Howdy, figure it's time to show off my big project for the last while, a nut and bolt replica of the Pennsey T1. With the utmost thanks to the T1 Trust, whom supplied the technical drawings in exchange for promotional renders for their cause.
So many details! It's really fantastic. I'm looking forward to seeing this textured. The biggest thing to watch out for is how well this will read when it's all textured up.
Again, too early to tell but from the close up of the wheel, it looks like the edges for the circular holes in the wheels (I know, very technical terminology here) might be too soft?
Comparing
to
But it could just be the angle we are seeing them from in this image.
I think that's a trick of the shading and a freshly cast, bare metal part that makes the 3d edges look softer and the real edges look sharper. Compare the textured wheel with 844's boxpok.
The design differs a little, but the basic radius of the hole rims chamfer seems to be the same
I like the more sedated, realistic lighting of overcast or otherwise meh weather/light. if it looks good and as it should under real world crap lighting, it'll look brilliant under god rays and uber refined lighting rigs.
It's divided up by assemblies basically. Driving gear as one, streamlining as one, frame/valves as one, boiler/steam pipe as one etc.
Painter handles a few million perfectly fine. It's vray converter doesn't understand what to do about reflection maps, but apart from that it's pretty much seamless
I see. And is the texturing - especially the mask generators - working without the normal/ao/curvature/etc. maps or do you generate those somehow?
I am thinking to texture few of my models what were made in Fusion 360. But I guess with those highpoly triangluated fbx the UV mapping will be pain in the... Did you use any auto UV solution what worked nicely or did you model with subd so you did uv the controlmesh only?
I just baked the mesh without a highpoly, it spat out enough maps to use the generators, with a little hand kneading/painting on top of that to get a sufficiently good result. My texturing is easily my weakest point, so the model itself does a majority of the work in the end.
As for the UV's, i just the UVing as i went along. Any time i made any repeating parts i made the one, mapped it and just duplicated it around the model. The bolts, wheels, rods, tanks, axle boxes, compressors etc. Or individual repeating parts in bigger parts, the piston packing/seal on the compressor being a good example. Ain't so bad if you chip away at it a little bit at a time
I'd say she's about done. Today I'm about to hit go on what's hopefully the first of the final turntable/pan shots, followed by the big expensive environment flyby animation.
Although if anyone has any qualms about the sheet metal, i'd love to hear them. I've spent a long time trying to get it right, and while i feel it works, there's something about it that could be better but i'm not sure what.
The weird seam on the frame there is from the mirrored UV map. To fix it i'd have to remap the frame which i'm kinda past doing.
Got about a minute and a bit of footage and most of the enviroment made. Organic stuff ain't my thing so stuff like trees, shrubs, grass etc is taking me a bit to figure out. She'll be 100% in time for the closing of the rookies though.
She's all done. The model's been done for over a year but Maya decided to make rendering as absolutely torturous as it could possibly be, and in future I'm going to absolutely refuse to use that god forsaken mess of a product.
But it's done now, and i'll post the final stuff with the reel, which is now waiting on my buddy to finish the texture for Stubby.
Late to the party here, and great work, but noticed the long pipes here have repetitive gloss maps. Especially evident in the banner image.
If you want a quick edit, you could simply skew the UVs, so the pattern is rotated gradually around the pipe. Old trick for mapping tree trunks.
Yeah i know. I didn't know much about high poly texturing at the time and i didn't know how bad of an idea that was going to be. If i was going to do it today i'd do a lot differently, that included.
I'm envious. I've always wanted to do a project where I had reference for all the pieces.
It's a pleasing thing. Though the documentation for anything PRR didn't build in house was spotty at best. I'm assuming the PRR drawings got to the Pennsylvania state archives since PRR was ultimately nationalized under conrail. But all the stuff from Westinghouse, Hancock, Franklin and even baldwin was not preserved at all, so i had to reproduce that from patent drawings and promotional stuff i dug up.
Really great job. Soo many mechanical parts and details. :-) For this kind of work if you have high res scans of sketches/blueprints I would consider to use SolidWorks or any other CAD sw from which you can generate these parts directly from 2D to 3D (I don't have experiences with that. I know only that some CAD have this feature). Can't say If it's precise enough.
Oh solidworks is great, i'm well versed in it. For a short stint i worked at a jewlery foundry, making patterns and cores for casting, using solidworks and CnCing those parts out in wood or 3d printing the oneoffs in wax for investment casting. Here's the valve head casting for a steam car i'm designing in my spare time, which i'm doing in solidworks
It's got double seat poppet valves related to the ones from the T1, and the 4 cylinder compound engine will drive the ring gear in the differential directly off the crank. One of these heads sits facing "out" at each end of the cylinder with the chain driven cambox in the center. They're designed to be modular so i can use the same casting for all 3 valve sizes. There's supposed to be a spring over the guide rods. But i wasn't sure how to do the weld for the faceplate, since the faceplate is supposed to be CnC'd. Might remove the bevel seal on the faceplate too, might not survive casting tolerances on the head.
I'm on the fence as to go with a vertical orientation for the head though, since it'd make the cam box mechanically simpler and as it is on the horizontal head, the live steam chamber is just a condensation trap. But the downsides of the vertical head is the actual casting for the cam box would be far more complicated, it'd need multiple seals between the valve tappet rockers and the outside air, the head casting would be complicated since you'd need to have the cylinder port where the face plate wants to be. And it'd make the motor just generally taller and bulkier, plus if even one valve breaks, you'd have to pull the whole cam box off the motor to take off one of the heads, as opposed to the horozontal type where you'd just have to pull the faceplate off of the head and pull the part in question. So bleh, maybe i'll stick with the condensation trap type.
But you can't really render that sort of thing, you can't really deform that sort of thing, and texturing an exported or even retopo'd bezier base model would be a bitch. And that wasn't the point of the asset/reel anyway, so there you go.
Well i actually intend to build the steam car so imma say reality is the best texturing suite of all. Procedural as all hell.
Gotta be careful with the random seed though, the majority of them just
break the false vacuum state or produce a wonky weak force which doesn't
allow chemical or elemental complexity.
Replies
I have no feedback to give other than this looks great so far. Looking forward to seeing the texture work!
Again, too early to tell but from the close up of the wheel, it looks like the edges for the circular holes in the wheels (I know, very technical terminology here) might be too soft?
Comparing
to
But it could just be the angle we are seeing them from in this image.
The design differs a little, but the basic radius of the hole rims chamfer seems to be the same
I like the more sedated, realistic lighting of overcast or otherwise meh weather/light. if it looks good and as it should under real world crap lighting, it'll look brilliant under god rays and uber refined lighting rigs.
You can also see the rig working away here. Albiet in a janky "hit record at 3am" kind of way.
Also started doing up the later sheet metal for the still shot. Pretty early on on that though
Crazy details!
How do you texture highpoly stuff in Substance Painter?
Painter handles a few million perfectly fine. It's vray converter doesn't understand what to do about reflection maps, but apart from that it's pretty much seamless
I am thinking to texture few of my models what were made in Fusion 360. But I guess with those highpoly triangluated fbx the UV mapping will be pain in the...
Did you use any auto UV solution what worked nicely or did you model with subd so you did uv the controlmesh only?
Thanks!
As for the UV's, i just the UVing as i went along. Any time i made any repeating parts i made the one, mapped it and just duplicated it around the model. The bolts, wheels, rods, tanks, axle boxes, compressors etc. Or individual repeating parts in bigger parts, the piston packing/seal on the compressor being a good example. Ain't so bad if you chip away at it a little bit at a time
I'd say she's about done. Today I'm about to hit go on what's hopefully the first of the final turntable/pan shots, followed by the big expensive environment flyby animation.
Although if anyone has any qualms about the sheet metal, i'd love to hear them. I've spent a long time trying to get it right, and while i feel it works, there's something about it that could be better but i'm not sure what.
The weird seam on the frame there is from the mirrored UV map. To fix it i'd have to remap the frame which i'm kinda past doing.
Got about a minute and a bit of footage and most of the enviroment made. Organic stuff ain't my thing so stuff like trees, shrubs, grass etc is taking me a bit to figure out. She'll be 100% in time for the closing of the rookies though.
It's happening.
She's all done. The model's been done for over a year but Maya decided to make rendering as absolutely torturous as it could possibly be, and in future I'm going to absolutely refuse to use that god forsaken mess of a product.
But it's done now, and i'll post the final stuff with the reel, which is now waiting on my buddy to finish the texture for Stubby.
It's bizzare to think it took longer to render than it did to make the bloody thing.
It's a pleasing thing. Though the documentation for anything PRR didn't build in house was spotty at best. I'm assuming the PRR drawings got to the Pennsylvania state archives since PRR was ultimately nationalized under conrail. But all the stuff from Westinghouse, Hancock, Franklin and even baldwin was not preserved at all, so i had to reproduce that from patent drawings and promotional stuff i dug up.
It's got double seat poppet valves related to the ones from the T1, and the 4 cylinder compound engine will drive the ring gear in the differential directly off the crank. One of these heads sits facing "out" at each end of the cylinder with the chain driven cambox in the center. They're designed to be modular so i can use the same casting for all 3 valve sizes. There's supposed to be a spring over the guide rods. But i wasn't sure how to do the weld for the faceplate, since the faceplate is supposed to be CnC'd. Might remove the bevel seal on the faceplate too, might not survive casting tolerances on the head.
I'm on the fence as to go with a vertical orientation for the head though, since it'd make the cam box mechanically simpler and as it is on the horizontal head, the live steam chamber is just a condensation trap. But the downsides of the vertical head is the actual casting for the cam box would be far more complicated, it'd need multiple seals between the valve tappet rockers and the outside air, the head casting would be complicated since you'd need to have the cylinder port where the face plate wants to be. And it'd make the motor just generally taller and bulkier, plus if even one valve breaks, you'd have to pull the whole cam box off the motor to take off one of the heads, as opposed to the horozontal type where you'd just have to pull the faceplate off of the head and pull the part in question. So bleh, maybe i'll stick with the condensation trap type.
But you can't really render that sort of thing, you can't really deform that sort of thing, and texturing an exported or even retopo'd bezier base model would be a bitch. And that wasn't the point of the asset/reel anyway, so there you go.
Gotta be careful with the random seed though, the majority of them just break the false vacuum state or produce a wonky weak force which doesn't allow chemical or elemental complexity.