. Forums getting overrun by spamswarms is a real danger.
    • A bunch of little quality of life things: Entirely up to you, since they'd need to be enumerated to comment on! :mrgreen:

    Hope that's useful!
    lewisniven
  • Beyond Principled: Tackling some of the weaknesses of the Principled BSDF

    Most Blender users are familiar with the Principled shader, and probably use it as the foundation of all of their materials. The Principled shader, based on a white paper put out by Disney enumerating their principles for creating a physically-based and artist-friendly shader, was a widely welcomed update to the convoluted node networks necessary to achieve similar results in Blender up to that point.

    Over time, though, cracks have appeared in the Principled shader's gleaming armor.

    For one, it is not an "energy conserving" shader; that is, it has the potential to reflect more light than it receives, in violation of what real-world materials do. While its energy gaining properties are relatively minor, it's easy to reproduce using a "white furnace" test: create a box with its normals facing inward and give it an white emission shader with a strength of 1, then place a sphere inside it with a white Principled material applied; the edges will be brighter than the rest of the sphere.

    fbad79bd57869f17e95023b4c48d4ef331cd33c0.png

    For another, although it correctly accounts for the Fresnel effect when blending between its albedo (diffuse) and reflective (specular, glossy) at various view angles, it does not properly account for the phenomenon of rough materials to seem much more smooth at glancing angles (c.f. a rough stone building exterior that, when viewed from an extreme glancing angle, takes on a mirror-like appearance). This effect is often referred to as micro-roughness. It can be forced by including a very fine noise texture in the Normals of a material, but this solution tends to be costly and artifact-prone. YouTuber Chocofur has a good video that describes a workaround for the problem, but in particular showscases some real-world surfaces that exhibit this property. The following images are extracted from the linked video, and I extracted the relevant 45 seconds into a bite- (megabyte-?) sized mp4 here

    micro_roughness_01.png micro_roughness_02.png micro_roughness_03.png micro_roughness_04.png

    Finally (well, for the sake of what interested me, at least!), the Principled shader granted no artistic control over the reflection color. In the "old days" of Diffuse + Glossy mix shaders, the Glossy shader's color could be independently controlled. While most dielectric (i.e. non-metal) surfaces do not exhibit colored reflections, some more-complex materials (such as interference paints!) can only be correctly represented by manipulating the reflection color. The Principled shader only allows tinting of the dielectric specular reflection, based on the Base Color input, which just isn't enough.

    These three things seemed sufficiently surmountable and important to producing more physically-correct materials that I set out to create a shader mix that properly accounted for both.

    >>>>> You can download the current version here. <<<<<

    This is meant as a more-or-less drop-in replacement for most basic materials using the Principled BSDF. It does not (currently) account for subsurface scattering, anisotropy, clearcoat, sheen, transmission, or emission! I may eventually add those as needs arise, but so far I have not needed them.

    If you don't want to know how it works and just want to give it a spin, enjoy!

    If you want to know more and the theory behind it, read on!

    Fundamentally, this node group does the same thing the Principled shader does in its OSL code: it computes a base color Diffuse shader, then adds (yes, adds) a Glossy shader on top of it.

    The first problem with the Principled shader, though, is that it doesn't correctly damp down the Diffuse contribution where the Glossy one kicks in. That leads to the Fresnel-influenced energy gain observed in white furnace tests of the Principled shader. It does this by calculating the "F0" value of the material, based on its Index of Refraction (IOR) and using it to compute how much of the Base Color to "add" to black, before passing that color into a Diffuse BSDF.

    What it does with the Glossy shader is a lot more involved. Rather than using a plain Fresnel term driven only by surface Normals, a "perturbed" Fresnel component, driven by the surface Roughness, is used as the Normal input to a Fresnel node. It also feeds the derived view angle (arccos of the dot product of the surface normal dot and the incoming camera vector) into a pair of albedo response curves, based on data from the Disney white paper, mapping between them depending on whether the surface is smooth (roughness < 0.1) or rough (roughness > 0.1). The albedo response curve for roughness > 0.1 adjusts the roughness down starting at around 60 degrees off-normal, reaching a minimum value of around 40% of the input roughness. This models the micro-roughness effect.

    specular_g_albedo.png

    And, because it's a Glossy BSDF, you have control over the reflection color the whole way.

    Actually mixing in the reflection color isn't solely the purview of the adjusted Fresnel calculation, though. The supplied Glossy color is "normalized" against the Fresnel (by multiplying the Glossy color by 1/Fresnel), which in turn is fed into a node that governs dielectic vs. metallic reflectance value. The Specular slider on the Principled shader goes from 0 to 1; this actually maps to a surface reflectance of 0% to 8%, corresponding to dielectric IORs of 1 (air) to about 1.79. The same effect results. When the surface is metallic, this control does nothing.

    The Fresnel component is then used as a factor that adds the computed Glossy color to black, which is at last fed into the Gossy BSDF.

    With both the Diffuse and Glossy BSDF accounting for the F0 "weight" they should each contribute in addition to the usual Fresnel component, energy is conserved. With micro-roughness accounted for, rough surfaces appear much more reflective at glancing angles. With a controllable Glossy shader, reflection color can be manipulated for complex material effects. Thus, you get a more physically correct, more controllable material.

    I'm always learning more and constantly striving to improve, so it's entirely possible that I'll revise how I have this shader assembled as time goes on. It's also possible that the needs it addresses will disappear as Blender continues to grow and evolve.

    I hope this has been informative and useful!
    GuerrillalewisnivenWarp Propulsion LaboratoryLizzy777xiaorobearInitium Thoth
  • SDF-1 Macross

    Many thanks, @Warp Propulsion Laboratory , @Gidiot , and @Zhor2395 !
    Barricade wrote: »
    But when you are done doing the DYRL/Post-2012 refit....will you also do the original series' version?
    No plan to, no. The DYRL version is the one that, for me, is the most "authentic" even though it's not the original. As outlined above, the plan is to replicate the 1/4000 model kit first, then go back and revise it based on all manner of sources (other larger-scale model kits, stills from DYRL, and probably some riffing of my own).
    Barricade wrote: »
    BTW, might want to get onto Macrossworld's forums too, as they have tons and tons and tons of images for viewers to look at, straight out of the Macross Chronicles magazine - which is official canon, and is second only to the Animes/OVAs/Movies for how detailed some of the images are.
    Oh, I lurk there regularly, rest assured. :mrgreen:

    Right! Now, that things have settled down in life and I've gotten the Ambassadorial monkey off my back, time to actually make some more progress on SDF-1!

    sdf-1_2022-01-23-2042.jpg

    I couldn't quite stick to a completely nondestructive workflow for that battery mount (which still has more detail yet to be added...and the battery itself!), but all of the "destructive" aspects have preserved source meshes, so can be revised rapidly. I've got some new, better calipers coming in the mail on Tuesday, too, which will hopefully make taking measurements from the kit pieces a little more reliable!

    I also migrated my revised PBR shader and compositing mix over from my Ambassador revamp, and then tweaked both further. For a bunch of procedural and stock images, I'm pretty happy with that surfacing--at least as a placeholder!
    Lizzy777lewisniven
  • Ambassador Facelift: Meet Aurora

    new_purple wrote: »
    Also loving the window texture...I've seen other models using set images for the windows, but I haven't tried it yet myself. Going to experiment once the windows are all in place.

    Bear in mind that it's not just a texture on the windows. The windows are actual "panes" of glass, with very simple room boxes behind them, mapped with set photos.

    uli7gtcexkeb.jpg

    Whether or not that's worth the effort depends entirely on how close you plan to get and if you ever plan to see the thing in motion, though. If you're only ever rendering stills, just slapping a texture on the window itself and calling it a day is generally fine. If you plan to see it move, or want to get close, though, having room boxes (or an equivalent) is pretty much a must-have, I think!
    Lizzy777Soul Scream
  • Ambassador Facelift: Meet Aurora

    m6rwkdgxor62.png

    Eight years ago, I finished my Ambassador-class model, then-named USS Trafalgar. A few months ago (mid October or so), I went to see what I could salvage out of the remnants of the files I still had after the hard drive failure I mentioned here, and then proceeded to give the model a facelift. I didn't want to rebuild it entirely; instead, my goal was to fix mesh errors, correct (some) topology issues, update (or recreate, where they had been lost entirely) the textures, and overhaul the shaders based on all the stuff I had learned working on Coro.

    I ended up doing far more than I originally planned, but I feel good about the end result. Meet USS Aurora.

    imgimgimgimg
    imgimg

    Full list of updates:
    • Modeling:
      • Fix nonmanifold geometry throughout
      • Add glass to window portals
      • Add ribs to deflector glass
      • Add a mount point for the belly strobe light
      • Recreate hull and stardrive "shells" to use as DataTransfer sources to correct normals
      • Create a new, conformal escape hatch based on the higher-detailed escape hatch used in Star Trek: Generations (previous version was loosely based on Voyager's hatches)
      • Remove some of the existing stardrive windows and replace them with others more consistent with the rest of the ship
    • Texturing:
      • Fix UVs, especially on the saucer
      • Switch textures to UDIMs
      • Re-create all original texture masters in Krita (formerly, Photoshop), including layer comps
      • Re-create all base color textures as vector files in Inkscape, sourcing colors from color-corrected photos of the studio model
      • Re-create the nacelle texture from scratch
      • Update the stardrive texture to remove some weird curves from the dark blue "grid" area
    • Shaders:
      • Revise every material the model uses
      • Ditch the Principled shader in favor of a custom-built shader mix network that uses "real" Fresnel to blend between Diffuse and Glossy, and also correctly accounts for microsurface apparent roughness falloff based on angle. Planning to write Wrote up a little guide on this one.
      • Update escape hatch material to use deliberate, rather than random, numbering, the way I did for Coro
      • Update all interior rooms to better match real-world lighting values, and to color-shift based on different alert statuses
      • Investigate (with major credit to @PixelMagic for doing real-world empirical research) how the interference paints on the refit Enterprise actually work and integrate those findings into the hull material. Planning to write up a little guide on this one, too.
      • Create a random-scatter "visual interest" texture mix that places small markings all over the hull (inspired again by an idea from @PixelMagic ). Yep, planning to do a little guide for this one, as well!
    • Miscellaneous:
      • Modernize the running light setup, based on Coro's
      • Pull "decals" out into their own hull-hugging objects so they don't contribute to hull shader complexity
      • Add a bunch of signage/registry illumination that I didn't have before and that didn't exist on the studio model, either
      • Revisit the compositor mix I developed for Coro and further refine it

    If I revisit the Ambassador class again in the future, though, I am definitely starting from scratch! :mrgreen:


    Enjoy!
    trekkiViperlennier1Soul ScreamWarp Propulsion LaboratoryLizzy777lewisnivenBridgercaveat_imperator