I'm not sure what you'd be comparing - the angle and environment matches a shot from Star Wars, but this R2's materials don't match any particular R2 dome/body set - not that there is one that's consistent for more than a couple cuts anyway. The level of dirt/polish changed radically during filming as the robots were cleaned, dirtied, repaired, even repainted constantly.
Thanks, everyone... and Chris, yes of course anyone can learn this. But I've said many times that the geometry is at least 60% of the thing. One thing I can't understand is why so few people embrace the asymmetry/randomness/humanity aspect of all real-world objects when they're modeling. Every real ship, and practical model, is just a giant assortment of wonkiness, and yet here, 99% of the models are mathematically precise, symmetrical, have perfect angles and surfaces. Nothing in the real world has this, and it immediately cripples your efforts.
Nothing on this R2 model is straight, or even, or symmetrical. It's basically a big pile of imprecision - just like the real ones. That makes all the difference in the world. I modeled him "perfect" and then spent a lot of time pulling all the panels out of whack, misaligning things, bending and distorting things, putting in bumps and scratches, twisting geometry, and more. THEN the texture detail comes in, but I almost never use (or have to use) bump maps, for example.
One thing I can't understand is why so few people embrace the asymmetry/randomness/humanity aspect of all real-world objects when they're modeling. Every real ship, and practical model, is just a giant assortment of wonkiness, and yet here, 99% of the models are mathematically precise, symmetrical, have perfect angles and surfaces.
It's easier to build half of a model, then mirror it. And depending on how you build objects, the "snap to grid" feature can be handy for placing primitives and drawing splines. That's the best explanation I can think of.
I'm self-taught, so I don't know if teachers stress that sort of thing or not.
I'm self-taught, too, and when I say "I don't understand," I don't mean that I don't understand why we do things like mirroring/snapping, etc., I just mean that if our goal is to make things look real, then our first job is to be aware of what really makes things look real in the first place. All the subtle cues in lighting, yes; the nature of the materials, yes; and the random imperfections inherent in every object, yes - those too!
I'm self-taught, too, and when I say "I don't understand," I don't mean that I don't understand why we do things like mirroring/snapping, etc., I just mean that if our goal is to make things look real, then our first job is to be aware of what really makes things look real in the first place. All the subtle cues in lighting, yes; the nature of the materials, yes; and the random imperfections inherent in every object, yes - those too!
One thing I can't understand is why so few people embrace the asymmetry/randomness/humanity aspect of all real-world objects when they're modeling. Every real ship, and practical model, is just a giant assortment of wonkiness, and yet here, 99% of the models are mathematically precise, symmetrical, have perfect angles and surfaces. Nothing in the real world has this, and it immediately cripples your efforts.
_Mike
Well said. That approach is something I've only recently started to adopt in my work. These machines of ours give us an extraordinary power to be as precise (or imprecise) as we have time to be. Truth is, there's a reason why hand-built models and physical VFX look better than some CGI...it's because there's a certain element of "error" that is contained within the work. That's not to say that CGI is a terrible thing, but keeping human error and the "fudge factor" in mind would do wonders for this industry.
I just bought a new rig - a 12-core Xeon system with 24GB of RAM. Finally enough power to do animations. I had never actually seen my R2 move before, and as soon as I started doing turntables, I saw things in the materials that weren't quite right - things you just can't tell until you see light and reflection play across a surface in motion. Little things, sure, but still important. Here's a leg post-tweak, and I'll put up some of the turnarounds soon.
I just bought a new rig - a 12-core Xeon system with 24GB of RAM.
*hurrk*
My god, how much did this rig RUN?!
And here I am, sitting in front of my single-core Celeron (more than 5 years old) with 640 MEGABYTES(!) of RAM. I feel like a shaky rat terrier at a bulldog convention . . . it's a wonder I can do ANYTHING with what I've got.:rolleyes:
12 core xeon?... Think you just explained why the hobby modellers here don't put imperfections in their models. Last one I did which was anywhere near finished was so badly made, it took 15 hours to do 150 frames... and that was without any special fx...
I just bought a new rig - a 12-core Xeon system with 24GB of RAM. Finally enough power to do animations. I had never actually seen my R2 move before, and as soon as I started doing turntables, I saw things in the materials that weren't quite right - things you just can't tell until you see light and reflection play across a surface in motion. Little things, sure, but still important. Here's a leg post-tweak, and I'll put up some of the turnarounds soon.
_Mike
Holy cow. I would never have believed that was digital - but again, its the imperfections which bring him alive!
Sorry if you answered at this Q before but: what did you used for unwrap this model? You've started to model in Rhino right? Then, you've exported the model to Maya in OBJ format, when I do that the model is really dense and the UV mapping/ UnWapping is a nightmare, how did you achieved this great result? Anyway anytime I see your model I can't take my eyes off of it
The last render is P E R F E C T, a photoreal render!!!
Almost without exception, once the model was in Maya, I re-projected UV's in various forms (cylindrical, planar, etc.) and then did a lot of manual moving of UV's to maximize normalized space and provide good layout for painting the maps. That can be more art than science, and takes some trial and error per part, but then once I had the UV's laid out the way I wanted them, I just did a UV Snapshot, and used the lines as guides for where to paint in Photoshop. In many cases, say with the feet, each face is individually projected, and then you just have to be very careful in Photoshop to create maps where the seams are seamless. The trick comes in places like the rounded edges on the feet: where is the division between front face and side face? Well, in most cases, I literally split the difference, and then could hide the division by clever copying/cloning in the map. For the feet, one might consider doing a cylindrical projection for the sides/font, and then a top projection for the top, but the distortion created by the cylindrical projection is just too annoying and severe to try and compensate for. Plus, my feet's geometry already matched the reference photos, so if I could do planar projections, I was already assured perfect alignment with the photos without much need for manual adjustment of UV's.
Almost without exception, once the model was in Maya, I re-projected UV's in various forms (cylindrical, planar, etc.) and then did a lot of manual moving of UV's to maximize normalized space and provide good layout for painting the maps. That can be more art than science, and takes some trial and error per part, but then once I had the UV's laid out the way I wanted them, I just did a UV Snapshot, and used the lines as guides for where to paint in Photoshop. In many cases, say with the feet, each face is individually projected, and then you just have to be very careful in Photoshop to create maps where the seams are seamless. The trick comes in places like the rounded edges on the feet: where is the division between front face and side face? Well, in most cases, I literally split the difference, and then could hide the division by clever copying/cloning in the map. For the feet, one might consider doing a cylindrical projection for the sides/font, and then a top projection for the top, but the distortion created by the cylindrical projection is just too annoying and severe to try and compensate for. Plus, my feet's geometry already matched the reference photos, so if I could do planar projections, I was already assured perfect alignment with the photos without much need for manual adjustment of UV's.
_Mike
Thanks for the tips, Mike, as I've supposed, there's a great work behind the texturing of this kind of model, your texures are seamless and fused perfectly, my compliments!!!
Posts
_Mike
_Mike
Nothing on this R2 model is straight, or even, or symmetrical. It's basically a big pile of imprecision - just like the real ones. That makes all the difference in the world. I modeled him "perfect" and then spent a lot of time pulling all the panels out of whack, misaligning things, bending and distorting things, putting in bumps and scratches, twisting geometry, and more. THEN the texture detail comes in, but I almost never use (or have to use) bump maps, for example.
_Mike
It's easier to build half of a model, then mirror it. And depending on how you build objects, the "snap to grid" feature can be handy for placing primitives and drawing splines. That's the best explanation I can think of.
I'm self-taught, so I don't know if teachers stress that sort of thing or not.
_Mike
viva la imperfection!
Well said. That approach is something I've only recently started to adopt in my work. These machines of ours give us an extraordinary power to be as precise (or imprecise) as we have time to be. Truth is, there's a reason why hand-built models and physical VFX look better than some CGI...it's because there's a certain element of "error" that is contained within the work. That's not to say that CGI is a terrible thing, but keeping human error and the "fudge factor" in mind would do wonders for this industry.
I just bought a new rig - a 12-core Xeon system with 24GB of RAM. Finally enough power to do animations. I had never actually seen my R2 move before, and as soon as I started doing turntables, I saw things in the materials that weren't quite right - things you just can't tell until you see light and reflection play across a surface in motion. Little things, sure, but still important. Here's a leg post-tweak, and I'll put up some of the turnarounds soon.
_Mike
the render looks great!
Chris
*hurrk*
My god, how much did this rig RUN?!
And here I am, sitting in front of my single-core Celeron (more than 5 years old) with 640 MEGABYTES(!) of RAM. I feel like a shaky rat terrier at a bulldog convention . . . it's a wonder I can do ANYTHING with what I've got.:rolleyes:
Love the Artoo renders. Keepin' it "real"! LOL
_Mike
Stunning model.
_Mike
Sweet looking leg.
Need to order me one of them there 12 cores
Jas
R2 MAYHEM Trilogy Pics > Post #570
Holy cow. I would never have believed that was digital - but again, its the imperfections which bring him alive!
btw, congrats mike!
you're insane - as usual. Lovin' it!
rofl
The last render is P E R F E C T, a photoreal render!!!
Bye
_Mike
Thanks for the tips, Mike, as I've supposed, there's a great work behind the texturing of this kind of model, your texures are seamless and fused perfectly, my compliments!!!
Bye
Episode 1 - Into the Future
Link To Full Size Pic
Episode 2 - Restless Spirits
Link To Full Size Pic
Episode 3 - Tortured Souls
Link To Full Size Pic
The Saga Concluded !! ........ for now ...
Look forward to seeing those turntables Mike ...
Jas