|
Megaspel posted:How far away is 3D scanning technology from being able to scan a model at 25 frames per second and store it for later playback? I've seen a lot of stuff with Kinect scan streaming, but that's still a bit poo poo at the moment. I don't know why we've not really seen any of that though. Still a ways. Even a quick, but high quality, scan takes at least an hour of processing time, and that's just for a single, non-moving object. Eventually things will be scanned with video, but most likely not for a few years at the very earliest. Photoscanning is still getting off the ground.
|
#
?
May 31, 2016 23:29
|
|
|
|
| # ? May 31, 2024 09:55 |
|
|
Back on VR talk, we just shipped delivered and got the final cheque for our 6th VR job. static 3d images. All of our clients always say 'i wish i could move around' - and then we invite them to our office to use the rift so they can move around and they go 'gently caress this big heap of tethered poo poo, our buyers will never put this on' and they go for the gear vr option. Our vive arrived today too, that was 7 days from order to delivery!
|
|
#
?
Jun 1, 2016 00:19
|
|
|
 So I made a hat. Because of course I did. What else do you use cone primitives for? Literally nothing. Ever. Rename it to hat primitive. Anyway. Is there an easy way to take a vertex, like the point of the cone, and "round" it off with extra faces? I tried using bevel and it just ended up folding itself up. I could cut it with the knife tool, delete the part I don't want and insert a face too, but that's too complicated. And Halfway through typing this I've realized I could just start with a cylinder and do the exact same thing and end with a flat end I can bevel...
|
|
#
?
Jun 1, 2016 01:41
|
|
|
Yep, avoid cone, use cylinder. I'll post a capping example when I get to my computer.
|
|
#
?
Jun 1, 2016 01:51
|
|
|
Insert 2 edge loops near the end, shrink one, select the last vertex and dissolve it.
|
|
#
?
Jun 1, 2016 02:23
|
|
|
Megaspel posted:How far away is 3D scanning technology from being able to scan a model at 25 frames per second and store it for later playback? I've seen a lot of stuff with Kinect scan streaming, but that's still a bit poo poo at the moment. I don't know why we've not really seen any of that though. Extremely near. It's not quite mature yet but it will be commonplace in 2 years. Really high fidelity models of course will be a little further out, but it's nearer than most people realize. (this is a field I'm currently working in, albeit somewhat tangentially)
|
|
#
?
Jun 1, 2016 03:17
|
|
|
What kind of office chair do you guys use? My neck is loving killing me. edit: And yeah I know there's a lot of other factors that can contribute to this, so what are y'alls healthy computering habits? BonoMan fucked around with this message at 04:38 on Jun 1, 2016 |
|
#
?
Jun 1, 2016 04:22
|
|
|
Steelcase and Aeron chairs are the way to go. You can find either of those used for $300-$400 when normally they're $900+ ...a lot of companies will buy a ton of them and then go out of business so usually there's warehouses full of them to sell in bulk or to individuals that hit them up. Check craigslist. I've bought 2 $900 chairs (one for work one for home) for $400 a piece and it's definitely worth it.
|
|
|
#
?
Jun 1, 2016 04:51
|
|
|
Crain posted:
The simplest option:  If it's not clear, I took the point of a cone and deleted every other edge around the circumfrance. I did this twice on the way up to the point to encourage smooth rounding, as an intersection of more than ~5 edges into one vertex or so starts to REALLY get messy even in smoothing. It helps if your number of side edges is divisible by 4 or 5 (depending on how many edges you want converging at the final vertex. Remember to keep the number of edges/sides in check! The fewer the sides, the more control you have. Only push the polycount at the end, and even then only push it as far as necessary. Edit: People are gonna call out my n-gons there on the 2nd row from the top, but most modern packages will smooth that just fine. There are fancier ways to terminate edges into perfect quad geometry, but this is the "quick and dirty" method. mutata fucked around with this message at 05:03 on Jun 1, 2016 |
|
#
?
Jun 1, 2016 04:55
|
|
|
mutata posted:The simplest option: Is there a setting for determining the number of starting faces for primitives like cylinders and cones? I know I gotta keep things simple but the cone started out with so many sides and trying to merge them initially was just a huge hassle.
|
|
#
?
Jun 1, 2016 05:00
|
|
|
Crain posted:Is there a setting for determining the number of starting faces for primitives like cylinders and cones? I know I gotta keep things simple but the cone started out with so many sides and trying to merge them initially was just a huge hassle. There absolutely is in any 3d package. Google it for whatever program you're using! 
|
|
#
?
Jun 1, 2016 05:02
|
|
|
BonoMan posted:What kind of office chair do you guys use? My neck is loving killing me. I got an Ikea Skarsta standing desk and made a little stand to raise my monitors another foot or so cos I'm tall. I freaking love it, standing at work feels great.
|
|
#
?
Jun 1, 2016 06:19
|
|
|
I wish more companies were like Allegorithmic. I'll always gladly drop money on software I believe is reasonably priced. The monthly payments adding up to a full buyout of the software is awesome, plus they're pushing updates fairly frequently. If it keeps up this will definitely be a standard in my workflow. It's already replaced 3D-Coat, xNormal, and Photoshop for me which is a godsend because going between all those programs blows. I know there's a The Foundry dude here, and they're doing good things for individual users as well which is nice. I'm just not a Modo or Mari user at the moment. And last time I used Modo it was missing some pretty essential game art tools which I'm sure has been mostly fixed by now. I typically will use whatever software our rigging and animation department is using and currently thats Maya. I absolutely loving hate Maya's pricing scheme now though. Speaking of Modo. Can anybody here who has used Modo and Max and/or Maya tell me how it is with working in UE3/UE4? One of the biggest issues at our studio is how different the FBX exporting is between Maya and Max. And just how shittily Max handles normal smoothing and triangulation. We have people who use Maya or Max or Modo and it's a real pain in the dick getting people to have consistent results. We're switching our engines over to Mikkt space which fixes the normals issue but triangulation on FBX export is still inconsistent between Max and Maya. ceebee fucked around with this message at 06:45 on Jun 1, 2016 |
|
|
#
?
Jun 1, 2016 06:40
|
|
|
EoinCannon posted:I got an Ikea Skarsta standing desk and made a little stand to raise my monitors another foot or so cos I'm tall. I made a standing desk for home out of Ikea shelves that I hacked together so that I have all sorts of cubbyholes and it's great. Standing on its own isn't apparently that great a health benefit but I find I'm more likely to walk away from my desk every 30 minutes or so if I'm standing, which is where I get the benefits from moving. If you can't find a stand to raise your monitor to eye level, put it on a stack of books. I have a steelcase Leap chair that I got for under $400 and it's quite excellent, you want to look for office surplus stores- these are the places that specifically deal in remaindered furniture from when businesses go under. Listerine fucked around with this message at 08:41 on Jun 1, 2016 |
|
#
?
Jun 1, 2016 08:39
|
|
|
speaking of 3D scanning what's the go-to next step up device after the Kinect in terms of price point/quality?
|
|
#
?
Jun 1, 2016 09:01
|
|
|
I saw a presentation from a company that was doing virtual presence teleconferencing. It would scan the 2 people in separate places and project 3d versions of them in VR. The scenario was a dad in one room playing with his daughter who was somewhere else. Edit: Not VR, AR via hololens. They call it "Holoportation" or some such. They use sensors to build and transmit a mesh in real time. Probably the closest thing to what people in here have been talking about. https://youtu.be/7d59O6cfaM0 mutata fucked around with this message at 10:10 on Jun 1, 2016 |
|
#
?
Jun 1, 2016 10:06
|
|
|
Crain posted:Is there a setting for determining the number of starting faces for primitives like cylinders and cones? I know I gotta keep things simple but the cone started out with so many sides and trying to merge them initially was just a huge hassle. If you're using blender, press T to bring up the left side window in the 3d viewport (or find the little + and press that) then add your new shape, you should see something like this down the bottom, which will let you set its initial values. 
|
|
#
?
Jun 1, 2016 13:02
|
|
|
Alan Smithee posted:speaking of 3D scanning what's the go-to next step up device after the Kinect in terms of price point/quality? DSLR with a cheap turntable.
|
|
#
?
Jun 1, 2016 13:40
|
|
|
Taffer posted:Extremely near. It's not quite mature yet but it will be commonplace in 2 years. I'd be interested in any leads on video scanning you might have. I've seen some decent normals from video scans but meshes have been quite poor. So far my tests with hand held scanners and "no processing" scanners have been pretty disappointing and nowhere near the results from photogrammetry. Would really like to see some video scanning solutions that come close to it.
|
|
#
?
Jun 1, 2016 14:12
|
|
|
Taffer posted:Extremely near. It's not quite mature yet but it will be commonplace in 2 years. I've seen some really cool stuff like this where they use a 3D scanner to project skin physics onto a standard biped rig which can be deformed to fit the actor and all sorts. If markerless mocap technology improves I could see some cool hybrid stuff coming about, though being able to straight up 3D record would be easier, if a bit more resource intensive maybe. Anyway, I'm looking forward to whatever comes out, make sure you do a good job you. Gearman posted:DSLR with a cheap turntable. I want to design some custom prosthetic stuff, it's a lot more difficult to stick a person on a turntable and make sure they don't blink or breathe. Is there any affordable options for body scanning? I saw some of these weird 3D scanning rectangles a while back which was just like 4 cameras and maybe a 6dof tracker inside? Dunno about the quality. mutata posted:I saw a presentation from a company that was doing virtual presence teleconferencing. It would scan the 2 people in separate places and project 3d versions of them in VR. The scenario was a dad in one room playing with his daughter who was somewhere else. I've seen an earlier vid of this but I guess they added the feature to murder your daughter by Doctor Manhattening her. curse of flubber fucked around with this message at 15:54 on Jun 1, 2016 |
|
#
?
Jun 1, 2016 15:49
|
|
|
Gearman posted:I'd be interested in any leads on video scanning you might have. I've seen some decent normals from video scans but meshes have been quite poor. So far my tests with hand held scanners and "no processing" scanners have been pretty disappointing and nowhere near the results from photogrammetry. Would really like to see some video scanning solutions that come close to it. The simple answer is machine vision is exploding. It's harder to get data out of video, but if you're able to get it it allows you a wealth of further information that can be used to refine and clarify everything. This is going to be magnified a ton as soon as dual aperture phones come out (one is coming out this month, and several more are on their way, notably the next iPhone.) Dual aperture provides a lot, from simple things like adjustable DoF after the fact (that doesn't look awful) to really detailed 3D analysis of a scene. You can use that for something as simple as placing a 3D digital object accurately on a table, all the way to doing really detailed mesh building real-time. Google is investing heavily in providing a foundation for this technology with Project Tango, which finally debuted for real at IO a couple weeks ago, and a thousand other players are working in the various niches of this technology already for everything from AR video games to marketing to home improvement. Basically specialized hardware is going away because it sucks, and will be replaced by mass produced components on every phone, just like so many technologies have in the past.
|
|
#
?
Jun 1, 2016 16:44
|
|
|
I had an idea a while ago to get higher resolution data from something like a kinect which has a 1080p camera as well as the depth cameras. Tell me if this is a complete waste of time or not, but could you not use the 1080p camera, find the edges, then if there's one big zpixel stretching over multiple pixels from the 1080p camera, you split it up and pixels on the outside of the line go to the nearest zdepth of the adjacent pixels, and the others do the same on the pixels nearer the front or whatever. It's not perfect, but it seems like it might improve the quality of some stuff. I'm thinking more of using a depth data for compositing, would save a lot of rotoscoping time. If you could just stick a cheap kinect or something onto a camera then refine the depth map you create from it, you could have a really efficient compositing workflow.
|
|
#
?
Jun 1, 2016 17:07
|
|
|
ceebee posted:I wish more companies were like Allegorithmic. I'll always gladly drop money on software I believe is reasonably priced. The monthly payments adding up to a full buyout of the software is awesome, plus they're pushing updates fairly frequently. If it keeps up this will definitely be a standard in my workflow. It's already replaced 3D-Coat, xNormal, and Photoshop for me which is a godsend because going between all those programs blows. I don't have answers for your questions, but I'm 100% in agreement with this. The substance suite has really done wonders for my hobbyist workflow, and knowing my monthly payments are working towards ownership is a weight off my shoulders. If this was a full time job I probably wouldn't care as much about payments, but as is there can be long stretches of not using whatever software when I'm focusing on a different area, and paying monthly with no end goal adds an extra layer of stress, a feeling that I'm wasting my money if I'm not using it constantly. I have to say though, the one huge goddamn benefit of all this modern subscription stuff is no more dongles. I have a drawer with a handful of various audio and gfx software dongles from over the years, and I don't miss playing switcheroo whenever I needed to switch workflows.
|
|
#
?
Jun 1, 2016 18:47
|
|
|
I have a model in Max with a UVW unwrap. When I apply a material the model is just taking in one color from the diffuse map and making it a solid color throughout the entire model. Any ideas what might cause this issue? I can take some screenshots if this is not clear enough.
|
|
#
?
Jun 1, 2016 18:55
|
|
|
Chernabog posted:I have a model in Max with a UVW unwrap. When I apply a material the model is just taking in one color from the diffuse map and making it a solid color throughout the entire model. Love to see the UVs, that sounds weird. Also perhaps the material properties rollout?
|
|
#
?
Jun 1, 2016 19:01
|
|
|
Sounds like an unwanted extra uv set with uvs crushed to 1 pixel?
|
|
#
?
Jun 1, 2016 19:04
|
|
|
KiddieGrinder posted:Love to see the UVs, that sounds weird. Also perhaps the material properties rollout?  I don't know why the textures are on repeat either, I haven't used max in a while but I don't remember them looking like that. mutata posted:Sounds like an unwanted extra uv set with uvs crushed to 1 pixel? That could be it but I don't know where that would be.
|
|
#
?
Jun 1, 2016 19:22
|
|
|
This is dumb non-artist troubleshooting but you might be able to swap your texture for a color gradient, which might tell you where the single pixel is, at least.
|
|
#
?
Jun 1, 2016 19:31
|
|
|
Chernabog posted:
Can you screenshot after you click on bolsasan_DIFF in the material editor?
|
|
#
?
Jun 1, 2016 19:34
|
|
|
The crushed pixel appears to be 0,0 And here's the other image. http://i.imgur.com/ugK1Qhp.png
|
|
#
?
Jun 1, 2016 19:39
|
|
|
Only other odd thing is the UV-Unwrap modifier seems to be working on channel 3, but your material is set to 1. Check the UVs for channel 1 and I bet theyre hosed up.
|
|
#
?
Jun 1, 2016 19:52
|
|
|
Yeah, I tried changing those but it's not working. I also removed all the UVWs and made a new one but it is still hosed.
|
|
#
?
Jun 1, 2016 20:44
|
|
|
I recently finished this spaceship I made as a exercise to learn substance painter and finally make the jump to modo as a modeling package. I'm really liking my new pipeline and the substance tools are amazingly useful. Just wanted to show it off and get comments so I can improve.  I put it up on sketchfab so you can check it out in 3d as well here: Cruiser on Sketchfab
|
|
#
?
Jun 1, 2016 21:04
|
|
|
Barcley posted:I recently finished this spaceship I made as a exercise to learn substance painter and finally make the jump to modo as a modeling package. I'm really liking my new pipeline and the substance tools are amazingly useful. Badass man, looks super busy and sharp. Hit it with a single source of light like a star in space, see how light rakes across it.
|
|
#
?
Jun 1, 2016 21:15
|
|
|
Barcley posted:I recently finished this spaceship I made as a exercise to learn substance painter and finally make the jump to modo as a modeling package. I'm really liking my new pipeline and the substance tools are amazingly useful. Looking good. As it stands right now though I think it lacks the correct sense of scale. I think lots of smaller things like antennae, sensors and other detail on the outside will give it a good scale. Also, that camera angle ain't doing it any favors! You're making it look like a sci-fi gun instead of a ship!
|
|
#
?
Jun 1, 2016 21:29
|
|
|
Chernabog posted:Yeah, I tried changing those but it's not working. I also removed all the UVWs and made a new one but it is still hosed. I think then it's your material; try unticking Use Real-World Scale, or adjust the size so it's not so big. I tried messing with one of my models and I changed it to use real-world scale (I never use that) and changed it to something large like yours and it does practically turn a solid color. Un-tick Real-World Scale, change it to tiling 1 x 1, and it should work.
|
|
#
?
Jun 1, 2016 22:12
|
|
|
Taffer posted:The simple answer is machine vision is exploding. It's harder to get data out of video, but if you're able to get it it allows you a wealth of further information that can be used to refine and clarify everything. This is going to be magnified a ton as soon as dual aperture phones come out (one is coming out this month, and several more are on their way, notably the next iPhone.) Funny you mention the Tango, I'm actually quite familiar with it, as Johnny Lee demo'ed my team's Tango app on stage at I/O this year (https://m.youtube.com/watch?v=yvgPrZNp4So&t=1030). Having been tinkering with the Tango for some time, the depth sensors are pretty good, and in general are "good enough" for placing things in a room and constructing a rudimentary 3D model of a space. But the resolution and quality is still pretty poor. The scanners that I've used, aimed at consumers and with little or no processing time, have all produced really poor and low quality 3D meshes. Photogrammetry is still much, much better than anything else right now, by a large margin. Video capture will eventually take over photogrammetry, but I think that's still three-to-five years away at the very earliest.
|
|
#
?
Jun 1, 2016 23:30
|
|
|
Megaspel posted:I've seen some really cool stuff like this where they use a 3D scanner to project skin physics onto a standard biped rig which can be deformed to fit the actor and all sorts. If markerless mocap technology improves I could see some cool hybrid stuff coming about, though being able to straight up 3D record would be easier, if a bit more resource intensive maybe. An array of raspberry Pi's will probably be your best option then, much like this: http://www.3ders.org/articles/20160208-student-designs-curatio-3d-hand-scanner-equipped-with-32x-raspberry-pi-cameras.html For more accurate results, you could use 6-8 point-and-shoot cameras or low-end DSLRs.
|
|
#
?
Jun 1, 2016 23:34
|
|
|
KiddieGrinder posted:I think then it's your material; try unticking Use Real-World Scale, or adjust the size so it's not so big. I tried messing with one of my models and I changed it to use real-world scale (I never use that) and changed it to something large like yours and it does practically turn a solid color. Oh sweet. That worked! Thank you.
|
|
#
?
Jun 2, 2016 01:43
|
|
|
|
| # ? May 31, 2024 09:55 |
|
|
Taffer posted:Extremely near. It's not quite mature yet but it will be commonplace in 2 years. https://www.youtube.com/watch?v=dQBJ0r5Pj5s 2:30
|
|
#
?
Jun 2, 2016 02:02
|
|