The Wonderful Dynamic Puzzle of 3D (or, What's my Work Flow)

1356

Comments

  • Joe CotterJoe Cotter Posts: 3,164
    edited April 2013

    .. "Poser lights load with shadows turned off" two years ago. :D

    I missed this the first time I read it. The fact they load with shadows off are actually a good thing for me. I turn shadows off on most of my lights, turning them on for the main light only (sun, moon...) That, combined with the UberEnvironment using AO and sometimes IBL for low ambient. It reduces the chance of competing shadows to pretty much nil, gives me the shadows I prefer, and cuts render time dramatically. If there was one tip on lighting in DAZ+3DLight I could give, that would probably be it.

    Post edited by Joe Cotter on
  • SickleYieldSickleYield Posts: 7,288
    edited December 1969

    Gedd said:
    .. "Poser lights load with shadows turned off" two years ago. :D

    I missed this the first time I read it. The fact they load with shadows off are actually a good thing for me. I turn shadows off on most of my lights, turning them on for the main light only (sun, moon...) That, combined with the UberEnvironment using AO and sometimes IBL for low ambient. It reduces the chance of competing shadows to pretty much nil, gives me the shadows I prefer, and cuts render time dramatically. If there was one tip on lighting in DAZ+3DLight I could give, that would probably be it.

    I'm not saying all the lights in a scene should cast shadows (eek, my poor processors :D). I also usually work with a main shadowcaster and shadows off on the fills (depending on the individual scene I might need point lights to cast, etc.). I just meant that if you didn't know that you could easily be confused. Usually it's just easier to create my own than fiddle with existing ones that weren't made for DS.

  • Joe CotterJoe Cotter Posts: 3,164
    edited April 2013

    Yes, I find the same usually, but at times they are good for positioning. For anyone that doesn't know it, we can select a poser light and then click on the icon for adding a new light (doesn't even have to be the same type) and we get a dialog box asking if we want to copy/replace selected.

    There are times I still use the poser lights themselves but it's rare. In the case of The Greek Bath, it was so well done for it's time that the lights have nice effects that are still relevant. In this case, I originally replaced the lights, but then after rendering missed the effects of the original lights, so I put them back in at half strength with my other lights. I could have even gone stronger if I wanted a more stylistic effect. Surprisingly, they added very little to the render time and gave a nice subtle touch.

    Post edited by Joe Cotter on
  • Joe CotterJoe Cotter Posts: 3,164
    edited April 2013

    Here's a site for post work that has some nice photo manipulation tutorials. A This one in particular is a nice one for demonstrating post work. All of the items in this case are photographic, but basically one could swap out 3D content for any part of this. Searching for cool photo manipulation techniques is a good way to learn some really interesting post work techniques.

    Post edited by Joe Cotter on
  • Joe CotterJoe Cotter Posts: 3,164
    edited April 2013

    Here's a quick example with the Sacred Barge prop. The first image comes out blown out with the default setup and lights.

    01.jpg
    800 x 800 - 80K
    Post edited by Joe Cotter on
  • Joe CotterJoe Cotter Posts: 3,164
    edited April 2013

    With just a little tweeking of the materials, we can get much better.. this example again is UE2 ambient with a single distant for sun. There are a number of things I would fix in this, but it is a good starting point. This isn't meant to be an example of the right way, or even the best, just an example of how one can go in and quickly tweak what is already there to get much better results without investing a lot of time. That is to say, if it doesn't look the way we want right out of the box, don't get discouraged... play with the settings :)

    02.jpg
    800 x 800 - 82K
    Post edited by Joe Cotter on
  • Joe CotterJoe Cotter Posts: 3,164
    edited April 2013

    The second image rendered in 2m17s.. it does still have some bad displacement under the small urns and some artifacts on the lounge still that would need to be fixed. I have been very lazy here and used diffuse maps for displacement, bump and specular maps. It is not the recommended method, but for a quick adjustment it can work depending on what you expect out of the image. In the case of the urns, if I was looking to actually use this, I would copy out the diffuse, adjust it in a 2D program and use that to quick fix the displacement problem. I also changed out the default 'plastic' for 'shiny metal' on the urns, skin for the carpets and base boat...

    Often getting away from all of the materials being set to plastic is a good starting point. It will effect how the specular and reflection settings work in the image. It might seem strange to use 'skin' on things like the carpet and boat, but I like using it on organic materials that aren't meant to be too reflective. The specular of skin I find works well for many organic surfaces. Setting metal to a shiny metal surface will do a lot for getting the reflections to actually work the way you expect. Changing the materials is usually one of the first things I do since many of the other settings are dependent on that.

    If anyone disagrees with this, or has any comments to make feel free to jump in. This is meant to be an open forum and my feelings won't get hurt as long as you're nice about how you rip it apart ;)

    Post edited by Joe Cotter on
  • Joe CotterJoe Cotter Posts: 3,164
    edited April 2013

    One of the things I like to do with a model is look at the base mesh in lit wireframe with the 'Scene Info' tab open. This lets us know what we're working with in regards to the mesh. Things like if there's enough polygons to support a displacement map, how hard it is going to be to adjust materials if we want, what kind of stretching to expect if the mesh is to be distorted by a morph in a given area... In the Scene info page, we can see if there are a lot of triangles vs quads, total poly count... Basically, these let us know what we have to work with, and how efficient the mesh will be.

    03.jpg
    736 x 654 - 92K
    Post edited by Joe Cotter on
  • Joe CotterJoe Cotter Posts: 3,164
    edited December 1969

    In this example we start off with an out of the box render of The Study with the Expansion Pack, It's a very nice prop, but it doesn't render as one would expect just putting the preloads in with a fireplace and default lights.

    01.jpg
    800 x 800 - 203K
  • Joe CotterJoe Cotter Posts: 3,164
    edited April 2013

    First we change the specular, bump and displacement of the various objects, that's the easy part. Then after getting rid of the original light set, we set up a UE light with Moonlight 4x, full power as it's a dark light. We convert the window, fire, and oil lamp globe to UA lights and set the light and ambient settings on them. These are the only surfaces we set ambient on as they emit light. The fire is set to an orange light with an associated map of the fire in the light's color channel, the oil lamp to a pale yellow, and the window light color is mapped to the window's diffuse with a pale blue. These help sculpt the light area we want and give added dimensionality to the scene, but they don't give us enough control over the exact lighting so we put in point lights in the fireplace and oil lamp and adjust them to get the final effect.

    The two things I would do with this image still is to one, use light gels on the fireplace point lights (there are two, a smaller orange and larger red) to sculpt the light inside the firebox better and two, adjust the reflection settings on the oil lamp.

    Render time was 17s on the first image and 33m17s on the second.

    All of the examples to this point have been on a bare DS install with just that single prop loaded in the runtime. No extra lights, no parts from other prop sets, and no textures that didn't come with the prop.

    Almost forgot, I left this picture with no postwork, not even a color balance or lighting adjustment. The main thing I would do in post is to fix the two problems I mentioned earlier (the firebox and lamp body) If I didn't fix them in the render. Fixing them post would be faster of course, but if I planned on using the prop again much it would make sense to consider fixing it in render if it didn't drastically increase the render time. Also, fixing it in the render would help with fine tuning our overall skills, so it has that benefit if we aren't on a time schedule.

    02.jpg
    800 x 800 - 65K
    Post edited by Joe Cotter on
  • Scott LivingstonScott Livingston Posts: 4,278
    edited December 1969

    Wow...lots of great stuff in this thread. Thank you so much for sharing, Gedd (and others)!

  • Joe CotterJoe Cotter Posts: 3,164
    edited December 1969

    Glad you like it. I hope more people join in and contribute. Especially if they include examples like Roberto Melo did :)

  • Joe CotterJoe Cotter Posts: 3,164
    edited December 1969

    One thing we notice is that many of these pictures jumped in render time to something that wouldn't work very well using current hardware for an animation. Even with new hardware, learning to become more efficient will go a long ways. So, if we create a still like this, figure out what is eating a lot of cpu cycles that we could do in post, specifically with a script in our photo editing program, and remove that. Then if we render our animation to a series of pngs we can import them into our photo editing program, run the script we created to fix the said issue, and bang... we just increased (drastically in some cases) the quality of our animation without increasing the render time to an unacceptable level. :)

  • Joe CotterJoe Cotter Posts: 3,164
    edited April 2013

    I also would mention, I saved none of these scenes. If someone were to ask me the exact settings I used, I couldn't tell them, and... I did that for a reason. These are exercises. We have the image, and the render time. If we go to recreate it we can use that as a benchmark to see if we can get better results with better render times. This was my first pass on these, and hopefully if I go back and do them again I will get better. If anyone decides to try these themselves, feel free to put up the finished result and render time.

    At some point, it would make sense to save the lighting and textures for use with the set so we don't have to go back and remake them every time, but that wasn't the point of these particular exercises. My take is, we don't want to ingrain in our mind something is a finished version when we are still working out concepts. The act of saving the file might contribute to that if it's just an exercise for some, like me. I want to keep it clear in my mind it was just an exercise and not a finished work. The other thing is, if we know we are going to just delete it after it's over, we don't tend to get caught up in trying to make everything perfect, as that is counterproductive when just working out concepts usually. The idea is like a quick sketch for the most part, get it down, analyze it, toss it out and move on to the next one. If I didn't have that mindset I might have spent more time on the fireplace then I wanted to at this point in the larger scheme of things. I've noted it, and will come back to it at some point on the future, and will hopefully be much more efficient at figuring out exactly what I want to do to fix it at that point.

    Having said all of that, one thing I learned the hard way from cooking is that we won't remember the exact steps to doing something if we don't record it, so if we get something particularly good, especially by happy accident, we might want to take the time to do a detailed set of notes along with saving presets. :)

    Post edited by Joe Cotter on
  • DaremoK3DaremoK3 Posts: 539
    edited December 1969

    Great thread, Gedd...

    I love the rendering example of the Sacred Barge prop. You really show how understanding lighting setups, and changing parameters can really make the all the difference. I have learned a lot that I will try to incorporate into my rendering workflow.

    However, I do find one thing you mentioned to be erroneous. I like what you said in post #68 above, but the displacement mapping in DAZ Studio/3Delight is not dependent upon mesh polygon density.

    The 3Delight RenderMan compliant renderer utilizes micro-poly displacement which is pixel based opposed to polygon based. Mesh density will not be a factor, but such things as map size, map detail, map pixelation, and pixel-shader rate all come into play.

    Basically, 3Delight creates the necessary polygonal density needed for the displacement. There is no need for you, or model author to create dense topology for mesh displacement.

    *From 3Delight_dot_com:

    "Shading Rate'
    Controls how finely the geometry is tessellated prior to shading. 3DELIGHT adaptively tessellates each primitive into very tiny elements called micro-polygons, those elements are then shaded and sampled. Usually, the size of one such micro-polygon is one screen pixel and this corresponds to a shading rate of 1.0. A shading rate of 4.0 will produce micro-polygons that cover approximately an area of 2x2 pixels and a shading rate of 0.5 will give two micro-polygons per pixel. It is recommended to leave this parameter to its default value of 1.0 and reasons why not doing so are explained in Rendering Guidelines."


    Here is a quick simple demonstration of 3Delight's displacement mapping within DAZ Studio. Image one contains three planes with varying mesh topology (single quad, 144 quads, and 10,000 quads). The second image shows the details used for testing.

    In image one, can anyone determine which mesh is the single quad plane, the 144 quad plane, and the 10,000 quad plane?

    MicropolyDisp01.jpg
    1050 x 800 - 148K
    MicropolyDisp.jpg
    1050 x 700 - 101K
  • Joe CotterJoe Cotter Posts: 3,164
    edited April 2013

    Thank you for that clarification. I hadn't actually researched what micro displacement was yet, just heard the term :)

    In general, I often use displacement and bump, but I use bump for finer detail and displacement for the less fine detail where I want it to actually stand out. Basically, for me bump is good for things like small grittiness in a surface vs actual geometry translation. If a surface is much smoother then we want, we can create a noise layer to put in the bump channel for instance whereas if we are creating something like bolts, rivets, etc.. a displacement looks much better.

    Since I am not familiar with micro displacement, this might not be the most efficient work flow, just the one I have been using so far.

    Post edited by Joe Cotter on
  • Joe CotterJoe Cotter Posts: 3,164
    edited April 2013

    In this example, we start off with the out-of-the-box render of the item and notice the texture is pretty flat, almost a toon shader.

    01.jpg
    800 x 800 - 69K
    Post edited by Joe Cotter on
  • Joe CotterJoe Cotter Posts: 3,164
    edited April 2013

    Often, it can be helpful to lay down the base textures in flat lighting as more advanced lighting can throw us off sometimes in setting up the various textures and we may end up with a prop that is good for that one lighting situation but not others.

    02.jpg
    800 x 800 - 71K
    Post edited by Joe Cotter on
  • Joe CotterJoe Cotter Posts: 3,164
    edited April 2013

    After getting the base textures down, it's a good idea then to set up a basic lighting environment the prop would be used in. In this case, basic UE with an off-white/blue and a pale yellow sun. Doing various test renders while moving the sun around to different angles helps us work out any small issues with the textures.

    Final render time: 17.4s

    03.jpg
    800 x 800 - 73K
    Post edited by Joe Cotter on
  • Joe CotterJoe Cotter Posts: 3,164
    edited May 2013

    This example was challenging. The prop used was the Wine Celler. The initial promo looks promising and even the initial render doesn't look to bad to start off with other then the strange cross hatch on the stone wine rack. But there was some forewarning to potential issues. The initial render was 20 seconds for what we see here, and that is a bit long compared to the other examples we've been through so far for a start render in flat light, especially considering the amount of geometry is not as much as say The Study.

    01_20s.jpg
    800 x 800 - 331K
    Post edited by Joe Cotter on
  • Joe CotterJoe Cotter Posts: 3,164
    edited December 1969

    The first think we look at is, what's going on with the textures, as these are what are going to add render time in a case like this. It ends up... there are a number of things. For one, there are normal maps in the displacement channel and not in the normal map section. The bump channel is at 200%, there is an ambient on everything. Basically, for my purposes they need to be totally reworked.

    After going through and cleaning up everything but before lighting, we come up with this image. The render time is 2m3s which is still a bit long considering there's no lighting.

    The bump may seem a bit strong, but this is going to be a dark scene so much of it will wash out and I want the displacement to catch some of the light along with a low specular setting so the whole scene doesn't get washed out.

    02_2m3s.jpg
    800 x 800 - 451K
  • Joe CotterJoe Cotter Posts: 3,164
    edited December 1969

    The final scene with lighting. Render time was 48m34s.

    Note, this is with a much simpler light setup then The Study example along with much less geometry and it's about 25% longer in render time. This has an impact on tweaking the scene as much had to be tweaked with the final lighting in this case so it was long times between minor changes. The overall result is that it was much longer to create this base scene. Because of this, I hated to throw away the settings, but it was still an exercise and we don't want to get overly attached to something just because it took a lot of work as that's a sure recipe for failure.

    03_48m34s.jpg
    800 x 800 - 358K
  • Joe CotterJoe Cotter Posts: 3,164
    edited May 2013

    This example was with no post work and only using the texture maps that came with the product. It is a good challenge, and so I put it out there as a challenge to anyone who is up to it. No post work, no ambient except on lights in the scene, post result and render time/hardware :)

    Some notes. There are many things I would tweak with this but the render time discourages me from doing as much as I otherwise would. The bottles are basically wasted geometry in a dark image like this but if one looks close they can see the reflections of them.

    It did inspire me to consider an image with someone sitting over to the right of the image on the floor with a bottle and candle light. It would also require reframing of course... I would change the falloff on the torch to create a darker area that could have it's own pool of light from the candle that would have high specular to reflect off of the bottle. I would also change the texture on the taps of the kegs to be a reflective copper. That along with a good candle shader and good character/outfit could make in interesting image. But, that will have to wait for another day.

    Post edited by Joe Cotter on
  • SickleYieldSickleYield Posts: 7,288
    edited December 1969

    Was it that red pre-render or is the lighting red?

  • Joe CotterJoe Cotter Posts: 3,164
    edited May 2013

    The first image is exactly how it comes out just loading and rendering. The second, with flat lighting but textures adjusted, and the third with lighting added. I used UE with Candelight set at default level which is a dark light, a US on the flame with the flame texture loaded in the color channel along with a light yellow, and a Linear Point Light with an off-white yellow, falloff on the flame lights set to 80/250 with raytraced shadows. I would probably adjust the falloff on the point light down some if I redid it, but I also would reframe for a wider shot to give room for adding characters and that would create a darker area off to the side so would have to test. I also don't usually have the same falloff params for the US and point light, and would probably unsync them here. But yes, I was going for a warm light in this case.

    One thing to note, having opposing ambient lights and detail lights tends to add contrast to the scene. If subtle, the viewer doesn't even notice it, it just has more depth/detail. Using lights from the same general tonal range tends to flatten out the image, so one has to be careful of this. I actually added a light olive green reflection and specular value to the walls, floor, and ceiling to counteract that, around 10%. Also, the kegs have a very pale red specular and hint of pale red reflection, around 15%

    So yes, all of the colors were set up in the various shaders for spec/reflect etc.. in the second image, but one didn't see them until the 3D when the lighting brought them out.

    Oh, and all of the images were rendered. The only ones so far that haven't been are ones in wire frame.

    Good question :)

    [Edit] I misunderstood the question at first, thus the long winded reply, but I ended up answering it eventually. The un-rendered view of the 3d image did not show all of the red.

    Post edited by Joe Cotter on
  • Joe CotterJoe Cotter Posts: 3,164
    edited May 2013

    Something related to the last question. My goal is to get something in neutral lighting that is neutral but with features like specular, reflection, etc.. that will come out in various settings in a way that works. So, if I wanted to change the lighting to something more cool, the materials should be already set up to be able to do that also and look good hopefully, with only minor tweaking if necessary. That's the goal anyways, which is why I set up in neutral flat lighting first. That's also why I don't believe in taking shortcuts if I can avoid them like adding ambient to something that wouldn't naturally have ambient to it.

    All of the things I add to specular, reflective, lighting, etc.. are things that I envision in a somewhat rl situation, just working with the render engine to bring that out. Now I don't try to match reality with a biased or unbiased render engine, as even in photography we tweak things.. extend the color range, etc.. If we really wanted reality we'd take our smartphone out back and snap a picture of the alley, complete with washed out colors and no real point of focus ;)

    Post edited by Joe Cotter on
  • FuselingFuseling Posts: 230
    edited December 1969

    Wow! There's a lot of good information in this thread. :)

    Gedd, you made another good point in your last post which I wanted to highlight. Often in 3d art and design, we're not trying to replicate reality (although we will argue fiercely over which setup produces the most "realistic" results). What most of us are trying to create is something that looks better than reality. Skin is smoother, metals are shinier, reflections are cleaner, grime is grimier, and nothing smells. What's difficult is figuring out exactly how to create the kind of unreality you really want, and threads like this can help with that.

  • SickleYieldSickleYield Posts: 7,288
    edited May 2013

    I wasn't questioning the decision, just curious about the specifics. I thought it was probably a red light preset, but if you somehow turned all the mats red by turning off the ambient, I wanted to know that. ;)


    Ambient is definitely something to be used with care and only where it's really "relevant" in a surface.


    Edit: Crosspost with Fuse! I was replying to Gedd's last, not hers. And she has a point. This is something we have in common with professional photography. Our medium doesn't represent the real in the same way the eye does. The question is not of representing a visual truth, but of choosing the visual story we want to tell.

    Post edited by SickleYield on
  • Joe CotterJoe Cotter Posts: 3,164
    edited December 1969

    Oh yes.. I didn't take it that way Sickle. It's just that it triggered a thought. That often happens with me so if it seems I'm reacting to something directly it might just be that you inspired me to another thought ;)

  • Joe CotterJoe Cotter Posts: 3,164
    edited December 1969

    Gilikshe has a thread in keeping with the idea of this topic related to toon shading, so I thought I'd post a link for anyone who might not have seen it yet.

Sign In or Register to comment.