Seawalls graphical issues

[imoutofusernameideas]

Hey everyone,

I'm a huge addict of seawalls (I gotta have 'em all  ) and while experimenting with this and that I've found the pros and cons of the most well known sets.
One of them is the Quais de Seine (by xannepan, I believe). It's absolutely great and while not being as extremely flexible as some others, it does allow for the most beautiful embankments and lovely curves, allowing the player to preserve as much of the coast's natural shape as possible. BUT, on Zoom 1 and 2 I've run into a little something that isn't so lovely:



Does anyone know what this is all about? And if so, can I fix this somehow? I've tested other seawalls on these Zoom levels too (It´s really only 1 and 2, the other 4 are fine) and Marrast's set experiences very similar issues. Nbvc's Marina and PEG's Industrial and Recreational Seawalls do not have this problem (though PEG's files have other problems, but that's a different matter entirely).

I'd very much appreciate it if someone could shed light on this! 

[art128]

There was someone who worked on a fix for different seawalls so they appear correctly at zoom 1 and 2, however I don't remember who it was, where it was and for which seawalls it worked. For the last point, I suppose it would be easy to make it work for other seawalls, just a matter of finding the good parameters, I suppose.

[Andreas]

The problem is caused by the BAT's incapability of rendering custom LODs for zoom 1 and zoom 2. Not sure if they can be added via tweaking, though.

[art128]

I found the "fix" I was referencing to: http://sc4devotion.com/forums/index.php?topic=16176.0

[imoutofusernameideas]

Hey art,

Thanks for the reply. The website you've linked does not however contain a fix for the quais de Seine... 

But I suppose the essence of how the fix works would be the same. How could I apply this to xannepan's walls?

[Missvanleider]

If knowledge of such a thing is reasonably straightforward i'd love to know as well, i'm marginally obsessed with these walls!

[mgb204]

Depends on your definition of straightforward really.

You can either do this in the reader, requires advanced knowledge of editing S3D files.

Or, thanks to Xannepann releasing the actual files used to create the models, you can simply re-render all the LODs with NVBC's script modifications for creating LOD 1 and 2. I say you can, but you'll need a copy of 3DS max, plus knowledge of how to use it. Even then, this is not really simple or quick, it's tedious work and a lot of it.

[Missvanleider]

Seems straightforwardly non straightforward by my standards! Thanks for the response, might set more realistic targets for myself

[woodb3kmaster]

Looking at SMTFrank's fix for the Marrast embankments, it appears that all he did was copy/paste the vertices and UV mappings from each piece's corresponding zoom 3 rotations to MacGyver a set of zoom 1 and 2 LODs. Tedious work, indeed.

Rather than go to all that trouble, I'm more inclined to apply NBVC's fix instead and re-render the Quais de Seine. I've been meaning to add new pieces to that set anyway. All I need to do is fix my issues with BAT4Max rendering Xannepan's models...

[mgb204]

Looking at SMTFrank's fix for the Marrast embankments, it appears that all he did was copy/paste the vertices and UV mappings from each piece's corresponding zoom 3 rotations to MacGyver a set of zoom 1 and 2 LODs. Tedious work, indeed.

Rather than go to all that trouble, I'm more inclined to apply NBVC's fix instead and re-render the Quais de Seine. I've been meaning to add new pieces to that set anyway. All I need to do is fix my issues with BAT4Max rendering Xannepan's models...

You might be surprised. Personally I'd be more inclined to do this job in the reader. Almost certainly it would be quicker if you are used to S3D files. The question I have, do the 1 and 2 LODs still use the Z3 textures? Or are they using the original z1/z2 textures? I guess it would have to be z3, since the z1/z2 ones are usually a blur? In which case, it's just a case of delete all z1/z2 files. Copy/Paste all z3 files twice, re-id to z1/z2 ID scheme. That's not as bad as getting into the S3D files and mucking about at least.

If the z1/z2 textures need to be used with the copied z3 LODs, that's another kettle of FSH. Because unlike normal texture MipMaps, renders aren't always scaled exactly by a factor of .5 from one zoom to another. Which leaves the tricky job of knowing how to edit the UV values to fit the textures to the adapted LODs.

[woodb3kmaster]

You might be surprised. Personally I'd be more inclined to do this job in the reader. Almost certainly it would be quicker if you are used to S3D files. The question I have, do the 1 and 2 LODs still use the Z3 textures? Or are they using the original z1/z2 textures? I guess it would have to be z3, since the z1/z2 ones are usually a blur? In which case, it's just a case of delete all z1/z2 files. Copy/Paste all z3 files twice, re-id to z1/z2 ID scheme. That's not as bad as getting into the S3D files and mucking about at least.

If the z1/z2 textures need to be used with the copied z3 LODs, that's another kettle of FSH. Because unlike normal texture MipMaps, renders aren't always scaled exactly by a factor of .5 from one zoom to another. Which leaves the tricky job of knowing how to edit the UV values to fit the textures to the adapted LODs.

Judging from the materials tab, it looks like the z1/z2 LODs use those zooms' textures. I'd have to copy/paste those textures into the DATs to judge how well the UV maps have been adjusted, but given the quality of z1/z2 textures, it might be hard to spot any real issues.

[mgb204]

Well as a quick test, I checked one of the S3D sets and copied the 2x (z3) models and set them up for z1/z2 textures. Sure enough, the UV mapping was totally off. And halving the z3 values for z2 simply didn't work.

So I copied the z3 textures, giving them the z2/z1 IDs to get around this. Honestly though, at this size I'm not totally sure if this is working or not. Certainly in my game this issue is nowhere near as bad as the screenshot above. Perhaps because I'm using HD resolution with SC4?

Of course, the issue with renumbering the textures, you can't release the modifications without permission.

:update:

I just took a comparison shot to try and see if the fix is doing anything:



It's really hard for me to say, because I'm not really seeing the issue in the first place. This image is full size (zoom 2), perhaps playing at higher resolutions doesn't really highlight the issue?

[imoutofusernameideas]

I'd be willing to test it and take some pictures. Will probably be more noticeable seeing as I have such extreme problems with it. I'm on skype, Rob. Contact me whenever you feel like it.

[art128]

because I'm not really seeing the issue in the first place.

1)You can see the water goes behind the seawalls from the ripple effect.
2)Try the diagonals now. The issue is more prominent on diagonals.

[mgb204]

Make what you will of the results here.

I've a feeling that a number of prop families are in use. Therefore it's really hard to test this without getting into it, which I've already stated I'm not inclined to do. Not to mention, the issue with releasing the fix makes it a non-starter before we begin.

But, seemingly the modifications I propose are having a positive effect. So it's possible I reckon.