649 réponses à ce sujet

[MerricksDad]

I find that I can't mix translucent and transparent on the same model properly. On certain models, you can use the txi command to "blend additive" which removes all the black from your texture. It appears that the function negates shininess functions, especially when your texture is all black, as is the mirror texture plc_reflect.

 

So what I did was create single tri faces and placed them where I needed them, not unlike glitter. I make the faces so small you can barely find them, so they only represent a single pixel at almost any camera angle/distance. I'll report actual size soon, so we can know those limits.

 

Right now, I'm working on determining if individual mesh per tri is more useful than all meshes merged to a single. Also trying to cut down the base rotation data, and making a mess.

 

Next picture will be 600 crystals per same space....

[MerricksDad]

Here's the 600 (actually 552) sparkles per 20m square.

 

 

This time, I applied a reset transform to the entirety of all attached faces. This is a single mesh over the single snow field mesh. It seems to load just as slow, but jerks less in game than 200 individual meshes. Again, the sparkly is not spectacular when spread over such a large area, but it is starting to approach reality. Imagine using this glitter method to coat a few crystal faces in your caves!

 

Just a note: the underlying snow texture has been toned down by giving it an approximately half-saturation reducer. This is done by simply changing the diffuse and ambient values in the model file to 150/256, rather than 256/256 (1). I used the default texture settings in gmax to accomplish this. Nor sure why they picked 150, but it doesn't matter. The point is, the highest color your display can show is white. So white sparkles on stark white illuminated snow will be invisible. So you need a snow that is not stark white. Alter as desired.

 

One might also play with the underlying envmap image, as they did in vanilla placeables for the crystals and gold. Nothing says you have to use the wavy silver scale image for your snow. Planar snow may come in various colors, or you may wish your snow to reflect the sky of your world, not silver white.

 

Glow in the dark snow is also not out of the question. Try adding self illumination to force your colors, and make sparkles available at night.

 

Also note the black pixels in the picture above are also the same frost faces, but with the darker portion of the envmap showing. I've found that increasing the number of faces per square meter actually makes more black spots than it makes light spots. I assume this is a drawback of using the base metal envmap image. Perhaps a rounded envmap would be more realistic than a wavy one in this case. If you really want some serious sparkle, make a high luminance, and very granular, envmap, like salt or sand.

 

OTR's package of alternative envmaps has a few I liked: vdu_envmap028, 32, 69, and 8. Especially 8! If you have the package, examine how index 69 has more white than the rest. This causes the flicker to never go off, just cycle through high luminance values. Index 8 does a great job of cycling through whites with a mix of sky colors. It also has enough near-gray colors to turn of the flicker as you walk away.

[Draygoth28]

This is shaping up quite nicely. Can't wait to see trees, rocky out crops and maybe???that fall brown grass like in Skyrim???

[MerricksDad]

Yeah, grass will be easy to change for seasonal variations. I'm looking for a summer first variant. We can all tag team additional variants after that I think.

[MerricksDad]

Another thing I intend to do is under-dummy every placeable-like object I put on the tiles. I've got a few scripts that respond to named dummies, and replace things like trees with another tree from my growing tree model library. I'd like to release that kit at some point in the near future (this year hopefully), and let people remake the tiles as they want, including trees, placeable rock types (from igneous to sediment of various colors), and base vegetation (because it won't be just tileset grass).

 

I find adding these dummies also helps me find stuff in treeviews, like model hierarchy views in NWexplorer.

 

Expect each tile to have a primary tree, as well as a secondary tree or shrub type. Also expect, in addition to tileset grass, wind-moveable plants in at least two varieties per tile, except intentionally bare tiles. The kit I release in the future might even have a command-line replacement tool, so you can quickly switch out the model parts for those tiles with parts for a separate season, at the same time replacing base textures to the next season.

[Draygoth28]

Any ETA on an updated HAK? Really excited about this project.

[MerricksDad]

I've been forgetting to repack. I can do that, minus the snow sparkles, probably tomorrow around lunch time. I'm unemployed again, so after this weekend, expect this project to skyrocket!

 

er...tuesday. Tuesday the boy goes back to school

[MerricksDad]

I had time to update. Dropbox is being nice this time, and kept the project link the same. 24megs it says now. The only thing in the package not used, I believe, is the tx_flow01.tga, which is what I used to line up the flow of water on some of the test stream tiles.

 

Here's the link again, and I also put it on the top post of this thread for future easy finding.

 

https://www.dropbox....khills.hak?dl=0

[MerricksDad]

And here is a peek at the frosted snow placeable

 

https://www.dropbox....stSnow.hak?dl=0

 

I think all you need to do with this one is make a placeable with appearance "Snow: Frosted". It will be automatically 25 cm high (off the ground), as it was for my testing purposes. Packed is the 552 glitter faces per 20m square section.

 

If you make the placeable static (in toolset), it will pick up far less light, having a dull appearance. Remove the static checkbox in the placeable creation panel so that it picks up all dynamic light.

[MerricksDad]

Off topic: "Sword Coast Legends" looks like something I would buy.

[MerricksDad]

Do you guys like this? I'm not certain I do. This is modeled somewhat after rocky walls in Kingsroad. It somewhat resembles some areas in the black hills, but not most.

 

[MerricksDad]

This one I like better. Other than color, it properly represents weathered granite.

 

[MerricksDad]

Or somewhere in between.

 

 

Obviously the top image has less polygons, but not by much. The bottom image, most complex basic tile, has only 1500-1600 faces. None of these tiles casts any shadows.

[Bannor Bloodfist]

If I am understanding you correctly, the middle of the 3 is somewhat less polygons than the last one?  That is strange as it looks to be the closest to a smooth edge of all 3 of them, which typically means a higher polygon count. Meaning to me it looks the best in my humble opinion anyway.  Of course. it may also be pointing to whatever smoothing options you chose?  Where they different between the three shots, meaning the source tiles were physically different in both polygon count AND smoothing?

[MerricksDad]

The one in the middle is about the same as the last, give or take 50 faces. It just has different noise, push, and smoothing patterns. The source tile, before 9 slightly changed modifiers, are the exact same tile.

 

I think if I put the same smoothing angle on the third picture, that will be much better than the second version, simply because of the edge mixing.

[Jedijax]

I'm sort of torn here. Both approaches are very interesting visual-wise. But if I had to choose, I think the second looks best. The first one looks way squared, not so much because of the actual models, but how lighting works on them. The second one looks great in terms of polys, and it flows marvelously; the texture work really shines. It's a matter of performance VS what you were going for, though, so it may make sense to go for the middle ground.

Close

[YeoldeFog]

Really nice, all of them. I think I like the bottom image most. Mostly because of the sharper edges. I don't know how much work it would take, but I'd like to see the first picture with more detailed rocks. Basically, the same form, but more irregularity in the irregularity!

[Draygoth28]

The middle smooth rock looks amazing. But the rough cut and weathered look is good too.

[MerricksDad]

The consensus seems to be that variety is good

[rjshae]

The combination of wear and fracturing looks the most natural, I think. I see that type of terrain quite a bit around these parts, although it's mostly a lighter granite.

[MerricksDad]

The combination of wear and fracturing looks the most natural, I think. I see that type of terrain quite a bit around these parts, although it's mostly a lighter granite.

I totally agree. The rock I used was from my upcoming rock textures package, number 38 or 35. Not sure. But I used it because it was a larger texture and had the most grain to show close up. I don't want to skimp. The darker textures I've tried looked too red.

 

I'm not sure If I want to keep that texture and lighten with the ambient/diffuse settings, or if I want to lighten the texture instead, but I like the texture. You are right though, it does not properly show the correct color of a quartz/feldspar pegmatite granite. And I think it might need some decals to create more variation here and there. Something for the future.

I think tonight, after I make potato leak soup (priorities), I'll go ahead and try that specific mod list on the crossers, see what I get. The variety shown above has the ledges at each increment of base height, and I think that looks good, if not just slightly too angular at the base height. Rubble and other detail can fill it in I suppose.

 

If anybody wants to play with the tiles in the last pack release, here is the mod list I am using:

 

• Edit Mesh: select all faces on the mesh which touch adjacent tiles, except faces which are at an increment of base level
• Face Extrude 0cm at scale 100 to clone out those faces for pulling pushing, yet leaving the edge seam underneath to fill any light holes later
• Tesselate faces via edge 1x at 25.0 tension
• Noise x/y at 400cm with a scale of 50 non-fractal seed 1
• Push 150
• Tesselate faces via edge 1x at 25.0 tension again
• Relax 0.5 x4 with no points retained
• Optimize with face thresh 8 edge 1 no retained boundaries
• Edit mesh to deselect faces
• UVWmap box of 1001 cubed with u/v tile 1: be sure to adjust the location to fit the bottom center of the tile affected (or at least one tile affected in a selected group) because the push modifier will make the tile taller than the box, making lining up textures tile-to-tile incorrect
• Smooth auto thresh 15 prevent indirect smoothing
• Trimesh render with no shadow and no texture rotation ( you really don't want that with the texture applied)

I have temporarily retained the original tile walkmesh, and will update when these are set in stone, pun intended.

[MerricksDad]

Same shape tiles as the third, with smoothing pulled in a bit, so less angular. Changed the texture, and added a little bit of yellow glossy for the light sources to pick up via the trimesh modifier.

 

[rjshae]

To me the dark texture looks more basaltic. Which is totally fine, and gives it a certain uniqueness as if you're heading into lava-blasted terrain.

 

The extra moss, mold and fungus on the new texture is a nice addition.

[MerricksDad]

Somebody had requested that I do a lava/volcano variety. I've got some definite basalt images, as well as some pinkish brown ryolite. Almost looks like a big chunk of chocolate. Found this cool rock in Wyoming at the border of the black hills. It seemed very misplaced, but my teacher told me it was from way way back, and the rest of the mass had degraded away already. South Dakota had very few volcanoes, but they were of interesting composition.

[MerricksDad]

Trying to go for this, but I think the area lighting is doing a good portion of the darkening.

 

 

But I would accept this:

 

They're only like 2 miles away from each other anyway. Just depends on the lighting how things get viewed.