|
Obliterati posted:So moving away from all these sweet game assets for a minute, I am trying to create a 3D image of a rock. Specifically, of this rock: For one thing, as far as I know, rotating the subject is a no-no. You really should try to walk around it yourself, I think the changing background helps the program identify what's the actual subject and what isn't. Secondly, I have no idea what else might help, I'm sure sigma will be able to answer better.
|
#
?
Aug 6, 2015 19:04
|
|
|
|
| # ? May 29, 2024 04:04 |
|
|
Obliterati posted:So moving away from all these sweet game assets for a minute, I am trying to create a 3D image of a rock. Specifically, of this rock: A few things: In an ideal world, you'd rotate the object a number of times (either 8 or 16 turns) taking photos from at least three different heights for each turn. You'd then flip the object over, and do the same thing (you can buy a very cheap turntable from a hardware store for like $5 if you don't want to build a lazy susan). Afterwards, you'd separate out those two photo sets, mask out the background from each image, and process them as separate chunks in Agisoft. If there are enough matching points of interest, you can even attempt to have Agisoft align both chunks and merge them in to a single mesh. If that fails, you can save out both meshes, align them in something like MeshLab, and then condense them down to a single mesh with two UV sets. If you have some working knowledge of 3D packages, you can go one step further and bake that down to a single UV and texture sheet. You can sometimes get away with scanning the top and bottom half of an object, processing both sets together, and getting a complete mesh from Agisoft, but it can be hit-or-miss. It's usually better to process as two separate chunks, align them, and condense down to a single mesh.
|
|
#
?
Aug 6, 2015 19:46
|
|
|
Wont that radial point circle underneath it be detrimental for trying to determine facings for the photos? It might be lacking in reference points.
|
|
#
?
Aug 7, 2015 05:47
|
|
|
It can, but if there are enough unique attributes in the rock, then Agisoft should align the photos just fine. It's important to check the alignment of the cameras in the alignment stage before moving on anyway, so you'd know pretty quickly if Agisoft is having trouble aligning them. If you're not seeing nearly-perfect circles of cameras, then you'll probably need to mask out everything around the subject. I also recommend putting objects on a little pedestal, or even a paper coffee cup, just to help reduce occlusion on the bottom. It really helps ensure that you get nice clean textures. Otherwise you'll usually end up with an odd color line in the texture, unless you mask out that bottom half of the object in every photo. For reference, by using a turntable, and the masking techniques mentioned earlier, I was able to get a full 360 scan of this small animal skull. I actually scanned it in four different positions, which helped to reduce a lot of texture occlusion: Texture:  Geometry: 
|
|
#
?
Aug 7, 2015 14:17
|
|
|
Gearman posted:For reference, by using a turntable, and the masking techniques mentioned earlier, I was able to get a full 360 scan of this small animal skull. I actually scanned it in four different positions, which helped to reduce a lot of texture occlusion: 1) How large is that skull? Was it a rat? 2) How easy would it be to do a larger object but with the same level of resolution, for example a human skull? 3) Are you in the LA area?
|
|
#
?
Aug 8, 2015 00:12
|
|
|
1. Kind of small, probably about the size of an orange. Not sure what animal it belonged to. 2. With the same process, I'd get a much better scan from a human skull with more detail. It's pretty much the perfect sized object for my rig. 3. Not in the LA area but feel free to PM me if you'd like to chat some more.
|
|
#
?
Aug 8, 2015 01:55
|
|
|
Looks like someone found a use for photoscans in the medical animation field.
|
|
#
?
Aug 8, 2015 13:14
|
|
|
Gonna need some help here on this one. I want to do a long camera shot that follows along all in one take. Essentially this here: https://www.youtube.com/watch?v=Uwarhzl76D8. How would I go about doing this?
|
|
#
?
Aug 9, 2015 00:43
|
|
|
Thanks for all this. I'm back in the lab on Monday, so hopefully I'll be able to sort this all out by then!
|
|
#
?
Aug 9, 2015 00:50
|
|
|
I've spent a lot of time trying to make this. Finally though I know I need more samples but I am an impatient man
|
|
#
?
Aug 9, 2015 01:23
|
|
|
echinopsis posted:I've spent a lot of time trying to make this. Finally though Nice! What renderer?
|
|
#
?
Aug 10, 2015 02:07
|
|
|
luxrender  it's free software and "modelled" in blenderno piracy in these posts
|
|
#
?
Aug 10, 2015 02:33
|
|
|
Stuff4and5 posted:Gonna need some help here on this one. I want to do a long camera shot that follows along all in one take. Essentially this here: https://www.youtube.com/watch?v=Uwarhzl76D8. How would I go about doing this? I don't know what you're asking here. Are you asking how to make an animated video? Because that is a very broad question. A long camera shot is the easiest and first thing you do in CG really, you just don't cut it?
|
|
#
?
Aug 10, 2015 03:43
|
|
Stuff4and5 posted:Gonna need some help here on this one. I want to do a long camera shot that follows along all in one take. Essentially this here: https://www.youtube.com/watch?v=Uwarhzl76D8. How would I go about doing this? Attach camera to motion path, or just set keyframes for movement of a camera. It's not hard really, like Megaspel said it's a pretty basic thing and you can find information on animating cameras in any 3D app ever. The scene you linked is all in 3D, unless you're talking about actual film/footage in which you'll probably need a camera, some sort of steadycam, and some tracking software if you can't do a steadycam. You really should not post requests in this thread without more information dude.
|
|
|
#
?
Aug 10, 2015 05:56
|
|
|
Gearman posted:1. Kind of small, probably about the size of an orange. Not sure what animal it belonged to. I've been modeling bones of the skull by hand for medical education, and that works okay when I'm going for a more simplified, illustrated look for teaching concepts. But it would be nice to have a perfect replica of a real human skull with all the sutures etc.; unfortunately some of the holes in the skull that are important are super small. Since you're not nearby and I can't invite you for a visit, I guess my follow up question would be what is the smallest resolution you think you'd be capable of capturing- like a square millimeter? I also have another unrelated question for everyone in the thread. I've been learning (slowly) Houdini, and I'm now starting to do some tutorials on material building and rendering. I've been interested in getting Arnold and with the recent price drop it's actually in my price range. If my end goal is to learn Arnold, should I bother with Mantra, or just invest in Arnold and learn that? In other words is it worth it to invest the time to learn more than one renderer? I'm excited to see Houdini 15 is finally getting a tweak tool.
|
|
#
?
Aug 10, 2015 23:38
|
|
|
Edit: nevermind
Kibbles n Shits fucked around with this message at 00:29 on Aug 11, 2015 |
|
#
?
Aug 11, 2015 00:04
|
|
|
Listerine posted:I've been modeling bones of the skull by hand for medical education, and that works okay when I'm going for a more simplified, illustrated look for teaching concepts. But it would be nice to have a perfect replica of a real human skull with all the sutures etc.; unfortunately some of the holes in the skull that are important are super small. Since you're not nearby and I can't invite you for a visit, I guess my follow up question would be what is the smallest resolution you think you'd be capable of capturing- like a square millimeter? If you're looking for actual geometry at the scale of a millimeter, it's going to be tough. I can usually get detail around a couple millimeters in size, but it tends to get noisy and much of the precision is lost due to the noise. If you use a DSLR, you can swap to a macro lens and that might get you there. Unfortunately, I haven't had much time to experiment with macro lenses, but I've seen some of Paul Debevec's work with them, and I'm pretty confident you could get sub-millimeter geometry from scans using a macro lens. It would just take a little bit longer, and the mesh would most likely exceed 10million polys in density, but you'd get a lot of detail. If you only care about texture at that scale, sub-millimeter is very possible, even with a standard kit lens on a DSLR. Doing scans with a macro lens is actually at the top of my todo list, so if/when I get there, I'll be sure to follow up.
|
|
#
?
Aug 11, 2015 00:07
|
|
|
Listerine posted:I've been modeling bones of the skull by hand for medical education, and that works okay when I'm going for a more simplified, illustrated look for teaching concepts. But it would be nice to have a perfect replica of a real human skull with all the sutures etc.; unfortunately some of the holes in the skull that are important are super small. Since you're not nearby and I can't invite you for a visit, I guess my follow up question would be what is the smallest resolution you think you'd be capable of capturing- like a square millimeter? Oh cool I didn't know Arnold had a price drop. It is all we use at work but at home I will never use anything but Redshift. I am spoiled by the speed. I would honestly learn Mantra for houdini. It is a workhorse renderer and is amazing for rendering volumes.
|
|
#
?
Aug 11, 2015 06:47
|
|
|
keyframe posted:Oh cool I didn't know Arnold had a price drop. It is all we use at work but at home I will never use anything but Redshift. I am spoiled by the speed. Yeah I got an email from them about siggraph and they mentioned the price drop, 1 node locked license with a year service is just under 800 bucks. I think it was like $1200 before? Listerine fucked around with this message at 08:17 on Aug 11, 2015 |
|
#
?
Aug 11, 2015 07:49
|
|
|
https://www.youtube.com/watch?v=0FLQ_h_LaqA Finished a new vid. Def learned a lot on this project. Happy with the results. On to the next one!
|
|
#
?
Aug 11, 2015 08:03
|
|
|
That was pretty rad! Heard rumours about Substance Painter 2.0 coming out sometime in the near future, anyone got some more info? What they're planning on adding to it and stuff, as i heard you're going to have to pay for the upgrade from 1.
|
|
#
?
Aug 11, 2015 16:45
|
|
|
Keket posted:That was pretty rad! Not surprised, they make you pay for pretty much every major update Allegorythmic comes out with in their line of products. I dunno if i'm just cheap or if they are kinda greedy. Edit : They're going to add a new viewport made in partnership with nvidia that's much closer to photorealism than the Yebis one they currently have. they also want to add a displacement shader that actually works in realtime in substance designer 5.2 (or 5.1 can't check right now). Odddzy fucked around with this message at 02:03 on Aug 12, 2015 |
|
#
?
Aug 12, 2015 01:54
|
|
|
I wonder how that works with their subscribe-to-own users. I got one payment in before PayPal hosed up and wouldn't make my payment so my account for suspended. I hope they come up with a better payment system too because gently caress PayPal.
|
|
#
?
Aug 12, 2015 02:25
|
|
|
gently caress substance painter. I hate that software with a passion. Have been playing with Mari 3 at work today and it is amazing. Just needs a price drop so I can justify buying it for home. 
|
|
#
?
Aug 12, 2015 03:16
|
|
|
Why?
|
|
#
?
Aug 12, 2015 03:30
|
|
|
I haven't messed much with Painter, but Designer is the new hotness in the game industry. I haven't heard of many game studios that use Mari, but I know it's wide in the film industry.
|
|
#
?
Aug 12, 2015 03:30
|
|
keyframe posted:gently caress substance painter. I hate that software with a passion. Have been playing with Mari 3 at work today and it is amazing. Just needs a price drop so I can justify buying it for home. Why do you dislike Painter? I've only used it for a few weeks but if we had a tech/enviro art team dedicated to Substance Designer I could see it being pretty drat amazing. I've even seen some cinematic companies using Painter and Designer in their stuff. Also allegorithmic's monthly to own payment system is dope, I wish more companies would do this. I'd be much more likely to subscribe than Adobe and Autodesk's stuff. The Allegorthmic dudes gave me a free copy of Painter but not Designer so I've mostly just been trying out Painter. An environment artist I know working on the new star wars games seems to like designer quite a bit. It would help if our engine was PBR based, but unfortunately we're still on lovely old UE3. I'm all for nondestructive workflows as it makes the character pipeline a bit shorter to handle, it's way too convoluted these days. To create a AAA next gen character easily takes 1-2 months at least. ceebee fucked around with this message at 04:23 on Aug 12, 2015 |
|
|
#
?
Aug 12, 2015 04:20
|
|
|
Ahah figured it out, thanks again for your help you guys.
Stuff4and5 fucked around with this message at 19:15 on Aug 12, 2015 |
|
#
?
Aug 12, 2015 06:32
|
|
|
I'm more interested in what they have planed for Painter, as if its just a viewport and a few tweaks i cant see much of a point in upgrading. Also to share the substance painter love, this took me no time at all compared to other methods, and would of been less but there was allot of loving around with masks and maybe a crash or three.     Really want to pick up Substance Designer sometime though, i heard it integrates really well with unity.
|
|
#
?
Aug 12, 2015 06:52
|
|
|
ceebee posted:Why do you dislike Painter? I've only used it for a few weeks but if we had a tech/enviro art team dedicated to Substance Designer I could see it being pretty drat amazing. I've even seen some cinematic companies using Painter and Designer in their stuff. Honestly I don't like anything about it. On my end it is extremely slow to paint with. I get brush lag with most things and I have a gtx980ti. The whole painting workflow is extremely rudimentary and lovely coming from mari. I understand it is good to throw a bunch of substance materials on a game model and call it a day but for what I do that won't fly. Even if I was making textures for game models I would prefer mari because it gives me total control and is extremely fast to work with. On the upside it is way more affordable than mari so there is that I suppose. I should give mari indie a try because I heard they made it less limited than the initial release version which was useless.
|
|
#
?
Aug 15, 2015 04:34
|
|
|
Weird, I don't get any slowdown when painting at all and I've been doing a lot of diffuse painting lately. You painting on a bunch of 8ks or something? Mari handles large sets of textures much better but if you're on games stuff I think Painter works pretty good. There's still a few weird features from 3DCoat I wish it had, but I found Mari pretty clunky to work with honestly. Also if you're having slowdowns in Painter, you can work at 1-2k and uprez to 4k or 8k (I think 8k is in now? I can't remember) for export. It keeps everything when scaling up which is pretty nifty. Edit: I completely forgot, you can't actually paint across multiple UV sets in Substance Painter. This could be a pain if you really need that function. ceebee fucked around with this message at 05:34 on Aug 15, 2015 |
|
|
#
?
Aug 15, 2015 05:17
|
|
|
ceebee posted:Weird, I don't get any slowdown when painting at all and I've been doing a lot of diffuse painting lately. You painting on a bunch of 8ks or something? Mari handles large sets of textures much better but if you're on games stuff I think Painter works pretty good. There's still a few weird features from 3DCoat I wish it had, but I found Mari pretty clunky to work with honestly. No I was painting 2k diffuse on a 20-30k poly model. Not across multiple uv sets either. I am not sure why I get so poo poo performance with it. I have 32 gb ram on the computer and my processor is pretty old (i7 920) but it is no slouch, every other software runs nicely on my computer except painter.  -- In other news check out this cool modeling stuff coming to maya in the future: https://www.youtube.com/watch?v=kdSx4a0tDQw
|
|
#
?
Aug 15, 2015 05:52
|
|
|
Yeah I just saw that, it's good that they're kinda ditching Mudbox imo. I don't know if Maya is the right platform for it, but seeing them use bones and skinning on a high res mesh would make posing high res stuff SO MUCH easier than ZBrush's garbage transpose master. Also, being able to comb curves on your high res mesh would prove to be a huge timesaver in the future for games that are using hair systems like Nvidia Hairworks and other curve-based generated systems. Could finally spell out the dealth of mother loving pain in the rear end haircards for current generation games. Overall, I think it's a good idea to have it in Maya. But the pure amount of topology that ZBrush can handle is still a huge factor, and the massive amount of tools. I'm a bit skeptic of Autodesk's integration techniques when it comes to merging software. It took them like 7 years to fully integrate NEX tools into Maya, so who the hell knows. My favorite parts of Mudbox over ZBrush was it's actual physically accurate camera, and having a legit 3D viewport where pivots and scale wasn't some stupid mystery number and location. ZBrush has awful basic object transforming/rotating/scaling compared to a traditional 3D viewport, and a non-destructive pivot system is definitely needed, transpose just barely cuts it. ceebee fucked around with this message at 07:08 on Aug 15, 2015 |
|
|
#
?
Aug 15, 2015 07:05
|
|
|
keyframe posted:
Hey that's Andrew Camenisch! Went to grad school with him at Mississippi State.
|
|
#
?
Aug 16, 2015 00:55
|
|
|
BonoMan posted:Hey that's Andrew Camenisch! Went to grad school with him at Mississippi State. Yea he is leading the maya modeling team afaik.
|
|
#
?
Aug 16, 2015 02:25
|
|
|
keyframe posted:Yea he is leading the maya modeling team afaik. Yeah I know he was part of the team that made mudbox at weta and then when he sold it to Autodesk it was part of his contract that he work there for a set amount of years. The guy is reallllly good. I remember he made a site for "the head tutorial" back in the early 2000s when we were grads at MSU that helped sculpt (pun intended) the way a lot of people approached head modeling.
|
|
#
?
Aug 16, 2015 03:18
|
|
|
Glad to see theyre utilizing all these good people to help improve Maya, even the bald guy doing presentations on all the new features for 2016 seems to know his poo poo.
|
|
|
#
?
Aug 16, 2015 03:39
|
|
|
The new modeling features in the last few versions of Maya have convinced me to start doing some personal projects using it, adesk doesn't give much of a gently caress about max anymore.
|
|
#
?
Aug 16, 2015 06:56
|
|
|
Are they ever going to fix transparency sorting though
|
|
#
?
Aug 17, 2015 22:48
|
|
|
|
| # ? May 29, 2024 04:04 |
|
|
Is ANYONE going to ever fix their alpha sorting?
|
|
#
?
Aug 17, 2015 23:19
|
|