Pre-rendered: Retro Sci-fi
polycounter lvl 8
Hey guys. I decided to do this piece pre-rendered in Vray because of the time frame I was under.
I made this piece for my Gnomon Master Class. I did an overview of all the steps it takes to create a believable environment efficiently .
It was heavily based on sci-fi movies from the 70s,80s and 90s.
And it really turned out being an homage to some of my favorite movies . Some reference to some of the movies can be seen in the final image.
Check out my gallery for more images here:
http://malibub0b.cghub.com/images/
Website: www.dev-xyz.com
portfolio
I made this piece for my Gnomon Master Class. I did an overview of all the steps it takes to create a believable environment efficiently .
It was heavily based on sci-fi movies from the 70s,80s and 90s.
And it really turned out being an homage to some of my favorite movies . Some reference to some of the movies can be seen in the final image.
Check out my gallery for more images here:
http://malibub0b.cghub.com/images/
Website: www.dev-xyz.com
portfolio
Replies
great job on the feel/mood of the sc
artquest: please let me know if you get a chance to watch the master class, would love to know what you think! I am totally willing to answer any questions you may have :-)
I do have a few questions for ya if you have some time.
1) You said you didn't do too much subdivision modeling for this piece, only adding it where it was really needed. I know you were pressed for time on this piece so I wanted to ask, how often does that happen in production for cinematics?
2) Also, since this is rendered in vray what kind of steps would you have to take/have done differently to make this ready to go into a next gen game? You seem to already have been very conscious of texture sizes and such so I'm just curious how close the pipelines are coming together between film and games.
3) I really liked the vray render setup/texturing explanation as well. For this piece did you mostly use color swatches in V ray and just tweak numbers in the shader to get the look for most materials? How often did you use a texture for specular/gloss and diffuse?
4) Also, how good is Vray at handling a linear workflow? I hate how in mental ray I still need to add a gamma correct node to almost every color swatch that doesn't have a texture plugged into it even after they've supposedly added in the global setting for color management.
1) well being pressed for time is pretty typical in most cases. But more often then not things are still modeled with sub d. It makes it easier when things need to be displaced and when things need to be changed (you dont have a million bevels to deal with)
2)well as I said we tend to pay a lot more attention to things like texture size, polycount and draw calls for games then we do for pre-rendered stuff. But, if I was to do JUST this scene in something like UDK I wouldnt do a lot different. For such a small scene UDK can handle everything I did for this just fine.
some main points though would be:
-how does the engine handle lighting? If it bakes it into a set of uvs, well then I would have to spend the time to set up 2nd uv sets for it (which obviously I didnt have to do for vray)
-if it was for production, meaning this was one room in a much larger level: I would really pay attention to the amount of textures I used, and made sure I could reuse as many as I could. Same goes for props. I would also see where it made more sense to bake things to normal maps on prop details, as apposed to modeling in every little detail
3)Vray is amazing because you can really get great looking materials if you know what areas to tweak, even without textures. This scene had a lot of really clean materials to it and I knew that I was going to do a photoshop paint over at the end. Because of this I knew that I didnt have to spend a ton of time on some of the more clean type materials. Normally at work, for our game engine we use specular and gloss textures pretty much 100% of the time. For this scene I ALWAYS tweaked the settings, but probably used actual textures in the channels on less than half the materials.
4)Vray is all set up to use a liner workflow. I didnt actually use it for this scene, but I have on past personal projects and used it on every project we used vray for at Blizzard.
Vray has its own frame buffer, which is great for viewing your liner renders, and you do not have to add a gamma correct node on to every file like you do for mental ray. You do however have to add tag to the file node, its similar, but a lot easier to use.
do a search for "maya vray liner workflow" and you should find some much better info.
hope that all helps!
please if you think of anything else , do hit me up!
website: www.dev-xyz.com
portfolio
Really impressive design and rendering.
What's your key and fill like in the scene? Or is there some other light set up?
here are some examples of my light passes that I composited in photoshop:
"kitchen lights"
"top lights"
"side lights"
I "added" (linear dodge) these, plus a few more passes together and ended up with a nicely lit scene that I could adjust as needed.
Hope that answers your questions! Let me know if you guys want to pick my brain for anything else :-)
Thanks for taking the time to answer our questions here on PC! It's very helpful.
I've thought of a few more if you have some more time!
1) When it comes to subdivision modeling do you ever use creased edges with pixar's open subdiv instead of manually adding holding edge loops to every edge? (either on a personal project or production at blizzard cinematics.)
2) Are there ever times where a model(s) is/are just left in an un subdivided state for a far shot like heaven from the diablo 3 cinematic? How detailed are scenes like that usually modeled out?
3) Do you currently use any procedural textures? (Something like substance designer or even procedural nodes in maya to get quick variation in gloss or spec maps as a jumping off point.) And if you don't currently, do you see yourself moving in that direction in the future at all?
Thanks again!
1)I personally use edge loops to hold the edged. I very recently have been trying to use creasing more, I am hoping it will help speed things up as well as help keep my geo cleaner going in and out of zbrush (less polys on the edge) At Blizzard I didnt model much that went through the "modeling pipeline". In the mattepainting department we kind of had our own way of doing things. I THINK they mostly used edge loops to hold edges, but I am not 100% on that
2)That scene was made from a lot of modular pieces. I think most of those actually had "baked in subdivisions" , meaning they weren't being subdivided at render time. Again, that scene was actually ran though the mattepainting vray pipeline. We used lots of vray proxys and 3dsmax instances for that scene so we were ok with having it be very very heavy geometry wise. As far as other complex scenes, well it depends on each scene and each shot. The nice thing about using sub-d is that it will subdivide more for things that are closer to camera and less for things further away, so it usually isnt too much of an issues. especially for renderman, which is what was normally used.
3)I dont use a lot of procedural at home or at work. I used to use them more often but have now found that even doing quick photo-based textures seems to work just fine. I have started playing with substance designer and have been really impressed with it. I hope to spend more time with it soon!
Jeff Parrott: Yea, I try and keep my lighting as straight forward as possible. The more "cheaty and tricky" i find it gets more and more easy to mess up the overall realism
I agree, I went a little crazy with the color correction. Between cooling the shadow color and adding some chromatic aberration (which I usually try to stay away from, but since this was retro style, I added some) The color got away from me a bit.
thanks for the crit and the nice words :-)
I think it works well to convey the theme
Oh, a side note for using creased edges (you may already know this but I figured i'd share it here for anyone else who's interested.), I've started using a few shelf buttons to speed up getting a high poly model out of maya with creased edges. I use these lines of mel to select all the hard edges on an object and then just add them to a crease set in maya.
polySelectConstraint -m 3 -t 0x8000 -sm 1; // to get hard edges
and
polySelectConstraint -m 0; // turn off constraint
From there if you need to sculpt the model you can just use "convert smooth mesh preview to polygons and export to Zbrush. (I use reconstruct subdivisions to get my lowpoly back in zbrush.)
It really makes for a fast workflow instead of selecting all the edges you want to crease by hand. Especially because as you continue to edit your model the creased edges tend to break as you change the geo.
(sorry for the tangent :P I just recently discovered this workflow so I'm pretty excited about it
You seem to have a lot of different shapes where light is coming from. Did you use any sort of geometry for the light shapes? Or did you just create point lights and treat it like light bulbs placed in the scene?
also, do self illuminating materials cast light into the scene in Vray?
awesome, I will be trying that out for sure, thanks for the tip!
As far as the lighting goes, I did convert a few pieces of the geo into lights. Mainly the lights for the plant prop (as you can see up there) But mostly I had large flat lights, so I just put area lights in all those areas. Also while having global illumination turned on, anything that I had set to self illuminate, or to be "emissive" cast some light as well. Though self illumination doesn't cast as nice of shadows as actually using real lights.
Stirls: sounds like a plan!
wester: Thanks man, I really appreciate the kind words! I think the artistic skills to make interesting environments definitely carries over from film to games. You must keep in mind a lot of similar things. But the over all technical limitations and tool sets for each, (though similar is some regards, with lots of crossover) can be a challenge to juggle at first. A lot of stuff to learn, remember and keep learning so you dont fall behind!
Did you render out those specific stills with different lighting and then just overlay magic in PS? Or were they all rendered together and then just overlay PS magic for post processing?
As far as the lighting is concerned, yes I rendered the images out separately, then combined them in photoshop. You just "add" (screen or linear dodge) them together. I did most of the lighting with all the lights turned on in the beginning, but for the final image I broke them up and rendered them out as passes for more flexibility.
hope that makes sense!
Any chance they'll be released separately or that you'll do more tutorials / training independent of Gnomon? I was really hoping to see how you went about making this.
MooseCommander: (awesome name): That really sucks the timing got screwed! I had talked to Gnomon about releasing the content as a dvd , with more added content as well. I am not really sure if that is going to happen though. Feel free to contact the Gnomon workshop and ask about it, couldn't hurt! ;-)
I am also hoping to do some smaller tutorials in the future, but nothing concrete yet. If you want to PM me with your email, I can let you know when that stuff goes up.