Just checked the updates: Cinemut?
-
- Expert Mercenary
- Posts: 893
- Joined: Thu Jul 08, 2010 11:43 pm
- Location: Sol III North American Continent
Just checked the updates: Cinemut?
So, with cinemut encode and decode tools, will we soon be seeing real video cut-scenes? If so, I may have a post in the modders help soon! (I really should get back to my creature and patch in some more WC stuff.)
Edit: I'm assuming yes for the moment. If that is the case, what do we want to see first? I think a video version of "A Dead Man's Ship" would be great.
Edit: I'm assuming yes for the moment. If that is the case, what do we want to see first? I think a video version of "A Dead Man's Ship" would be great.
Last edited by travists on Fri Sep 21, 2012 10:19 pm, edited 1 time in total.
-
- Elite
- Posts: 7243
- Joined: Mon Apr 18, 2005 2:40 pm
- Location: LS87, Buenos Aires, República Argentina
Re: Just checked the updates: Cinemut?
Ehm... cinemut is a shader set and texture format. It's not for cutscenes.
We've had video-based cutscene support for a while now.
We've had video-based cutscene support for a while now.
-
- Expert Mercenary
- Posts: 893
- Joined: Thu Jul 08, 2010 11:43 pm
- Location: Sol III North American Continent
Re: Just checked the updates: Cinemut?
Really, last I saw was animated sprites with horrible timing. Reason I thought that was a search for cinemut returns youtube video sharing.
-
- Elite
- Posts: 7243
- Joined: Mon Apr 18, 2005 2:40 pm
- Location: LS87, Buenos Aires, República Argentina
-
- Expert Mercenary
- Posts: 893
- Joined: Thu Jul 08, 2010 11:43 pm
- Location: Sol III North American Continent
Re: Just checked the updates: Cinemut?
Ah... Spelling/typing error on my part.
-
- Elite
- Posts: 1567
- Joined: Tue Jan 26, 2010 2:03 am
Re: Just checked the updates: Cinemut?
careful with those typing mistakes.. may end up on some off the beaten path pron site.. Worked with a gal who tried using dogpyle (spelling? search engine) but misspelled it and had a merry time getting away from pron pop-up hell. Real kicker, she did this at work so we all knew about it and got to tease her about it for a few weeks.
Because of YOU Arbiter, MY kids? can't get enough gas. OR NIPPLE! How does that mkae you feeeel? ~ Halo
-
- Expert Mercenary
- Posts: 893
- Joined: Thu Jul 08, 2010 11:43 pm
- Location: Sol III North American Continent
Re: Just checked the updates: Cinemut?
Or the infamus whitehouse .com/.gov one.
-
- Elite
- Posts: 7243
- Joined: Mon Apr 18, 2005 2:40 pm
- Location: LS87, Buenos Aires, República Argentina
Re: Just checked the updates: Cinemut?
So... back to cinemut... I've been thinking of adding dual UV support to bfxm, so we can start producing better light maps for our models and perhaps, finally, put the PRTs in place.
Right now, the biggest hinderance to all this is that diffuse-unwraps are very bad AO-unwraps and viceversa. If we can have dual UVs, we can unwrap both separately and live the high life.
Do we have any artists willing to do this? I cannot run xNormal to bake stuff, because xNormal is windows-only, sadly. I'd like to code my own baking tool in mesher, but it's quite complicated so xNormal will have to do till then.
If there's any artist that knows about dual UVs... how do modelling tools export those?
Right now, the biggest hinderance to all this is that diffuse-unwraps are very bad AO-unwraps and viceversa. If we can have dual UVs, we can unwrap both separately and live the high life.
Do we have any artists willing to do this? I cannot run xNormal to bake stuff, because xNormal is windows-only, sadly. I'd like to code my own baking tool in mesher, but it's quite complicated so xNormal will have to do till then.
If there's any artist that knows about dual UVs... how do modelling tools export those?
-
- Elite
- Posts: 1363
- Joined: Sat Aug 04, 2007 3:42 pm
Re: Just checked the updates: Cinemut?
I don't know that you should assume that an AO friendly UV map is very bad for diffuse maps. From the perspective of some map types (like diffuse), you only waste texels this way when you feel you can get away with identical repeating structures or surfaces.klauss wrote: I've been thinking of adding dual UV support to bfxm, so we can start producing better light maps for our models and perhaps, finally, put the PRTs in place. Right now, the biggest hinderance to all this is that diffuse-unwraps are very bad AO-unwraps and viceversa. If we can have dual UVs, we can unwrap both separately and live the high life.
I'd recommend starting with (new) meshes that have a single set of UVs that are designed for the newer maps like AO, (seamless) Normal and these new PRT things, and see how it goes. Start simple and then you can always add dual UV support down the road if it turns out to be important.
You should ask chuck that, but my guess is poorly.If there's any artist that knows about dual UVs... how do modelling tools export those?
-
- Elite
- Posts: 7243
- Joined: Mon Apr 18, 2005 2:40 pm
- Location: LS87, Buenos Aires, República Argentina
Re: Just checked the updates: Cinemut?
Well, yeah. And that's bad. Wasting texels means there's less detail when there could be more.Deus Siddis wrote:I don't know that you should assume that an AO friendly UV map is very bad for diffuse maps. From the perspective of some map types (like diffuse), you only waste texels this way when you feel you can get away with identical repeating structures or surfaces.klauss wrote: I've been thinking of adding dual UV support to bfxm, so we can start producing better light maps for our models and perhaps, finally, put the PRTs in place. Right now, the biggest hinderance to all this is that diffuse-unwraps are very bad AO-unwraps and viceversa. If we can have dual UVs, we can unwrap both separately and live the high life.
Yes, but no. That'd be a dead end, then we'd depend on full new texture jobs for the other ships, and I don't think we have the manpower for that.Deus Siddis wrote:I'd recommend starting with (new) meshes that have a single set of UVs that are designed for the newer maps like AO, (seamless) Normal and these new PRT things, and see how it goes. Start simple and then you can always add dual UV support down the road if it turns out to be important.
I may. But, you know, he learned modelling for VS, so he probably stayed within VS limits (ie: no multi UV)Deus Siddis wrote:You should ask chuck that, but my guess is poorly.If there's any artist that knows about dual UVs... how do modelling tools export those?
-
- Elite
- Posts: 1363
- Joined: Sat Aug 04, 2007 3:42 pm
Re: Just checked the updates: Cinemut?
But you'll need to weigh the benefits against the costs. You will have to develop the engine support for this feature, and there will need to be very good content pipeline support from mesher as well as either blender or xNormal. Then it will take content creators more time to build the second texture arrangement and UV layout for each model.klauss wrote: Well, yeah. And that's bad. Wasting texels means there's less detail when there could be more.
And the reward is you will save texels only for some types of textures, and only on repetitive, symmetrical structures.
If the goal is to improve the visual quality of the game, then there is only so far you can nurse along primitive 512x512 texture jobs anyway.Yes, but no. That'd be a dead end, then we'd depend on full new texture jobs for the other ships, and I don't think we have the manpower for that.
IMHO, with the work it takes to produce modern content, and this project's loss in content creation power over the years, the only realistic way to improve the look of the game further is to downsize the number of ship models you try to maintain. You really need to start thinking about initiating work on a much smaller VS content v2.0. A more or less clean break from most of the venerable 3D content, that focuses on full new feature support and modern levels of detail.
-
- Elite
- Posts: 7243
- Joined: Mon Apr 18, 2005 2:40 pm
- Location: LS87, Buenos Aires, República Argentina
Re: Just checked the updates: Cinemut?
Well, engine support wouldn't be that hard to achieve. I've been musing on this already, and though I haven't checked the actual code, from what I remember it would not be that hard.Deus Siddis wrote:But you'll need to weigh the benefits against the costs. You will have to develop the engine support for this feature, and there will need to be very good content pipeline support from mesher as well as either blender or xNormal.klauss wrote: Well, yeah. And that's bad. Wasting texels means there's less detail when there could be more.
The content pipeline is the biggest worry here, that's why I was asking how it works for modelling tools. Ideally, mesher would read whatever blender exports and that's it. Not sure about xNormal, but we can always ask Santiago to work on supporting it (he's usually quite receptive).
Well, that's the beauty of dual UVs. AO or PRT bakings work perfectly well with automatic unwraps. Creating them would be really really simple if you know the tools.Deus Siddis wrote:Then it will take content creators more time to build the second texture arrangement and UV layout for each model.
If we want high quality big ships or stations, we need repetitive textures. It's the only way to get the level of detail needed without gigabytes of textures. We're already hitting the hardware limits with planets.Deus Siddis wrote:And the reward is you will save texels only for some types of textures, and only on repetitive, symmetrical structures.
Sure, but I'd be OK with just adding PRT bakes on them until we get a new consistent content set in place. PRT bakes can do a lot for quality, you'd be surprised. Dynamic, soft shadows for one. And we've had the shaders for a while now.Deus Siddis wrote:If the goal is to improve the visual quality of the game, then there is only so far you can nurse along primitive 512x512 texture jobs anyway.Yes, but no. That'd be a dead end, then we'd depend on full new texture jobs for the other ships, and I don't think we have the manpower for that.
IMHO, with the work it takes to produce modern content, and this project's loss in content creation power over the years, the only realistic way to improve the look of the game further is to downsize the number of ship models you try to maintain. You really need to start thinking about initiating work on a much smaller VS content v2.0. A more or less clean break from most of the venerable 3D content, that focuses on full new feature support and modern levels of detail.
And, in any case, supporting repetitive diffuse textures while applying ambient occlusion and PRT bakes really requires separate UV unwraps. I'm quite sure one of the big reasons why we get so little contributions, besides the fact that people tend to contribute whatever they want without regard for what we readlly need, is that when they do try to contribute they get overwhelmed with all the requirements.
Right now, it's tough to grasp all the requirements, let alone meet them. Making it easier for unwraps can only be a good thing... right? And separate unwraps do make it easy.
Right now, an unwrap has to:
- be consistent in its winding (ie: always clockwise or always counter-clockwise) for connected faces, because otherwise you break normal maps.
- not overlap any faces, or you break AO maps
- must keep differently exposed areas separate or MIP mapping will bleed and cause artifacts
- yet it must assign lots of texels to visually important areas
Now, if you can separate the two texturing coordinate sets, you only require:
- consistent unwinding for diffuse+normal
- automatic unwrap for AO and PRT (keeping different areas separate)
-
- Elite
- Posts: 1363
- Joined: Sat Aug 04, 2007 3:42 pm
Re: Just checked the updates: Cinemut?
Well it turns out Blender does support this.klauss wrote: The content pipeline is the biggest worry here, that's why I was asking how it works for modelling tools. Ideally, mesher would read whatever blender exports and that's it.
But the next leg of the journey runs from blender to mesher, which will require a file format that both programs recognize and that itself supports this feature. I'm guessing .OBJ wouldn't support something as new as multiple UVs per vertex?
I agree. Though a second reason could be that the hordes of mediocre 3D content that does nothing to demonstrate the engine features, turns away the highly talented volunteer content creators you see working on mods of commercial games. But I digress.I'm quite sure one of the big reasons why we get so little contributions, besides the fact that people tend to contribute whatever they want without regard for what we readlly need, is that when they do try to contribute they get overwhelmed with all the requirements.
IMHO, with artificial objects that big, they can't look good unless they have individually, visibly destroyable and damageable sections. They really have to use sub-units, which also solves the efficient repetition issue.If we want high quality big ships or stations, we need repetitive textures. It's the only way to get the level of detail needed without gigabytes of textures. We're already hitting the hardware limits with planets.
Weird. I wonder how perfectly though.AO or PRT bakings work perfectly well with automatic unwraps.
If an auto unwrap really is just as good then that makes a huge improvement to the content development cost of this feature.
-
- Elite
- Posts: 7243
- Joined: Mon Apr 18, 2005 2:40 pm
- Location: LS87, Buenos Aires, República Argentina
Re: Just checked the updates: Cinemut?
I don't really know.Deus Siddis wrote: Well it turns out Blender does support this.
But the next leg of the journey runs from blender to mesher, which will require a file format that both programs recognize and that itself supports this feature. I'm guessing .OBJ wouldn't support something as new as multiple UVs per vertex?
Well, not really. Though what you say makes sense, lots of subunits is inefficent for the hardware. So it's not the right way either, although it's probably something similar. Instanced geometry comes to mind, but supporting it in VS is a lot harder than multiple UVs.Deus Siddis wrote:IMHO, with artificial objects that big, they can't look good unless they have individually, visibly destroyable and damageable sections. They really have to use sub-units, which also solves the efficient repetition issue.If we want high quality big ships or stations, we need repetitive textures. It's the only way to get the level of detail needed without gigabytes of textures. We're already hitting the hardware limits with planets.
In any case, our AO or PRT-capable meshes show remarkable inefficiency in texture usage. That's because they can't easily repeat. Remember most of the units you see are old-fashioned, you have to look at the newest ones, like the diplomatic center. It's quite nice, but mostly because the shaders are so complex. If you look, there's scarce texture detail. Try doing the commerce center like that: no way you get half the detail.
Even with subunits, improving on this would be a win I think.
-
- Elite
- Posts: 1363
- Joined: Sat Aug 04, 2007 3:42 pm
Re: Just checked the updates: Cinemut?
But if large ships already have all those turrets using AI and such, relatively how much more would it cost to have ~ 6-24 subunits?klauss wrote: Well, not really. Though what you say makes sense, lots of subunits is inefficent for the hardware. So it's not the right way either, although it's probably something similar. Instanced geometry comes to mind, but supporting it in VS is a lot harder than multiple UVs.
At that scale you might need repeating detail textures that fade in as you get closer. IIRC, the engine supports a very primitive version of this, something like a fixed 16x16 repeat mapping.In any case, our AO or PRT-capable meshes show remarkable inefficiency in texture usage. That's because they can't easily repeat. Remember most of the units you see are old-fashioned, you have to look at the newest ones, like the diplomatic center. It's quite nice, but mostly because the shaders are so complex. If you look, there's scarce texture detail. Try doing the commerce center like that: no way you get half the detail.
-
- Elite
- Posts: 7243
- Joined: Mon Apr 18, 2005 2:40 pm
- Location: LS87, Buenos Aires, República Argentina
Re: Just checked the updates: Cinemut?
Yes, but for turrets, you pay the cost because you need the benefit: you need each turret to have an AI. It's not the same for destructible parts.Deus Siddis wrote:But if large ships already have all those turrets using AI and such, relatively how much more would it cost to have ~ 6-24 subunits?klauss wrote: Well, not really. Though what you say makes sense, lots of subunits is inefficent for the hardware. So it's not the right way either, although it's probably something similar. Instanced geometry comes to mind, but supporting it in VS is a lot harder than multiple UVs.
Current support for detail textures is a lot more advanced than people give it credit. It supports two layers of detail applied with an arbitrary projection unwrap each. I'd add cube projection support to make it uber, but still it's quite useful as it is. We just lack the proper detail textures.Deus Siddis wrote:At that scale you might need repeating detail textures that fade in as you get closer. IIRC, the engine supports a very primitive version of this, something like a fixed 16x16 repeat mapping.In any case, our AO or PRT-capable meshes show remarkable inefficiency in texture usage. That's because they can't easily repeat. Remember most of the units you see are old-fashioned, you have to look at the newest ones, like the diplomatic center. It's quite nice, but mostly because the shaders are so complex. If you look, there's scarce texture detail. Try doing the commerce center like that: no way you get half the detail.
In fact, I've been working on more advanced detailling techniques for planets. The only one that worked out pretty well until now is city detail, but I was working on cloud and terrain detail too. Each material needs a different kind of detail work, so, we need better detail techniques. But that's a shader thing, the engine's support is already 80% of the way there.
-
- Elite Mercenary
- Posts: 1329
- Joined: Thu Mar 27, 2003 11:51 pm
- Location: Manchester, UK
- Contact:
Re: Just checked the updates: Cinemut?
Maybe somebody should email chuck and see if this gives him enough desire to get back into it?
Free Gamer - free software games compendium and commentary!
FreeGameDev forum - open source game development community
FreeGameDev forum - open source game development community
-
- The Shepherd
- Posts: 5841
- Joined: Fri May 13, 2005 8:37 pm
- Location: Ottawa
- Contact:
Re: Just checked the updates: Cinemut?
Chuck will be getting a new broadband connection this weekend he has been limited to net access from work where he is part time ATM.charlieg wrote:Maybe somebody should email chuck and see if this gives him enough desire to get back into it?
Enjoy the Choice
my box::HP Envy i5-6400 @2Q70GHzx4 8 Gb ram/1 Tb(Win10 64)/3 Tb Mint 19.2/GTX745 4Gb acer S243HL K222HQL
Q8200/Asus P5QDLX/8 Gb ram/WD 2Tb 2-500 G HD/GF GT640 2Gb Mint 17.3 64 bit Win 10 32 bit acer and Lenovo ideapad 320-15ARB Win 10/Mint 19.2
Q8200/Asus P5QDLX/8 Gb ram/WD 2Tb 2-500 G HD/GF GT640 2Gb Mint 17.3 64 bit Win 10 32 bit acer and Lenovo ideapad 320-15ARB Win 10/Mint 19.2