Project Reality Forums

Ye increasing the size of the rear irons UV is a good move, but not the front, front sight could possibly be scaled down even.

Also think you can scale down the inside of the tube more, view it against a checker map on the model while doing so since you will also want to bring the far end back some more otherwise you will not have square UVs.
Also the inside end of the breech you can scale down a bit too, right now you've only scaled down the whole thing but not its end too.

[R-DEV]Rhino wrote: front sight could possibly be scaled down even.

But its worth it? I mean i understand scaling inside surface of back & front tube, but those already small enough and not taking lot of space. Also i want keep most shells scale as close as possible. ops:

w0lf3k wrote:But its worth it? I mean i understand scaling inside surface of back & front tube, but those already small enough and not taking lot of space. Also i want keep most shells scale as close as possible. ops:

Yes, for 1st person, unless that part of the model is going to be seen close to the player (accounting for animations as well).

Having high rez textures for parts you're always going to be far away from is pointless because when you're far away from a surface where you can no longer see its texture in its highest quality, the mip-maps kick in and you only see a low rez texture of it anyways, and if you save the texture with no mips (as you should for a 1p model texture to save on memory), and that part of the texture is still to far away for it to be seen in its full rez, then you will get a z-fighting like effect where pixels of the texture are fighting to be rendered since they all can't fit into a single pixel on the players display and it looks really bad which is why some badly UVed 1p texture have to be saved with some mips, significantly increasing the memory usage and with no benefits, where you get a big benefit of saving UV space for having small UVs for things that are always going to be far away from the player.

For 3rd person models however, yes you want pretty much everything to be at the same scale as the player can get up close to pretty much all parts of the model and the bits they can't like the interior of a vehicle they physically can't get up-close to (without entering the vehicle and then seeing its 1p model), you can down scale those yes.

Ironsight/ruler with distance scaled up to 1.5x. Inner surfaces down to 1.5x.



playing with normal bakes of rifling

Back


Front


Actually i'm not use springs, i just copy inner surface, extrude some polys, add loops for smooth modifiers and use Twist modifier

Looking good

So, should i start packing all things together and start texturing?

Packing yes, but don't start texturing until we have had a chance to give you feedback on your UVs.

[R-DEV]Rhino wrote:Packing yes, but don't start texturing until we have had a chance to give you feedback on your UVs.

Well, i was planned to pack with script(not 3ds max), but keep together some parts like ironsight and clamping for ironsight. Is that OK?

Providing the end result is good, yes, but post your packed UV here so we can see if it is or isn't.

You need file?

Ye like every other UV packing script I know of, the packing is really bad... loads of wasted space and things all jumbled up all over the place...

I would seriously recommend you manually pack this from scratch, starting with all the big items and working your way down to the small ones, placing bits that fit nicely together and also have the same kinda texture to avoid pixel bleeding issues and so on.

Quick example of me packing my Frigate's UVs if it helps:

well, then i need scale up a bit all shells, right?

It's all about packing your UV map as efficiently as possible to maximise the texture map usage so you get the highest quality textures out of the least amount of memory used. I'm not aware of any decent tutorials but at the end of the day, it's like a puzzle with trying to fit all the bits together as well as possible, finding the bits that sit next to each other well, while also leaving a bit of a gap between each UV so you don't get any pixel bleeding.

If anyone knows of any decent UV Packing tutorials please post them here but at the end of the day, each UV is unique and it's really a matter of playing around with the packing until you get it right and over time with experience, it will come naturally.

Like I said above, start with the big bits, and work your way down to the small bits, and watch that little video above to see how I packed my Frigate for some examples.

tile coverage 93% 2 pixels bleeding

Still could be much better, loads of empty space that could be used by more efficient packing and scaling up some UVs to use the extra space too.

Well, i repack it 3 times and this my best attempt. Last time i even scale down, because lot of shells doesnt fit in tile.

'[R-DEV wrote:Rhino;2169503']

Also, you show me this as example, but its atm has 78% tile coverage. Mine 93%

w0lf3k wrote:Also, you show me this as example, but its atm has 78% tile coverage. Mine 93%

The Bren was an example for how to set textures up, not packing. The Bren's UVs are nowhere near perfect either, especially for its 1st person texture since it had massive spacing between it's UVs between 10 and 20px, compared to your 2px, but for the most part, things are packed together in quite a logical way using up leaving no huge dead areas, if you accounting for the 10 to 20px spacing

But yes for a 1st person texture, you only want 2px of spacing between the UVs and as such, the Bren's UVs are inefficient because of this, but not because of how things have been packed.

But still, i dont know how to improve it. As i said shells was bigger, but they doesnt fit in tile, so i decrease their scale. Also most empty spaces is around circle shells.

Post a solid face UV sheet with full transparency like I did above, and I'll give you some feedback later.

[R-DEV]Rhino wrote:The Bren was an example for how to set textures up, not packing. The Bren's UVs are nowhere near perfect either,

Then we have a problem that we dont have any models to show as example to starting modellers neither they can learn from extensive modelling resulting in community models of enough quality getting forbidden